Inkjet-Printed Light-Emitting Devices: Applying Inkjet Microfabrication to Multilayer Electronics
by
Peter D. Angelo
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Chemical Engineering & Applied Chemistry University of Toronto
Copyright by Peter David Angelo 2013 Inkjet-Printed Light-Emitting Devices: Applying Inkjet Microfabrication to Multilayer Electronics
Peter D. Angelo
Doctor of Philosophy
Department of Chemical Engineering & Applied Chemistry
University of Toronto
2013
Abstract
This work presents a novel means of producing thin-film light-emitting devices, functioning according to the principle of electroluminescence, using an inkjet printing technique. This study represents the first report of a light-emitting device deposited completely by inkjet printing. An electroluminescent species, doped zinc sulfide, was incorporated into a polymeric matrix and deposited by piezoelectric inkjet printing. The layer was printed over other printed layers including electrodes composed of the conductive polymer poly(3,4-ethylenedioxythiophene), doped with poly(styrenesulfonate) (PEDOT:PSS) and single-walled carbon nanotubes, and in certain device structures, an insulating species, barium titanate, in an insulating polymer binder.
The materials used were all suitable for deposition and curing at low to moderate (<150°C) temperatures and atmospheric pressure, allowing for the use of polymers or paper as supportive substrates for the devices, and greatly facilitating the fabrication process.
ii
The deposition of a completely inkjet-printed light-emitting device has hitherto been unreported.
When ZnS has been used as the emitter, solution-processed layers have been prepared by spin- coating, and never by inkjet printing. Furthermore, the utilization of the low-temperature- processed PEDOT:PSS/nanotube composite for both electrodes has not yet been reported.
Device performance was compromised compared to conventionally prepared devices. This was partially due to the relatively high roughness of the printed films. It was also caused by energy level misalignment due to quantization (bandgap widening) of the small (<10 nm) nanoparticles, and the use of high work function cathode materials (Al and PEDOT:PSS). Regardless of their reduced performance, inkjet printing as a deposition technique for these devices presents unique advantages, the most notable of which are rapidity of fabrication and patterning, substrate flexibility, avoidance of material wastage by using drop-on-demand technology, and the need for only one main unit operation to produce an entire device.
iii
Acknowledgements
The Pulp & Paper Centre in the Department of Chemical Engineering & Applied Chemistry at the University of Toronto has provided me with an environment in which I have been able to develop and refine both my project and my understanding of the fundamentals of chemical and materials engineering. The Department, the Estate of W.H Rapson, the School of Graduate Studies, the University of Toronto, the SENTINEL Bioactive Paper Network, and the Government of Ontario have all financially supported my continued perseverance in my doctoral studies, and I am deeply grateful to each. I would also like to acknowledge the generosity and companionship of all my colleagues within the Pulp & Paper laboratories and other laboratory groups within the Department. My understanding of several new aspects of chemical engineering was also deepened by my supervising committee members, Dr. Edgar Acosta and Dr. Timothy Bender. The laboratory of Dr. Ning Yan in the Faculty of Forestry was particularly welcoming, and always willing to provide access to both analytical equipment and technical expertise. The encouragement of my supervisor, Dr. Ramin Farnood, has been instrumental in keeping me focused on my work. His continued forward drive and enthusiasm, as well as all of the knowledge he has readily imparted to me, have led me to the successful conclusion of this long process. I feel very fortunate to have worked with such a helpful and driven person.
My family’s perpetual support has buoyed me from the first day of my degree and throughout. I cannot express my thanks sufficiently to them for their love and help. My father’s ready willingness to discuss my work and give advice taken from his long-time familiarity with electrical engineering has given me insight and inspiration on many occasions. Finally, my wife Emily has kept me level, confident, and determined throughout my studies, while bringing me joy each and every day. Her love has made persevering to the end worth every moment spent, and taught me to look to the future with excitement and hope.
iv
Nomenclature
Abbreviations 2D two-dimensional 3D three-dimensional AC alternating current AFM atomic force microscopy ALD atomic layer deposition ASTM American Society for Testing & Materials BSE backscattered electron CRT cathode ray tube CVD chemical vapour deposition DC direct current DLS dynamic light scattering DOD drop-on-demand DPI dots per square inch EL electroluminescence ELD electroluminescent display FPD flat-panel display HOMO highest occupied molecular orbital IR infrared LED light-emitting device/diode LUMO lowest unoccupied molecular orbital OLED organic light-emitting diode/device PDP plasma display panel PEL powder electroluminescent device PL photoluminescence PLE photoluminescent excitation PVD physical vapour deposition QD quantum dot R2R roll-to-roll SEM scanning electron microscope TAPPI Technical Association of the Pulp & Paper Industry TEM transmission electron microscope TFEL thin-film electroluminescent device TFT thin-film transistor ToF-SIMS time-of-flight secondary ion mass spectrometry UV ultraviolet XRD X-ray diffraction
v
Terminology dynamic viscosity (cP) surface tension/energy (mN/m) zeta-potential (mV) dp particle diameter (nm) dielectric constant (unitless) resistivity ( m) R resistance () conductivity (S/cm) wavelength (nm) c contact angle (°) Oh Ohnesorge number (unitless) Z-1 inverse Ohnesorge number (unitless) En Energy number (unitless) V voltage (V) I current (A) C capacitance (F) Eg energy band gap (eV) work function (eV) L luminance (cd/m2) luminous efficiency (lm/W) M molarity (mol/L) Mw polymer molecular weight (g/mol) h Planck’s constant (4.13566733 × 10−15 eV·s) ħ reduced Planck’s constant (6.58211899 × 10−16 eV·s) −12 0 vacuum permittivity (8.854187817620 × 10 F/m) 8 c speed of light (2.99792458 × 10 m/s)
vi
Materials 3-MPA 3-mercaptopropionic acid AA acrylic acid - Ac acetate ion (CH3COO ) AKD alkylketene dimer ATO antimony tin oxide C60 fullerene CdS cadmium sulfide CdSe cadmium selenide DMSO dimethyl sulfoxide HW hardwood ITO indium tin oxide MMA methyl methacrylate MWCNT multi-walled carbon nanotube PAA polyacrylic acid PAc polyacetylene PAni polyaniline PDADMAC poly(diallyldimethylammonium chloride) PEDOT poly(3,4-ethylenedioxythiophene) PEDOT:PSS poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) PEG polyethylene glycol PEI poly(ethyleneimine) PET polyethylene PMMA poly(methyl methacrylate) PPV poly(p-phenylene vinylene) PPy polypyrrole PT polythiophene PTFE polytetrafluoroethylene (Teflon) PVDF polyvinylidene fluoride PVK poly(N-vinylcarbazole) PVP polyvinylpyrrolidone SHMP sodium hexametaphosphate SLS sodium lauryl sulfate SW softwood SWCNT single-walled carbon nanotube TGA thioglycolic acid S-SWCNT ultra-short single-walled carbon nanotube ZnS:X doped zinc sulfide, X being dopant atom
vii
Units a.u. arbitrary units at. % atomic percentage wt. % weight percentage, weight basis vol. % volume percentage, volume basis mol. % molar percentage, molar basis VAC alternating current voltage VDC direct current voltage RPM revolutions per minute / ohm/square, sheet resistance for a square sample
viii
Table of Contents
Acknowledgements ...... iv
Nomenclature ...... v
Table of Contents ...... ix
List of Figures ...... xiv
List of Tables ...... xx
List of Appendices ...... xxi
List of Original Papers ...... xxii
Chapter 1 ...... 1
1 Introduction ...... 1
1.1 Motivation ...... 4
1.2 Related work on printed LEDs ...... 6
1.3 Approach ...... 9
1.4 Thesis structure ...... 11
Chapter 2 ...... 14
2 Inkjet printing ...... 14
2.1 Inkjet printer types ...... 14
2.1.1 Continuous (CIJ) ...... 16
2.1.2 Thermal (DOD) ...... 16
2.1.3 Piezoelectric (DOD) ...... 17
2.2 Piezoelectric inkjet printing: fluid dynamics ...... 20
2.2.1 Ink ejection ...... 20
ix
2.2.2 Droplet formation ...... 21
2.2.3 Droplet impact with substrate ...... 22
2.3 Inks ...... 24
2.3.1 Typical composition ...... 24
2.3.2 Fluid properties ...... 26
2.3.3 Orthogonal solvent systems ...... 33
2.4 Print quality ...... 34
2.4.1 Resolution ...... 34
2.4.2 Roughness & topography ...... 36
Chapter 3 ...... 40
3 Materials for inkjet-printed electronics ...... 40
3.1 Conductors ...... 41
3.2 Semiconductors ...... 49
3.2.1 Inorganics ...... 50
3.2.2 Organics ...... 55
3.3 Insulators ...... 57
3.4 Encapsulants ...... 60
3.5 Substrates ...... 60
Chapter 4 ...... 64
4 Light-emitting devices ...... 64
4.1 Light-emitting diodes (LEDs) ...... 65
4.2 Electroluminescent devices (ELDs) ...... 69
4.3 Suitability for printing ...... 71
4.4 Characterization ...... 73
x
Chapter 5 ...... 75
5 Approach & method development ...... 75
5.1 Materials selection model: semiconductor ...... 75
5.2 Ink formulation model: conductor ...... 76
5.3 Film formation model: insulator ...... 77
5.4 Multilayer device model: LED ...... 78
Chapter 6 ...... 80
6 Materials & methods ...... 80
6.1 Materials selection ...... 80
6.1.1 Electrodes ...... 81
6.1.2 Insulators ...... 81
6.1.3 Charge-transporters ...... 82
6.1.4 Emitters ...... 82
6.1.5 Substrates ...... 86
6.2 Ink formulation ...... 88
6.2.1 PEDOT:PSS inks ...... 90
6.2.2 BaTiO3 ink ...... 92
6.2.3 ZnS inks ...... 92
6.3 Jetting ...... 94
6.4 Drop spacing and film formation ...... 95
6.5 Ink distribution and print quality ...... 96
6.6 Functional testing of individual layers ...... 96
6.6.1 Conductive ink ...... 96
6.6.2 Emissive ink ...... 97
xi
6.6.3 Insulating ink ...... 98
6.7 Device fabrication and testing ...... 99
6.7.1 Interlayer dissolution ...... 101
6.7.2 Electrical characterization ...... 101
Chapter 7 ...... 102
7 Results & discussion ...... 102
7.1 ZnS synthesis ...... 102
7.1.1 Mn2+ loading ...... 102
7.1.2 Zn2+ to S2- ratio ...... 107
7.1.3 Post-synthesis capping ...... 108
7.1.4 Reaction temperature and time ...... 110
7.1.5 Reduction of particle size & improvement of dispersion ...... 113
7.1.6 Optimized synthesis procedure ...... 115
7.1.7 Synthesis of ZnS:Cu nanoparticles ...... 116
7.1.8 Characterization of ZnS:Mn, ZnS:Cu nanoparticles ...... 118
7.2 Other materials ...... 123
7.2.1 BaTiO3 ...... 123
7.2.2 PEDOT:PSS & CNTs ...... 124
7.2.3 Substrates ...... 129
7.3 Ink formulation ...... 137
7.3.1 Conductive ink ...... 137
7.3.2 ZnS inks ...... 152
7.3.3 BaTiO3 ink ...... 155
7.3.4 Optimized ink formulations ...... 156
xii
7.4 Ink film formation ...... 158
7.4.1 Drop spacing ...... 158
7.4.2 Film topography ...... 161
7.4.3 Interlayer interactions ...... 166
7.5 Functional testing of individual layers ...... 172
7.5.1 PEDOT:PSS/SWCNTs ...... 172
7.5.2 ZnS ...... 173
7.5.3 BaTiO3 ...... 174
7.6 EL device testing ...... 175
Chapter 8 ...... 182
8 Conclusions ...... 182
8.1 Major findings ...... 183
8.2 Recommendations & future work ...... 186
References ...... 190
xiii
List of Figures
Figure 1.1. Example of a subtractive photolithography process ...... 2
Figure 1.2. Example of an additive printing process ...... 2
Figure 1.3. Flowchart outlining general experimental approach...... 10
Figure 2.1. Schematic representation of the three main inkjet printer types...... 15
Figure 2.2. Sample model fluid waveform and a typical waveform used in this study...... 17
Figure 2.3. Fujifilm-Dimatix DMP2831 Dimatix Materials Printer ...... 19
Figure 2.4. Schematic of acoustic wave propagation in a piezoelectric nozzle...... 21
Figure 2.5. Drop formation from a piezoelectric inkjet nozzle...... 22
Figure 2.6. Jetted drop behaviour on a substrate, showing the “coffee-ring” effect...... 23
Figure 2.7. Fluid properties, and effects of deviation on jetting...... 28
Figure 2.8. Schematic representation of “peak and valley” topography formed during printing. 37
Figure 2.9. Schematic representations of common line morphologies...... 38
Figure 2.10. Drop spacing of QD/polymer/water ink on slide glass...... 38
Figure 3.1. Carbon nanostructures...... 44
Figure 3.2. PAni structure ...... 47
Figure 3.3. PEDOT (left) and PSS- (right) structures...... 47
Figure 3.4. Simplified electronic band structures...... 50
Figure 3.5. Film formation in inks containing inorganic nanoparticles...... 57
xiv
Figure 3.6. Microcracking in sol-gel-derived BaTiO3 spun films on glass ...... 59
Figure 4.1. Schematic of an OLED and its energy level diagram...... 65
Figure 4.2 Typical planar LED structures...... 67
Figure 4.3. Improvement of carrier mobility by polymer embedding of QDs used in QDLEDs. 67
Figure 4.4. ELD structures...... 69
Figure 4.5. PM LED array for testing: schematic (left) and setup (right)...... 74
Figure 5.1. Printed-coated ACPEL on paper ...... 79
Figure 6.1. Solvent selection for different device component inks...... 89
Figure 6.2. Light-emitting device builds prepared in this study...... 99
Figure 6.3. Schematic drawing showing a bird's-eye view of device construction...... 100
Figure 7.1. PL emission from ZnS:Mn nanoparticles (1.5 at.% Mn)...... 103
Figure 7.2. PL intensity vs. Mn added (at.%) ...... 104
Figure 7.3. PL of uncapped ZnS:Mn nanoparticles suspended in water...... 105
Figure 7.4. PL intensity at 608 nm vs. actual Mn content...... 106
Figure 7.5. PL emission in uncapped ZnS:Mn particles (50% Mn), different Zn:S ratios in reaction solution...... 107
Figure 7.6. Effect of capping agents added after synthesis on PL of ZnS:Mn...... 109
Figure 7.7. PL spectra of ZnS:Mn nanoparticles capped with SHMP ...... 110
Figure 7.8. PL spectra of ZnS:Mn nanoparticles capped with SHMP aged for varying times ... 111
xv
Figure 7.9. PL spectra of ZnS:Mn nanoparticles capped with varying amounts of SHMP...... 112
Figure 7.10. ZnS:Mn nanoparticles in water (0.1 w/w%), different stabilizers ...... 114
Figure 7.11. Finalized synthesis method of water-dispersible 3-MPA-capped ZnS:Mn ...... 115
Figure 7.12. Comparison of ZnS:Mn dispersion in water (1 w/w%). In all of the images, the vial on the left contains ZnS:Mn capped with SHMP; the right vial, ZnS:Mn capped with 3-MPA. 116
Figure 7.13. PL and PLE spectra of ZnS nanoparticles capped with 3-MPA...... 117
Figure 7.14. Mechanisms of light emission in ZnS:Mn and ZnS:Cu...... 119
Figure 7.15. XRD spectra of ZnS nanoparticles and bulk material ...... 121
Figure 7.16. TEM micrographs of ZnS:Mn nanoparticles...... 122
Figure 7.17. DLS scans of ZnS:Mn and ZnS:Cu nanoparticles in water (ZnS:Mn) and toluene (ZnS:Cu)...... 123
Figure 7.18. DLS-obtained particle size distribution of BaTiO3 (5 w/w%)...... 124
Figure 7.19. DLS-obtained particle size distribution of aqueous PEDOT:PSS ...... 125
Figure 7.20. STEM micrographs of MWCNTs ...... 126
Figure 7.21. Zeta-potential of CNT/SLS solutions and 50/50 (v/v) PEDOT:PSS/CNT mixtures...... 127
Figure 7.22. Printed PEDOT:PSS conductivity differences between different commercial paper types...... 129
Figure 7.23. Conductivity of printed PEDOT:PSS (single layer) as a function of added filler. . 130
Figure 7.24. ToF-SIMS maps of relative distribution of PEDOT, PSS, PEI, and TiO2 ...... 132
xvi
Figure 7.25. Estimated conductivity of PEDOT-SWCNT ink on SW fibres...... 133
Figure 7.26. Relative distribution of PEDOT:PSS and PDADMAC in HW sheet ...... 134
Figure 7.27. Estimated conductivity of PEDOT-SWCNT printed ink on HW fibres ...... 134
Figure 7.28. Cross-sections of printed SW handsheets (30% filler) showing PEDOT:PSS ink penetration...... 135
Figure 7.29. Greycale ToF-SIMS images of PEDOT distribution on unfilled (0% TiO2) ...... 136
Figure 7.30. Effect of added water on PEDOT:PSS suspension viscosity ...... 138
Figure 7.31. Effect of added DMSO on PEDOT:PSS/glycerol mixture’s viscosity ...... 139
Figure 7.32. 2-point resistance of cast-coated PEDOT:PSS/glycerol/water films ...... 139
Figure 7.33. Surface tension and viscosity in PEDOT:PSS inks with different surfactants ...... 141
Figure 7.34. PEDOT:PSS ink droplet formation during ejection from DMP2831 cartridge nozzles...... 142
Figure 7.35. Printed patterns of PEDOT:PSS inks on acetate ...... 143
Figure 7.36. 2-point estimated conductivity of printed PEDOT:PSS inks on cellulose acetate . 145
Figure 7.37. Raman spectra (excitation wavelength = 785 nm) of PEDOT:PSS inks...... 147
Figure 7.38. Conductivity of printed PEDOT:PSS-SWCNT ink (SLS surfactant) at varying SWCNT loadings ...... 149
Figure 7.39. Conductivity of inkjet-printed PEDOT:PSS-carbon composites on acetate ...... 150
Figure 7.40. ToF-SIMS maps of PEDOT and substrate component distribution ...... 151
Figure 7.41. Printed ZnS:Mn/AA ink on cellulose acetate, 10 printed layers...... 153
xvii
Figure 7.42. Droplet formation of ZnS inks at 10 µs intervals...... 154
Figure 7.43. Optical microscope imaging of printed ZnS:Mn/PVP ...... 155
Figure 7.44. Droplet formation of BaTiO3 ink at 5 µs intervals...... 156
Figure 7.45. (a) Jetted drops of BaTiO3 ink on ITO PET; (b) edge of BaTiO3/PMMA film on ITO glass...... 156
Figure 7.46. Drop sizes of BaTiO3/PMMA ink on various substrates...... 158
Figure 7.47. Printed lines of BaTiO3 ink (single jet, single layer) at different drop spacings. ... 159
Figure 7.48. 3-D profile (left) and 2-D linescan (right) of printed BaTiO3 on slide glass...... 162
Figure 7.49. SEM micrographs of printed BaTiO3 ink ...... 162
Figure 7.50. BaTiO3/PMMA average dried ink film thicknesses on slide glass...... 163
Figure 7.51. Topography of single printed layers of all inks on slide glass...... 165
Figure 7.52. Multiple layers of ZnS:Mn/PVP ink showing edge splattering with excessive overprinting...... 167
Figure 7.53. ITO dissolution by ZnS:Cu/PVK ink, containing 3-MPA, and ZnS:Mn/PVP ink, containing TGA ...... 168
Figure 7.54. Optical profilometry of printed ink layers, on the surfaces they would cover in printed devices ...... 170
Figure 7.55. Reduction of “peak-and-valley” topography in ZnS:Cu/PVK films with successive overprints on slide glass...... 171
Figure 7.56. PEDOT:PSS/SWCNT ink conductivity when printed on slide glass...... 172
Figure 7.57. Estimated relative dielectric constants of printed BaTiO3 films ...... 174
xviii
Figure 7.58. Current-voltage (I-V) and luminance-voltage (L-V) characteristics of PEDOT:PSS/SWCNT – ZnS:Cu/PVK – Al LED ...... 176
Figure 7.59. Current-voltage (I-V) and luminance-voltage (L-V) characteristics of PEDOT:PSS/SWCNT – ZnS:Cu/PVK – PEDOT:PSS/SWCNT LED ...... 176
Figure 7.60. Electronic band structures (top), device architecture (middle) and EL emission (bottom) of printed ZnS:Cu DC-LEDs ...... 177
Figure 7.61. Current-voltage (I-V) and luminance-voltage (L-V) characteristics of PEDOT:PSS/SWCNT – ZnS:Mn/PVP – Al DCPEL ...... 180
xix
List of Tables
Table 2.1. Typical inkjet ink composition ...... 25
Table 4.1. ELD/LED materials and layer properties...... 72
Table 4.2. Characteristic properties of light-emitting devices...... 73
Table 6.1. Capping agents for ZnS:Mn nanoparticles, aqueous synthesis...... 83
Table 6.2. Selected properties of commercial substrates...... 87
Table 6.3. Selected properties of lab-made handsheets...... 88
Table 6.4. Surfactants tested in PEDOT:PSS ink and their CMCs...... 91
Table 7.3. PEDOT:PSS inks’ fluid properties...... 142
Table 7.5. Finalized ink formulations...... 157
Table 7.6. Summary of drop sizes and line spacing for all inks...... 160
Table 7.7. Printed film roughnesses and peak-to-valley differences in ZnS and PEDOT:PSS/SWCNT inks...... 164
Table 7.8. Ink layer thicknesses and layers required for device construction...... 166
xx
List of Appendices
APPENDIX A. ZnS nanoparticle synthesis & dispersion ...... 215
APPENDIX B. Procedure for conductivity estimation in PEDOT/SWCNT films ...... 223
APPENDIX C. Film thickness estimation ...... 226
APPENDIX D. Drop and line spacing optimization ...... 228
APPENDIX E. Paper substrate preparation ...... 231
APPENDIX F. Impermeable substrate preparation ...... 232
APPENDIX G. Paper substrate characterization ...... 233
APPENDIX H. Detailed ink formulations ...... 234
APPENDIX I. Ink iterations ...... 237
APPENDIX J. Jetting waveforms ...... 251
APPENDIX K. ToF-SIMS fragments analyzed and construction of molecular maps .. 253
APPENDIX L. Printed PEDOT:PSS/SWCNTs on paper ...... 255
xxi
List of Original Papers
1) Angelo, Peter & Farnood, Ramin (2010): Poly(3,4-ethylenedioxythiophene): poly(styrene sulfonate) inkjet inks doped with carbon nanotubes and a polar solvent: the effect of formulation and adhesion on conductivity, Journal of Adhesion Science & Technology. 24(3), 643.
2) Angelo, P.; Farnood, R. (2012): Inkjet-printed PEDOT:PSS-SWCNT films: the effect of surfactants on jetting and electrical performance, submitted (Journal of Materials Chemistry).
3) Angelo, P.; Cole, G.; Sodhi, R.; Farnood, R. (2012): Conductivity of inkjet-printed PEDOT:PSS/SWCNTs on uncoated papers, Nordic Pulp & Paper Research Journal 27(2), 486.
4) Angelo, P.; Farnood, R. (2012): Conductivity of PEDOT-CNT composites, submitted (Journal of Materials Research).
5) Angelo, P.; Farnood, R. (2012): Inkjet-printed BaTiO3/PMMA dielectric films, Ceramics International, in press.
6) Angelo, P.; Sweeney, C.; Farnood, R. (2012): Paper-based electroluminescent devices prepared using conventional printing/coating, submitted (Journal of Display Technology).
7) Angelo, P.; Farnood, R. (2011): Photoluminescent inkjet ink containing ZnS:Mn nanoparticles as pigment, Journal of Experimental Nanoscience, 6(5), 473.
8) Angelo, P.; Chlebowski, M.; Farnood, R. (2012): Mn2+ incorporation into ZnS nanoparticles prepared in aqueous solution, submitted (Journal of Nanoparticle Research).
9) Angelo, P.; Kronfli, R.; Farnood, R. (2012): Synthesis and inkjet printing of aqueous ZnS:Mn nanoparticles, Journal of Luminescence, in press.
xxii
Chapter 1 1 Introduction
Electronic devices of all types are ubiquitous in today’s society. The production of electronics has therefore been established over the past several decades as an integral industry for the sustenance and development of the modern world in virtually all of its aspects (Sedra & Smith, 1997). Devices such as integrated circuits, sensors, displays, and lighting systems comprise a tremendous fraction of current technology and are critical to industrial, social, and scientific endeavours. In particular, display devices of all types are necessary to establish an interface for technological interaction and for the conveyance of information. Markets have shifted away from conventional cathode-ray-tube (CRT) technology towards flat-panel displays (FPDs) such as liquid-crystal displays (LCDs), plasma display panels (PDPs), and electroluminescent displays (ELDs), pushing peripheral technologies involved in display fabrication forward (DFF 2008, Mentley 2002). The global market for these displays is expected to reach a value of over US $100 billion by the year 2015, with the ELD market in particular experiencing the largest compound annual growth rate out of the FPD classes – 5.3% over a 4-year analysis period, based on studies of 249 different companies (Electronics.ca Publications 2011). It is self-evident from these figures that the production and sale of FPDs and their components represents a tremendously lucrative market. Research into the reduction of materials or manufacturing costs, as well as the development of novel platforms for display applications, is therefore an area attracting well-deserved interest. Furthermore, many of the components used to construct displays, such as the conductive electrodes and the semiconducting phosphors, may also be used in the construction of other electronic devices (Dimitrijev 2005, Chen et al. 2011), meaning that their development can contribute to markets beyond that of FPDs.
The gradual development of increasingly complex devices, as well as their miniaturization, has necessitated constant invention of new fabrication techniques and equipment (Jaeger & Balock 2003). Naturally, these methods have also themselves become complex and involved as electronic components and their materials of construction shrink to the micro- or nano-scale (Mahalik 2006). Production has become an intricate process involving multiple deposition steps
1
of a variety of high-quality materials, often at high temperature or vacuum, utilizing aggressive or toxic chemicals, in a cleanroom environment (Campbell 2001). For the production of FPDs, which are capable of demonstrating ever-increasing pixel counts, extremely careful patterning to improve the display resolution necessitates the use of labour-intensive photolithography methods (Jaeger 2002). In all cases of electronics manufacturing, some degree of patterning is required, which generally is achieved by either masking areas to prevent material from being deposited on them, or by selective removal of material after deposition (Figure 1.1). Both of these methods
4) rinse solubilised oxide 1) clean substrate photoresist
substrate
5) acid-etch oxide photoresist 2) coat photoresist
mask/UV 3) expose to UV 6) dissolve remaining through photomask photoresist
Figure 1.1. Example of a subtractive photolithography process requiring both masking and etching. imply material wastage; in the former case, wasted material is deposited on the mask, and in the latter, a “subtractive” method, material is etched away and removed. They also imply additional processing stages – namely mask fabrication, swapping out of masks between stages, maintenance of masks, and etching. Throughput, efficiency, and economic return would all benefit from the use of an “additive” method, where material is deposited in the correct pattern at each stage of fabrication, with little to no material wastage (Figure 1.2).
printhead carrier solvent wet precursor bare substrate 1) clean substrate 2) print pattern (computer- 3) heat or UV cure controlled)
Figure 1.2. Example of an additive printing process using digital patterning rather than masking.
2
The standard substrates of silicon or glass upon which much of the microelectronic industry is built cannot take advantage of the unique opportunities offered by roll-to-roll (R2R)-type processing, despite their generally desirable electronic or surface properties. Naturally, their stiff and brittle nature makes them unsuitable for R2R printing in its conventional sense. However, even without considering the issue of flexibility, silicon and glass suffer from one major technological chokepoint – they are limited in terms of size. Silicon wafers generally cannot be made larger than ~300 mm diameter (O’Mara et al. 1990), and the most advanced new 10th- generation display-grade glass cannot be cut in sizes larger than ~2 3 m (American Ceramic Society 2010). As large as this glass might be, its growth is limited by the rate of advancements in materials science and practical considerations of handling. Scale-up requires extensive work on improvement of the physical properties of the substrates, and tremendous investment in infrastructure for processing, including new dies, chambers, masks, and so forth, and it is generally more common simply to reduce microelectronic feature size while retaining smaller substrates (Moore 1965). Scale-up in R2R processing, conversely, is almost entirely digital: patterns can be remade and scaled with a keystroke, and substrates of practically limitless size can be readily handled.
With interest in flexible electronics constantly increasing (Whitmarsh 2005), the possibility of R2R fabrication is a particularly attractive one, due to its high-throughput, continuous process model. The use of R2R processing implies the use of flexible substrate materials to support electronic devices, which in turn implies certain restrictions on the process, such as a limited temperature range, usually < 200°C (to accommodate polymers, textiles, and paper). These substrates present both advantages and challenges, ranging from mechanical flexibility to biodegradability in the former, and from mechanical robustness to porosity in the latter. Innovative materials selection and integration, for both the liquid precursors and the substrates, therefore, is the dominant factor in the realization of R2R-printed electronics.
For increasingly complex printed devices, more and different materials will be required, necessitating the establishment of ink “libraries” applicable to different substrates and underlying layers at a variety of processing conditions (Magdassi 2010). These materials must, by definition, be suitable for application in the liquid phase, forming features of tightly-controlled
3
size and morphology, while retaining functionality. Also, in most cases, these materials are nano-sized structures or particles due to the restrictions on particle size imposed by the use of inkjet nozzles (Tekin 2008). So, it becomes immediately evident that a thorough understanding of a wide range of subjects, from microelectronic construction to colloid chemistry, is involved in the realization of printed electronic devices.
1.1 Motivation
Inkjet printing is currently only sporadically used for the production of electronic devices, generally to provide a single layer, most often an electrode. However, the ever-growing library of jettable materials presents the possibility of combining inkjetted layers into more complex devices. Naturally, the specific material requirements determine which materials must be incorporated into inks. The ink formulation process may be very complex, depending on the material being used, its compatibility with carrier solvents and other ink components, and its ease of dissolution or dispersion. Ink formulation therefore becomes an often-overlooked but absolutely crucial aspect of functional material deposition and device fabrication. This is an involved process, requiring careful control of fluid properties, wetting behaviour, drying behaviour, interaction with a given substrate, maintenance of dispersion, and above all, retention of functionality. A consideration of the level of retained functionality, relative to conventionally- produced electronics, is a natural component of this process. To summarize, there is a large gap between a simple suspension or solution and an actual ink suitable for producing electronic structures (Caglar 2009, Magdassi 2010). An effort to establish and refine the ink formulation process would prove to be an invaluable tool for the rapid development of the aforementioned material libraries.
The potential applications of such materials are nearly limitless. Certain applications might require simple single-layer printed structures, some multilayer stacks. Some concrete examples of the former type include radio frequency identification (RFID) tags, antennas, or simple printed wiring using conductive material. They might also include printed patterns of light-emitting or absorbing species. If biomolecules are printed, they might function as pathogen-detecting or
4
even pathogen-deactivating sensors. Multilayer devices could include any conventionally- prepared planar electronic device, ranging from thin-film-transistors (TFTs) to photovoltaics (PVs) to light-emitting devices (LEDs). In this work, the LED served as an ideal model device for the demonstration of the inkjet deposition process as it pertained to microelectronic fabrication.
Therefore, there was one overriding objective in this study – the production of a functional LED using inkjet printing as the sole unit operation, as a demonstration of the possibility of inkjet deposition of a range of materials. Within this larger objective, several smaller goals were delineated. Firstly, the development of several customized inkjet inks, each bearing a particular functional material, was necessary. A methodology for analyzing their performance during jetting, upon contact with the substrate, and upon curing or drying was integral to this development. While these analyses of performance occasionally represented relatively simple characterization (e.g. fluid properties), the interconnectedness of each variable in ink formulation meant that this first objective was not a trivial one. Other work has reported the use of simple dispersions or solutions of materials as inks (as is discussed in the following sub-section); optimal jetting, drying, and electronic behaviour were highly dependent on thorough analysis of more detailed ink formulations, however.
A second sub-objective was to utilize these optimized inks to produce films and test their functionality, particularly in comparison to materials not deposited (or able to be deposited) by inkjet. As will be discussed in more detail in Chapter 3, these structures included conductive, insulating, and semiconducting layers. So, in a multilayer device, it was a necessity to establish each layer’s electrical performance as sufficient for device application. Furthermore, in certain cases, the substrate was expected to have a bearing on layer performance. So, any materials likely to be in contact with an underlying layer with which they might physicochemically interact had to be tested for functionality on a variety of different substrates to establish the optimal surface upon which to deposit them.
The final sub-objective was to stack those sufficiently functional layers into an LED. There being many different types of LED, multiple structures were expected to be tested. Also, due to
5
the likelihood of interlayer interactions, certain materials were only applicable in combination with certain other materials. Emission of light was considered proof of concept for a printable LED – optimization of its electrical properties was beyond the scope of this work.
1.2 Related work & state of the art
The notable contributions of this work were to further develop the science behind inkjet-ready materials development, and to examine the potential and performance of such materials in fully- printed devices. The challenges associated with such a novel product as a printed LED necessitated the development and optimization of specialized processes for materials preparation and handling.
1.2.1 Inkjet-ready electronic materials
As will be discussed in detail in the body of the thesis, inkjettable materials are specialized in several ways – the two most notable being sub-micrometre particle size and liquid dispersibility. Of course, many materials with these properties have already been prepared and studied: conductive metallic/polymeric colloids, carbon nanostructures, ceramic nanoparticles/sol-gels, and colloidal semiconductors (quantum dots). Quantum dots, in particular, have been extensively studied as useful materials for optoelectronic devices. In order to control particle size and dispersibility, these materials are generally prepared using a “bottom-up” approach – chemically reacting their component atoms into nanoparticle seeds that are grown in solution, with growth being arrested by the presence of a compound coating their surface, referred to as a “cap”. The cap, depending on its chemistry, then determines the stability of a dispersion of the quantum dots in a given carrier solvent.
Many different types of quantum dots have been prepared for solution-based processing. The most famous of these are the CdS or CdSe quantum dots, which can be tailored to produce light emission/absorption at a variety of wavelengths. For displays or light-emitting diodes, conventional technology has often relied on a different type of semiconductor, however – one containing a dopant (impurity) atom, from which a single characteristic wavelength is emitted.
Some very common examples of this include yttrium-based phosphors (such as Y2O3) and zinc-
6
based phosphors, including ZnS, which is ubiquitous in electroluminescent display technology, and was therefore considered a prime candidate for development of a light-emitting device.
The science behind inkjet deposition, however, is more complex than the simple production of liquid droplets. As will be discussed in the following chapter, inkjet fluids must meet certain rigorous criteria in order to jet at all, and to form smooth films on a given substrate. Moreover, the dispersion of quantum dots in a given solvent – which must then be generally mixed with other components to produce an ink with suitable properties for jetting – is paramount to maintain printer function. In the literature, the myriad synthesis methods for quantum dots rarely consider redispersion, let alone redispersion into a multicomponent mixture. There has not been a work, to date, in which as-synthesized quantum dots are dispersed into a stable ink and jetted while retaining photoluminescent and electroluminescent brightness. In this work, such methods as exist for the production of doped nanoparticles are more closely examined. More importantly, a refined synthesis drawing upon these previous methods is presented, by which true, monodisperse quantum dots of doped ZnS are produced, with size and surface functionality suitable for jetting (in aqueous and organic media) and subsequent retention of function. This has hitherto not been demonstrated for doped ZnS nanoparticles of any type.
1.2.2 Inkjet-printed displays
Inkjet printing of electronics has attracted interest in the past several years. It is a useful means for the complete fabrication of several film-based electronics, such as transistors (Chung et al. 2010, 2011; Tseng & Subramanian 2011, Liu et al. 2005), sensors (Molina-Lopez et al. 2012), and capacitors (Lim et al. 2010) which often not only require careful patterning, but also involve the use of expensive or exotic materials. Some of the advantages afforded by inkjet printing include single-step processing and low firing and curing temperatures, with relatively high device resolution (Calvert 2001, Tekin et al. 2004, 2008; Sirringhaus & Shimoda 2003, de Gans et al. 2004, Yoshioka & Jabbour 2006). Attempts at similarly preparing LEDs have also been undertaken (Wood 2009, Haverinen 2010); however, fully-printed LEDs have not been realized. Because inkjet printers require very small particle sizes in the inks (Tekin et al. 2004, Meixner et al. 2008), printable LEDs require specialized materials as light-emitting species. Often, these
7
take the form of nanoparticulate semiconductors, also often referred to as quantum dots (QDs) when below a certain size (Sun 2005), or certain organic polymers or molecules. In the existing research and development conducted on printed displays, the emissive layer alone was printed, with the remainder of the device fabricated by more conventional means.
The types of materials suitable for inkjet printing of such displays have been solution-processed by means other than printing in several studies. Well-known cadmium-based nanoparticulate semiconductors, such as CdS, CdSe, or CdTe, have been used in LEDs studied by Gaponik et al. (1999), Kumar et al. (1997), Colvin et al. (1994), Yang & Holloway (2003), Mattousi et al. (1998), and many others. In these cases, the nanoparticles were suspended in a conductive or semiconducting binder material, or densely packed, by solution processing – usually spin- coating. Similarly, zinc-based nanoparticles (such as ZnO and ZnS) have been used. ZnS:Mn has been applied by dispersing it in a cyanoresin paste and screen printing the composite (Adachi et al. 2007, 2008). Schrage et al. reported a spin-coated single-layer ZnS:Cu/PVK composite LED (2010). A similar structure using ZnS:Mn, spin-coated with no polymer binder but on top of a spin-coated PVK layer, was also reported by Yang et al. (2003). A more complex structure, using PVP-capped ZnS quantum dots with multiple vacuum-deposited charge transport layers was reported by Manzoor et al. (2003). Another simple single-layer structure using ZnS quantum dots in polymeric matrices was reported in a patent by Hieronymas (2002). The motivation for the use of nanoparticulate inorganic semiconductors rather than organic emitters has generally remained the same: their narrow emission peaks with ~20 nm full-width at half- maximum (Dabbousi et al. 1997) versus the wider emission spectra of organics, at ~100 nm full- width at half-maximum (Xing et al. 2005). Besides this, organics may be extremely expensive, difficult to synthesize, and challenging to process and maintain (Kwong et al. 2005).
With the efficacy of such materials established in these studies, a more elegant and rapid means of patterning – inkjet printing – appeared to be well-suited to applying them. However, the realization of printed LEDs was not as simple as expected. Initial attempts simply used the photoluminescent properties of printed QDs to produce light of various colours, by exciting them through electroluminescent emission of an adjacent bulk phosphor layer (Wood 2009, Taylor 2007). Although this method did not actually cause light emission from the QDs themselves, and
8
furthermore, required deposition of a bulk phosphor material, it did allow for the use of alternating-current (AC) drive. Eventually, Haverinen et al. (2009, 2010) applied CdSe QDs by inkjet as the sole emissive layer in a complex device stack incorporating several charge transport layers, showing promising device performance with several different colours of QDs. While it is evident from that study that the inkjet production of an LED from such luminescent materials has been at least partially realized, this work attempts to further describe the role of the printing process in such a fabrication. Moreover, in one of the iterations of the devices reported in this thesis, the entirety of the device is printed, rather than only the single emissive layer. This represents the first demonstration of a simple light-emitting device prepared entirely by inkjet at low-moderate temperatures and atmospheric pressure. This clearly represents a major advancement in the field of printed electronics and opens the door to the realization of solution- based R2R systems as the sole unit operations in electronics manufacturing.
1.3 Approach
Initially, the focus of the work was to use conventional printing and coating techniques to produce LEDs (in particular, ELDs) using paper as a substrate. This entailed several limitations on materials which would be pertinent throughout the remainder of the study. These included, most notably, the use of liquid precursors and low processing temperatures. Therefore, when the decision to attempt inkjet printing of different functional materials was made, the constraints on the experimental approach of outlining a roadmap for ink development and application were well-understood. When studying printable materials which might be applicable on a variety of substrates (potentially including paper), these two specifications of liquid-phase precursors and low-temperature processing guided the research plan.
Because several different materials and variations of those materials were used, as well as their respective inks, studies were carried out in a “layer-by-layer” fashion. In other words, the formulation, jetting, and functional testing of each material’s ink was initially carried out one ink at a time. When layers of inks were stacked together, a similar approach was used to observe jetting performance, film formation, and so forth, when in contact with the other materials. Figure 1.3 shows a simplified flowchart of the research method for this study. Because of the
9
large number of interconnected variables associated with ink formulation and deposition, the process was often an iterative one.
Materials selection Suitable material No Establishment of functionality available?
Yes
Establishment of Synthesize materials to dispersion/solution jettable specifications
Dispersible No /soluble?
Yes
Ink formulation
Printing studies Droplet formation
Functional testing
Jettable? No
Yes Functional as Film optimization desired? Layer characterization No
Forms desired No Yes structures?
Printed device Interlayer interaction Yes Device characterization
Figure 1.3. Flowchart outlining general experimental approach.
Firstly, suitable materials were chosen to produce the requisite layers. Secondly, an attempt was made to incorporate those materials into a liquid carrier to serve as the basis of an ink. If the
10
materials could not be sufficiently well-incorporated by dispersion or dissolution, the first step of materials selection was repeated. Next, the dispersion was reformulated into an ink with specific fluid properties for jetting. Chemical incompatibilities became a major concern at this stage, as certain ink components and additives compromised dispersion or dissolution. The ink was then introduced to the inkjet printer, where printer and cartridge settings had to be tailored to each specific ink to produce ideal droplets. If an ink was unable to jet properly, ink formulation was repeated. The exact print specifications required to produce a printed structure of the desired conformation (a film) were then established. If the structure could not be successfully formed, reformulation and possibly materials reselection were warranted. Finally, successfully jetted materials were studied for functionality, which was affected by the various iterations described above. Lastly, the interaction of adjacent layers was considered, as deleterious effects in this regard would have a bearing on LED function.
It would be redundant to outline in this document all of the iterations of all of the inks formulated (although they are appended to the work in APPENDIX I). Indeed, the individual inks served unique purposes in outlining the transition from a bulk material to an inkjet-printed layer. Their exact roles will be elaborated upon within the body of the thesis. In brief summary, a conductive ink served as a model for the intricacies of precise formulation and its effects on function, a semiconducting ink as an example of the difficulty of materials selection and incorporation, and an insulating ink as a model for film formation. Finally, as has been indicated above, an LED structure was used to study the challenges posed by interlayer interaction, film structure, and so forth, and as a proof-of-concept of the suitability of this novel development scheme for inkjet- printed electronic materials.
1.4 Thesis structure
The thesis is divided into eight chapters. The first four of these chapters are intended to review the fundamental knowledge required to develop inkjet-printed electronic devices. This first chapter has served to give an overview of the motivation and objectives of the work, as well as a brief background into the idea of printed electronics. Chapter 2 is concerned with outlining the
11
scientific concepts behind inkjet printing and inks. Chapter 3 provides an overview of some of the materials which are suitable for use in printed electronic applications, dividing them into the most common classes, such as electrodes, insulators, and so forth. Substrates, including flexible ones (polymers, paper) are also discussed here. Chapter 4 discusses the principle of light emission from LEDs of various types, their typical structures, and means of characterizing them. The ELD and the actual light-emitting-diode LED are focused on in particular, as they were the best-suited to inkjet application.
The discussion of the inks prepared and deposited for this study begins in Chapter 5. This chapter deals with the development of a process to move from a desired structure and its component materials (covered in Chapters 3 & 4) to actual inks containing those materials, ready for assembly into an ELD/LED. The experimental methods necessary for this development are elaborated upon in Chapter 6, taking each material into account in particular for functional testing of printed films. Each material used had a particular purpose which required a very different test platform to observe its performance. Certain of the experimental methods requiring more detailed descriptions are elaborated upon in the Appendices. Chapter 7 describes the results of these experimental methods, comparing the observed performance of the various inks both in isolation and in the context of a full ELD/LED. Demonstrations of device functionality are also presented in this chapter. Finally, Chapter 8 summarizes the major conclusions, the findings presented in Chapter 7, the limitations of the methods and devices presented, and recommendations for future research. The Appendices illuminate some of the more detailed experimental methods, exact ink formulations, novel experimental setups, and so forth.
Finally, the original publications listed in the prefatory material can be considered complements to the thesis. Certain portions of these publications are included in the body of the thesis; these sections of the thesis may be considered a distilled version of the work conducted in these publications.
PAPER 1 describes a first attempt at formulating a conductive inkjet ink, and its application to a flexible substrate (paper and acetate). The mechanical resilience of the resulting layer is then studied, and the possibility of using a flexible substrate to support this layer established.
12
PAPER 2 describes the more detailed aspects of refining the formulation of the conductive ink, comparing the effects of different surfactants in an otherwise constant formulation. Changes in conductivity attributed to both the print quality and chemical interactions induced by different surfactants are discussed.
PAPER 3 outlines the introduction of carbon nanotubes (CNTs) into the conductive film and their role in conductivity enhancement. .
PAPER 4 examines the wide variability in electrical resistance in the conductive printed layer on different paper substrates, by a comprehensive analysis of several different hand-made paper sheets’ effect on conductivity. An optimal surface for conductive ink deposition is established.
PAPER 5 describes the formulation and application of an insulating ink, to function as a dielectric layer. Topography and dielectric behaviour are observed to establish the feasibility of using such an ink in an ELD.
PAPER 6 summarizes the fabrication and testing of a preliminary ELD on a flexible substrate, using a printed conductive layer and coated insulating and emissive layers.
PAPER 7 describes an initial attempt at synthesizing and depositing ZnS nanoparticles to function as a light-emitting layer in a fully-printed ELD. Nanoparticles are characterized for photoluminescence, crystallinity, and size, and an ink is formulated and printed using an acrylate polymer as a binder.
PAPER 8 further explores the synthesis of these nanoparticles, with a focus on characterizing and improving the doping level of the nanoparticles to increase luminescent yield. Several different pre- and post-synthetic treatments are considered to improve emission from the nanoparticles, with an optimal synthesis method being described.
PAPER 9 describes a refined version of the synthesis method used in PAPER 8, which is used to produce truly monodisperse nanoparticles of doped ZnS that are readily dispersible in water and suitable for inkjet printing. Issues of print quality and substrate interaction are briefly explored, and the ink is considered a suitable candidate for application to a rudimentary LED.
13
Chapter 2 2 Inkjet printing
Printing has been used as a patterning technique since ancient times, and until recently, has been entirely based on physical contact between a transfer part and a substrate (Eisenstein 1997). The transfer part, whether it is a patterned roll, stamp, block, or screen, functions on the basis of some physical phenomenon – usually pressure, surface energy gradients, or masking (Skotheim 2006). In terms of printing technologies, inkjet printing is unique in that it is a non-contact method, dispensing material directly from reservoir to substrate in a predefined pattern (Le 1998, Bucknall 2005). Because this pattern is generally defined using software, inkjet printing is considered a “digital” printing method, which allows for limitless flexibility in pattern definition and rapid pattern adjustment. The substrate or the nozzle(s) can be moved in two or three dimensions as drops are formed to form the pattern. By selecting exactly where to deposit material, the printer can avoid wastage almost entirely and can deposit very small amounts of ink in a tightly controlled fashion (Zhouping et al. 2010, Haverinen 2010). The digitally-controlled inkjet printer can also readily be realigned to a particular location on a substrate, allowing for over-printing of new materials and patterns with pinpoint accuracy (Curling 2006). The latter two features, in particular, make inkjet-printing particularly well-suited to the fabrication of many different types of electronic devices.
2.1 Inkjet printer types
Inkjet printers can be broadly divided into two classes: continuous (CIJ) and drop-on-demand (DOD). Each of these classes can then be further subdivided into several subclasses, including Hertz and microdot continuous printers, and thermal and piezoelectric DOD printers (Caglar 2010). The primary difference between the two main classes involves the production of droplets; as is evident by their name, CIJ printers produce a continuous stream of droplets from their nozzles, whereas DOD printers eject drops only when impelled to do so. Figure 2.1 shows a schematic of the three main printer types – CIJ, thermal-DOD, and piezoelectric-DOD.
14
heating element piezo element vapour voltage nozzle bubble source
charging droplet plates
deflector plate voltage ground source
gutter substrate
thermal (bubble-jet) piezoelectric continuous Figure 2.1. Schematic representation of the three main inkjet printer types.
Certain similarities exist between the printer types. Of course, all have some system of nozzles to allow the flow of ink and its formation into droplets. Generally, all also have some type of pump, which can provide suction to prevent ink from dripping out of the nozzles when not printing, and positive pressure to flush the nozzles with ink or another cleaning fluid. More importantly, though, all require a set of electrical motors and servos to correctly position the nozzles above the substrate to produce a pattern, according to the digital specifications of that pattern. These motors can either move the nozzle assembly (printhead) itself, or move the substrate underneath a fixed nozzle. Commonly, there is motion in both the printhead and the substrate – a desktop printer being a good example, where the printhead traverses the sheet laterally, printing a line at a time, and the paper sheet moves forward through the printer.
15
2.1.1 Continuous (CIJ)
In CIJ systems, a nozzle connected directly to a reservoir is used to produce a continuous stream of droplets, either by simple Plateau-Rayleigh instability (Papageorgiou 1995) or by induction of a high-frequency vibration using a piezoelectric crystal (Croucher & Hair 1989), or both. As the drops fall, a fraction of them are electrostatically charged by passing between charging plates. As they approach the substrate, the charged droplets are selectively forced towards a grounded deflector plate by a large potential applied across a second plate; these deflected droplets are collected in a gutter and recycled while undeflected droplets impact the substrate (Le 1998). This is the most mature inkjet technology, likely because of the simplicity of the nozzle assembly and lack of microfabricated parts. Its relatively complex apparatus, including pumps, reservoirs, and several charged plates and power sources, however, means that DOD systems are often more compact and simple, and therefore more commonly used (Caglar 2010).
2.1.2 Thermal (DOD)
Thermal inkjet printers function by rapidly (i.e. in a few µs) heating a small amount of ink inside a reservoir with an open nozzle using a resistor connected to a power supply. As the heat is applied, the ink vapourizes, creating a bubble which applies a high pressure (> 1 MPa) wave to the fluid inside the reservoir, forcing a drop out through the nozzle (Le 1998, Croucher & Hair 1989, Rembe et al. 1999). The heater then cools rapidly, before another drop is produced. An advantage of this technology is the relatively low cost, as it involves only a resistor and a reservoir, easily fabricated by existing MEMS techniques with low-cost materials (Molesa 2006). Therefore, thermal inkjet systems are commonly used as household printers, where printer cost and regular printhead replacement are relevant. An issue associated with thermal systems is their limited applicability – usually, inks must be water-based and low-viscosity (~1 cP) to be easily vapourized and jetted (Zhouping et al. 2010). On this same note, the high thermal stresses to which inks are subjected may destroy sensitive ink components, such as biomolecules or polymers, or otherwise alter the functionality of suspended materials (Haverinen 2010). Thermal inkjet printers are therefore better-suited to the printing of conventional inks for graphics and text.
16
2.1.3 Piezoelectric (DOD)
Piezoelectric printers, by avoiding thermal cycling issues, are suitable for a broader range of ink formulations and materials (Clymer & Asaba 2004, Magdassi 2010). A mechanical deformation of a small ink reservoir induces a pressure wave in the ink, forcing out a droplet (Le 1998, Ridley et al. 1999). The component of the printhead which causes the deformation is a piezoelectric crystal, which changes shape upon the application of potential; the use of this specialized material means that the manufacture of piezoelectric printheads is significantly more difficult and costly than thermal ones (Molesa 2006). Another difference between this printhead type and a thermal printhead is the degree to which drop size and shape can be controlled – careful adjustment of peak voltage, voltage pulse length, and even the shape of the voltage-time waveform applied to each piezo element can alter drop formation (Fujifilm-Dimatix 2006, Molesa 2006, Haverinen 2010, Caglar 2010). An example of such a waveform is shown in Figure 2.2.
rise time echo time
model step width waveform
V
slope (“slew rate”)
typical waveform
t (µs)
pulse width
Figure 2.2. Sample model fluid waveform (blue) and a typical waveform used in this study (red).
17
The “rise time” refers to the length of the waveform which increases up to a peak voltage; the “echo time” or “fall time” refers to the portion in which voltage drops to its minimum value. These two regions of the waveform affect the fluid mechanics inside the print nozzles, which are described in Section 2.2.1. If a waveform is segmented into several increases/decreases and plateaus in voltage, each of these segments has a temporal width referred to as the “step width”. The slope of the rising or falling segments is referred to as the “slew rate”, generally expressed in units of V/µs. Finally, the temporal length of the entire waveform is called the “pulse width”.
Piezoelectric nozzles may also include features such as temperature control for more viscous fluids, high-frequency vibration to prevent liquid “skins” from forming, and purging settings that flush the nozzles with fluid. They may also include cameras to assist in realignment of the printer between layers (Curling 2006), and to observe and optimize drop formation by adjusting the voltage waveform. Because almost any material can be used in the piezoelectric printhead without concern over its thermal stability or heat of vapourization, fluids with relatively high viscosity – up to 40 cP, in some cases – can be jetted (Zhouping et al. 2010). Nozzles are usually arranged in a line in numbers ranging from less than ten up to several hundred (Le 1998, Clymer & Asaba 2004), and can be individually driven by different waveforms (Fujifilm-Dimatix 2006). The flexibility and almost universal applicability of the piezoelectric inkjet printer to functional materials made it ideal for a study of ink development for electronics deposition. As will be discussed in Section 2.3, however, the ink itself must meet certain criteria before the unique advantages offered by piezoelectric DOD printing can be realized.
2.1.3.1 Fujifilm-Dimatix DMP2831 Dimatix Materials Printer
An example of a testbed piezoelectric printer, produced by Fujifilm-Dimatix (formerly Dimatix), is the DMP2831, shown in Figure 2.3 (images courtesy of Fujifilm-Dimatix). This bench-scale printer offers piezoelectric inkjet printing capability for a limitless number of inks and substrates. The 1.5 mL cartridges of this printer may be filled and refilled with any type of ink, based on any solvent system; more rugged cartridges exist for aggressive solvents like toluene and mineral acids. The printheads’ sixteen 21.5-µm-diameter, 254-µm-spaced nozzles deposit droplets of either 1 pL or 10 pL in volume, depending on type, allowing for the printing of relatively high-
18
resolution patterns. The droplets’ size, shape, and speed can all be controlled by adjusting the voltage waveform applied to the nozzles. Spacing of droplets can be controlled by angling the line of nozzles along the printhead. The printer also provides control over suction (meniscus) to prevent ink from dripping, high-frequency “tickling” to keep the nozzles wet, temperature of the nozzle plate (from ambient to 80°C), temperature of the substrate (from ambient to 60°C), and the type and frequency of cleaning cycles. A mounted fiducial camera allows for observation of the printed surface, landmarking for alignment, and adjustment of the print origin and angle. A second camera, the drop watcher, provides real-time imaging of drop ejection from the nozzles; voltage waveforms and other cartridge settings can be actively adjusted while monitoring drop formation in order to find optimal values. Finally, the substrate support platen can be angled and tilted as desired to print more complicated structures. The DMP2831 therefore contained all of the features required for ink development and application in this study and was an ideal tool. Indeed, the considerable capability of the DMP2831 has been widely and successfully used in several other studies on printed electronics. Some of these include the deposition of PEDOT:PSS by Mire et al. (2011) and Lopez et al. (2008), PEDOT:PSS/SWCNTs by Mustonen et al. (2007), silver by van Osch et al. (2008), quantum dots by Haverinen et al. (2009) and Small et al. (2010), and a polymer-fullerene blend by Hoth et al. (2007).
Figure 2.3. Fujifilm-Dimatix DMP2831 Dimatix Materials Printer; magnification shows a disassembled cartridge and printhead assembly.
19
2.2 Piezoelectric inkjet printing: fluid dynamics
The fluid dynamics occurring during printing involve the formation of pressure (acoustic) waves due to the movement of the piezo element, which force droplets out of the nozzles, where they fall to the surface. Each of these steps – drop formation, detachment, and impact – involves unique fluid mechanics.
2.2.1 Ink ejection
The effect of the movement of fluid inside the ink chamber to produce jetting was described by Shield et al (1986). A schematic of their model is presented in Figure 2.4. The motion of the ink begins with the application of potential to the piezo element. Generally, potential is applied such that the ink chamber enlarges, creating a negative pressure inside it, and generating two negative pressure waves inside it, moving in opposite directions from the piezo element. These waves are reflected from the ink reservoir and the nozzle opening, the former changing its sign during reflection to a positive acoustic wave and the latter reflecting back as a negative acoustic wave. As the second part of the voltage waveform (the voltage drop) occurs, the piezo element expands, producing two positive pressure waves in opposite directions. One of these annihilates the reflected negative pressure wave, and the other doubles the amplitude of the reflected positive pressure wave. This propagating pressure wave forces out a droplet of ink when it is sufficiently strong to overcome the viscous drag and surface tension acting on the ink at the nozzle.
The entire process – and therefore, the length of a typical jetting waveform – is about 20 µs (Wijshoff 2008). As the length of the channel and the speed of the pressure wave (i.e. the applied voltage) will have some bearing on how long this process takes, the waveform length will vary with applied voltage and the exact printer being used. In some cases, an “echo” waveform is used to dampen the pressure waves in the ink chamber after the drop is ejected, allowing the chamber to refill for the next drop; an unfilled chamber will fail to jet properly (Molesa 2010, Wijshoff 2008). An example of an “echo” waveform is shown in Figure 2.2.
20
1) First voltage rise, +ve potential 2) Dwell time piezo element
reservoir nozzle
+ve potential: contraction, -ve pressure wave reflection from reservoir, tip
2) 3) Voltage fall, -ve potential 1)
V 3)
t amplification of +ve pressure wave, jetting
Figure 2.4. Schematic of acoustic wave propagation in a piezoelectric nozzle.
2.2.2 Droplet formation
As ink is ejected from the nozzle, it forms elongated droplets which fall to the surface as spherical drops, ideally (Lee 2003, Meixner et al. 2008, Wijshoff 2008), shown in Figure 2.5. The surface tension of the ink determines how easily it can form a drop: too low, and it will simply spread onto the nozzle plate and cover the nozzle with a skin of liquid, also known as overfill (Shin & Smith 2008); too high, and the pressure wave in the nozzle will be insufficient to overcome the surface tension and eject a droplet. Surface tension will be elaborated upon in Section 2.3.
Similarly, other fluid properties like viscosity determine how well droplets form. High-viscosity inks form droplets with long, pronounced tails – tail length scales with viscosity, which is why inks containing large proportions of polymers tend to display the extreme of this effect – the so- called “bead on a string” (Mauthner et al. 2008). As long as the tail merges with the body of the droplet during flight, or at least detaches and descends directly behind the main droplet, the
21
droplet should strike the substrate where intended. However, excessively long tails may break into several satellite droplets, which then follow different trajectories to the surface, completely compromising print resolution (Dong et al. 2006, Meixner et al. 2008, Lee 2003). Viscosity alone is not solely responsible for satellite formation; excessively high driving voltage or unsuitable voltage waveforms may also cause splattering of satellite droplets by ejecting the main droplet at too high of a speed (Haverinen 2010).
t1 t2 t3 t4 t5 t6 t7 t8
0 µs 5 µs 10 µs 15 µs 20 µs 25 µs 30 µs 35 µs
100 µm
Figure 2.5. Drop formation from a piezoelectric inkjet nozzle. Top: schematic of drop detachment and formation into spherical drop. Bottom: BaTiO3/PMMA ink drop forming.
2.2.3 Droplet impact with substrate
When the drops strike the substrate, they will wet it according to Young’s Equation, which describes the relationship between the surface energy of the substrate and that of the fluid:
훾푠표푙𝑖푑 −푙𝑖푞푢𝑖푑 + 훾푙𝑖푞푢𝑖푑 −푣푎푝표푢푟 휃푐 = 훾푠표푙𝑖푑 −푣푎푝표푢푟
c represents the contact angle, shown in Figure 2.6, between the fluid and the substrate, and the surface energy. As drops impact the surface, they will deform (because of their speed), and
22
eventually, begin to evaporate. So, again, the fluid properties and voltage waveform come into play, with regards to jetting speed, wetting ability, and finally, boiling point and volatility.
Solvent evaporation
photoresist Capillary flow or “Coffee-ring” formation momentum deflection
c
2) Drying 3) Dried droplet 1) Impact with substrate
Figure 2.6. Jetted drop behaviour on a substrate, showing the “coffee-ring” effect.
Upon impact, the droplet will spread to a diameter several times its original diameter (Tekin et al. 2008) and deform into a shape not unlike that shown in Image (3), in Figure 2.6, with thick edges and a thin centre, due to outward deflection of momentum (Soltmann & Subramanian 2008). If the ink dries rapidly after impact, this morphology will remain. So, an ink with high volatility and low boiling point – such as one containing a volatile organic solvent as its carrier medium – will likely suffer from this morphological problem. This is, of course, if the ink is able to jet, as a highly volatile solvent will also evaporate in the nozzles, clogging them with the solids it leaves behind.
This is one manifestation of the “coffee-ring effect” (Fukai 1993, 1995). Another mechanism by which it might occur is more rapid solvent evaporation at the edge of the droplet, and capillary flow of solvent to the edge to replace it. The thinner layer of liquid at the edge of the droplet compared to the centre of the droplet causes the rate of evaporation there to increase, and solvent to migrate by capillary flow to the that region (Craster et al. 2009). Subsequent Maragoni flow of solute to the region results in a ring with a raised edge of concentrated solute, and almost no solute at the centre of the drop. In both cases, the mixing in of co-solvent(s) with different boiling points than the primary carrier medium may assist in reducing this effect, by evening evaporation rates throughout the entire droplet and reducing evaporation rates at the edge of the droplet.
23
In any case, the control of ink rheology and surface tension determines the morphology of resulting printed features. A higher surface tension means less spreading, resulting in thickner films with higher resolution, but poorer adhesion; a higher viscosity means less-pronounced edge thicknening, but a generally thicker film; and so forth. Depending on the application, inks must be carefully formulated to achieve a desired outcome on a particular substrate.
It is also worthwhile to mention that porous substrates, with paper being a prime example, have entirely different properties when it comes to drop impingement. These structures will absorb the ink into their bulk almost immediately, as soon as inertial forces in the drop dissipate (Croucher & Hair 1989). After this point, the ink will sink into the surface, at a rate determined by the pore size, pore structure, and the surface energy of the pore walls themselves (Gane 2004, Holman 2002). These substrates present difficulties for printed electronics, as the functional material is no longer in a cohesive film, but distributed into the substrate itself.
2.3 Inks
As has already been discussed in the two previous sections, inkjet printers require particular fluid properties to jet reliably and consistently. Although these may vary somewhat between printers, viscosity, surface tension, specific gravity, and particle size are always controlled (Magdassi 2010). Stable inkjet inks may also require control over dispersion (zeta-potential), foaming, microbial growth, and pH (Karsa 2003). An ink formulation often must juggle successful jetting and drop formation, wetting and drying behaviour on the substrate, and functional performance. However, there are certain guidelines to ink formulation that can prove useful in designing a suitable inkjet ink for a given application.
2.3.1 Typical composition
Although inks may be formulated in many different ways, the composition of the ink is generally similar to that given in Table 2.1. The primary components of the ink are the “pigment” – which, when applied to electronically functional inks, implies the functional material – and the carrier medium or solvent. It is unlikely that a functional material will be readily available and dispersible in a solvent with appropriate viscosity and surface tension for printing, so usually, the
24
ink must contain at least one other component to make it jettable. In some cases, this can be as simple as a second solvent – for example, mixtures of water and common alcohols like EtOH, MeOH, or IPA can produce solutions with low surface tension, one high-boiling solvent (water) and one low-boiling solvent (alcohol) (Vazquez et al. 1995). However, this example of a mixture does not address the issue of viscosity, which would be too low. Therefore, another co- solvent is often added, with a relatively high viscosity. Such a material may also function as what is called a humectant, which prevents rapid drying of the ink in the printer nozzles (Croucher & Hair 1989). A surfactant may also be required to adjust surface tension to a suitable level. So, as the ink components become more and more sensitive to the presence of additives, or more specialized materials are used, the ink formulation may become increasingly complex. In certain cases, such as the printing of biomolecules (Di Risio & Yan 2007), formulation and ink functionality become very closely entwined, and an iterative approach to achieving the proper viscosity and surface tension is necessary.
Table 2.1. Typical inkjet ink composition, as described by Tekin et al. (2004, 2008); Le (1998), Croucher & Hair (1989), Zhouping et al. (2010), Magdassi (2010), and Calvert (2001).
Component Function Loading (w/w%)
Dye, pigment, or Key component 0.1 – 10 functional material
Solvent Dispersion/dissolution medium 50 – 90
Controls drying (“coffee-ring”) Co-solvent(s) Viscosity modification 0 – 50 Surface tension modification Modifies surface tension Surfactant 0 – 5 Improves wetting
Viscosity modifier Generally, increases viscosity < 1 (dissolved)
Humectant Low volatility, prevents ink drying in nozzles 0 – 20
pH buffer; biocide; fungicide; dispersant; defoamer; binder Other < 1 (polymer)
25
There are some pitfalls that commonly arise during the selection of ink components. Firstly, the use of dissolved polymers can be problematic, as they tend to modify the rheology of the ink at high concentrations and molecular weights, causing poor jetting, spreading, and film formation (Tekin et al. 2004, 2008; de Gans et al. 2004, Mauthner et al. 2008). Dissolved polymers also tend to increase viscoelasticity, producing long filaments of ink during jetting which subsequently may break up into small droplets and “splatter” the substrate. Secondly, inks that require the use of aggressive solvents, including toluene/tetrahydrofuran (THF) mixtures, or concentrated acid, can damage the print apparatus, and are to be avoided. Finally, and most importantly, any suspended particles in the ink must be below a critical size to jet uniformly and avoid clogging of printer nozzles. The packing of particles into a nozzle, even particles that are smaller than the nozzle diameter by many times, may still result in an irreversible blockage of that nozzle if they are above a certain size (Valero et al. 2007). As a large amount of particles is passing through the nozzle at a given time, the maximum allowable particle size is generally significantly lower than the actual nozzle diameter. Therefore, not only particle size, but dispersion of the particles becomes an issue in pigmented inks, which are generally thermodynamically unstable and only held in suspension by chemical or electrostatic treatments (Wang 2002). Added chemical dispersants keep particles separated by steric or electrostatic hindrance, overcoming the attractive London or van der Waals forces. Dispersion may therefore be compromised by chemical or energetic stressors (Spasic & Hsu 2006), so once again, ink formulation becomes a balance of fluid properties and ink stability. Unstable inks are not suitable for jetting, especially when using expensive and non-recoverable piezoelectric printheads as the deposition mechanism.
2.3.2 Fluid properties
As has been discussed in Section 2.2, viscosity and surface tension have a major impact on drop size, drop velocity, satellite formation, and droplet morphology upon impact. Through repeated attempts at ink formulation, certain restraints have been determined for fluid properties. Ranges of values for these key properties are shown in Table 2.2; the effects of deviating from these values are also listed in this table, Figure 2.7, and are further discussed below.
26
Table 2.2. Restrictions on fluid properties in jettable inks based on the work of Fujifilm-Dimatix 2006, Croucher & Hair 1989, Magdassi 2010, Calvert 2001, Meixner et al. 2008, Zhouping et al. 2010, Wang 2002.
Property Value Problems Controlled by
Minimum Maximum <2 cP: Solvent Droplet breakup Co-solvent Excessive spreading Viscosity modifier Viscosity (µ) 2 cP 20 cP >20 cP: Droplet tails Cannot exit nozzle <30 mN/m: Solvent Nozzle plate wetted (cannot Co-solvent jet through film) Surfactant Surface tension () 30 mN/m 40 mN/m >40 mN/m: Poor drop formation Unable to wet most surfaces
1% of nozzle >1% nozzle diameter: Dispersant Particle size (d ) 0 nm p diameter Rapid clogging Surfactant Solvent <1: Co-solvent Ink backflows under Humectant Specific gravity meniscus pressure ~1 1.5 (SG) >1.5: Ink drips out (insufficient meniscus pressure)
Zeta-potential (), As high as <40 mV: Dispersant 40 mV absolute value possible Precipitation of solids Surfactant
>1: Drop velocity Energy number (En) 0 1 Ink splatters on impact with Other fluid properties substrate
<1: Same issues as µ > 20 cP Inverse Ohnesorge Nozzle size -1 1 10 >10: number (Z ) Other fluid properties Satellite formation Poor drop formation
Many inkjet inks are water-based (Le 1998), and so achieving the relatively high viscosities ideal for piezoelectric inkjet printing is problematic. Even organic-solvent-borne inks face the challenge of increasing viscosity, as many common solvents such as toluene have viscosities of <1 cP as well. Increasing viscosity can best be accomplished by introducing a co-solvent of higher viscosity (Calvert 2001) or some sort of polymeric additive, like a cellulose compound (Di Risio & Yan 2007). However, in electronic devices, the presence of a dried polymer after ink curing may interfere with or completely compromise device function. Furthermore, high-
27
viscosity co-solvents are often high-boiling materials as well (e.g. glycerol, ethylene glycol), which may require high-temperature or low-pressure curing to completely remove from the printed layer. In certain special cases, this may be acceptable, such as with glycerol in printed PEDOT:PSS films, where it actually serves to improve electrical conductivity (Lia et al. 2003).
unstable bead-on- no drops a-string droplets form
flooded nozzles
µ (cP) (mN/m)
ideal immediate stable jetting clogging dispersion
precipitation
dp (nm) (mV)
Figure 2.7. Fluid properties, and effects of deviation on jetting.
In general, however, foreign compounds are detrimental to electronic performance. This implies that viscosity modifiers must be cleverly chosen so as to be readily removed, or to remain as integral parts of the device. For example, a conjugated polymer such as poly(n-vinylcarbazole) might be dissolved in an organic solvent-based ink as a host material for an emissive molecule or nanoparticle, while also increasing viscosity. If high-temperature curing is not an issue for the substrate or the functional material, all of these concerns are nullified – any liquid viscosity modifier may be chosen that can be burned off by heat treatment after deposition.
Surface tension reduction presents a unique challenge for aqueous inks, as was mentioned in the previous section. Most other solvents have surface tension values somewhere in the jettable range specified in Table 2.2, although some have lower surface tension values than desired, such as acetone and certain alcohols (Colclough 1968). Therefore, water-based inks must contain
28
either a large proportion of some co-solvent, or a surfactant. The ionic charge present on many surfactants must be suited to the dispersed species in the ink – if the species is dispersed by electrostactic repulsion, as is often the case, a surfactant with the opposite charge will interact strongly with it, sometimes even causing flocculation (Karsa 2003). A surfactant with the same charge as the dispersant material will compete for adsorption sites on the dispersed particle, which may be beneficial or detrimental to dispersion. In many cases, the most reliable means of reducing surface tension without compromising dispersion is the addition of non-ionic surfactants, which only weakly interact with charged particles or their adsorbed dispersant layers (Karsa 2003). As the non-ionic surfactants remain dissolved in solution without much interaction with any dispersed materials, they are generally loaded in very small amounts, well- below their critical micelle concentrations (Tadros 1987, Capek 2006). Another advantage offered by non-ionic surfactants is that they cause considerably less foaming of the ink than their ionic counterparts, especially anionic surfactants (Davison & Lane 2003). This can help to prevent refill issues in ink nozzles that become flooded with air and cannot jet (Haverinen 2010), and avoid the addition of generally immiscible silicon-based defoamers (Karsa 2003), although even non-ionic surfactants can cause some degree of foaming. Foaming can be readily observed by a simple shake test, followed by measuring the height of foam formed and its lifetime before settling. If a large volume of foam is formed which does not dissipate rapidly, either a different surfactant or a defoamer must be added to the ink (Tracton 2005, Karsa 2003).
Dynamic surface tension, the surface tension of a surfactant-laden liquid at different times after the formation of a new interface, is often characterized in inks as well. Surface tension, as described in Table 2.2, is equilibrium or static surface tension – i.e., the surface tension value of an ink as a long-term asymptote of its dynamic surface tension values. However, dynamic surface tension also plays a role in droplet formation in inks containing surfactants. As the droplets are forming a new liquid-vapour interface (with the surrounding air) when jetting occurs, and a certain amount of time is required for surfactant to diffuse to the interface, the surface tension upon leaving the nozzle and that upon impact with the substrate will differ (Eastoe & Dalton 2000). The higher value of dynamic surface tension (at time 0, or immediately upon drop formation) compared to equilibrium surface tension may result in poor drop formation
29
or decreased substrate wetting. Furthermore, as the drops impact, the velocity of the spreading fluid along the plane of the substrate may be faster than the diffusion velocity of the surfactant molecules to the new interface formed with the solid (Aytouna et al. 2010). Therefore, to ensure the best droplet formation and surface wetting in inks containing surfactants, it is ideal to utilize surfactants which have dynamic surface tension values close to their equilibrium values at a given concentration (Titcomb 1981). This often entails a careful consideration of the chemistry of the surfactants in question, as the head group, molecular weight, and charge of a surfactant may affect its rate of diffusion through the solvent to the interface (Manglik et al. 2001).
Perhaps the most critical fluid property is particle size. For piezoelectric printers – and definitely for the DMP2831 – a good rule of thumb for maximum particle size is 1% of the nozzle diameter (Fujifilm-Dimatix 2006a, 2006b). With particles larger than this size, clogging of the nozzles will inevitably occur, requiring replacement of the printhead and process downtime. It is theoretically possible to print larger particles with larger-diameter nozzles, but they will produce larger droplets, which may be undesirable for high-resolution printing (Lee 2003). In general, the smallest particles possible are ideal, to avoid any issues with clogging of nozzles, which has spurred the development of inks containing such miniscule particles as quantum dots (QDs) and metal nanoparticles (Tekin et al. 2004, 2008; de Gans et al. 2004).
Specific gravity (SG) of an ink has less bearing on jetting performance and more on the maintenance of good printer function. Most printers contain a pump or similar means by which the meniscus of the ink while in the nozzle can be controlled (Molesa 2006). Meniscus refers to the curvature of the ink layer in the nozzle before jetting – a high meniscus pressure provides a convex bulge of ink out of the nozzle, and a low meniscus pressure draws the ink further back into the reservoir. Most printheads are designed with an SG of ~1 in mind, as most common solvents – and particularly water – have SG values of approximately 1. So, for the meniscus pump to function properly on an inkjet printer, preventing both the backflow of ink into the reservoir and the outflow of ink onto the nozzle plate, SG should be ~1, or slightly greater (Fujifilm-Dimatix 2006a, 2006b; Molesa 2006).
30
Zeta-potential () is an expression of the magnitude of repulsive or attractive forces acting between individual particles, and generally increases in absolute value as the thickness of the charged layer around the particles increases (Davison & Lane 2003). A larger absolute value (negative values applying to negatively charged systems) correlates with better dispersion stability. At || of around 12-14 mV, immediate flocculation no longer occurs and a dispersion begins to become stable. However, a dispersion is not considered indefinitely stable until || > 40 mV (Spasic & Hsu 2006). These values are generally only applied to suspensions of roughly spherical particles; particles with different shapes and hydrodynamic radii will be stably dispersed at different values of . A detailed discussion of the theory behind is not necessary for this study; let it suffice to say that provides a good indication of the stability of an inkjet ink containing dispersed particles. Flocculation of dispersed particles may lead to irreversible damage to printheads through clogging. It may also reduce the amount of functional material being delivered to the substrate, as some of it is settled out of the ink before jetting, and removed during pre-printing filtration stages. In some cases, it may even be a result of a chemical reaction in which the functional material agglomerates or grain size grows. An example of this was observed by Small et al. (2010) when attempting to print ZnS:Cu dispersed in water with an organic acid containing a sulfide group, resulting in the formation of CuS, a water-insoluble compound which rapidly precipitated out of solution. In any of these cases, jetting performance is compromised, and functionality after deposition may also be compromised. Therefore, measurement of and maintenance of jetting conditions (temperature, pH) at a value that keep in the stable range is important for replicable, high-quality inkjet deposition.
It is important to note that is generally a term applied to aqueous (or, more generally, polar) suspensions, which comprise the vast majority of inks. The measurement and definition of in hydrocarbons is less well-defined. The dielectric constants of hydrocarbons are so low as to prevent any ionization of solvated species, which in turn prevents the development of electrostatically charged layers around dispersed particles (McGown & Parfitt 1967). In the absence of charged layers, adsorbed molecules with hydrocarbon-soluble tails provide dispersion and sterically prevent agglomeration (van der Waarden 1950). Weak van der Waals/London forces also play a role in maintaining dispersion stability (Hamaker 1937).
31
There are three more ink properties which are often calculated to determine whether or not an ink is suitable for jetting. These have been defined as the dimensionless quantities We, En, and Oh, which refer to the Weber number, Energy number (Meixner et al. 2008) and the Ohnesorge number (Ohnesorge 1936, Derby & Reis 2003, Jang et al. 2009), respectively. The Weber number establishes the likelihood of liquid splashing on the substrate; We expresses the ratio of inertial to surface tension forces (Bergeron et al. 2000). This is very similar to the Energy number, which has been specifically derived for inkjet inks. En is defined as the ratio of the kinetic energy of a falling ink droplet to its surface energy, i.e.
1 2 2 3 2 푚푣 휌휋푟 푣 휌푟푣2 퐸푛 = 2 = 3 = 4휎휋푟2 4휎휋푟2 6휎
where is the ink density, is the ink’s surface tension, and r and v are the droplet’s radius and downwards velocity, respectively. En also determines the likelihood of splashing of a droplet on the substrate, which may result in reduced printed resolution. A drop may split into two (or more) droplets upon impact if the gain in surface energy is lower than the kinetic energy of the flying drop, so En should be <1 to prevent this from happening – although droplet splitting may not always occur, if a sufficient amount of energy is dissipated thermally (Cibis & Krueger 2005).
The Ohnesorge number is in turn defined as a relationship between capillary and viscous forces acting on a droplet (Ohnesorge 1936). It can be expressed as the ratio of the Reynold’s number, which expresses the ratio of inertial to viscous forces, and the root of the Weber number. Oh – or, more commonly, the inverse of Oh, often denoted by Z or Z-1 – is a means of expressing droplet stability for a fluid of given properties being ejected from a nozzle of a certain size by the following expression:
32
휌푣푑
−1 푅푒 휇 휌휎푑 푍 = 1 = 1 = 푊푒 2 휌푣2푑 2 휇 휎
where µ is the ink viscosity, and d is the diameter of the nozzle. This dependence on nozzle size implies that the same ink may not jet from every different printer or printhead type, and ink reformulation may be necessary for transferring inks between inkjet platforms. What is meant by droplet stability has been previously outlined in Section 2.2.2 – the establishment of a spherical droplet with no pronounced satellite droplets or a very long tail. Meixner et al. (2008), building on the work of Derby & Reis (2003), suggested that an ink/nozzle combination with an inverse Ohnesorge number with a value between 1 and 10 would be suitable for the formation of stable droplets. Jang et al. (2004), as later outlined by Zhouping et al. (2010), more rigorously specified that values of Z-1 in the range of 2 to 4 resulted in droplets where the tail had sufficiently high velocity to recombine with the droplet quickly, whereas when 6 < Z-1 < 13, the tail detached from the droplet, forming a secondary drop. However, in this latter case, the tail and its resultant droplet still retained high enough velocity to recombine with the larger droplet before impact. Values of Z-1 > 14 represented printing systems where ink droplets fell at a high velocity, leaving behind tails which dispersed into several satellite droplets and splattered irregularly on the substrate. The greater degree of viscous dissipation of energy (and thereby, velocity) in inks with lower values of Z-1 allows the droplet tails to recombine with droplets, preserving print quality. Therefore, a lower value of Z-1 – which often implies a higher viscosity – is ideal for printing.
2.3.3 Orthogonal solvent systems
A final consideration to make during ink formulation is of the ink’s compatibility with adjacent layers, if a multilayer structure is to be deposited. Each solvent must preferably undergo minimal interaction with the layer below it during jetting. What this usually implies is that each layer must alternate its solubility; a water-soluble material, for example, cannot be overprinted
33
with a water-based ink without some damage to the underlying layer and resulting compromise of layer integrity. This often requires selection of dispersants for functional materials which can solubilize the materials in different solvents, and may even preclude the use of certain materials which cannot be made suitably soluble. Where alternating aqueous/organic layers are not feasible, the issue of interlayer dissolution may be also be alleviated by selecting materials which are minimally soluble (e.g. a hexane-borne layer overprinted with an alcohol-borne layer), or by post-treatment of deposited layers, such as sintering or cross-linking, to reduce solubility.
2.4 Print quality
The concept of print quality encompasses many different characterizations of printed media, such as sharpness, raggedness, optical density, print mottle, and so forth, as are described by Oitennen & Saarelma (1998). These characterizations are applied to conventional printed media, typically on paper surfaces, where they can be summarized into two basic descriptors of a quality print – resolution and colour intensity/trueness (Oitennen & Hannu 1998). With printed electronics, the latter descriptor is not particularly relevant – printed layer functionality is a more valuable measure of print quality. However, the former issue, of resolution, is of major concern when manufacturing printed electronics, as the feature size will be limited to the maximum resolution deliverable by the printer. An issue that arises with printed electronics, which is of no concern in conventional printed matter, is the alignment of individual layers when overprinting a multilayer device. Finally, although the roughness of printed media can have a bearing on their optical properties (Daniel & Berg 2006), roughness is a matter of vital concern in printed electronics, where functional layers are often only a handful of nanometres thick, or have a tendency to catastrophically fail at localized thin or thick points.
2.4.1 Resolution
Resolution, in its most rigorous definition as applied to printing, refers to the number of distinct “dots” or drops printed along a fixed length, such as dots per inch (DPI) or pixels per inch (PPI) (Lee et al. 1981). Resolution may be observed by image analysis of printed films – optical microscopy usually being sufficient, as drop sizes are generally in the micrometre range. For a
34
printed display, pixel count may be an important measure of quality, but for other printed electronics, the only meaning of resolution that is significant is that of feature size. Many conventionally-processed electronic devices have feature sizes of 50 nm or smaller (Mahalik 2006), allowing the packing of many transistors, for example, onto a single silicon wafer.
It is unlikely that a DOD printing process will be able to match this miniscule feature size in the foreseeable future. The main reason behind this is that the minimum feature size producible by an inkjet printer is limited to the size of the impacted droplet that it can deposit. Moreover, unless the desired feature is circular, the minimum printable feature size must be larger than the droplet size, utilizing multiple droplets to provide lines, corners, and so forth. The impacted droplet size will be larger than (but related to) the nozzle diameter (Tekin et al. 2008), so narrower nozzles can deliver smaller droplets and therefore smaller feature sizes. The narrowing of nozzles causes changes in jetting behaviour, and fluid properties, including particle size, must be accordingly adjusted to produce stable drops and avoid clogging. Realistically, for printing suspended nanoparticles such as QDs, which have minimum particle sizes around 1 nm (Haider et al. 2009, Borovitskaya & Shur 2002), the minimum nozzle diameter would be 100 nm (Fujifilm-Dimatix 2006a, 2006b). If a droplet formed with diameter 100 nm, it would produce a circular film of ~200 nm (Tekin et al. 2008), still significantly larger than photolithography- produced features – not to mention any issues arising with jetting, as indicated by the very small value of Z-1 associated with such a small orifice diameter. Factors like surface energy of the substrate would also have an effect on spreading, and the degree of spreading would have a bearing on resolution in the vertical dimension as well as in the horizontal plane.
Therefore, in its current incarnation, inkjet printing is not particularly well-suited to the production of extremely high-resolution electronics on the order of those currently produced by conventional means. However, by delivering droplets that are a handful of micrometres in diameter, reasonably well-resolved rudimentary electronics may be produced. Furthermore, the high throughput and limitless substrate size associated with R2R manufacturing somewhat alleviate the need for extremely small, tightly-packed arrays of pixels, transistors, and so forth.
35
2.4.2 Roughness & topography
Conventional processing methods for electronics can produce extremely smooth films – almost atomically smooth in some cases, such as with atomic layer deposition, or ALD (Jaeger 2002). There is good reason for this: electrical properties can be affected by local variations in thickness, when films are used in electronic devices (Campbell 2001). Key properties such as resistance, mobility, and dielectric breakdown strength can all be changed by changing the geometry of a layer. Moreover, localized regions of lower thickness may lead to electrical arcing or shorting in devices driven by an elevated voltage (Ono 1995). Devices containing films of rough or irregular topography may therefore either fail immediately, or not function at all. Films with localized fluctuations in thickness may allow current to channel through the thin regions, or direct current flow to thicker regions, tunnelling through adjacent films (Wood 2000). As a rule of thumb, the films should be as smooth as possible to prevent such issues from causing a device to lose functionality (Haverinen et al. 2009, 2010). As significant a problem as surface roughness is the presence of pinholes. Any holes in a printed layer present a localized channel across which the driving current may flow without actually driving the device, or where an arc or short may cause catastrophic failure (Ono 1995). Therefore, there must be some degree of overlap between printed drops to prevent the presence of holes. Too much overlap, however, and the thickness of the overlapping regions will increase dramatically (Haverinen 2010).
Producing smooth, uniform films is relatively simple when using methods like ALD, CVD, PVD, epitaxy/MBE, and so forth, because the target substrate is exposed to a beam or cloud of molecules or atoms that deposit all over the surface at the same rate, ideally (Campbell 2001, Seshan 2002). Film quality and reproducibility are best when using slow (Å/s) deposition rates with high-purity materials, rather than introducing relatively thick films of materials containing solvents and surfactants. However, the inkjet printer, by definition, is a solution-processing tool. With an inkjet printer, the film is formed of individual droplets deposited one at a time from each nozzle, producing overlapping discs on the substrate (Figure 2.9). In an ideal situation, each disc merges with its neighbours, forming continuous lines, which in turn merge with adjacent lines. However, depending on fluid properties, substrate surface properties, drying rates, ink
36
formulation, and so forth, the same issues with capillary/Marangoni flow and “peak and valley” formation may manifest themselves in the printed films (Figure 2.9).
Different line morphologies, including “stacked coins”, scalloped, and “bulging”, may appear (Soltmann & Subramanian 2008), as well as the ridged structure caused by coffee-ring formation. For a given substrate, the different morphologies are dependent primarily upon drop spacing and drying rate. Some of these morphologies are schematically shown in Figure 2.8.
Solvent flow Peak and valley formation during drying Solvent flow
Spreading
Drop spacing Printed line droplet
1) Printed lines, birds-eye 2) Wet printed ink, x-section 3) Dry printed ink, x-section
Figure 2.8. Schematic representation of “peak and valley” topography formed during printing.
When qualifying an ink’s performance, printed film morphology may be the most important determinant in whether or not it can be used for a functional layer. Therefore, establishing proper drop spacing and drying conditions is very important in ensuring layer (and device) function. If smooth layers are not formed under any variations of these conditions, then the ink may have to be completely reformulated. An example of drop spacing testing is shown in Figure 2.10, where an aqueous ink containing QDs and a polymer was printed onto glass and observed with an optical microsope. Drops did not begin to coalesce into continuous lines until the drop spacing was < 75 µm, and the adjacent lines did not coalesce until the drop spacing was < 65 µm. Below this spacing, the lines overlapped excessively, leading to the “peaks and valleys” topography shown in Figure 2.8. There is a narrow range of drop spacing values which will allow smooth film formation, and these will be different for every ink and on every substrate or underlying layer.
37
Proper drop Drops too Drops too far Fluid proper- Drops too spacing, T close, or T apart, or T too ties/T causing far apart too high low coffee-ring
Normal Stacked coins Bulging Ridged Scalloped
Figure 2.9. Schematic representations of common line morphologies (Soltmann & Subramanian 2008). “T” refers to the substrate temperature, and thus the drying rate of the ink.
90 µm drop spacing 80 µm drop spacing 75 µm drop spacing 65 µm drop spacing
100 µm
Figure 2.10. Drop spacing of QD/polymer/water ink on slide glass.
Thickness is another issue to consider during inkjet deposition. Film thickness is often a determinant of exactly how a device functions (Dimitrijev 2005, Wood 2000). All of the layers of a thin-film light-emitting device, for example, are generally of controlled thickness to maximize performance while minimizing the amount of material deposited and the resistive losses across the device (Adachi et al. 2007, 2008; Schrage et al. 2010, Manzoor et al. 2003, Hieronymas 2002, Cho & Cha 2009). Increasing thickness may lead to poorer device performance and higher power draw (Schrage et al. 2010).
The inkjet printer is capable of controlling thickness via ink formulation (wetting and spreading of drops) and by depositing multiple layers, or “print passes”. Although the volume of ink deposited by the inkjet printer can be easily observed from the size of individual droplets, the actual film thickness after printing and drying is entirely dependent on the spreading of droplets
38
on the surface, the density of the dried film, the amount of solvent removed, and so on. The volume of individual drops is usually significant enough that film thickness is on the order of a few tens of nanometres, at least. Consider, for example, a 10 pL drop from a 25 µm nozzle, producing a 50 µm circle on the substrate. Assuming that 90% of the drop is solvent, 1 pL of solids remain after drying. This results in a thickness of ~500 nm. Functional layers in PVs, LEDs, TFTs, and so forth are often significantly thinner than this (Mahalik 2006, Chen et al. 2011). A droplet that spreads wider to reduce thickness also reduces resolution, and may encounter issues with coffee-ring formation. Coffee-ring or peak-and-valley formation presents difficulties not only in increasing the potential for device failure, but also in estimating film thickness, as it is not uniform. A droplet with reduced solids content will form a thinner film, but may not contain enough functional material to operate as desired. So, as was the case with balancing fluid properties, compromises must be made to produce high-quality films by inkjet.
39
Chapter 3 3 Materials for inkjet-printed electronics
Although theoretically any material that can be dispersed or dissolved into an ink can be used as a functional species in an inkjet ink, the constraints on fluid properties and film-forming behaviour limit materials selection. In particular, the necessarily small dispersed particle size restricts available materials to colloids (Singh et al. 2010, Tekin et al. 2008, Zhouping et al. 2010, Caglar 2006), polymers – either dissolved or micellar (Tekin et al. 2004, de Gans et al. 2004, Meixner et al. 2008), and other nanostructures such as nanotubes (Mustonen et al. 2007) or fullerenes (Hoth et al. 2007). There are, then, several materials which are particularly well- suited to inkjet printing of thin-film electronics which will be described in this section. These can be broadly divided into three classes – conductors, semiconductors, and insulators (Ohring 1991). There are other materials as well which serve auxiliary or supplementary purposes, such as sealing films or barrier layers. Finally, a very important material to be considered as part of a device is the substrate upon which it rests, which in the case of printed electronics, may be any conventional rigid substrate, or a R2R-processable flexible substrate.
Dissolution of inorganic dielectrics, semiconductors and metals, is not usually possible at atmospheric conditions with solvents suitable for printing. Conversion of a precursor to the desired material during/after jetting can be accomplished, as with AgNO3-based inks producing Ag films (Liu et al. 2005) or sol-gels for ceramics and oxides (Zhou et al. 2008, Lima et al. 2007, Sharma et al. 2000, Atkinson et al. 1997, Harizanov et al. 2004, Sharma & Sarma 1998, Sharma & Mansingh 1998), but the production of sintered or crystalline films requires heat or hydrothermal treatment (Zeng et al. 1999). Metallic colloids often require high temperatures after printing, for sintering, to demonstrate any electrical conductivity (Tekin et al. 2008). In the case of polymers, inkjet deposition of dissolved material is possible, although loading of an ink with more than a small amount of polymer can cause undesirable effects in drop formation, especially the bead-on-a-string effect mentioned in the previous chapter (Mauthner 2008, Magdassi 2010, Fujifilm-Dimatix 2006a). So, dispersed particles of either inorganics or polymers which require no aggressive post-treatments offer the simplest deposition route.
40
In this work, materials which were functional upon deposition were used. What this means is that the materials were dispersed in the ink with their desired properties already established – i.e. no chemical, thermal, hydrothermal, or other post-treatments were to be used to induce functionality in the printed films. Drying of the inks was of course necessary, but high- temperature annealing or sintering were not used. The reasoning behind this decision was that the inks were meant to be usable on a variety of substrates, with minimal unit operations (ideally, just the inkjet printer) involved in their processing, under atmospheric conditions.
3.1 Conductors
Electronic devices, by definition, utilize the passage of electrons to provide a desired function. In order to convey electrons to and through a device, an electrode material which conducts them well, with low losses in voltage or current due to resistance, is required. In virtually all cases, a set of two electrodes or interconnects is used to provide a complete circuit for electrons flowing to and from a power source. These electrodes/interconnects may be composed of the same material, or, in some cases, two different materials, depending on the electrical demands of the device in question. There are many different electrically conductive materials in existence, but only some of them are readily available in a form suitable for inkjet printing. The three primary groups of these materials are metallic colloids, carbon nanostructures, and conjugated polymers. The former two groups are dispersible species, whereas conjugated polymers may be either dissolved or dispersed.
Metallic colloids have become increasingly common as inkjet-printable materials, and include such conventionally-used electrode materials as Ag (Dearden et al. 2005, Wu et al. 2007, van Osch et al. 2008, Perelaer et al. 2008, Nguyen et al. 2007, Jeong et al. 2010, Kim & Kim 2010, Jung et al. 2007, Ryu et al. 2005), Au (Jensen et al. 2011 Chow et al. 2009, Ko et al. 2007, Cui et al. 2010, Lee et al. 2008b), Cu (Li et al. 2009, Hong 2000, Jang et al. 2010, Lee et al. 2008a), Ni (Li et al. 2009), Mg (Aguey-Zinsou & Ares-Fernandez 2008), and Al (Meziani et al. 2009), as well as conductive oxides, such as indium tin oxide (ITO) (Yarema et al. 2012, Gan et al. 2006, Cho et al. 2006, 2009; Al-Dahoudi & Aegerter 2006) and antimony tin oxide (ATO) (Cho et al.
41
2009, Yang et al. 2007). In many of the works listed, conductivity comparable to bulk values was achieved with proper post-treatment of the colloids. Usually, this post-treatment entails a high-temperature curing phase to remove solvent and decompose the organic ligands on the nanoparticles, which were used to stabilize the colloidal suspensions (Caruso 2004). The solvent and encapsulating ligand or polymer (also referred to as the “cap” or “capping agent”) which stabilized the particles in solution prevent sintering from occurring until they are removed. The solvent should ideally evaporate or decompose at a relatively low temperature, as should the cap, leading to the use of such materials as nitrocellulose (Nguyen et al. 2007), which decomposes around 135°C (Selwitz 1988). After the removal of the cap, the particles are generally still non- conductive, being physically separate and therefore preventing effective electron transfer (Tekin et al. 2008, Mei et al. 2005, Zhouping et al. 2010). They are usually then sintered at high temperature to induce grain growth and particle merging. The high surface energy to volume ratio of these extremely small particles can cause a reduction in melting temperature upwards of 500°C (Huang et al. 2003, Hostetler et al. 1998) – however, this still implies relatively high sintering temperatures for most metals and even higher for oxides, well above 200°C, the usual cutoff point for flexible substrates. If temperature is not a concern – for example, on a glass substrate – these colloidal metals and metal oxides make excellent candidates for inkjet-printed conductive layers. Conductivity in these materials approaches bulk performance, and they are largely chemically inert, although atmospheric oxidation may be a problem over time for Al and Cu in particular. There are also several commercial inks available which contain Ag or Cu, and a few others containing ITO or similar transparent conductive oxides (TCOs); this completely eliminates the need for ink formulation and voltage waveform tailoring. However, these inks are expensive – sometimes upwards of $100 US/mL – and even in-house synthesized nanoparticles for custom-formulated inks require expensive chemical precursors containing ionized metals. So, colloidal metals and TCOs are acceptable for use in small amounts, as in interconnects, but usually only on glass or comparable substrates. However, recent attempts have been made to reduce the sintering temperatures of such materials, opening the door for their use on flexible substrates. Some of the techniques used include the preparation of a metallorganic Ag precursor rather than nanoparticles with a low reduction temperature of 125°C (Dearden et al. 2005); the use of more weakly adsorbent ligand species to cap Ag nanoparticles (Perelaer et al. 2008); and
42
avoiding sintering entirely by encouraging close packing of negatively charged PAA-capped Ag nanoparticles using cationic PDADMAC solution (Magdassi et al. 2010). Also, photonic or electrical methods such as infrared exposure (Tobjork et al. 2012, Denneulin et al. 2011) or high- voltage treatment (Allen et al. 2008) can also produce low-temperature-sintered films, suitable for use on more thermally fragile substrates, at the cost of introducing more complex processing to the deposition process.
Carbon nanostructures – including nanotubes, nanosheets (graphene) and nanoballs (fullerene) – are another major category of printable electrode materials. Amorphous carbon is intrinsically electrically conductive due to the delocalization of its valence electrons within covalent C-C - bonds (Cutnell & Johnson 2008). Different allotropes of carbon may lead to different values of conductivity: diamond, for example, having very low electrical conductivity, whereas graphitic structures have moderate conductivity. Conductivity in carbon is often significantly lower than in metals. Electrical conduction in carbon relies on electron transfer along the C-C bonds in a single plane, whereas in many metals electrons are delocalized in every direction/plane within a metallic crystal lattice, providing many adjacent conduction sites (Collins & Avouris 2000). Within the aforementioned groups of nanostructures there is also wide variation in conductivity, attributable to changes in the electronic wavevectors of the structure in question. If a gap exists between the occupied and unoccupied energy states of the material, which may occur with a certain size, bonding structures, and chirality, semiconducting behaviour will result; an infinitesimally small gap will result in metallic behaviour (Dresselhaus et al. 2001). Observation and control of the metallic-to-semiconducting ratio of carbon nanostructures has been reported by groups with an interest in producing either highly conductive (Wang et al. 2008b, Blackburn et al. 2008) or purely semiconducting nanomaterials (Naumov et al. 2009, Kanungo et al. 2009). Each of these types may be useful in different electronic devices – the former being ideal for electrode formation. The mechanical strength and flexibility of carbon nanostructures also makes them well-suited to flexible substrates, increasing their appeal over more brittle conductive species like ITO (Chen et al. 2002).
The most commonly inkjet-printed carbon species are carbon nanotubes (CNTs), both single- (SWCNTs) and multi-walled (MWCNTs), fullerene, and graphene. Graphene is a monolayer-
43
thick sheet of carbon atoms bonded hexagonally by hybridized sp2 bonds (Geim & Novoselov 2007). SWCNTs are formed by “rolling up” a graphene sheet into a hollow tube; MWCNTs are several concentric SWCNTs of increasing diameters (Dresselhaus et al. 2001).
Fullerenes are spherical carbon nanostructures of a “soccer-ball” shape, comprising both hexagons and pentagons of C atoms – based on the original buckminsterfullerene (C60) synthesized by Kroto et al. (1985). Schematics of these three structures are shown in Figure 3.1.
(a) (b)
(c) (d)
Figure 3.1. Carbon nanostructures: (a) SWCNT; (b) MWCNT; (c) graphene; (d) buckminsterfullerene.
Nanotubes and graphene are more widely used as printed electrode materials than fullerenes because they can layer into stacked sheets or networks, whereas fullerenes must be packed tightly (like metallic colloids) to allow electrons to hop between them. This behaviour was
44
observed in printed layers of MWCNTs (Fan et al. 2005) where overprinting of successive layers was sufficient to improve conductivity greatly, as more MWCNTs filled in the network of overlapping conductive paths. Several other groups also successfully printed MWCNTs (Wei et al. 2007, Sumerel et al. 2006), SWCNTs (Song et al. 2009, Nobusa et al. 2011, Gracia-Espino et al. 2010), and graphene (Te et al. 2011, Torrisi et al. 2012, Huang et al. 2011) as electrodes or interconnects in different electronic devices, including thin-film transistors (TFTs) and field- effect transistors (FETs). In most cases, the sheet resistance of CNT or graphene films was still several k/square, 2-3 orders of magnitude higher than that of TCOs like ITO and many orders higher than metals. However, carbon does present some unique advantages over these conventional materials. Firstly, the extremely strong nanostructures allow for a degree of flexibility in printed devices. Secondly, the processing of CNT or graphene dispersions does not require high-temperature sintering like TCO or metal nanoparticle suspensions – only the solvent has to be removed to allow the materials to conduct electricity, although the dispersant or surfactant may also need to be decomposed to improve conductivity. Indeed, carbon will not sinter under atmospheric conditions regardless of temperature (Saavatimskiy 2005). Surface functionalization of nanotubes or graphene sheets with organic functional groups may achieve dispersion in a variety of solvents without the need for surfactants, eliminating this problem. Thirdly, the small, controllable size of carbon nanomaterials means that they can be functional as conductors or semiconductors while retaining optical transparency, presenting a potential replacement for ITO or ATO. Finally, while carbon nanostructures are hardly inexpensive due to their relatively complex purification procedures (Dresselhaus et al. 2001), they are certainly less expensive than colloidal TCOs, and many times less expensive than colloidal metals. So, CNTs and graphene are ideal inkjet materials to produce conductive layers, although perhaps best- suited to the replacement of TCOs.
A point worth addressing about CNTs and graphene sheets is their aspect ratio. These materials are considered “nanomaterials” due to the fact that at least one of their dimensions is in the nm- range. However, they may be quite large in the other dimensions – nanotubes are often several µm long, as are graphene sheets, in both length and width. This does not present a problem for their functionality, but it is of concern for inkjet printing, where particle sizes are restricted.
45
Also, their stability in solution is often poor (Kim & Ma 2011). Both of these issues should be taken into consideration when including these materials in an inkjet ink.
The final significant class of jettable conductors is that of conductive polymers. Like metals, there are many different types of conductive polymers that are suitable for jetting. Unlike metals (and carbon), these are not always in the dispersed state, but may be dissolved into an ink. The basis of electrical conduction in these materials is the same as that in carbon nanostructures – delocalization of overlapping p-electrons along conjugated -C=C-C=C- (or similar) bonds within the polymer’s “backbone” (Skotheim & Reynolds 2006). The conjugated structure may also contain other atoms, such as N, O, or S, which also have p-orbitals. When oxidized or reduced by doping or electrochemical treatment, the mobility of the delocalized orbitals is greatly increased, resulting in their relocation to other energy states, partially emptying out an electronic band and allowing electrical conduction along the backbone (Skotheim 1997). The disordered nature of these polymers, when compared to more highly ordered inorganics, can lead to structural irregularities within the polymer’s conjugated chain, comparatively reducing mobility and conductivity (McGinness 1972). Therefore, these polymers are generally not nearly as conductive as metals, TCOs, or even carbon, as carbon structures are materially homogeneous, although improving synthesis and purification methods may help to close this gap.
Some of the best-known conductive polymers are polyaniline (PAni), polypyrrole (PPy), polyacetylene (PAc), poly(p-phenylene vinylene) (PPV), and polythiophene (PT). These polymers have been extensively modified, substituted, and combined to produce many different conductive species. The two types which have arisen as the most commonly used, due to their ease of processing and relatively high conductivities, are PAni and PT – and in particular, a variant of PT known as poly(3,4-ethylenedioxythiophene), or PEDOT (Molesa 2006). The structure of these polymers is shown in Figures 3.2 and 3.3. Processing of these types of polymers is greatly simplified by their ability to be synthesized and dispersed as a micellar suspension in a solvent by using a charged surfactant molecule, whereas other conductive polymers must be dissolved (Skotheim 1997). This limits the use of these other polymer types, since conductive polymers, and conjugated polymers in general, are difficult if not impossible to dissolve in most solvents (Nguyen et al. 2001). Furthermore, as was mentioned above, suitably
46
high conductivity is only achieved in conductive polymers by doping (usually oxidative, i.e. p- doping), and these doped polymers are generally nearly insoluble in most solvents (Nalwa 2000). In the case of doped PAni (shown in Figure 3.2) a variety of surfactants may be used to disperse nanoparticles prepared by emulsion polymerization methods (Eftekhari 2010), including an acid species which induces solubility in water (Lee et al. 2005b), although the dispersions often have short shelf lives (Li & Kaner 2006). PEDOT (shown in Figure 3.3) has been cleverly dispersed in an aqueous suspension using a material which also functions as a dopant – the anionic polymer poly(styrene sulfonate), or PSS- – a combination discovered at Bayer (Bayer AG, Leverkusen, Germany), producing gelled particles of PEDOT/PSS- with an average particle size ranging from 30 nm (Changneng et al. 2005) to an average of 100 nm (Lee et al. 2005a).
Figure 3.2. PAni structure, where y = 1-x. If y = 1, leucoemeraldine; y = 0.5, emeraldine; y = 0, pernigraniline.
Figure 3.3. PEDOT (left) and PSS- (right) structures.
47
Since the discovery of PEDOT:PSS, as it is now called, it has been extensively studied as an electronic material due primarily to its ease of processing by wet methods, including inkjet printing (Tekin et al. 2008, Yoshioka & Jabbour 2006, Ouyang et al. 2004). PEDOT:PSS being an aqueous dispersion, solvent removal and film curing is a relatively simple matter, and definitely feasible on flexible substrates. The conductivity of PEDOT:PSS is also relatively high for a polymer, and has been reported as more than 500 S/cm in a cured PEDOT:PSS film (Crispin et al. 2003, 2006), with Heraeus claiming conductivity up to 1000 S/cm in films of their CLEVIOSTM product, which contains PEDOT:PSS as the conductive species. Some other advantages offered by PEDOT:PSS are its transparency in the oxidized state, making it an appropriate material for devices requiring a transparent electrode (Ouyang et al. 2005), the stability of both the dispersion and films (Crispin et al. 2003), its mechanical flexibility (Polasik & Schmidt 2005), and its relatively low cost.
A final useful feature of PEDOT:PSS is that its conductivity can be improved by adding certain polar solvents to the dispersion before printing, or post-treating films with certain oxidizing species. Some of the solvents that have been found to improve conductivity are dimethyl sulfoxide (DMSO) (Fehse et al. 2007, Dobbelin et al. 2007, Kim et al. 2002, Xue & Su 2005), tetrahydrofuran (THF) (Kim et al. 2002), N-methyl pyrrolidone (NMP) (Jonsson et al. 2003), and glycerol (Lia et al. 2003). Anionic surfactants (Fan et al. 2008) and certain salt solutions (Xia & Ouyang 2009) have also been shown to improve conductivity. The mechanism of conductivity enhancement has been described in several ways, including conformational change of the PEDOT molecule (Ouyang et al. 2004, 2005) and washing away of non-conductive PSS- anions (Hsiao et al. 2008). Post-treatment of printed films with formic acid was also observed to improve conductivity by further oxidation of the polymer (Daniel & Fotheringham 2007). Finally, combining two of the conductive material classes, the addition of a small proportion of carbon nanotubes (CNTs) to the PEDOT:PSS suspension significantly increases conductivity in the resulting films (Kymakis et al. 2007, Mustonen et al. 2007, Wang et al. 2008a, 2008b; Ham et al. 2008, Moon et al. 2005, Bhandari et al. 2009), although the films were only printed in the work by Mustonen et al. For inkjet printing of electrodes, of all of the conductive polymers available, PEDOT:PSS suspension is by far the best-suited due to its small dispersed particle size
48
and ready ability to mix with a variety of solvents and surfactants, many of which are even beneficial to its conductive performance.
3.2 Semiconductors
Semiconductors usually form the key functional layers of an electronic device. Light-emitting layers, light-collecting layers, diode junctions, and transistor stacks are all formed of different semiconductors. The property which makes these materials suitable for use in all of these applications is their small-to-moderate discrete electronic bandgap. The bandgap represents a region of forbidden energy levels which cannot be occupied by electrons in that particular material, lying between the unoccupied conduction band and the occupied valence band (Neamen 2002). Upon some sort of energetic stimulus, electrons can be excited across the band gap into higher energy states in the conduction band or higher energy bands. These electrons are free to move in the conduction band, resulting in limited electrical conductivity through the semiconductor. Electrons that jump to the valence band leave behind an unoccupied space which is usually referred to as an “electron hole” or simply a “hole” (Neamen 2002). When this happens, an electron adjacent to a hole in the valence band will fill the hole, leaving a hole, which will then be filled, and so on – so an apparent flow of holes in the valence band and actual flow of electrons in the conduction band occurs. The excited electrons can relax to a lower energy state by jumping back across the bandgap and recombining with holes, losing energy which is released as either heat or light (Sedra & Smith 1997). As is shown in Figure 3.4, insulators also have a bandgap, albeit a wide one, meaning that a very large amount of energy is required to excite an electron into the conduction band – rendering them effectively non- conductive (Huang 2009). Metals, conversely, have overlapping bands, allowing facile transfer of electrons and high conductivity. It is also worth noting that the terms conduction and valence band are generally applied to inorganic semiconductors, whereas organic semiconductors’ bands are characterized as the lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO).
The bandgap (Eg) is an important property in all semiconductor-based microelectronics, where it determines the amount of current or potential required to emit light (LEDs), the current needed to
49
modulate conductivity (TFTs), or the wavelength and intensity of light required to excite electrons into the conduction band, where they can flow and be collected (PVs) (O’Mara et al. 1990). Semiconductors are defined as materials with moderate bandgaps (> 0 – 4 eV). Certain materials are better-suited to each of these applications, with silicon (Eg = 1.11 eV) often being a “workhorse” material due to its applicability to many types of microelectronics (O’Mara et al. 1990). However, when inkjet printing is the deposition method, rather than vapour deposition techniques most commonly used for silicon processing (Jaeger 2002), other materials which are solvent-dispersible or soluble (which silicon is not) must be considered. These materials can be inorganic or organic semiconductors, and each type displays unique properties and useful qualities.
Vacuum level Electron affinity
Work function
Conduction band Bandgap Fermi level
Valence band Electron (ev) energy
Conductors Semiconductors Insulators
Figure 3.4. Simplified electronic band structures.
3.2.1 Inorganics
Inorganic semiconductors are made up of compounds of different inorganic atoms, often classified by the periodic group in which the element is located – e.g. GaAs would be a III-V semiconductor. They may also include C, which is not technically an inorganic species, such as in SiC. They may include any number of inorganic molecules in any proportion, in so-called “solid solutions” of inorganic atoms (Berger 1996).
When considering inkjet printing of such materials, the likelihood of their synthesis and dispersion as nanoparticles with size below the cutoff specified by the printer becomes remote as
50
the complexity of the molecule increases. A controlled synthesis method for producing solution- dispersible nanoparticles with good semiconductive performance is a difficult prospect even for simple binary compounds. It is likely partially for this reason that certain materials have attracted a great deal of attention as solution-processable inorganic semiconductors – most notably, the chalcogenides of Zn, Cd, Pb, and Y. In particular, CdS and ZnS have been prepared as nanoparticulate suspensions by controlled precipitation of nanocrystals (Chander 2005). The crystalline nature of these materials means that they are theoretically functional upon wet deposition, with no post-treatment, which was specified earlier as a key property of ink components in this study. However, to allow charge transfer from nanocrystal to nanocrystal, these materials need to be either sintered at high temperature into monolithic films, incorporated into a conductive polymeric binder, or self-assembled into a packed film (Bakueva et al. 2003).
CdS and ZnS (as well as their selenides) are generally prepared using a competitive precipitation process, in which soluble salts of the metal cation and chalcogen anion are mixed in the presence of a solvent, a ligand, and sometimes, a dopant ion. Under controlled conditions of temperature and pH, nuclei of the semiconductor material are formed and gradually grown until their growth is arrested by double-layer repulsion, which prevents free ions still in solution from reacting on the surface of the colloids (Chander 2005). The positively-charged counterion from the chalcogen source, which does not participate in the reaction, forms a positively-charged boundary layer which prevents initial agglomeration of the particles (Warad et al. 2005). The presence of ligands which attach strongly to the colloids’ surface also assists in keeping them dispersed and preventing Ostwald ripening or agglomeration (Capek 2006). In some cases, more conventional colloidal chemistry, such as reverse micelle synthesis, has been used to prepare these particles (Yang & Bredol 2008). The feature of these synthesis methods that makes them so well-suited to preparing inkjet-printable materials is control over the ligand species and solvent system. With the proper ligands, nanoparticles can be made soluble in a variety of solvents and solvent mixtures, greatly facilitating ink formulation.
Of the species mentioned above, one of the most widely studied is ZnS – and specifically, doped ZnS. Doped ZnS is a well-known photoluminescent and electroluminescent material (Yen 2004) which is almost ubiquitous in ELDs and some LEDs. Doped ZnS can also be prepared in a
51
nanocrystalline form with a minimal number of reagents in a variety of solvents, including water. Doped ZnS nanoparticles have been prepared in aqueous or alcoholic solutions by many groups (Althues et al. 2006, Igarashi et al. 1997, Adachi et al. 2007, 2008; Konishi et al. 2001, Hwang et al. 2005, Karar et al. 2004, Takahashi & Isobe 2005, Mu et al. 2005, Vogel et al. 2000, Manzoor et al. 2003, Warad et al. 2005, Yang et al. 2003, Yu et al. 1996, Jindal & Verma 2008) using acetate or chloride salts of the metallic precursors. Doping ions in these studies included Cu+, Mn2+, Al3+, and several halogens. The reason ZnS is particularly notable is that its synthesis is inherently facile – it requires only a Zn/dopant soluble salt, a suitable ligand or cap, and easy-to- handle solvent system. This stands in contrast to the more complex methods of synthesizing such colloidal semiconductors as CdSe/CdS/CdTe (Hines & Guyot-Sionnest 1996, Talapin et al. 2001) or PbS (Bakueva et al. 2003). These usually require elevated temperatures (and therefore non-aqueous solvents), multiple injections of precursor materials, expensive or exotic salts and ligands, and, of course, the use of biologically and environmentally harmful Cd and Pb. The use of non-aqueous solvents precludes their inclusion in a water-based ink without some sort of ligand exchange, as well. However, the aqueous synthesis of doped ZnS has been observed to yield unpredictable doping levels (Peng et al. 2005), variable ratios of Zn:S (Althues et al. 2006), and clustering of dopant atoms at the surface of the nanoparticles (Yu et al. 1996, Bulanyi et al. 2002, Bulanyi et al. 1998), if temperature and pH are not controlled similarly to the Cd-based synthesis. So, preparing inkjet-ready materials is a process which must be carefully adapted to each ink type and particular application.
Another consideration when preparing or choosing jettable inorganic semiconductors is their greatly increased surface area to volume ratio, due to small particle sizes. In certain processes, such as light emission or absorption, the splitting or recombination of charge carrier pairs determines device function. So-called surface states – different or “bent” electronic band structures at the semiconductor’s surface, resulting from the transition between the crystal lattice and the vacuum/atmosphere – tend to trap charge carriers with energy barriers (Shik 1997). The larger relative surface area in nanomaterials means that surface states are abundant. If charge carriers are trapped, they may not recombine (emitting light) or move (providing current), or may recombine inside another material, away from a light emitting centre (Vilms & Spicer 1965).
52
For this reason, another surface layer, called a “shell” or “cap” is often applied to Cd- and Zn- based nanoparticles, in a process commonly referred to as “passivation”. This secondary surface layer prevents the electronic band structure at the underlying particle (usually called the “core”) surface from changing and forming surface trap states (Bhargava et al. 1994). In order to keep the charge carriers confined to the core material and prevent them from travelling or recombining in the shell or on the shell’s surface, a wide bandgap material is often chosen for the shell, such as ZnS (Eg = 3.6) for CdS (Eg = 2.4) (Yang & Holloway 2004). This wide bandgap material presents an energy barrier which prevents carriers from leaving the core. Shelling can present a problem in materials with already wide bandgaps, such as ZnS. However, shelling with an insulator like ZnO or SiO2 has been successfully attempted by several groups, resulting in improved photoluminescent emission from ZnS:Mn nanoparticles (Karar et al. 2004a, Jiang et al. 2009, Haranath et al. 2005). The insulating nature of many polymers also makes them useful for capping nanoparticles, such as poly(acrylic acid)-capped ZnS:Mn (Igarashi et al. 1997, Konishi et al. 2001, Althues et al. 2006).
3.2.1.1 Quantum behaviour
A unique feature of semiconducting materials (and technically, all nanomaterials) is observed when one or more of their physical dimensions become smaller than some characteristic length – such as the de Broglie wavelength, the electron mean-free-path, the Bohr exciton radius, and so on (Shik 1997). Carriers become confined within potential wells with physical sizes smaller than these characteristic lengths in either one dimension (i.e. a thin film), two dimensions (i.e. a nanoribbon or nanowire), or all three dimensions (i.e. a nanoparticle, or more commonly, quantum dot or QD) (Kelly 1995, Harrison 2000). When so confined, the wavefunctions of carriers in the confined dimensions are no longer represented by continuous probability density functions, but rather by discrete energy states occurring at integer multiples of the wavefunctions’ wavelength, much like states within a single atom (Kelly 1995). Therefore, as the confined dimensions change in size, and the wavefunctions’ wavelengths also change, so do the discrete energy levels: this alters the electronic band structure of the material (Gaponenko 1998). In other words, by using nanosized semiconductors, bandgap and electron affinity can be
53
altered by adjusting one or more physical dimensions below a critical value. This is referred to as “energy quantization” or simply “quantization”; materials below whatever characteristic length is required to achieve quantization are called “quantized”.
In the case of the nanoparticles typically incorporated into inkjet inks, the radius of the nanoparticles determines whether or not they are quantized. The critical radius below which strong quantization occurs is referred to as the exciton Bohr radius, or a * (Harrison 2000). If the b particle radius, R, is smaller than a *, the bandgap of the material will change according to the b equation given below (Sun 2005, Caruso 2004):
휋2ℎ2 퐸 = 퐸 + 𝑔푛푎푛표푠푡푟푢푐푡푢푟푒 𝑔푏푢푙푘 2휇푅2
1 1 1 = ∗ + ∗ 휇 푚ℎ 푚푒
* * where Eg is the electronic bandgap of the material, and h is Planck’s constant. mh and me refer * to the effective masses of electrons and holes in a given material – for example, in ZnS, mh is * 0.61me, and me is 0.40me (Vogel et al. 2000), where me is the effective mass of a charge carrier (9.11 10-31 kg) (Berger 1996). As the radius shrinks, the bandgap of the material widens. Because the emission of light, for example, occurs when an excited electron from the conduction band recombines with a hole in the valence band, losing energy Eg in the process, the magnitude of Eg will determine the wavelength of the light emitted () (Neamen 2002):
ℎ푐 퐸 =
where c is the velocity of light travelling in a vacuum (3 108 m/s). Therefore, wider-bandgap (smaller diameter) nanomaterials will emit light upon electrical or photonic excitation at a higher energy, and a lower wavelength. As has been demonstrated extensively with Cd-based QDs, the
54
full visible range of colours, as well as UV and IR, may be emitted by QDs of varying sizes * below ab . This property may be useful for LEDs and PVs, but the small nanoparticle sizes required – usually below 10 nm (Shik 1997) – are not necessary for inkjet printing. However, minimization of particle size is helpful for preventing clogging and maintaining dispersion, meaning that QDs are a good starting point for ink formulation.
3.2.2 Organics
The electrical properties of organic semiconductors are very similar to those of inorganics, with a few subtle differences in terminology and carrier transport mechanisms. As was mentioned previously, the conduction band and valence band do not exist in organics; instead, the comparable values of LUMO and HOMO, respectively, refer to the edges of organics’ electrical bandgaps. Also, charge transport within organics is not simply a matter of electrons moving through a relatively empty conduction band; it is often modeled as being the result of electrons hopping between different trap states and overcoming energy barriers to do so (Vissenberg & Matters 1998). This implies behaviour unlike that in inorganics with respect to temperature and input potential, which both assist electron hopping by providing energy (the opposite occurring in inorganics).
Organic semiconductors present certain advantages over inorganic semiconductors, not the least of which is the relative ease of incorporating polymers and small molecules into solution when compared to inorganic nanoparticles. Of course, issues with solubility already raised in the section on conductive polymers also apply to semiconductive polymers, which have similar conjugated chemical structures, often including aromatic rings (Skotheim & Reynolds 2006). Small molecules may be soluble in particular solvents; however, to allow charge transport between molecules, a conductive polymer binder (or “host”) is still required, as with inorganic nanoparticles. Also, the tight molecular packing of solid-phase small molecule semiconductors, due to interactions between the delocalized charged regions characteristic of conjugated species, can sometimes mean that they are surprisingly difficult to dissolve (Dimitrakolopous & Mascaro 2001). Functionalization of the molecules or polymers by adding side groups and chains can assist in improving solubility, but the reduced packing caused by such treatment can then
55
compromise carrier mobility (Chang 2005). A less-packed, more-soluble layer is also more prone to oxidation and moisture damage. Further functionalization to improve packing and ordering upon deposition and heat-treatment may improve mobility somewhat, but the more chemical pretreatment steps that are required, the “dirtier” the final product (Vissenberg & Matters 1998). Ideally, a molecule or polymer must be built from the ground up, adding functional groups to the initial structure that allow solubility and encourage packing upon deposition and drying.
Common semiconductive polymers and molecules are typically built around aromatic rings or their derivatives (thiophene, pyrrole, etc.) A few have already been listed in the section on conductive polymers – most semiconductive polymers are similar to these, being substituted polythiophenes or poly(phenylenevinylene)s, but others include polypyrroles and the commonly- used poly(n-vinylcarbazole) or PVK (Yang et al. 2003). Some of the most common small molecules are pentacene, copper phthalocyanine, tris(8-hydroxyquinolinato)aluminum (Alq3), and fullerene – all of which include numerous aromatic rings, and can be substituted and functionalized into many different variations (Mullen & Scherf 2006, Cheng et al. 2009).
In many cases, the relatively simple processing and infinite customizability of organic semiconductors makes them a superior choice for electronics fabrication. However, inorganic materials do present several advantages, the most important of which are their better environmental stability (Ono 1995) and, in LEDs, narrower emission spectra (Xing et al. 2005). In the following chapter, the intricacy of organic LEDs when compared to conventional inorganic LEDs or ELDs will also be outlined. When constructing an all-printed device, the number of different inks – all of which must be based in orthogonal solvents – directly correlates to the difficulty of fabrication and the likelihood of device failure. Therefore, realistically, a 3- layer device has a better chance of being successfully fabricated by inkjet than an 8-layer one, which is a significant advantage for inorganic-based inks.
56
3.3 Insulators
Insulators are materials that allow almost no current to flow through them. Similarly to the previous two classes of materials, there are both inorganic and organic insulators; and again, inorganic insulators can be made inkjettable by their dispersion or synthesis as colloids, whereas organics can be dissolved or dispersed. However, inorganic insulators have dielectric constants that can be orders of magnitude larger than the best-insulating organic materials (Ono 1995, Matsumodo 1997, Vetanen 1997). In many cases, devices do not require high-dielectric (“high- k”) layers, and so the moderate dielectric constants of many readily soluble polymers are sufficient. Most polymers which are not intrinsically conductive (i.e. conjugated) conduct electricity very poorly, and function well as insulating layers – such materials as poly(vinylidene) fluoride (PVDF) (Vetanen 1997, Jung et al. 2010), poly(methyl methacrylate) (PMMA) (Jung et al. 2010), cyanoethylcellulose (Saad 1994), and polyimide (Park et al. 2005). The dielectric constants for these materials are usually in the range of 5-11 (Reddish 1962). These and many other polymers can be dissolved in common solvents and deposited using inkjet printing, although loading/molecular weight may both have a bearing on jetting quality.
Solvent evaporation
Interparticular forces
Grain merging (<200°C) (>1000°C) (<200°C)
Sintering Binder Self-assembly
Figure 3.5. Film formation in inks containing inorganic nanoparticles.
Improved insulating performance can be achieved by using inorganic materials, with dielectric constants sometimes in the thousands when crystalline. Crystallinity is a prerequisite for good insulating behaviour (Jia et al. 2000), meaning that crystalline nanoparticles are needed.
57
However, like inorganic semiconductors, inkjet-printed layers must be sintered at high temperature to form a monolithic film, suspended in a binder, or self-assembled (Figure 3.5)
Inorganic insulators, usually being oxides, have extremely high sintering temperatures (>1000°C) which are not suitable for R2R processing, particularly not on conventional substrates. If a binder is used, that binder must also be relatively insulating to prevent localized tunneling of current. Self-assembly requires monodisperse particles with controlled intermolecular or interparticular forces on an extremely flat substrate (Aubry et al. 2008).
Another means of producing an inorganic insulating film is to use sol-gel chemistry. Sols are monodisperse colloidal suspensions, prepared from precursor materials, which may be gelled by the removal of solvent, and further cured and densified by the elimination of the liquid phase (Brinker & Scherer 1990). The resulting solids are glassy and amorphous, requiring post- treatment to become crystalline. Post-treatments can be as simple as high-temperature heating, or may utilize techniques such as lower-temperature hydrothermal exposure (Xu et al. 2006) or chemical post-treatment (Zeng et al. 1999, Matsuda et al. 2000). Insulating films of several species have been prepared and sintered using sol-gel chemistry, particularly high-dielectrics like barium titanate (BaTiO3 or simply BT), lead titanate (PbTiO3), lead-zirconate titanate (PZT), and barium strontium titanate (BST) (Yakovlev 2004, Zhou et al. 2008, Lima et al. 2007, Sharma et al. 2000, Atkinson et al. 1997, Harizanov et al. 2004, Sharma & Sarma 1998, Sharma & Mansingh 1998). However, sintering temperatures in all of these films were well above 200°C, making them unsuitable for many flexible substrates. Also, the relatively low solids content of sols means that they shrink a great deal when dried (Brinker & Scherer 1990), causing cracks (shown in Figure 3.6). These cracks would only be worsened by the bending of a flexible substrate. Finally, exposure to a strong base or an alcohol over several hours for hydrothermal sintering is a technical hassle, requiring a specialized chamber. In some cases, these materials can potentially solvate or damage other layers and the substrate itself.
58
200 µm 20 µm
Figure 3.6. Microcracking in sol-gel-derived BaTiO3 spun films on glass. These films were produced in the laboratory using sol-gels prepared similarly to those described by Sharma & Mansingh (1998).
Self-assembly is an interesting prospect for forming films of nanoparticles, since very small nanoparticles can be densely packed into structures resembling monolithic solid films. A fundamental requirement of self-assembled layers, however, is that they must be made up of monodisperse particles. Although sol-gel chemistry can produce colloid-sized – or smaller – particles of insulators, these are not crystalline, and so their dielectric constants are not much better than polymers. Normally, ceramic particles are ground down from bulk materials, so their minimum size is limited. Synthesis of sufficiently small/monodisperse insulating nanoparticles for self-assembly is not yet well-researched, although it has recently been accomplished by Ould- Ely et al. using a sol-gel/hydrothermal reaction setup (2011). Redispersion of these nanoparticles with suitable ligands to allow self-assembly has not been reported, however; there is substantial work to be done before self-assembled insulating films become a reality.
Therefore, inorganic insulating inks generally contain suspended ceramic nanoparticles of insulators such as titanium dioxide (TiO2), BaTiO3, or barium strontium titanate (BaSrTiO3) with an accompanying binder polymer (Kim & McKean 1998, Sakai et al. 2006, Ding et al. 2004, Tseng et al. 2006). This means of producing moderately insulating films is robust, simple, and applicable to many different substrates – although the usual challenges associated with ink formulation and film morphology, of course, are present.
59
3.4 Encapsulants
Electronic devices sometimes include other, inert layers for physical protection. These layers may be used to prevent moisture from infiltrating the device, oxygen (and other oxidants) from reacting with device components, and physical/thermal/electrical shock (Ardebili & Pecht 2009). Devices using organic components, which have many functional groups with which oxygen or water can react, require particularly effective encapsulation. Conventional encapsulants are most often two-part epoxies or laminated polymer films, which are rolled on to planar devices (Ono 1995). Jettable encapsulants are generally UV-cured thermoset resins with low permeability and some degree of mechanical toughness, some of which are commercially available as inks. Similar materials may also be used in the novel field of paper-based electronics, where the rough and permeable surface of paper presents a problem for device deposition, function, and environmental stability. The paper surface may be coated with a material like spin-on-glass (Kim et al. 2010) or another impermeable resin to improve smoothness and reduce permeability.
3.5 Substrates
Technically, inkjet printers can deposit materials on any substrate, from the usual paper sheet to a three-dimensional object to a textile. For microelectronic fabrication, the same substrates as are commonly used in conventional electronics processing – glass and silicon – can be inkjet- printed. In an R2R process, flexible substrates may also be printed, including paper, fabric, and polymers. Each substrate presents unique advantages and limitations; as a rule of thumb, the rigid substrates already used in electronics manufacturing support the best-functioning devices, and flexible/permeable substrates currently lag behind them in terms of device functionality.
The traditional support for many types of electronics is silicon. Silicon is a useful substrate because it can also be a functional layer – a TFT, for example, can be prepared by oxidizing the surface of the Si in a defined area, producing an insulator (SiO2), and source and drain electrodes applied with evaporated metal (Bucknall 2005). Silicon wafers may also be made almost atomically smooth by controlled crystal growth and cleaving (O’Mara et al. 1990). Silicon can also be easily doped with an electron donor (n-doped) or electron acceptor (p-doped) with high
60
spatial resolution (e.g. by ion implantation or epitaxy) to produce diodes and other p-n junction- based devices (Jaeger 2002). For many types of electronics, and especially for integrated circuits and TFTs, silicon wafers are the ideal substrate, providing the necessary electronic and physical properties for maximal functionality. However, many of these advantages of silicon are nullified when using atmospheric-condition inkjet printing as a fabrication technique. The first and most important reason why silicon is not particularly useful for printed electronics is that printing is an additive technique. Functional layers are deposited on an inert substrate, and the substrate itself is not deliberately altered by the inks resting on it. A hybrid method of fabrication involving wafer doping or oxidation followed by printing of interconnects, insulators, and so on is of course feasible; however, the necessity to use both high-vacuum deposition techniques and printing somewhat defeats the purpose of either deposition method. If a device is going to be partially constructed at high vacuum by conventional means, a likely better-functioning device could just be constructed using only conventional means; likewise, if a device is being printed to increase throughput and facility, having vacuum stages nullifies both of these gains. In other words, silicon has its place as a substrate in conventionally-processed electronics, but it is not a sensible candidate for printed electronic substrates.
Glass, on the other hand, still retains its usefulness, even when used in printed electronics. Glass is widely used in the fabrication of many types of device, especially those requiring light input or emission (silicon being opaque). The smooth topography and high-temperature processability of silicon are also present in electronics-grade glass. However, glass offers a further advantage for printed electronics: it is largely chemically inert, meaning that almost any ink system can be deposited on its surface. This chemical inertness also means that glass can be effectively surface-treated to control wetting by any given ink (Menawat et al. 1984). Of course, the major limitation of glass is that it is brittle and inflexible, like silicon, making it useless for R2R applications. However, for processing of displays and solar arrays in a batch fashion, inkjet printing onto glass has been established as a rapid means of production which may displace conventional fabrication techniques (Caglar 2006, Singh et al. 2010, Kim & Han 2010). ITO- coated glass is also almost ubiquitous with LEDs and PVs, presenting similar advantages and limitations as bare glass, but of course, with an added conductive layer. The types of glass
61
available for electronics fabrication are usually borosilicates or soda-limes, depending on the processing conditions (Chen et al. 2011).
The next logical step after glass in alternate substrates is towards polymers. Polymers offer many of the advantages of glass, with relatively smooth surfaces, tunable surface chemistry, and optical transparency. Like glass, ITO-coated versions of certain polymers – particularly polyethylene (PET) – are available. Unlike glass, they offer mechanical flexibility, light weight, low cost, and, as a result, R2R processability. All of these advantages come at one major cost: reduced temperature tolerances. Not many transparent, inexpensive polymers exist with glass transition temperatures (Tg) much greater than 200°C, and many have decomposition temperatures even lower than that (Zhouping et al. 2010). Some high temperature-tolerant polymers do exist, such as Mylar and silicone, but these are generally not transparent or inexpensive. Furthermore, unlike glass, polymers are soluble in many different solvents and can be chemically reactive. So, when using polymers, solvent and additive selection during ink formulation is crucial to ensure that a device will function and not damage the substrate. Several research groups and corporations have produced different types of flexible electronics, sometimes using inkjet printing, on many different polymer types (Carter & Gardiner 2009).
Paper is an emerging substrate for electronics, well-regarded due to its renewable and biodegradeable nature, flexibility, and relatively low cost. Certain film-based devices, such as transistors, batteries, and ELDs, have been successfully prepared on paper (Tobjork & Osterbacka 2011). Although offering these advantages, paper is a problematic substrate for electronics. In terms of processability, it is similar to many polymers: most paper will scorch below 200°C, and may burn only a few degrees above this temperature (FPL 1964), while being lightweight, inexpensive, and R2R-processable. However, it is also opaque, rough, irregular, and porous. The opacity may be overcome in PVs and LEDs by reversing the order of the layer stack in the device. The surface properties, however, require paper sheets to be treated to minimize the deleterious effects of their rough, absorbent surfaces. As ink penetrates into a paper sheet, non- functional paper components become ingrained in the functional layers of the device, and fibres swell and roughen (Xie et al. 2008) – both of these occurrences might cause a device to fail completely. Also, physicochemical interactions, like adsorption, may occur between functional
62
materials and paper components (Montibon et al. 2009, 2010; Di Risio & Yan 2009). There are certainly means of smoothening and sealing paper, usually with mechanical calendering, sizing, or coating (Smook 1992). Despite the difficulty of using paper to support thin-film electronic devices, it has presents certain unique features which make it particularly attractive. Paper is a biodegradeable, disposable, renewable material upon which much of the existing infrastructure for R2R printing is based. Therefore, using paper as a substrate for simple, rudimentary electronics is economically and environmentally desirable. However, because the roughness of most sheets exceeds the typical thickness of LED, PV, or TFT films, its use for producing such devices is limited. Nevertheless, TFTs (Lim et al. 2009, Kim et al. 2004), primitive capacitors (Pushparaj et al. 2007), and displays (Kim et al. 2010, Andersson et al. 2002) have been produced on paper, albeit with few or no inkjet-printed layers.
63
Chapter 4 4 Light-emitting devices
Light-emitting devices come in many different forms and utilize a myriad of materials, many of which were described in the previous chapter. Some of these devices were listed in Chapter 1. Several display types – in particular, the LCD, CRT, and PDP – are not usually classified as light-emitting displays, as they do not emit light solely from electrically exciting the semiconducting species within the device, but rather by filtration of backlight, excitation of phosphors with an electron beam, or with plasma, respectively. Although CRTs and PDPs are technically “emissive”, with the phosphor (luminescent species) being excited by an energy source and luminescing, the phosphors do not luminesce due to simple applied current. The architecture of these displays also requires either a vacuum tube or small gas-filled cells, meaning that thin-film processing – e.g. inkjet printing – cannot be applied to their manufacture (Chen et al. 2011). The solid-state, planar display types, ELDs and LEDs, can be built from films of materials, and utilize the phenomenon known as electroluminescence (EL) to produce light. EL is the emission of light from certain materials when exposed to an electric field or electrical current. Unlike thermoluminescence, bioluminescence, or chemiluminescence, EL is entirely a result of electrical excitation of materials, wherein high-energy electrons fall to lower- energy states, releasing energy as a characteristic wavelength of light (Matsumoto 1984). The means of exciting the electrons varies depending on the device type and voltage drive type, and therefore so does the device structure and component materials. It is for this reason that there exist several different configurations of functional layers for devices which utilize EL.
For inkjet-printed display devices, stacked-film structures are ideal. The relatively simple quasi- 2D structures of ELDs and LEDs allow them to be much more easily fabricated using solution processing, by overlaying successive layers of functional materials. Structurally, ELDs and LEDs are similar, but ELDs are excited by accelerated electrons and LEDs are excited by current flow. This distinction is particularly important depending on the materials used, as certain excitation mechanisms are only available to some materials. The different structures and typical materials are outlined in this chapter.
64
4.1 Light-emitting diodes (LEDs)
The LED is a well-established technology based on an extremely simple principle of operation. A diode is a junction between a p-doped and an n-doped semiconductor (a p-n junction), with an anode connected to the p-doped material, and a cathode connected to the n-doped material. As direct current (DC) flows from the cathode into the n-doped side, high-energy electrons in the conduction band (or LUMO) recombine with holes in the valence band (or HOMO) flowing from the p-doped side, causing the electron energy to drop across the bandgap and releasing light of a characteristic wavelength (Sedra & Smith 1997). In the simplest LEDs, a single layer which transports both electrons and holes and emits light can be used. The use of both a hole- transporting (p-doped) and electron-transporting (n-doped) layer, one of which is the light- emitting material, improves device function and efficiency (Crone et al. 1998). However, a range of additional layers may be included to provide optimal efficiency (see Figure 4.1), including carrier-blocking layers to keep carriers confined to the emissive region (Adamovich et al. 2003) and current-limiting layers to prevent device burnout (Ono 1995). To further improve functionality, electrodes may be chosen which have specific carrier energy levels. The difference between the energy level and the vacuum level (0 eV) is the electrode’s work function. The anode, for example, which conveys holes, should have a work function large enough to contain holes with energy close to the valence band/HOMO of the layer into which the holes are being injected (He et al. 2004).
blocking LUMO values
- - e - e e VDC source
anode cathode - cathode work e electron-transport layer function work function h + hole-blocking layer emissive molecule film Energy(ev) h + electron-blocking layer h + hole-injection layer h + transparent anode substrate HOMO values blocking
Figure 4.1. Schematic of an OLED (right) and its energy level diagram (left), showing movement of charge carriers.
65
The complexity of such devices can be significant. Using any fabrication method, this increases processing time and cost. In particular, with inkjet printing, each additional layer entails formulation and functional testing of another inkjet ink. More significantly, the need for an orthogonal solvent system dictates that each layer must not solvate the layers below it; this can be problematic for a seven-layer stack in which several conjugated polymers are used, as not many of these are water-soluble. This problem was addressed by Haverinen (2010), who used cross-linked layers to minimize interlayer dissolution in a seven-layer stack – however, in that work, only one of the layers was inkjet-printed and another was spin-coated and cross-linked, the rest being vacuum-deposited. This problem presents the greatest hurdle to the application of multilayer LED technology by inkjet. Single-layer devices (which are in fact three layers, including electrodes) require only one insoluble layer to be placed between the other layers, greatly simplifying fabrication, at the cost of device functionality. So, depending on the materials used and the facility of their incorporation into orthogonal inks, a suitable device structure may be chosen. Some of the common structures are shown in Figure 4.2; note that the they may not include all of the layers shown, but definitely must include electrodes and an emissive species.
The classic LED structure is very similar to another structure which is particularly well-suited to inkjet deposition: the quantum dot LED or QDLED. This structure does not necessarily have to include QDs as the emissive species when inkjet-printed, as long as the particles are sufficiently small to be printed. When using inorganic materials, the advantage of this structure over a standard LED is that it does not require a monolithic semiconductive film, but rather nanoparticles in a conductive binder. A self-assembled single film of QDs has been deposited in a QDLED-type structure without a binder by Haverinen et al. (2009, 2010), Yang et al. (2003), and Hieronymas (2002), but such structures suffer from issues with charge transfer between QDs. The ligands (and sometimes, shells) of the nanoparticles are usually wide-bandgap materials – often insulators – preventing charge carriers from moving between nanoparticles without being provided with significantly more energy (i.e. potential). By incorporating the QDs into a polymer with a higher LUMO value than the conduction band/LUMO of the ligand and shell materials, it becomes energetically favourable for electrons to transfer to the QD to
66
recombine with holes (Coe-Sullivan et al. 2003, 2005). A schematic of this improved charge transfer is shown in Figure 4.3, based on a similar diagram produced by Haverinen (2010).
+ - + -
VDC source VDC source
cathode cathode current-limiting layer QDs/conductive binder emitter transparent anode transparent anode substrate substrate
LED QDLED
+ - + -
VDC source VDC source cathode electron-transport layer cathode hole-blocking layer electron-transport layer emissive molecule/host emissive polymer film hole-transportfilm layer hole-transport layer hole-injection layer transparent anode transparent anode substrate substrate
PLED OLED
Figure 4.2 Typical planar LED structures.
large energy barrier energetically favourable - LUMO e
e -
CB
core
shell - ligand
layer(CTL) e Energy(ev)
Energy(ev) + charge transport h VB + + h h HOMO QD Packed QD-only layer CTL-embedded QD layer
Figure 4.3. Improvement of carrier mobility by polymer embedding of QDs used in QDLEDs.
67
Layer thicknesses in conventional LEDs are on the order of a few hundred nanometres (Ohring 1991). Film thickness is often a determinant of exactly how a device functions (Dimitrijev 2005, Wood 2000). All of the layers of any light-emitting device are generally of controlled thickness to maximize performance while minimizing the amount of material deposited and the resistive losses across the device (Adachi et al. 2007, 2008; Schrage et al. 2010, Manzoor et al. 2003, Hieronymas 2002, Cho & Cha 2009). Increasing thickness may lead to poorer device performance and higher drive voltage and current (Schrage et al. 2010). The topography of LED films also plays a major role in their performance. Films with localized fluctuations in thickness may allow current to channel through the thin regions, or direct current flow to thicker regions, tunnelling through adjacent films (Wood 2000). As a rule of thumb, the films should be as smooth as possible to prevent such issues from causing a device to lose functionality (Berger 1996, Haverinen et al. 2009, 2010).
LEDs are composed of a wide variety of materials, especially organic LEDs (PLEDs and OLEDs), which can incorporate many of the conjugated polymers/small molecules discussed in the previous chapter. Inorganic LEDs can contain any semiconductor with a suitable bandgap for emitting the desired colour of light. In glass-based LEDs, the light emission occurs through the “front” of the device, i.e. the glass layer. Therefore, the front electrode or anode must be made of a transparent or translucent conductive film. The most widely used film for this purpose is ITO, which has a relatively high conductivity of 2.5 103 – 5 103 S/cm (Phillips et al. 1995) and a high work function of ~ 4.65 eV (Parker 1994, Bakasybramanian & Subrahmanyam 1991). Other conductive oxides such as zinc-indium-oxide (ZIO) or antimony-tin-oxide (ATO) are also used for this application, both of which have comparable resitivities. PEDOT:PSS is also commonly used, although sometimes only as an overlying layer on ITO for hole transport, where its smooth surface assists in hole injection (Yoshioka & Jabbour 2006). Lithium fluoride is often employed in a similar role at the cathode, where it assists in electron injection (Nalwa 2003). The cathode is traditionally composed of evaporated metal with high conductivity and low work function (Ono 1995). The many other possible layers included in LEDs are composed of specialized polymers, for the most part; some of the more common materials used for charge transport and blocking layers are listed in Table 4.1.
68
4.2 Electroluminescent devices (ELDs)
The mechanism of inducing EL is different in what are classically referred to as ELDs than the mechanism in diodes. Rather than allowing electrons and holes to recombine radiatively inside a semiconductor by passing a current through a diode, ELDs rely on a strong electric field to accelerate electrons through the entire device stack. Above a certain voltage, known as the threshold voltage or Vth, electrons will tunnel through the device layers, gathering enough kinetic energy to strike and excite luminescent dopant centres in the emissive layer, similarly to a CRT’s electron gun but in the solid state (Ono 1995). This requires a high voltage; therefore, these devices contain insulating rather than charge-transporting layers to prevent catastrophic dielectric breakdown (Kitai 2008, Hart 1999). This also requires specialized materials: light emission occurs at the dopant sites in the emissive layer, meaning that the emissive materials – or phosphors – must be doped (Bredol & Dieckhoff 2010, Vlasenko & Popkov 1960). Typical phosphors are chalcogenides doped with <5 at.% of impurity, often rare-earth metals, transition metals, or halogens (Yen 2004, Yen 2006). A summary of ELDs is shown in Figure 4.4.
+ - + - VAC source VAC source cathode cathode insulator insulator phosphor film phosphor/insulating binder insulator transparent anode transparent anode substrate substrate ACPEL ACTFEL
+ - + -
VDC source VDC source cathode cathode current-limiting layer electron barrier phosphor/insulating binder phosphor film transparent anode transparent anode substrate substrate
DCPEL DCTFEL
Figure 4.4. ELD structures.
69
Unlike LEDs, ELDs can be both AC- and DC-driven. This presents several logistical advantages, but also means that the same materials rarely function in both types of device. However, a unique advantage offered by AC drive is that the work function of the two electrodes can be the same (Wang et al. 1996), offering the possibility of a completely transparent device. The two common structures (four total, since each can be both AC- and DC-driven) are described below.
A common and robust ELD type is the AC-driven powder ELD, or ACPEL. The term “powder” refers to the inclusion of particulate phosphor material within an insulating binder, rather than a solid film of phosphor material. The core of this device structure is an insulating polymer resin incorporating phosphor grains, protected from dielectric breakdown by an insulating layer, all sandwiched between two electrodes. The other AC-driven device, the AC thin-film ELD or ACTFEL, contains no binders, but solid films of phosphors and insulators. The core of the device is a micrometre-thick stack composed of a transparent insulating layer, followed by a phosphor layer, and finally a second transparent insulating layer. The necessity for solid, crystalline films of both phosphors and insulators again means that this structure is not well- suited to solution processing. The DC-driven equivalents of the ACPEL and ACTFEL are similar in structure; however, current is limited in these devices rather than voltage. In the
DCTFEL, for example, an electron barrier and current limiting layer – of ZnSe and MnO2 respectively – are typically coated between the phosphor layer and the cathode in order to prevent catastrophic failure (Ono 1995). In the DCPEL, no insulating layer is included, and the phosphor grains are considerably smaller (0.5 – 1 µm diameter) and more densely packed to allow electron transfer between the individual particles. These devices generally only work with Cu-doped phosphors, as the chalcogenides of Cu are still moderately conductive and form conductive surface layers on the particles when exposed to a strong electric field (Ono 1995). The magnitude of the applied field can be quite large – on the order of 107 V/m, requiring typical drive voltages well over 100 V. However, the current required is comparatively low, and ACPELs, for example, draw only a few mW of power. This high operating voltage has limited the application of ELDs in many types of electronics, not only due to power supply issues, but also because any layer faults will lead to rapid dielectric breakdown.
70
Doped semiconductors and high-k dielectrics therefore make up the majority of ELD materials. For powder EL structures, the phosphor material can be suspended in a dielectric polymer resin with a high dielectric constant (in the range of 10-15) to form a luminescent layer. Some suitable resins are cyanoethylcellulose or fluorinated polymers such as PTFE, PVDF, or PVDF:TFE (Matsumodo 1985). Electrode materials remain the same as those used for LEDs. A summary of common ELD (and LED) materials is given in Table 4.1.
4.3 Suitability for printing
In general, the DC-driven LEDs are more widely reported when using nanoparticles, which are required for inkjet printing. There are several reasons for this. The principle of DC drive is based on the flow of current through the device, and the recombination of electrons and holes within the emissive material. In the case of AC drive, emission occurs due to the acceleration of high-energy or “hot” electrons within the electric field and their collision with dopant sites. Brighter emission is achieved when recombination occurs within the bulk of an emissive material, rather than at its surface in both cases, due to previously described surface states. To counter this effect, all ACPEL phosphors are encapulsated with inorganic oxide shells (such as
Al2O3 or SiO2), and TFEL phosphor layers are sandwiched between ceramic layers (Bredol & Dieckhoff 2010) – this is similar to nanoparticle shelling. In nanoparticles, which have an extremely high specific surface area, the likelihood of dopant clusters resting on the surface of the particles is proportionally higher. There are two major problems with encapsulation of nanoparticles. In bulk ZnS, a charged group on the surface of the phosphor particle is not necessary to maintain dispersity. This is not the case with ZnS quantum dots, which are almost always covered with an organic cap or ligand – which is rarely electrically conductive – in order to maintain dispersion. The electric field is therefore not strong enough around nanoparticles to induce electroluminescence by injection of hot electrons (Adachi et al. 2007, 2008; Bredol & Dieckhoff 2010). Even if the particles were dispersed without a ligand, there would still be issues with encapsulation. Firstly, encapsulation of nanoparticles with ceramics is almost always achieved using sol-gels (Bredol & Dieckhoff 2010). However, sol-gels require high-temperature
71
Table 4.1. ELD/LED materials and layer properties.
Layer Present in Typical thickness Materials Properties
Metals (Al, Mg, Ca, Ag) High conductivity, Cathode All devices > 100 nm Carbon (AC-drive) low work function Polymers (AC-drive)
500 nm (TFEL) Perovskites (BaTiO , Insulator AC-ELDs 3 High > 5 µm (PEL) PbZrTiO3 and variants)
1000 nm (TFEL) Doped: ZnS, Y2O3, CdS, Phosphor ELDs, LED grains > 20 µm (PEL) Gd2O2S, ZnO Electroluminescent 50-200 nm (LED) Undoped: GaAs, CdS
Insulating Fluorinated polymers ACPEL > 20 µm Moderate binder Cyanoethylcellulose
Conductive Moderate QDLED 50-200 nm Conjugated polymers binder conductivity
Current-limiting DCTFEL 500 nm MnO Resistive layer 2
Electron barrier DCTFEL 500 nm ZnSe Large bandgap
Emissive MEH-PPV, P3HT, PVK, PLED OLED 100-200 nm Electroluminescent polymer polyfluorene
Emissive Alq , PVK containing OLED 50-100 nm 3 Electroluminescent molecule fluorescent dye
Electron LED High electron < 100 nm TPBi, Alq , LiF transport layer PLED OLED 3 mobility (n-doped)
LED Hole blocking PLED 50-100 nm BCP, PBD, Liq Poor hole mobility layer OLED LED Hole transport High hole mobility PLED 50-100 nm PVK, TPB, NPB layer (p-doped) OLED LED Hole injection Smooth film-forming, PLED < 100 nm PEDOT, CuPc layer conductive OLED ITO Moderate Anode All devices 15-50 nm PEDOT conductivity, high CNTs work function treatment to achieve crystallinity, which leads to irreversible nanoparticle sintering, migration of atoms like sulphur, and oxidation of metallic ions. Moreover, sol-gel-derived shells suffer from extensive pore networks and cracking left behind by reactants removed during drying and the
72
escape of gases produced by the reaction (Bredol & Dieckhoff 2010). Closing the pores requires high temperature, which leads again to the problems described above. DC-LEDs only require that the nanoparticles be dispersible in the binder polymer matrix, to facilitate energy transfer from the matrix to the particle, allowing current flow and resulting luminescence. Although AC EL has been reported from ZnS:Mn nanoparticles in a typical ACPEL structure by Adachi et al. (2007, 2008), the emission was very weak and required extremely high voltage to induce a strong enough electric field for charge injection into the nanoparticles. Therefore, for a printed device, which by definition uses nanoparticles, the LED structure must be used.
4.4 Characterization
Table 4.2. Characteristic properties of light-emitting devices.
Parameter Symbol Unit Significance
2 Intensity of visible light emission from Luminance L cd/m LED at a given voltage
2 Threshold voltage Vth V Voltage at which 1 cd/m is observed
Voltage at which diode behaviour is Turn-on voltage V V 0 observed
2 Power required to light the device at a Power density P W/m in given voltage over unit area
L ouput per unit power, taking into Luminous efficiency lm/W account the solid angle
Ratio of the device illuminance to Contrast ratio CR unitless ambient illuminance
Length of time “on” before a device Lifetime t h 1/2 loses 50% of its initial L value
Light-emitting devices are characterized using several specific parameters. These parameters define the quality of the visual output, as well as the overall efficiency and durability of the device. There are generally typical values for these parameters which are highly dependent upon the devices’ structures and materials. Table 4.2 provides a summary of these parameters and
73
their significance. All of these properties may depend on ambient conditions, such as temperature and humidity, and device structure, such as layer thickness, so these values are recorded when reporting on the characteristics of the device.
A primary distinction between devices is the matrix type. Some devices are active-matrix (AM), some are passive matrix (PM), and some are simply lights or “lamps”, with no pixels assigned. AM devices have individual circuits behind each luminescent pixel, which can alter colour and brightness – the ubiquitous TFTs. PM devices simply operate by addressing one row and one column at a time, lighting up the pixel where the two cross. Lamps do not have any rows or columns, and the entire luminescent area is one large pixel. The PM architecture is utilized for testing many light-emitting devices, in order to keep the connections to the power source removed from the fragile layers of the device itself. An example of such a setup is shown schematically and photographically in Figure 4.5.
+ -
emissive anode layer stack
Cathode
substrate
Figure 4.5. PM LED array for testing: schematic (left) and setup (right).
74
Chapter 5 5 Approach & method development
The background research into the technique of inkjet printing, as well as the materials and electronic structures best suited to it, strongly suggested certain materials and structures as the best candidates for printing an LED. A detailed study of each ink and device to be created was considered necessary for the realization of this project. However, each material presented unique opportunities for study; some of the materials, such as PEDOT:PSS suspension, for example, were readily available, but had not been comprehensively studied when included in a printable ink. Others, such as quantum dots, were not readily available, and so had to be studied from the ground up, including methods of their synthesis. Thus, the concept of using different materials and electronic devices as “models” for stages of a complete method of developing all-printed electronics was envisioned. The major stages in this method were briefly outlined in Section 1.3. In this chapter, a more comprehensive consideration of the first major contribution of the work is undertaken, discussing the stages of ink development and the “models” used to comprehensively study each of these. A description of the experimental methods used for each study is provided in the following chapter. It must be noted that each of these stages was included in the development of all of the inks; however, the inks which were not the “models” for that stage followed the groundwork laid by the “model” material rather than beginning from scratch.
5.1 Materials selection model: semiconductor
The prevalence of conductive and insulating materials in dispersion, as was described in Chapter 3, meant that these materials were readily available for incorporation into inks. However, semiconducting materials were not similarly available. Also, the requirement for materials with a certain bandgap – either to emit or absorb light at a fixed wavelength, or operate at a certain voltage to behave as a transistor – meant that in-house tailoring of such materials was preferable to using commercial substitutes. Perhaps most importantly, direct control over particle size and dispersibility could be exercised if semiconductors were prepared in the lab. Therefore, the use
75
of the semiconducting species as a model for materials selection/synthesis was a sensible approach. This model study included the following component parts:
i. Development of a method by which liquid-dispersible semiconductor nanoparticles, polymers, or molecules with a desired bandgap could be reliably prepared;
ii. Control of the particle size (where applicable) to <200 nm for clog-free jetting;
iii. Characterization of the relevant property or properties of the semiconductor, such as mobility, PL intensity/spectrum, EL intensity/spectrum, quantum efficiency, etc.
iv. Adjustment of synthesis method or post-treatment to optimize the relevant properties;
v. Dispersion of the material in a liquid carrier and testing of its compatibility with other ink components, at a desired loading in the ink.
5.2 Ink formulation model: conductor
Because many different dispersions of conductive materials exist, many of which were described in Chapter 3, the conductive species served as a better model for ink formulation. Indeed, many conductive species are already included in commercial inks of unknown formulations. So, the steps required to produce such an ink from a ready-made material (like Ag nanoparticles or SWCNTs) were studied using the conductive species as the model.
Ink formulation was an involved and iterative process, originally undertaken after the early stages of the project, when PEDOT:PSS and ATO inks were being used to produce AC-driven PELs on paper. It was noted that both reformulation of the PEDOT:PSS ink and the use of different substrates, which PEDOT:PSS wet differently, had a major effect on device performance. Therefore, the conductive ink became the first to undergo an in-depth study of its formulation, and the effect of even minor components on functionality. The methods used in this study would then be carried over to all of the other inks which were prepared. The major steps involved in the ink formulation study are as follows:
76
i. Selection of a suitable solvent system; measurement and optimization of viscosity within a target range (2-12 cP for the DMP2831);
ii. Addition of surfactant (or another modifier) to adjust surface tension to an appropriate value for jetting – this may also change viscosity (Jansen et al. 2001), so reexamination of viscosity is appropriate here;
iii. Treatment of ink with other necessary components, such as humectants, biocide, pH buffer, binder polymer, etc. – re-examine viscosity and surface tension and return to Step (i) if no longer suitable for jetting;
iv. Particle size examination and particle stability study (-potential) – again, if unsuitable, return to Step (i);
v. Jetting: optimization of voltage waveform and drop formation;
vi. Print quality: examination of printed patterns and distribution of ink components, where necessary.
5.3 Film formation model: insulator
The sole remaining ink to use as a model was the insulator ink. Therefore, this ink was used to outline the process for optimizing film structure on the substrate, for functional layer deposition. This role was particularly appropriate to this material because any flaws in an insulating film might cause catastrophic failure of a printed device, unlike in the other films where they might simply reduce functionality. Also, a key aspect of film formation – film thickness – was of major concern in an insulating layer, which might prevent sufficient charge carrier transfer if too thick (Ono 1995). The issues with drop spacing, line merging, coffee-ring effect, and so forth that were described in Chapter 2 were studied using this ink as model, and expanded to be carried out on the other printed layers, as overall device smoothness and uniformity was expected to be paramount in determining whether or not a device would function. The steps in the film formation study are outlined as follows:
77
i. Establishment of droplet volume, drop size, and drop topography upon impact with the substrate – if unsuitable, reformulation would be needed;
ii. Using the drop size obtained in Step (i) as a starting point, establish ideal drop spacing to produce uniform films with no holes and the smoothest possible topography;
iii. Establishment of exact film topography using profilometry, microscopy, etc.;
iv. Reformulation of ink to improve topography and film formation if necessary – improvement of leveling, wetting, etc.;
v. Establishment of film thickness as a single layer, and when overprinted several times, and optimization of print parameters to yield films of appropriate thickness for device application
5.4 Multilayer device model: LED
As was mentioned in Section 5.1 above, the initial work conducted in this study was the preparation of an LED on a paper substrate (as shown in Figure 5.1). This was a proof-of- concept approach, to demonstrate that conventional electronic materials and technologies could be adapted to an unusual substrate and processing conditions. However successful this study was, it was still fraught with difficulties in terms of the deposition process. In that work, an ACPEL device prepared by inkjet patterning of the electrodes and Meyer-rod coating of the other two layers was constructed on paper. Both of these deposition techniques are continuous processes that could be readily applied to roll-to-roll fabrication of paper-based ACPELs. They also avoided the limitations and waste associated with the batch process of screen-printing, which is more commonly used (Satoh et al. 2007, Kim et al. 2010). However, they also required masking of the substrate, as coated layers cannot be patterned. They also required two separate unit operations. Both of these issues made fabrication very tedious, and the ease of deposition offered by inkjet printing – which had already been applied to the two electrodes in the ACPELs – was an attractive prospect. Furthermore, the coatings, although viscous and containing large particles, were still in the liquid-phase, and could be used as models for inkjet inks.
78
VAC source
EL emission PEDOT:PSS (printed) ZnS:Cu,Cl/PVDF (coated) BaTiO3 /PVDF (coated) PEDOT:PSS (printed) paper substrate
Figure 5.1. Printed-coated ACPEL on paper. Top: schematic of device. Bottom: Device under driving voltage; off, on, and on (darkened room)
Therefore, a similar, inkjet-printed structure – either an ELD or an LED – was used as the final “model” in the method described in this chapter. The ELD or LED would serve as a demonstration of the potential of inkjet printing an entire electronic device. The utility and visual appeal of such a device also made it ideal as a stepping stone for the development of all- printed electronics. The approach taken to prepare and characterize this ultimate “model” is outlined below:
i. Determination of suitable structures that utilize the materials/inks studied in the previous stages of the project; ii. Deposition/characterization of variants of these devices, using conventional materials (ITO, vacuum-deposited metals) to determine the functionality of individual layers before printing entire ELDs/LEDs.
79
Chapter 6 6 Materials & methods
Production of planar electrolumenscence light-emitting devices using inkjet printing techniques required the formulation, jetting, and functional testing of several materials’ ink on various substrates. Each series of experiments was focused on a particular ink that served as a model species for one of the stages in ink development, as was explained in the previous chapter. Each ink underwent the basic steps of ink formulation, however; the model inks simply expanded upon each stage. The experimental methods used were also somewhat different for each ink, as each ink had a different intended function. Each of the experimental techniques will be outlined in this chapter.
6.1 Materials selection
Materials selection was based upon the typical materials used in ELDs/LEDs (described in Chapter 4) and refined by choosing materials that were also jettable (as discussed in Chapter 3). A summary of the characteristics of these materials follows. All reagents were supplied by Sigma-Aldrich Canada, except where otherwise noted.
Several other materials were considered for use with little success, either due to difficulty of incorporation into an ink, unsuitable processing conditions, non-functionality upon jetting, or a variety of other reasons. In the interest of brevity, these materials will not be discussed here in any detail. Let it suffice to say that materials selection, like ink formulation, is something of an iterative process, and that theoretically suitable materials may not be practically useful. Some of these other materials included colloidal Ag (processing temperature too high), aqueous CdS QDs (unstable in the presence of certain ink additives, especially alcohols), sol-gel-derived emitters and insulators (non-crystalline), and TiO2 (poor insulating performance). TiO2’s relatively low dielectric constant of 86 (Sears et al. 1982) meant that it was not sufficiently insulating to perform well under the conditions of this study.
80
6.1.1 Electrodes
The electrodes were to be inkjet-processable and suitable for curing at a low temperature.
Therefore, the primary conductive species used was aqueous PEDOT:PSS suspension (Mw =
8073, 1.3 w/w% in H2O, PEDOT:PSS ratio = 1:2). This deep blue-black suspension contained PEDOT particles, stabilized by PSS- sodium salt, swollen with water. The dark colour of the PEDOT:PSS was also ideal for establishing ink penetration depth on porous substrates.
As was mentioned in Section 3.1, carbon nanostructures have been incorporated into
PEDOT:PSS layers to improve conductivity. In this work, SWCNTs, MWCNTs, and C60 were all mixed with PEDOT:PSS for this purpose. SWCNTs (1.2-5 nm diameter, 2.5 µm length),
MWCNTs (7-15 nm diameter per bundle, 0.5-10 µm length), C60, and shortened single-walled nanotubes (S-SWCNTs) were used. The ratio of semiconductive CNTs to metallic CNTs was 2:1, as specified by the supplier. The S-SWCNTs were prepared from the as-supplied SWCNTs using a chemical cleaving method described in by Chen et al. (2006). All of the carbon species
(SWCNTs, MWCNTs, S-SWCNTs, and C60) were added to water at 0.04 w/w% and stabilized with 0.5 w/w% sodium lauryl sulfate (SLS) as suggested by Kymakis et al. (2007). Any undispersed agglomerates were removed by centrifugation. These dilute carbon solutions were mixed into the electrode inks as needed.
In some cases, ITO was used as the anode. Ready-made ITO glass slides were used, of two varieties: the first being an unpatterned 1” 1” slide (Rsheet = 70-100 /), and the second being a patterned 2” 2” slide (Rsheet = 15 /, Kintech Company). The patterned slides had 38 strips of ITO, in two columns of 19 each, with dimensions of 1 mm 22 mm per strip, and 1.5 mm strip spacing. ITO PET was also used for some tests (Rsheet = 100 /). Finally, when needed, evaporated Al was used as a cathode material.
6.1.2 Insulators
The high-k dielectric BaTiO3, being a widely-used insulating layer in ELDs, was utilized. The existence of a ready-made BaTiO3 nanopowder (≤ 25 nm particle size) made this a relatively simple material selection to make. Other insulating materials included as binders were methyl
81
methacrylate (MMA), which could be polymerized in-situ into PMMA, and poly(vinyl pyrrolidone) (PVP), with a molecular weight of 1 300 000.
6.1.3 Charge-transporters
Both PEDOT:PSS and PVK were used as charge-transporting species. PVK was used in two forms: from Sigma-Aldrich (Mw = 25 000 – 50 000) and synthesized in-house by Brett Kamino, a fellow student, using a living polymerization method described by Higashimura et al. (1980).
6.1.4 Emitters
The emissive layer was the “model” for materials selection, where a desired functional material had to be tailored to inkjet application by controlling its particle size and solubility/dispersibility. This implied the use of dispersible, nanosized, dispersed semiconductor particles, or a dissolved semiconductive polymer/molecule. Because of its successful application in the printed/coated paper-based ACPEL described in the previous chapter, doped ZnS was the material primarily considered for inkjet deposition. As has been mentioned in Section 3.2, doped ZnS has been widely prepared in a nanoparticulate form, with many of the notable syntheses summarized by Chander (2005). However, inkjet deposition of these nanoparticles has not been as well- documented. Cd-based nanoparticles, which have been inkjet printed, were also considered as possibilities, but their more difficult synthesis processes and aqueous indispersibility were major impediments to their use. Even a surface-functionalized aqueous CdSe QD solution was incompatible with viscosity modifiers/surfactants necessary for printing, rapidly flocculating. The inkjet fabrication of an entire device also precluded the consideration of most organic materials, as printing of several different orthogonal layers would be an extremely difficult task. More importantly, the extreme sensitivity of organic materials to contaminants – such as surfactants and other ink additives – made them problematic for inkjet processing. A single-layer inorganic or hybrid device structure was thus considered ideal.
Firstly, doped ZnS had to be synthesized. The synthesis method, using hydrated acetate salts of 2+ 2+ 2+ 2- Zn , Cu , and Mn as the cation/dopant source and Na2S or thiourea as the S source, is described in detail in APPENDIX A. In order to arrest particle growth and maintain nanoscale
82
sizes for jetting, as well as to passivate surface states to improve PL and EL, different capping agents were used, to varying degrees of success. Some of these capping agents were polymers: sodium hexametaphosphate (SHMP), chitosan, poly(acrylic acid) (PAA), and PVP. The remaining agents were polar molecules: mercaptoethanol, thioglycolic acid (TGA), 3- mercaptopropionic acid (3-MPA), unpolymerized acrylic acid (AA), and citrate ion (from sodium tricitrate). Bare particles, with no capping agent, were also prepared, relying on double- layer repulsion to keep them dispersed (Warad et al. 2005). Capping agents likely to yield a transparent, nearly monodisperse suspension were considered preferable, as they would be more thermodynamically stable (Capek 2006); prior work mentioning monodispersity was of particular interest to this study. Table 6.1 lists the capping agents used in this study and their relative amounts in the reaction solution.
Table 6.1. Capping agents for ZnS:Mn nanoparticles, aqueous synthesis.
Capping agent Zn2+:cap in solution Source
SHMP 10:1 (w:w) Warad et al. 2005
PVP Manzoor et al. 2003 10:1 (w:w) Mw = 10 000 Porambo & Marsh 2009
Citrate ion 5:3 (w:w) Peng et al. 2005
Chitosan 20:1 (w:w) Warad et al. 2011
Mercaptoethanol Not specified: 1:3 used (mol:mol) Vogel et al. 2000
TGA 0.5:1 (mol:mol) Zhang & Lee 2010
Klausch et al. 2010, Schrage et al. 3-MPA 1:3-4 (mol:mol) 2010, Zhuang et al. 2003
AA 1:36 (mol:mol) Konishi et al. 2001
PAA Konishi et al. 2001 ~4:1 (w:w) Mw = 100 000 Hwang et al. 2005
83
It is worthwhile to note that each of these capping agents provided water dispersibility, allowing particles to be synthesized in an aqueous medium. However, the particles were then only water- dispersible after synthesis. In order to produce organic-dispersible ZnS, the capping agents had to be chemically altered. This was not feasible in all cases, but in species with a reactive acid or alcohol group in solution, phase-transfer was accomplished by reaction of the ligand with octylamine, producing a long lipophilic tail. This phase transfer has been previously described by Klausch et al. (2010).
Besides the capping agent, other variables were considered during synthesis. One of the more important of these was dopant concentration. For example, 1.5 at. % Mn2+ has been shown repeatedly to be the optimum loading to achieve the maximum emission intensity from the nanoparticles, whereas for Cu2+ doping, 1 at.% is desirable (Chander 2005). However, dopant ions primarily populate the surface of the resulting nanoparticles (Igarashi et al. 1997, Adachi et al. 2007, 2008; Yu et al. 1996, Bulanyi et al. 1998, 2002), where light emission is quenched at surface states. Light can also be emitted by different mechanisms if excessive dopant is present on the surface, which emit at different wavelengths and intensities than expected. One such mechanism is the Mn–Mn exchange interaction which occurs in ZnS:Mn doped too heavily with Mn2+ (Yang et al. 2003). With aqueous synthesis, the solubility differences between dopant and Zn2+ ions in water has been reported to lead to unpredictable atomic percentages of dopant (Peng et al. (2005). The entire amount of dopant dissolved in the precursor solution may therefore not be effectively incorporated. Similarly, solubility differences between Zn2+ and S2- in aqueous solution may lead to non-ideal ratios of Zn2+ to S2-, where insufficient S2- is present in the particles and deep blue emission dominates from S2- vacancies (Yen 2004). So, different amounts of dopant were added to the reaction solution to determine an ideal doping level. Subsequently, the ratio of Zn2+:S2- in the reaction solution was varied (keeping dopant concentration fixed) to achieve (Zn2+ + dopant):S2- = 1:1. The concentration of each species in the final product was determined using inductively coupled plasma atomic emission spectrometry (ICP-AES). Dried ZnS:dopant nanoparticles were dissolved in 25 v/v% concentrated (15 M) HNO3 and 75 v/v% concentrated (12 M) HCl – aqua regia – at 80°C. 200 mg of the nanoparticles were added to 4 mL of acid mixture, dissolved, and diluted to 1 L with
84
deionized water. The Zn2+, S2-, and dopant percentages of the resulting solution were determined using an Optima 7300 ICP-AES.
Reaction conditions for preparation of doped ZnS were initially ambient, and the nanoparticles were removed from the reaction broth for isolation by centrifugation immediately after mixing the reagents. However, it has been reported that in the synthesis of ZnS nanoparticles, time and temperature play a role in the development of luminescent intensity (Manzoor et al. 2003, Zhuang et al. 2003). Therefore, after having established the optimal proportions of reagents, reaction time and temperature were both varied and optimized.
When the ZnS nanoparticles – using different capping agents, different amounts of dopants, different reaction conditions, and so forth – were prepared, they all possessed several different characteristics which were ultimately important to device function. These were: crystallinity and crystalline structure, PL emission intensity, and perhaps most importantly for printing, particle size. Dried nanoparticles’ crystalline structures were observed using a Philips PANalytical PW1830 X-ray diffractometer (XRD) and compared to reference spectra. PL emission was measured in solution: particles were redispersed in water at 0.1 w/w% using the pigment dispersing additive ZetaSperse 1200 (tetramethyl-5-decyne-4,7-diol-2,4,7,9-propylene glycol), provided as a sample by Air Products (0.05 w/w%). The PL of these solutions was observed qualitatively first using a UVP UVM-57 ultraviolet lamp with a 302 nm emission wavelength. PL intensities and spectra were then quantified using a Perkin-Elmer LS-55 spectrofluorophoto- meter and polystyrene cuvettes. Finally, particle size was measured in four separate ways. It was initially estimated from the XRD spectra using Scherrer’s Equation,
퐾 퐿 = 훽 cos 휃 where L is the particle diameter, is the wavelength of the X-rays (1.5406 Å), is the width of the diffraction peak at the Bragg angle at half-maximum intensity, K is 0.9 and is the Bragg angle at which the peak is located (Cullity & Stock 2001). Secondly, particles in aqueous solution were passed through a 0.2 µm syringe filter to determine if they made the cutoff for printing on the DMP2831. Filtered solutions were observed under UV to determine if any
85
particles had passed the filter. Thirdly, particle size distributions were obtained using a dynamic light scattering (DLS) apparatus, a Malvern Zetasizer Nano ZS. Lastly, particle size was directly observed using transmission electron microscopy (TEM) with a FEI Tecnai 20, by casting a drop of dispersant-free nanoparticles onto a holey carbon grid.
Particles were also treated with additional capping agents post-synthesis to improve dispersibility. These were AA monomer, PAA, ZnO, TGA, and SHMP. To prepare particles capped with AA polymerized in-situ, 200 mg of dried ZnS nanoparticles were added to 22.5 mL
DI H2O and 5.5 g AA. The resulting solution was heat-treated at 80°C for 24 hours while stirring vigorously to polymerize AA on the nanoparticles’ surface (Konishi et al. 2001, Althues et al. 2006, Liu et al. 2008). Particles capped with PAA were prepared similarly, but 0.32 g of
PAA solution (35 w/w% PAA in H2O) diluted in 5 mL of DI H2O was added instead of AA as described by Konishi et al. (2001), followed by ultrasonication for 2 hours. ZnO-capped ZnS nanoparticles were prepared by adding 200 mg of dried nanoparticles to 200 mL DI H2O, which were dispersed using ultrasonication and vigorous stirring. 10 mL of 0.05 M Zn(NO3)2 × 6H2O solution was then added to the suspension, followed, dropwise, by 10 mL of 0.05 M NaOH (Karar et al. 2004, Jiang et al. 2009). For TGA treatment, particles were dispersed in 0.023 M TGA, and the pH was adjusted to ~9 with 0.5 M NaOH to obtain a nearly transparent dispersion (Yang & Bredol 2008). Cu-based nanoparticles were not dispersible using TGA (it reacted with the Cu dopant), so 1% SHMP was used instead. A detailed description of these dispersion methods can be found in APPENDIX A.
6.1.5 Substrates
Several different substrates were used upon which to build the devices. These included slide glass, ITO glass (patterned and unpatterned, described in Section 6.1.1), cellulose acetate (Avery), and several paper types. The paper substrates were used after observing that the functionality of the ACPELs deposited on paper changed depending on paper type, which was attributed to the interaction of the sheet with the printed electrode layer, as that layer was in direct contact with the paper sheet. Therefore, PEDOT:PSS-sheet interactions were considered in more detail. Both commercial paper sheets and lab-prepared handsheets were used. The
86
detailed handsheet preparation procedure is given in APPENDIX E, along with the handling procedure for commercial sheets. Handling and cleaning procedures for the impermeable (non- paper) substrates are outlined in APPENDIX F. Sheet thicknesses were measured using a TMI micrometer at 10 different points on the sheet. Contact angle was estimated using an aqueous solution of crystal violet dye (test ink) prepared according to TAPPI Standard Method T431 for measuring ink absorbency into paper. The test ink had a surface tension of 62 mN/m. 30 µL of this ink was dropped with a calibrated micropipette onto a handsheet, and the resulting drop was immediately photographed from the side using a Canon Rebel XT-ME DSLR camera with a MP- E 65 mm macro lens. Finally, absorbency of the sheets was observed by measuring the time for complete absorption of a 30 µL sample of the same test ink into the surface. During this test, the samples were placed directly under a 60 W incandescent lamp elevated 30 cm from the test specimen’s surface. Complete absorbency was defined according to Test Method T431 as the point at which light reflection from the droplet on the surface was no longer visible. A summary of the sheets’ properties is given in Table 6.2.
Table 6.2. Selected properties of commercial substrates.
Absorbency Type Brand Thickness (µm) Contact angle (°) (µL/min)
Avery Inkjet Cellulose acetate 159 26 0 Transparency VWR Plain Slide glass 960 51 0 Microslide Glass: 1090 PET: ITO (glass, PET) Sigma-Aldrich 48 0 140 Epson Premium Photo-paper 259 25 0.63 Photo
In the handsheets, different components were added to examine their effect on the performance of printed films on their surface. These included filler (TiO2), fixation agent (PDADMAC), internal sizing (alkylketene dimer, or AKD), and filler retention aid (poly(ethyleneimine), or PEI), as well as two fibre types – kraft chemical hardwood (HW) and softwood (SW). A summary of the handsheets’ properties is given in Table 6.3, and a description of characterization methods used on these sheets is provided in APPENDIX G.
87
Table 6.3. Selected properties of lab-made handsheets.
Pulp Filler (w/w%) AKD PDADMAC Thickness (µm) Contact angle (°) Absorbency (µL/min)
93 81 19 102 100 0.34 0 110 59 33 102 108 0.33 124 82 14 SW 119 102 0.46 8.2 113 44 16 109 107 0.37 133 47 18 13.6 111 74 4.1 123 30 14 119 94 1.7 92 52 51 97 100 0.35 0 105 31 43 101 107 0.33 108 50 62 HW 98 114 0.39 10.3 90 40 30 85 102 0.36 108 57 25 16.1 116 92 0.46 139 53 28 127 97 0.37
6.2 Ink formulation
Ink formulation required iterative testing of fluid properties while adjusting the specific components and proportions of those components. Because each ink contained different functional materials, the ink components also varied in order to achieve and maintain dispersion while retaining functionality. The choice of the most voluminous ink component, the primary solvent, had the most bearing on the remainder of the ink components, which had to be chosen to be compatible with that solvent system. Incompatibility was defined as the situation where the addition of a particular component caused either an (undesired) chemical reaction, precipitation of suspended/dissolved material, or liquid-phase separation (i.e. immiscibility). However, the choice of the solvent, was restricted by the necessity for orthogonal solvents in a layer-by-layer
88
printed LED (Figure 6.1). In the most practicable device structures (summarized in Chapter 4), namely the DC-LED or PEL, this implied different choices of solvent. In both cases, the electrode material of choice was PEDOT:PSS, which was water-borne. Therefore, in the single- layer DC-LED structure, the ZnS layer had to be insoluble in water. In the PEL structure, the ZnS layer was more problematic, as the front and rear electrodes were both water-borne. Ideally, the rear electrode would not be printed onto a layer which was water-soluble, as it required many print passes to build up sufficient thickness (i.e. conductivity), which would prolong the time for the PEDOT:PSS ink to solubilize the underlying layer. Therefore, an insoluble water-borne ZnS layer and an organic-borne BaTiO3 layer were necessary for a quasi-orthogonal structure. To avoid the ZnS layer solvating the water-soluble PEDOT:PSS anode, non-printed ITO was considered as an alternate anode.
PEDOT:PSS aqueous PEDOT:PSS aqueous BaTiO3 organic organic ZnS + conductivebinder ZnS + insulating binder aqueous PEDOT:PSS aqueous PEDOT:PSS or ITO aqueous substrate substrate
DC-LED DC* or ACPEL
*No BaTiO3 layer necessary in the DCPEL.
Figure 6.1. Solvent selection for different device component inks, to produce an orthogonal structure.
These considerations necessitated the formulation of an aqueous PEDOT:PSS ink, an organic
BaTiO3 ink, and both an organic and an aqueous ZnS ink, containing a conductive and an insulating binder, respectively.
With the solvent systems so defined, the remaining ink components were added to adjust the fluid properties of the ink. The necessary ink properties for jetting on the testbed of choice, the DMP2831, were given in Chapter 2. Many different reagents were used to adjust the relevant properties of the inks to the acceptable values summarized in Table 2.2, including multiple solvents, surfactants, and stabilizers. The exact materials used for each ink will be described in the following sub-sections. As these materials were added, the fluid properties were monitored using viscometry, tensiometry, zeta-potential analysis, and particle size analysis.
89
Prior to testing, inks were ultrasonicated in an ice bath to thoroughly mix their components. Viscosity was measured using a capillary viscometer (Kimax, size 100, grade B17) at 25°C. Equilibrium (not dynamic) surface tension was measured using a Sigma 700 Wilhemy platinum plate tensiometer (KSV Instruments), at 25°C. Particle size was determined using a DLS apparatus (Brookhaven Instruments) and a 2 W Ar laser, after 10 dilution to allow the laser to pass through the darker-coloured inks, such as PEDOT:PSS. DLS was used for particle sizing because the drying of some ink components – particularly surfactants – on a TEM grid would obscure the particles of interest. Stability of the similarly-diluted inks was observed using a Brookhaven ZetaPlus zeta-potential analyzer. Stability was also qualitatively observed by monitoring the formation of precipitates either over time or immediately upon addition of an incompatible ink component.
6.2.1 PEDOT:PSS inks
PEDOT:PSS served a similar role to ZnS’s role in materials selection: it was a model species for ink formulation, upon which the ink formulation method was developed and applied to the other inks. Therefore, the effect of precise adjustments to a formulation on its functional performance was studied in greater depth for the PEDOT:PSS inks. Also, the indication of its printed performance – electrical conductivity – was measured, as described later in this chapter.
PEDOT:PSS inks were formulated using previously described aqueous PEDOT:PSS suspension as the basis. The viscosity of this suspension was 14.4 cP, and its surface tension was slightly lower than that of water (61 mN/m). Therefore, it required treatment with both diluent and surfactant. Inks were diluted with deionized water (the primary solvent) and DMSO to control viscosity, improve conductivity, and to incorporate a co-solvent to avoid undesirable effects in film topography, such as the coffee-ring effect. Water and DMSO loadings were systematically varied to obtain a suitable viscosity for printing and maximum conductivity enhancement from the DMSO. To prevent premature drying of the ink in the printhead nozzles, glycerol was added to the mixture as a humectant. Glycerol has also been shown to be a mild conductivity enhancer for spin-coated films of PEDOT:PSS (Lia et al. 2003). To adjust surface tension for improved
90
jetting and substrate wetting, several surfactants were added to the above ink formulation to provide several ink types. SLS, Triton X-100 (Union Carbide, laboratory grade), Zonyl FS-300 fluorosurfactant (E.I. DuPont de Nemours et. co.), Igepal CA720 (General Dyestuff Co.), and ZetaSperse 3700 (Air Products) were added to the ink in amounts ranging from 0.01 w/w% to 2.0 w/w%. An ink containing no surfactant was also studied. Cationic surfactants were not studied; the addition of benzalkonium chloride caused flocculation through suspected reaction with the PSS- anion. Each surfactant was loaded into the ink until its critical micelle concentration (CMC) was reached. The CMC was estimated in the PEDOT:PSS/water/glycerol/DMSO system as the concentration above which further addition of surfactant yielded no decrease in surface tension. Surfactant CMCs (in this system) and their respective inks are summarized in Table 6.4. All inks were treated with Surfynol DF-110D defoamer (Air Products) at 0.5 w/w% after final formulation to control foaming caused by the surfactants.
Table 6.4. Surfactants tested in PEDOT:PSS ink and their CMCs.
Surfactant Type CMC (w/w%)
None n/a n/a Triton X-100 Non-ionic 0.1 Zonyl FS-300 Fluoro- 0.02 Igepal CA720 Non-ionic 0.2 ZetaSperse 3700 Anionic 0.2 SLS Anionic 0.5
As was mentioned in Section 6.1.1, carbon species (SWCNTs, S-SWCNTs, MWCNTs, and C60) were used as conductivity enhancers. The 0.04 w/w% solutions of these materials were also added to the PEDOT:PSS inks, displacing water (the C solutions were mostly water). Loadings from 0-10 w/w% C solution were tested in the PEDOT:PSS inks.
The effect of different components on the conductivity was considered in several ways – as a function of drop formation, print quality, and chemical interaction. These were each examined in turn using the printer-mounted camera, optical microscopy (Leica DM-LA), and Raman spectroscopy (Raman Micro 300). Alterations in conductivity were predicted based on changes in the distribution of double bonds in the PEDOT structure, and consequent shifting of the
91
primary PEDOT peak at Raman shift 1450 cm-1, (Ouyang et al. 2004, 2005), or changes in physical conformation of PEDOT chains (Fan et al. 2008).
6.2.2 BaTiO3 ink
The BaTiO3 ink was organic-based in order to be suitable for use in the ACPEL structure. The
BaTiO3 nanopowder (≤ 25 nm particle size) was dispersed in a several solvents, including water, isopropanol, ethanol, MMA, and ethylene glycol, of which the best-suited organic solvents were ethanol and MMA. Several dispersants were also tested, including oleic acid, terpineol, Disperbyk 111 (Byk Chemie), Surfynol CT-324 (Air Products). , and ZetaSperse 1200 (Air Products). A mixture of Surfynol CT-324 and Disperbyk 111, both dispersants for inorganic oxides, was used to aid in dispersion. Methyl methacrylate (MMA) was also added to the ink to be polymerized in situ into an insulating binder after printing. Finally, as a humectant and viscosity increaser, poly(ethylene glycol) with an average molecular weight of 300 (PEG 300) was added. The low surface tensions of the organic solvents meant that no surfactant was necessary in this ink. The use of MMA precluded the use of immiscible water-based materials. A dissolved binder polymer was not used as it was expected that a relatively large amount (>5 w/w%) of BaTiO3 would be required in the ink to demonstrate acceptable insulating performance, and adding a dissolved polymer would likely be problematic for jetting.
6.2.3 ZnS inks
Where ZnS nanoparticles have been used in electroluminescent devices, they are generally loaded into the precursor solvent for spin-coating at a concentration of ~1 w/w% (Yang et al. 2003, Schrage et al. 2010, Hieronymas 2002). Therefore, this concentration was targeted as an optimal value for the inks. It is worth noting that as the ink is not composed entirely of colloidal suspension but also of additives to control viscosity, surface tension, and drying rate, the “stock” ZnS colloidal suspension forming the basis for the ink would necessarily be of a higher concentration than 1 wt. %. The functional material, ZnS, was dispersed after synthesis either into toluene, using oleylamine and 3-MPA, or into water, using TGA and NaOH (see
92
APPENDIX A and Section 6.1.4). These dispersions were stable when they contained 2.5 w/w% ZnS. Therefore, the inks had to contain 40 w/w% ZnS “stock solution” as a starting point.
The organic ZnS ink was loosely based on a model CdS ink described in a patent by Cho & Cha (2009). The patent suggested the use of a high-boiling organic solvent such as toluene or chlorobenzene as a primary solvent, a low-boiling co-solvent like tetrahydrofuran (THF) or cyclohexane, and a viscous surfactant to adjust surface tension and viscosity. So, the organic ZnS ink used chlorobenzene as a primary solvent, and cyclohexane as a co-solvent. Because a conductive binder was required, PVK was dissolved into the ink at a low concentration (<0.1 w/w%). The dissolved polymer was also expected to increase viscosity. The choice of this polymer was important for device function: without dispersibility in the polymer layer, energy transfer between the binder matrix and the nanoparticles is limited or non-existent (Schrage et al. 2010, Hieronymas 2002). Again, no surfactant was required, due to the already-low surface tension of the organic solvents.
The aqueous ZnS ink was formulated with a similar template to the organic one. Unlike PEDOT:PSS, the viscosity of the ZnS stock solution was not already high – it was approximately that of water. Therefore, co-solvents like glycerol and PEG were considered. Ethylene glycol butyl ether (or butoxyethanol), with its boiling point of ~170°C, moderate viscosity of 3 cP, and low volatility, was an ideal candidate. Viscosity was also strongly affected by the presence of the dissolved polymer, which was the insulator PVP in the case of the aqueous inks. In order to avoid issues with “bead-on-a-string” jetting, a low concentration (<1%) of PVP (Mw = 1 300 000) was used. To reduce surface tension, a surfactant was needed – however, the tendency of surfactants to interact with suspended particles and polymers (Jansen et al. 2001) was of concern in an ink containing both. So, isopropanol was introduced to control surface tension (Vazquez et al. 1995). This avoided issues with foaming and introduced a third solvent (along with water and butoxyethanol) to further limit coffee-ring formation.
93
6.3 Jetting
Finalized inks were ultrasonicated in an ice bath for 30 minutes, filtered at 0.45 µm and then at 0.2 µm, and injected with a syringe into 1.5 mL DMP cartridges. Using the Dimatix Model Fluid Waveform as a starting point, drop formation from the nozzles was observed using the Drop Watcher software and camera included with the DMP2831 printer. Drop formation was controlled by adjusting the voltage waveform applied to the piezoelectric nozzles – see APPENDIX J for the jetting waveforms. In certain cases, an ink formulation failed to jet at all and the formulation step had to be repeated. As in the previous section, for the sake of brevity, these attempts are not described here. Ideal drop formation, described in Section 2.2.2, was the goal. Once a suitable waveform was found, the ink was ready for jetting and observation of jetted drops. All inks were jetted from a print height of 1 mm onto the substrate, and dried on a hotplate at 150°C for 30 minutes.
Because particle size was carefully controlled during the formulation stage, theoretically none of the functional materials were removed during filtration. However, the long CNTs were hypothesized to have been at least partially removed. Passage of the as-supplied SWCNTs and MWCNTs through the filters and print nozzles was evaluated using UV-visible spectrometry. For filtration tests, the aqueous solutions of CNTs/SLS were passed through a 0.2 µm nylon filter. For print nozzle tests, the previously filtered solutions were jetted using the DMP2831 at a jetting frequency of 8 kHz into a collection well for 10 minutes. Changes in CNT concentration across these steps were evaluated by obtaining absorbance spectra of the filtered and jetted CNT solutions with a UV-visible spectrometer (Perkin-Elmer Lambda 20), utilizing the strong visible- range absorption of CNTs at an arbitrary wavelength of 500 nm.
Isolated CNTs which passed filtration/jetting were observed using scanning transmission electron microscopy (Hitachi HD-2000 STEM), using dried droplets of CNT solution on copper grids. It was also expected that zeta-potential (ζ) of a CNT dispersion would increase in absolute value as more CNTs were removed and the stability of the dispersion thereby increased, stability often being closely related to the concentration of the dispersed phase (Tadros 1987). A certain absolute value of ζ – usually 30 mV (Lin et al. 2010) – can be considered to represent a stable
94
colloidal solution, although a lower value of 25 mV has been reported (Kim & Ma 2011). ζ of the filtrates and jetted solutions was expected to become increasingly negative in this case as CNTs were removed, due to the anionic surfactant.
6.4 Drop spacing and film formation
Drop size and spread had a bearing on how films formed (see Section 2.4.2). Drop size testing was performed by printing a spread of droplets of each ink at a spacing of 254 µm onto each substrate. Films of the finalized inks were to form functional layers in an eventual working device. Therefore, their wetting performance was important not only on a model substrate like glass, but also on the surface of other materials. For example, the emissive layer was to rest on PEDOT:PSS or ITO. So, glass substrates covered in each of these materials (except ITO) were prepared by spin coating the inks (2000 RPM, 30 s). The resulting drops were imaged (without drying) on the substrate using the optical microscope and their diameters measured.
With the drop size established, the drop spacing can be determined. Using a single nozzle, several lines of each ink were printed onto each substrate at different drop spacings in decrements of 5 µm, starting at the drop size measured in the previous step. The resulting films were observed using the microscope. Ideal drop spacing was considered to be when lines were fully merged (i.e. no holes) but not overlapping, as discussed in Section 2.4.2 and APPENDIX D.
Using the optimum drop spacing, films of 1-10 layers of each ink were printed on each substrate and dried as described above. The films were then characterized for thickness and topography using a Veeco WYKO optical profilometer at the edge of each film. The detailed procedure for measuring thickness from the WYKO profiles is given in APPENDIX C. Given the roughness of the printed films, thickness of film thickness varied from “valley” to “peak”. For the sake of consistency, thickness was measured in the valleys of these layers. The optical profilometry measurement also gave an idea of film morphology and the presence of pinholes, which could compromise film function.
95
6.5 Ink distribution and print quality
A major factor in the electrical performance of any functional layer is its contiguity. As was discussed in Chapter 2, inkjet printing produces sequential drops which form lines or blocks, repeating this mechanism until the print is finished. The contiguity of the printed patterns in this study was dependent on the properties of the ink, the nature of the substrate, and the successful establishment of drop spacing. The patterns were optically examined for print quality in terms of edge resolution, the presence of line overlap or inter-line holes, and minimum feature size. To observe the print quality of ink layers, images were taken of the layers on their relevant substrates using both the Canon Rebel XT-ME DSLR (MP-E 65 mm macro lens) and the Leica DM-LA optical microscope. In the case of the PEDOT inks, which were in direct contact with the substrates (including paper), a closer examination of the distribution of conductive material was undertaken, in order to better understand the mechanism of conductivity change between different substrates. These ink layers were examined using time-of-flight secondary-ion mass spectrometry, or ToF-SIMS, which provided a molecular map of the substrate surface. An ION-
ToF ToF-SIMS IV apparatus was used to perform the measurements. A Bi3 primary ion gun was used to induce ion ejection and fragmentation. The detailed ToF-SIMS analysis procedure is described in APPENDIX K.
6.6 Functional testing of individual layers
6.6.1 Conductive ink
The main property of concern in the case of the conductive ink was conductivity. This was not directly measured; resistance was instead measured using a multimeter, and by determining the physical dimensions of the layer, conductivity could be obtained. Square samples were printed and bus bars of carbon paste (DuPont MCM) were applied to two of their sides. 2-point measurement across the sample yielded bulk resistance (R). This value is independent of sample geometry. Therefore, to obtain a more meaningful figure – conductivity – resistance was used to calculate resistivity (), the inverse of which is conductivity (). Resistance is related to conductivity by the following expressions (Heaney 2000):
96
푅푤ℎ 1 휌 = 휎 = 푙 휌 where w, h, and l are the width, thickness, and length of the conductive layer, respectively. With a square sample, w = l and = (Rh)-1. Therefore, conductive layer thickness was required to estimate conductivity. This was obtained by cross-sectioning or profilometry of the printed samples. A detailed description of these methods is provided in APPENDIX B and APPENDIX C.
When using paper as a substrate, it is worth noting that “conductivity” in the case of a void- filled, absorbent substrate, containing macropores, micropores (in the fibres themselves), and hygroscopic, hollow fibres, was an estimate at best. Conductive material was not present as a film, per se, but rather as a layer coated onto the fibres, and occasionally filling the voids between them. Furthermore, the “film” was filled with a large volume fraction of non- conductive regions – fibres, filler, and empty voids. Surface resistance/sheet resistance might appear to have been a better metric by which to measure the electrical properties of such sheets, as a result. However, the penetration of the ink into the bulk of the sheets suggested that not only the sheet surface was involved in the conduction of electricity. Rather, the conductive volume was less of a film and more of a composite block. For this reason, the bus bar carbon material was diluted with DMSO to facilitate deeper penetration into the sheets and permit conduction over the entire volume.
6.6.2 Emissive ink
The first means of ensuring that an emissive species was present and still functional in an ink was observing its PL. This was observed using the aforementioned UV lamp (302 nm excitation) and quantified using the fluorimeter, for inks in solution. Both PL and PLE (PL excitation) spectra were obtained for the inks, as well as time-resolved PL, to confirm that emission occurred from the dopant centres (Bube 1953). PL was also observed after printing and drying of the ink. However, although this confirmed the presence of successfully synthesized ZnS, it did not establish whether or not it was electroluminescent (the characteristic of interest
97
for an LED). In order to establish that the as-synthesized ZnS quantum dots were indeed electroluminescent, the “stock solution” dispersions, 2.5 w/w% ZnS in water or toluene, were diluted to 1 w/w%, mixed with 0.1 w/w% of their binder polymers (PVP and PVK, respectively), and spin-coated onto unpatterned ITO slides using a Laurell WS-400-6NPP 150 mm spin coater. These were spun at 2500 RPM for 15 seconds, followed by 200 RPM for 45 seconds, and dried at 150°C for 30 minutes. 100 nm-thick circular Al cathodes were applied on top of the spin- coated films by vacuum deposition. EL was observed by connecting the ITO and Al electrodes to a power source (DC in the case of PVK-bound ZnS, and both AC and DC for PVP-bound ZnS) and increasing voltage. The DC power source was an MPJA 0-50V, 3A benchtop power supply. The AC source was built in-house; it was capable of delivering 50-200 VAC and 0.1A at 160 Hz frequency, using a near-square waveform with rise and fall times of <100 µs, a period of 16 ms, and a pulse width of 8 ms. Inks in devices displaying EL were considered suitable for printed device fabrication.
6.6.3 Insulating ink
The primary measure of an insulating film’s function is its dielectric constant. This is closely related to the film’s geometry and capacitance. BaTiO3 ink was printed onto ITO glass with different numbers of overprinted layers to observe film structure, thickness, and capacitance. Samples had to be dried at 250°C for 60 minutes on a hotplate in air to ensure the evaporation of all of the glycol and to polymerize the MMA. Capacitance of the resulting films was estimated by first applying an Ag electrode using an Ag pen, covering the surface of the samples. Then, capacitance was measured between the Ag and the ITO using an Agilent U1701A capacitance meter, treating the sample as a parallel plate capacitor. Dielectric constant () was estimated from capacitance according to the following formula:
퐶푑 = 휀0퐴
98
where C is capacitance, d is the distance between plates (film thickness), o is the vacuum permittivity, and A is the sample area (ASTM International 2004). In this case, the sample area was the size of the printed samples, which were 2 mm 2 mm squares.
6.7 Device fabrication and testing
PEDOT:PSS 1 µm Al 300 nm Al
ZnS:dopant/PVK 200 nm ZnS:dopant/PVK ZnS:dopant/PVK ITO ~100 nm PEDOT:PSS 100 nm PEDOT:PSS substrate substrate substrate
DC-LED
Al 100 nm Al PEDOT:PSS 1 µm
BaTiO3/PMMA 1 µm BaTiO3/PMMA BaTiO3/PMMA ZnS:dopant/PVP 20 µm ZnS:dopant/PVP ZnS:dopant/PVP ITO ~100 nm PEDOT:PSS 100 nm PEDOT:PSS substrate substrate substrate
ACPEL
PEDOT:PSS 1 µm Al 100 nm Al
ZnS:dopant/PVP 20 µm ZnS:dopant/PVP ZnS:dopant/PVP ITO ~100 nm PEDOT:PSS 100 nm PEDOT:PSS substrate substrate substrate
DCPEL
Figure 6.2. Light-emitting device builds prepared in this study. Layer thicknesses were based on the following: Al cathode, Hieronymas (2002); ZnS/PVK, Hieronymas (2002), Schrage et al. (2010), and Yang et al. (2003); BaTiO3 and ZnS/PVP, Adachi et al. (2007, 2008); PEDOT:PSS anode, Yang et al. (2003). ITO thickness was specified by the supplier. PEDOT:PSS cathode thickness was arbitrary.
The device structures most suitable for the materials and printing method – the DC-LED and the two PEL structures – have already been discussed in Section 6.2 on ink formulation. The simplicity of these structures and their incorporation of binder polymers (rather than self-
99
assembled or monolithic layers) made them ideal for inkjet deposition. In order to minimize the number of variables which might affect device function after printing, glass was used as the substrate for all the device builds.
The two electrodes of the printed devices could be deposited conventionally as well – i.e. ITO as the anode (as supplied), and metal (typically Al) as the cathode, using vapour deposition. So, DC-LED devices utilizing either one printed layer (the ZnS/binder), two printed layers (PEDOT:PSS anode and ZnS/binder), or all three layers printed were prepared. ACPEL devices were prepared similarly, as were DCPELs. The resulting printed structures and the thicknesses of their layers are shown in Figure 6.2. The layer thicknesses for DC-LEDs were based on those described by Hieronymas (2002), Schrage et al. (2010), and Yang et al. (2003). The layer thicknesses for the PELs were based on those described by Adachi et al. (2007, 2008). Layer thickness was controlled by setting the number of print passes, based on the thickness of individual films determined by optical profilometry. These previous works also utilized ZnS nanoparticles as the emissive species. Each of the layers of each device, besides the ITO or vacuum-deposited Al, was printed using its optimum line spacing, with thickness controlled by the number of layers deposited (see Section 6.4). Layers were all dried at 150°C for 30 minutes in air and the full devices were again dried at 200°C for 30 minutes to ensure the removal of any remaining solvents. Figure 6.3 shows a bird’s-eye view of a device being built.
ZnS/binder
BaTiO3/PMMA PEDOT: or PSS Al PEDOT: substrate 1) Print PEDOT:PSS anode, dry 2a) Print ZnS/binder 3) Print PEDOT:PSS cathode, (or prepare ITO) 2b) Overprint BaTiO3/binder dry (or vacuum-deposit Al)
Figure 6.3. Schematic drawing showing a bird's-eye view of device construction.
100
6.7.1 Interlayer dissolution
The simplest method for establishing whether overprinted layers dissolved one another was a solvent rub test, specifically the test outlined in ASTM Standard D5402-06 for the solvent resistance of organic coatings (2006). In this case, the method was adapted to use printed films of each layer, and the solvent rub was replaced with the overprinted ink (e.g. PEDOT:PSS ink films were rubbed with ZnS ink, ZnS films with PEDOT:PSS and BaTiO3 inks, and so forth). Single layers of each ink were printed onto slide glass and rubbed once across with a wipe which had been soaked in the overlying ink (s). Rub resistance was considered acceptable when the ink film remained with no visible damage after the rub. Any visible damage to the film constituted a failed rub test. In the case of the aqueous ZnS/PVP ink, overprinting with both BaTiO3 ink and PEDOT:PSS ink (for the PEL builds) was necessary. PVP is highly soluble in many solvents; even the orthogonal BaTiO3 solvent system contained ethanol, which solvates PVP. To prevent this from occurring, the aqueous ZnS films containing PVP were UV-treated to induce cross- linking to prevent their dissolution by overprinted layers according to the method described by D’Errico et al. (2008). UV treatment was carried out with a Trojan UV lamp (254 nm) over 4 hours to achieve complete cross-linking without decomposing the samples.
6.7.2 Electrical characterization
For the DC-driven devices, a programmed DC power source (Keithley 647) and its coupled lightmeter (Minolta LS-110) were used to detect I-V characteristics, luminance, and efficiency simultaneously on a HP 4140B multimeter and its accompanying software. Voltage was increased at 0.1 V/s to a maximum of 50 V, with a maximum drawn current of 10 mA. AC- driven devices were connected to the lab-made AC power source described in Section Emissive ink6.6.2. Because this voltage source did not have a coupled lightmeter, luminance was measured with a Sekonic Flashmate L-308S light meter at a distance of 1 cm directly above the sample in a darkened room.
101
Chapter 7 7 Results & discussion 7.1 ZnS synthesis
ZnS nanoparticles were prepared based on competitive precipitation method with the reaction of hydrated acetate salt of zinc and Na2S, Na2S2O3, or thiourea, using two different dopants. The first was Mn2+, producing ZnS:Mn, which emits in the orange range at 585-595 nm (Brus 1986). The second was Cu2+, producing ZnS:Cu, which emits in the green range at 470-480 nm (Bowers & Melamed 1955). Doping with Ag+ to produce blue-emitting nanoparticles was attempted, but the reaction mixture inevitably formed insoluble Ag2S, which precipitated out and was presumably not incorporated into the ZnS crystal lattice. As is described in APPENDIX A, the synthesis for both Mn- and Cu-doped nanoparticles was similar; for the sake of brevity, a study of the synthesis of Mn-doped nanoparticles is described in this section in more detail. For clarification, it should be noted that atomic percent (at.%) refers to the percentage of Zn atoms substituted with Mn atoms in the ZnS lattice. In a material like ZnS, with one atom of Zn per atom of S, at. % and mol % are interchangeable terms. The at. % Mn2+ reported here represents the mol % of Mn2+ in the ZnS:Mn theoretically yielded by the reaction.
7.1.1 Mn2+ loading
Synthesis of ZnS with several different capping species, including citrate, chitosan, and PVP, at the suggested Mn2+ loading of 1.5 at.% (in the reaction solution), resulted in weak PL. This weak PL from Mn2+ centres was expected to be in the characteristic orange range, corresponding 2+ 4 6 to the energetic transition of Mn from the T1 state to the A1 state (Karlin 2005). In Figure 7.1, weak, red-shifted PL was observed, suggesting both poor Mn2+ incorporation and poor surface passivation. Red-shifting of PL emission in ZnS:Mn nanoparticles due to lower quantum efficiency has been observed in the past (Bhargava et al. 1994); moreover, the lack of a sufficiently passivating layer on the nanoparticles may have led to surface recombination events, and caused a reduction in PL emission energy (i.e. longer wavelength). The occurrence of a similar shift even in the particles passivated with polymers suggested that polymeric capping
102
failed to sufficiently cap the particles. Particle size also appeared to be large, as none of the particles could pass through a 0.2 µm filter, and they presented as a sedimented white precipitate.
PL spectra (stacked plot)
(d)
(c)
(b)
(a) Photoluminescent intensity (arbitrary units)(arbitrary intensity Photoluminescent
530 540 550 560 570 580 590 600 610 620 630 Emission wavelength (nm)
Figure 7.1. PL emission from ZnS:Mn nanoparticles (1.5 at.% Mn). (a) no cap; (b) citrate cap; (c) PVP cap; (d) chitosan cap. Spectra are stacked in order to compare emission at 595 nm.
The PL of the nanoparticles was drastically improved for all the polymeric caps when using higher concentrations of Mn in the reaction mixture, although it was also red-shifted and eventually dropped off as more non-luminescent MnS was formed (see Figure 7.2, Figure 7.3). In general, below 10% Mn in the precursor solution, no notable PL emission was observed, and PL intensity did not peak until 50% Mn or higher at 608 nm. Stronger luminescence in the red region at higher loadings was attributed to Mn-Mn interactions, as well as surface recombination events resulting from an excess of Mn2+ ions and MnS on the surface of the nanoparticles (Yang et al. 2003). The high surface to volume ratio of the nanoparticles would favour the population of the surface with a large number of MnS molecules and Mn2+ ions in highly Mn-doped precursor solutions.
103
1 100% No capping agent
80% 60% 50% 40% 20% 10% 5%
2% Photoluminescenct intensity, normalizedintensity, Photoluminescenct
0 520 540 560 580 600 620 640 660 680 Emission wavelength (nm)
1
100% PVP cap
80% 60% 50% 40% 20% 10% 5%
2% Photoluminescenct intensity, normalizedintensity, Photoluminescenct
0 520 540 560 580 600 620 640 660 680 Emission wavelength (nm)
Figure 7.2. PL intensity vs. Mn added (at.%), comparing uncapped and PVP-capped particles. Spectra are normalized to the highest-intensity PL emission (100 at.%).
104
UV Visible
0% 0% 1.5% 10% 20% 50% 100% 1.5% 10% 20% 50% 100% Mn loading
Figure 7.3. PL of uncapped ZnS:Mn nanoparticles suspended in water under 302 nm UV excitation.
The actual loading of Mn in the particles, measured using ICP-AES, confirmed that the peak PL at ~608 nm in the orange range (i.e. not red-shifted to 630 nm, as emission was in the case with loadings > 50%) occurred when Mn loading was ~1.5% (Figure 7.4). As expected from the PL spectra, the actual Mn2+ content of the nanoparticles was significantly different from that added during synthesis (Table 7.1). The increased solubility and higher activity of the Zn2+ ion relative to the Mn2+ ion made Mn2+ incorporation difficult in an aqueous solution at room temperature. The inclusion of citrate, which has been reported to assist in solubilizing Mn2+ (Peng et al. 2005), naturally led to higher concentrations of Mn in the citrate-capped particles.
Table 7.1. Actual Mn content of ZnS:Mn nanoparticles determined by ICP-AES.
Mn2+ retained in particles (at. %) Mn2+ added (at. %) No capping agent PVP cap Citrate cap
0.5 0.14 0.02 0.10 1 0.17 0.04 0.20 2 0.30 0.30 0.28 5 0.61 0.62 0.65 10 0.74 0.42 1.95 20 1.24 0.79 6.27 30 0.74 0.85 4.77 40 1.24 1.26 3.46 50 1.47 1.85 6.14 60 2.10 1.82 8.09 80 2.15 2.60 8.96 100 5.10 3.80 10.74
105
It was hypothesized that the inclusion of citrate in the precursor solution may have resulted in nanoparticles so heavily loaded with Mn that non-emissive MnS would form in large clusters rather than incorporating as a sporadic dopant in the ZnS lattice. Emission from Mn-Mn interactions would therefore be very strong, which explained the increasing PL intensity in the red region shown in Figure 7.4. This hypothesis was supported by the XRD studies, presented subsequently, which showed small shoulder peaks in the citrate-bearing nanoparticles’ spectra, as well as by the texture and morphology of the particles. The citrate-bearing ZnS:Mn was not a dry powder even after vacuum drying, but retained a waxy consistency. It was expected that this was a result of the presence of citrate complexes of either Mn2+ or Zn2+, which may have affected results of ICP-AES and PL measurements.
1 No capping agent PVP cap
Citrate cap
None PVP Cit. Photoluminescenct intensity, normalized intensity, Photoluminescenct
0 0.01 0.1 1 10 100 Actual Mn content of ZnS:Mn nanoparticles (at.%)
Figure 7.4. PL intensity at 608 nm vs. actual Mn content. Inset: 50% Mn (added during synthesis) containing ZnS:Mn nanoparticles in water under 302 nm UV excitation. PL intensity is normalized to the highest value (no capping agent, ~1.5 at.%).
106
7.1.2 Zn2+ to S2- ratio
Besides the amount of Mn2+ doped into the nanoparticles, the ratio of Zn2+ to S2- in the precursor solution has been noted to affect the PL intensity and spectrum. Generally, a ratio of Zn2+:S2- of 2:1 has been recommended for optimal particle size and emission intensity (Adachi et al. 2007, 2008; Althues et al. 2006). With the particles produced in this study, the same trend in higher emission intensity was observed for particles with this precursor ratio, as can be seen in Figure 7.5. Nanoparticles prepared with ratios greater than 2:1 generally showed little to no PL emission in the orange range, although they did still emit in the blue range (characteristic of undoped ZnS (Yen 2004, 2006). This was due to the large excess of Zn2+ in solution which consumed the S2- before it reacted with Mn2+; 3:1 retained weak orange PL, but 4:1 and above had insignificant orange PL. Also, the large amount of Zn2+ resulted in numerous S2- vacancies in the particles, which are responsible for blue emission (Yang & Bredol 2008). Nanoparticles prepared with ratios less than 1:1 emitted very weakly in all ranges, suggesting that this ratio was the lower limit to achieve PL in ZnS:Mn prepared in aqueous solution, because insufficient host material (i.e. ZnS) was present to allow PL emission from luminescent centres.
1
0.25:1
1:1 2:1 3:1 5:1
10:1 Photoluminescent intensity, normalizedintensity, Photoluminescent
0 360 410 460 510 560 610 660 Emission wavelength (nm)
Figure 7.5. PL emission in uncapped ZnS:Mn particles (50% Mn), different Zn:S ratios in reaction solution.
107
The difference in emission intensity between 2:1 and 1:1 Zn2+ to S2- ratios was small and within the range of experimental or instrumental uncertainty. However, the 2:1 emission colour was also slightly blue-shifted relative to the 1:1 spectrum. This could be because of the expected effect of Zn2+ on particle size, observed in a previous study (Althues et al. 2006). Excess Zn2+ preserved a smaller particle size in that study, leading to a slight blue-shift in the emission spectrum. Also, the smaller size of the crystallites implied that they were more likely to remain suspended in water (for PL testing), so it is possible that the slightly brighter PL observed for 2:1 Zn2+:S2- nanoparticles was simply because they were better dispersed than 1:1 nanoparticles. Also, the smaller particle size would lead to less diffuse scattering from large agglomerates and a higher perceived PL intensity.
7.1.3 Post-synthesis capping
Regardless of the Zn/S/Mn ratio, all of the particles synthesized with the caps initially tested (i.e. PVP, citrate, and chitosan) displayed red-shifted emission and large particle size (i.e. were not able to pass through a 0.2 µm nylon filter). The red-shift was due to poor surface passivation, which also likely resulted in excessive particle growth. So, other caps were applied after synthesis to see if they could reduce this effect. Capping of the nanoparticles with AA or PAA did have a beneficial effect on PL intensity (Figure 7.6). The effect may have been the result of better passivation of surface states, as was noted in our previous work (Angelo & Farnood 2011) and by other groups using AA (Igarashi et al. 1997, Adachi et al. 2007, Konishi et al. 2001, Althues et al. 2006). AA capping has been highlighted as ideal for ZnS:Mn due to the excitation at 350 nm of C=O bonds which are bonded to S2- in the ZnS structure, resulting in enhanced PL through additional energy transfer to the Mn2+ luminescent centres (Konishi et al. 2001). PAA capping did not show this effect due to the lack of C=O double bonds. So, in this case, where PAA capping improved PL the most of any treatment, it is more likely that PAA simply served as a better dispersant, retaining a larger fraction of ZnS:Mn in solution and increasing apparent emission intensity, or reducing scattering. PAA-capped nanoparticles also emitted with a slight blue shift (to 605 nm) compared to bare ZnS:Mn. However, AA-capped nanoparticles showed a further blue shift (to 595 nm emission) compared to all the other capped particles as well as bare ZnS:Mn nanoparticles. The ZnS:Mn-AA solution was also significantly
108
more transparent than the others (see inset of Figure 7.6). This effect closely resembled that observed with the 2:1 Zn2+:S2- nanoparticles versus the 1:1 nanoparticles. Therefore, it was hypothesized that AA served to better water-disperse and isolate the nanoparticles than the other capping agents.
1 No capping agent
AA cap ZnO AA PAA
PAA cap
ZnO cap Photoluminescent intensity, normalizedintensity, Photoluminescent
0 360 410 460 510 560 610 660 Emission wavelength (nm)
Figure 7.6. Effect of capping agents added after synthesis on PL of ZnS:Mn nanoparticles (Zn:S:Mn = 2:1:0.5, i.e. 50% Mn in solution). Inset: Capped ZnS:Mn nanoparticles in water under 302 nm UV excitation.
The addition of an inorganic shell, ZnO improved PL likely due to, passivation of surface states, as was previously reported (Karar et al. 2004, Jiang et al. 2009). However, reduction of surface recombination events was not sufficient to blue-shift the particles as in the case of PAA or AA. Also, the further addition of PAA and AA over the ZnO layer did not compound the effects of the two caps. PL was quenched somewhat; the reason for quenching of PL due to the ZnO shell was not well-understood, although it was suggested that the presence of ZnO precluded AA or PAA molecules from chemically bonding with the ZnS:Mn metal or S2- ions, thereby diminishing their beneficial effects on PL due to bond excitation.
109
7.1.4 Reaction temperature and time
Most syntheses of ZnS reported in the literature were undertaken at room temperature, and when these procedures were replicated (as described above with PVP, citrate, and so on), they immediately produced visible precipitates of unsuitably large size. However, increased temperature during particle nucleation, along with increased pH and higher reagent concentrations was known to possibly yield smaller nanoparticles (Capek 2006). Therefore, temperature was maintained at an increased value for the duration of the reaction.
PL spectra (stacked plot) 70°C
RT Photoluminescent intensity (arbitrary units)(arbitrary intensity Photoluminescent
520 540 560 580 600 620 640 Emission wavelength (nm)
Figure 7.7. PL spectra of ZnS:Mn nanoparticles capped with SHMP at RT and 70°C, reacted 16 h. Inset shows particles redispersed in water under 302 nm UV excitation. The spectra are stacked to compare emission at 595 nm.
As is shown in Figure 7.7, particles prepared with another capping agent, SHMP, demonstrated greatly increased PL intensity when held at 70°C throughout their synthesis. The synthesis was also carried out for 16 h (see the following section on reaction time). This was likely the result of both better dispersion (smaller particle size) and better Mn2+ integration due to faster diffusion through the boundary layers. More rapid diffusion of Mn2+ towards the particles’ centres also likely reduced the number of surface states, resulting in emission which was not as red-shifted as that observed with room-temperature (RT)-prepared ZnS:Mn. Increases in temperature below
110
70°C did not have much notable effect, and rapid evaporation became a problem at more elevated temperatures. A similar result observed for ZnS:Cu nanoparticles was described by
Klausch et al. (2010).
PL spectra (stacked plot) 0.5 h 1 h 4 h
16 h Photoluminescent intensity (arbitrary units)(arbitrary intensity Photoluminescent
520 540 560 580 600 620 640 Emission wavelength (nm)
Figure 7.8. PL spectra of ZnS:Mn nanoparticles capped with SHMP aged for varying times. All were prepared at
70°C and neutral pH in H2O. Inset shows particles in water under 302 nm UV excitation. The spectra are stacked to compare emission at 595 nm
It had been suggested that increasing the reaction time of doped ZnS nanoparticles leads to major improvements in luminescent intensity (Klausch et al. 2010, Zhuang et al. 2003). This is likely due to the longer time available for dopant ions to diffuse into the bulk of the particle. Allowing longer time and higher temperature to improve diffusion of Mn was necessitated by the solubility difference between ZnS and MnS (Yang et al. 2005). ZnS, having a solubility constant of 1.1 -24 - 10 , will immediately form as insoluble nanoparticle “seeds”, not allowing MnS (Ksp = 2.6 10 13) to co-precipitate inside the crystal lattice. Instead, MnS forms as a surface layer, quenching luminescence. Control of temperature and time may have overcome this effect somewhat. Furthermore, oxidation of the surface by dissolved oxygen may also have passivated the surface
111
(Zhuang et al. 2003); passivation (and therefore capping) playing a major role in development of PL intensity. In this vein, by observing PL emission from ZnS:Mn nanoparticles capped with SHMP at different times after the mixing of the S2- and Zn2+/dopant solutions, the positive effect of longer reaction time on photoluminescence was easily observed (Figure 7.8). Dopant-related orange emission at ~595 nm was drastically increased by aging of the reaction broth before the removal and isolation of the nanoparticles. Also, the emission was initially red-shifted (after 4 hours) due to the presence of a large number of Mn2+ defects on the surface of the nanoparticles (Yang et al. 2003). After aging for 16 hours, the emission was blue-shifted in the more typical orange range, resulting from the quenching of these effects by Mn2+ diffusion into the particle and, possibly, by improved dispersity and reduced particle size. Further aging of the particles led to excessive particle growth and agglomeration, as is predicted by the La Mer model for particle nucleation and growth (the Ostwald ripening phase). After >17 h aging, the majority of the
particles had settled to the bottom of the reaction flask, making them unsuitable for inkjetting.
5:1 Zn:SHMP PL spectra (stacked plot) 10:1 Zn:SHMP 100:1 Zn:SHMP
200:1 Zn:SHMP Photoluminescent intensity(arbitrary units) intensity(arbitrary Photoluminescent
520 540 560 580 600 620 640 Emission wavelength (nm)
Figure 7.9. PL spectra of ZnS:Mn nanoparticles capped with varying amounts of SHMP. Ratios are in wt:wt. All were prepared at 70°C for 16 h in water. Inset shows particles in water under 302 nm UV excitation. Spectra are stacked to compare PL emission at 595 nm.
112
In many of the works describing previous studies of this type, the amount of capping agent used is often left unmentioned. How much capping agent is used can determine PL intensity, as well as dispersion and stability, since it is directly responsible for maintaining dispersion and passivating the particle surfaces (Chander 2005, Caruso 2004). However, it was noticed that using large amounts of any cap led to early precipitation of the particles, and that the solubility limit of materials like SHMP or PVP was quickly reached.
In fact, decreasing SHMP concentration was sufficient to greatly increase PL (Figure 7.9). This came at the cost of dispersion and passivation, as the particles containing little SHMP – i.e. only 1/200 of the Zn2+ loading – began to show red-shifted emission spectra, Emission at 585-590 nm was strongest in particles containing 100:1 weight ratio of Zn2+ precursor to SHMP. The presence of additional capping agent was thought to retard Mn2+ diffusion and incorporation into the nanoparticles, and to prevent S2- from filling sulphur vacancies on the particle surface. This lead to a large number of surface states and enhanced undesirable blue emission. Solubility differences between the component ions, especially when passing through a thick boundary layer of dissolved SHMP, also may have prevented ZnS from crystallizing properly. A similar effect was seen when adding excessive amounts of Zn2+ (in Figure 7.5).
7.1.5 Reduction of particle size & improvement of dispersion
After establishing the ideal reaction conditions, molar ratios, and cap amounts to produce the most highly luminescent nanoparticles, it was found that they were universally still unable to pass a 0.2 µm filter. Particles capped with PVP, chitosan, citrate, PAA, AA, and ZnO were all brightly photoluminescent, but not suitable for printing. Furthermore, the maximum loading of any of the nanoparticles to achieve stability (i.e. no settling or precipitates) was < 1 w/w%. It was evident even during synthesis that the particles were forming large agglomerates that were not redispersible – the reaction solution became opaque and white upon the addition of the S2- solution, instead of transparent and unaffected by light scattering, as would be the case with a monodisperse nanoparticle suspension (Figure 7.10). It was at this point that the addition of a more suitable capping agent was considered, with the highly-reactive mercaptans being prime candidates. As Zhuang et al. (2003) suggested, the use of mercaptans might provide better
113
passivation by actively filling S2- vacancies on the nanoparticles’ surfaces with the thiolated end of the molecule, forming a strong bond.
Therefore, an attempt was made to optimize particle size by using the mercaptan stabilizers. However, the issue of complexes forming between the reactive S-bearing ends of the molecules and the Zn2+ ion before the S2-/dopant sources were introduced was still present. The key to successfully dissolving the mercaptans in the Zn2+/dopant precursor solution and leaving the cations free to react with the S2- source lay in the protonation of the thiol group. According to Adachi (2008), by adjusting the solution pH to a basic value to induce re-protonation of the thiol group, the Zn2+/dopant-mercaptan complex could be dissociated , and as suggested in Klausch et al. (2010) and Zhuang et al. (2003), the Zn2+ complex became soluble and the precursor solution clarified.
Visible
(a) (b) (c) (d) (e) (a) (b) (c) (d) (e)
UV (302 nm) unfiltered filtered (0.2 µm)
(a) (b) (c) (d) (e) (a) (b) (c) (d) (e)
Figure 7.10. ZnS:Mn nanoparticles in water (0.1 w/w%), different stabilizers: (a) none; (b) chitosan; (c) PVP 10000; (d) SHMP; (e) 3-MPA.
Using all three mercaptans (mercaptoethanol, TGA, and 3-MPA) at pH values > 8 (adjusted using NaOH and buffered), reaction mixtures were formed which showed no white colour or agglomeration upon mixing of the Zn2+ and S2- precursors but displayed strong photoluminescence after reaction. 3-MPA was settled upon as the easiest to use of the three
114
stabilizers simply because of its significantly reduced respiratory toxicity compared to the other two. Particles were so well-dispersed that they could not be removed from the reaction broth by simple centrifugation, but rather had to be “crashed” out of solution using acetone, in which the stabilizing ligands were insoluble. Figure 7.10 shows the comparative dispersity and PL intensity qualitatively between the 3-MPA-capped particles and those using the other caps.
7.1.6 Optimized synthesis procedure
Zn(Ac)2 Zn[SCH CH COOH] 2+ 2 2 2 Zn pH = 10.3
Mn2+ Mn(Ac) H2O Mn[SCH2CH2COOH]2 NaOH 2 - (1) CH3COO - CH3COO + - 3-MPA (H )(SCH2CH2COOH )
Na2S pH = 10 S2- S2- H2O pH = 10 Δ + + (2) pH 10 Na Na buffer T = 70°C
2+ Zn O
2+ Zn2+ Mn S OH O O Ac- S2- S2- OH - OH Ac - 2+ ZnS Ac 2+ Mn Zn Δ ZnS ZnS 16 h ZnS:Mn (1) + (2) S2- 2- O ZnS S OH MnS - - OH T = 70°C Ac Ac S2- O HO Zn2+ Zn2+ 2+ O Zn
Figure 7.11. Finalized synthesis method of water-dispersible 3-MPA-capped ZnS:Mn quantum dots.
A finalized synthesis method, to produce these 3-MPA-capped particles, is given in APPENDIX A. The use of the mercaptan species allowed the S-bearing end of the molecule to react with
115
Zn2+ during particle growth, thus filling the S2- vacancies and forming a strong passivating layer, as is shown in Figure 7.11 (Zhuang et al. 2003, Vogel et al. 2000, Small et al. 2010).
Another advantage offered by using a thiolated acid was that the tail end of the molecule, being a carboxylic acid group, was water soluble. Furthermore, the electronegativity of the O atoms present in the group polarized the molecule, providing a highly negatively-charged region around each nanoparticle to prevent agglomeration. Finally, the acid group could be reacted with an amine (in this case, octylamine) to produce a long, lipophilic tail, rendering the particles dispersible in non-polar solvents, such as hexane, chlorobenzene, and toluene (Klausch et al. 2010). A detailed explanation of the phase-transfer process is also provided in APPENDIX A. Improvement in the dispersion of 3-MPA-capped ZnS:Mn versus SHMP-capped ZnS:Mn can be easily seen in Figure 7.12 where both are dispersed using 0.023 M TGA at pH 9.
Visible 302 nm UV
Figure 7.12. Comparison of ZnS:Mn dispersion in water (1 w/w%). In all of the images, the vial on the left contains ZnS:Mn capped with SHMP; the right vial, ZnS:Mn capped with 3-MPA.
7.1.7 Synthesis of ZnS:Cu nanoparticles
An important advantage of this synthesis method was that it was readily applied to another dopant, Cu2+. In experiments conducted with both of Mn2+ and Cu2+dopants, the solution became immediately black and unstable upon the addition of the Na2S solution due to irreversible reaction of these dopants with S2- before incorporation into the ZnS lattice. The introduction of a different S2- source, thiourea was suitable for forming Cu-doped ZnS due to its 2- slower dissociation into S ions (pKa = 2.03) and lower reactivity (Klausch et al. 2010, Schrage et al. 2010). The method used was identical to that outlined in Figure 7.11, with a few notable exceptions.
116
ZnS:Mn emission ZnS:Cu emission ZnS:Mn excitation ZnS:Cu excitation
1
Photoluminescent intensity, normalizedintensity, Photoluminescent
0 200 250 300 350 400 450 500 550 600 650 700 Emission wavelength (nm)
Figure 7.13. PL and PLE spectra of ZnS nanoparticles capped with 3-MPA (2.5 w/w% in water). Insets show 3-MPA- capped particles in water/SHMP or water/TGA/NaOH solutions under 302 nm UV excitation. PLE and PL spectra are normalized to their maximum values to show emission colour clearly.
2- Firstly, the Na2S was replaced with excess thiourea, due to its slow release of S anion. Secondly, the Mn2+ dopant was replaced with Cu2+, also in the form of a hydrated acetate salt. Lastly, the pH was adjusted to 8 and buffered at 8. Finally, based on the work of Klausch et al. (2010), heat treatment was performed at 95°C rather than 70°C to maximize PL intensity for ZnS:Cu. In that work, 95°C was used as a reaction temperature presumably to maximize the diffusion rate of dopant into the nanoparticles, while avoiding boiling of the aqueous solvent. Therefore, this temperature was used for this synthesis as well. ZnS:Cu nanoparticles of similar dispersity to the ZnS:Mn nanoparticles were thus also prepared. The PL spectra of both of these materials dispersed in water are shown in Figure 7.13. Both materials were readily dispersed in
117
water at high loadings (2.5 w/w%) and passed through a 0.2 µm filter with no reduction in PL intensity. The use of TGA/NaOH, that proved to be successful for the dispersion of ZnS:Mn nanoparticles in water, caused rapid precipitation of ZnS:Cu due to the immediate formation of -36 highly insoluble CuxS (Ksp = 1.27 10 ). Hieronymas (2002) suggested the use of polyphosphates to disperse ZnS nanoparticles; based on those proportions and methods, a transparent suspension of ZnS:Cu was readily formed using a solution of 1% SHMP in water.
7.1.8 Characterization of ZnS:Mn, ZnS:Cu nanoparticles
Luminescent emission from doped nanoparticles is slightly different than that from pure materials. While “true” quantum dots like pure CdS emit characteristic colours due to exciton formation across their bandgap, doped materials emit colours associated with certain characteristic energy levels, rather than just the conduction and valence band levels. In ZnS:Mn, the emission results from the d-electron states in the dopant centres interacting with the s-p 4 electron states in the ZnS host, and the resulting transition of an electron in Mn’s T1 energetic 6 state to the A1 energetic state (Karlin 2005). The energy difference between these states is approximately 2.1 eV, resulting in characteristic orange emission at 585 nm. These energetic states both lie within the bandgap of the host material, ZnS. In ZnS:Cu, emission is attributed to 2 the transition of an excited electron from the conduction band to the lower energy T2 state, above the valence band (Srivastava et al. 2010). A simplified schematic of these mechanisms is shown in Figure 7.14. So, although the mechanism is similar to that for excitonic emission across the bandgap, the energy levels between which carriers move are not the same – and therefore, widening of the bandgap due to quantization will not have the same bearing on emission colour as it does in undoped materials.
From Figure 7.13, PL spectra of the ZnS:Mn/water/TGA/NaOH and ZnS:Cu/water/SHMP suspensions peaked at 591 nm and 495 nm, respectively – having bulk emission wavelengths of 585 and 520 nm, respectively. The large blue shift in emission from the ZnS:Cu nanoparticles is attributed to their quantization. As the bandgap widens, the conduction band tends to shift while the valence band remains essentially fixed (Robel et al. 2007). Since the mechanism of emission
118
2 in ZnS:Cu involves carrier transition from the conduction band to the T2 state, movement of the conduction band widens this energetic gap and causes the observed blue-shift. Using the energy quantization expression discussed in Section 3.2.11 for ZnS:Cu, the blue shift corresponded to a bandgap of 4 eV (versus 3.7 for bulk ZnS), or a primary particle size of 2 nm. However, the emission from ZnS:Mn was actually red-shifted. This has been hypothetically attributed to a large amount of either surface states or electron-phonon coupling, in previous studies on ZnS:Mn (Yang et al. 2005). With the polymeric capping agents reported above, where the surface states were not as efficiently passivated, the red shift was more pronounced, as is discussed in Paper 7.
- e e- ZnS:Mn e- ZnS:Cu Conduction band
e- e- Shallow trap states (Zn2+ vacancies)
4 T1 state hv (2.5 eV) hv (2.7 eV) hv (2.1 eV) (tunable)
6 A1 state 2 T2 state
Shallow trap states (S2- vacancies)
Valence band h+ h+ h+
Figure 7.14. Mechanisms of light emission in ZnS:Mn and ZnS:Cu.
The small differences in peak locations for the ZnS:Mn particles dispersed in water vs. NaOH/TGA vs. ink were attributed to improved passivation by TGA. No blue shift in the characteristic blue emission (at ~430 nm) from S2- vacancies on the nanoparticles’ surface was observed either. Indeed, since the Bohr exciton radius of ZnS:Mn is estimated to be approximately 2.5 nm (Bhargava et al. 1994), the particle size of the ZnS:Mn in this study may have been insufficiently small to observe quantum effects. The location of the PLE peak (at 338
119
nm) was similar to that of bulk ZnS:Mn. A brief consideration of the UV-visible absorbance of the nanoparticles yielded an estimated bandgap of 3.71 eV; again, similar to that of bulk ZnS:Mn. Regardless, both the ZnS:Mn and ZnS:Cu particles remained suitable for jetting due to their relatively small size. Although the particles had already passed through a filter and were therefore suitable for jetting, their small size was ideal for clog-free jetting and the production of uniform printed films.
As was mentioned above, PLE peaked at 348 nm for the ZnS:Mn nanoparticles and 322 for the ZnS:Cu nanoparticles, representing a large Stokes shift. For the purpose of obtaining visible emission using an interrogating UV lamp or laser, this large Stokes shift is ideal, as the emission colour is very different from the relatively narrow range of UV wavelengths suitable to excite it, and the excitation spectra are narrow, requiring very specific wavelengths. The blue emission at 2- 2+ 4 6 ~425 nm in the ZnS:Mn nanoparticles, attributed to S vacancies rather than the Mn T1– A1 energy level transition (Yen 2004), was weak compared to the orange emission, suggesting a well-passivated surface with S2- vacancies occupied by 3-MPA or TGA molecules (Small et al. 2010). The extremely bright visible PL from the dispersions supported this suggestion, as PL is quenched by unpassivated surface S2- vacancies – such as that seen in Figure 7.10, where emission from ZnS:Mn not capped with 3-MPA is visibly blue or violet.
Aside from the strong PL emission visible in the 3-MPA-capped nanoparticles, the successful synthesis of the desired cubic ZnS was confirmed by XRD (Figure 7.15). Broadened peaks at 2 ~ 28.5°, 47.5°, and 56.6°, characteristic of nanosized ZnS particles (Takahashi & Isobe 2005, Mu et al. 2005) were clearly visible. According to the Scherrer Equation, the peak widths corresponded to a mean crystallite size of 2.7 nm for ZnS:Mn and 2.3 nm for ZnS:Cu. This and the PL spectra implied that the ZnS:Cu dispersion was actually close to a dispersion of individual primary crystals, while the ZnS:Mn dispersion contained agglomerates of a few crystals.
120
28.5° ZnS:Mn [111] lattice plane ZnS:Cu
Bulk ZnS
47.5° (220) lattice plane
Counts (arbitrary units)(arbitrary Counts 56.6° (311) lattice plane
20 30 40 50 60 Bragg angle, 2 (°)
Figure 7.15. XRD spectra of ZnS nanoparticles and bulk material. Inset: cubic crystal structure of luminescent ZnS.
TEM imaging of the 3-MPA-capped ZnS confirmed the primary crystallite size as approximately 2-3 nm (Figure 7.16). The particles were generally clumped into loose nanometer-sized agglomerates on the TEM grid, where they had likely grouped during the drying process. The dispersion of the particles in solution was obviously not visible using TEM, so DLS was used to establish the degree of dispersion in solution – both aqueous and organic (Figure 7.16). DLS suggested that the particles are present as small agglomerates in solution, although the hydrodynamic radii of the dispersed particles were likely somewhat larger than the particles themselves, due to the presence of the 3-MPA or 3-MPA/octylamine ligands. However, the hydrodynamic particle size was still an order of magnitude smaller than the maximum allowable size (200 nm) for inkjet filtration and printing. This was a marked contrast to the ZnS:Mn nanoparticles capped with the other agents and those prepared using in-situ polymerized acrylic
121
acid (Angelo & Farnood 2011), of which barely, if any, passed through a 200 nm filter and those that did pass through the filter were generally not emissive at the characteristic 595 nm peak for ZnS:Mn. The dispersity and small size of the particles, even if somewhat agglomerated in the ink, allowed for bright PL before and after filtration and jetting.
It may be inferred that the appearance of a white precipitate during synthesis of ZnS precludes the redispersion and eventual jetting of ZnS nanoparticles, as only the 3-MPA-capped ZnS:Mn’s reaction broth remained transparent during synthesis. Successful redispersion of ZnS for inkjet printing where the broth contained precipitate was only reported by Small et al (2010) – however, a mercaptan-derived acid was also used in that case as a stabilizer, although it was added after synthesis, citrate being used during synthesis. Also, in that work, inks did not remain well-dispersed over time, and ZnS:Cu inks succumbed to reaction with the thiol group in the stabilizer (mercaptosuccinic acid, in that case). In this work, the dispersion remained after many weeks of shelf life at ambient conditions; in fact, the DLS measurements shown in Figure 7.17 were actually carried out on ZnS dispersions that were several weeks old. So, through careful study of the reaction conditions and reagents, reliably inkjet-printable ZnS-based luminescent nanoparticles were produced for the first time.
(a)
(b)
10 nm 20 nm
Figure 7.16. TEM micrographs of ZnS:Mn nanoparticles, dried from dispersion in water. (a) Agglomerate, ~10-20 nm diameter; (b) primary crystallites, 2-3 nm diameter.
122
ZnS:Mn
ZnS:Cu Intensity (arbitrary units)(arbitrary Intensity
1 10 100 Hydrodynamic diameter (nm)
Figure 7.17. DLS scans of ZnS:Mn and ZnS:Cu nanoparticles in water (ZnS:Mn) and toluene (ZnS:Cu).
7.2 Other materials (c)
7.2.1 BaTiO3
The crystalline BaTiO3 nanopowder was best dispersed in an ethanol/MMA mixture using a mixture of commercial dispersants, Disperbyk 111 and Surfynol CT324, yielding the particle size distribution shown in the DLS results (Figure 7.18). Similar dispersion was achieved in water with the same dispersants, but the immiscibility of water and MMA precluded its use. Furthermore, the high surface tension of water would have necessitated the addition of a surfactant to render the dispersion jettable. Using ethanol, with an already-low surface tension, avoided the use of a surfactant, which may have adverse effects on dispersion and shelf life of a given ink. According to the specifications of the (a)nanopowder used, this solvent and dispersant (b) mixture produced a nearly monodisperse suspension of BaTiO3. Dispersions loaded with more
BaTiO3 nanopowder demonstrated sedimentation of some of the BaTiO3 over time; the 5 w/w% ink was stable without sedimentation for several weeks. Although some sedimentation in the ink did eventually occur, ultrasonication of the ink for 30 minutes was sufficient to redisperse it and re-establish similar shelf life to that observed in the fresh ink.
123
in ethanol/MMA
in water
Intensity (arbitrary units)(arbitrary Intensity
1 10 100 Hydrodynamic diameter (nm)
Figure 7.18. DLS-obtained particle size distribution of BaTiO3 (5 w/w%) dispersed with Disperbyk 111 and CT-324 in ethanol/MMA and in water.
7.2.2 PEDOT:PSS & CNTs
Although inkjet printing of PEDOT:PSS has been widely reported, the PEDOT:PSS suspension was still tested for particle size to make sure that it would pass through the printer without significant loss of conductive material. Figure 7.19 shows the DLS-obtained PSD of the as- received PEDOT:PSS, confirming that the PEDOT:PSS particles were sufficiently small to meet the 200 nm cutoff for particle size. However, when compared to the other materials, the PSD showed quite large particles, and some degree of clogging was expected when printing. Indeed, it was noted when the inks were finally formulated that PEDOT:PSS did cause more clogging than the other inks; however, the fact that PEDOT:PSS was a polymer rather than a rigid particle meant that clogs could be more easily cleaned by re-dissolving the polymer with a suitable solvent, alleviating this problem somewhat.
124
Intensity (a.u.) Intensity
1 10 100 1000 Hydrodynamic diameter (nm)
Figure 7.19. DLS-obtained particle size distribution of aqueous PEDOT:PSS suspension, diluted 100x.
A more pressing concern was the size and shape of the CNTs being added to the PEDOT:PSS ink. DLS was not suitable for establishing their PSDs, because of their irregular hydrodynamic radii. However, with both SW- and MWCNTs, 0.2 µm filtration removed a large proportion of the nanotubes, due to the small pore size of the filters relative to the length of the nanotubes (Table 7.2). Even more of the nanotubes were removed when passing the CNT solution through the printhead. In the case of SWCNTs, fewer than 10% of the nanotubes passed through the filter, although a greater number of the nanotubes were able to pass through the print nozzle (without prior filtration) than were MWCNTs. The increased passage through the printhead without filtration was likely a result of the relatively large nozzle size of 25 µm as compared to the CNTs (5 µm length), and the low concentration of the CNTs. Passage of higher concentrations of particles with sizes greater than 200 nm is compromised by crowding and packing of the particles in the nozzles (Fujifilm-Dimatix 2006).
Table 7.2. Concentration of CNTs measured using UV-visible spectrometry.
Concentration (ppm) CNT type Untreated Filtered only Filtered & jetted Single-walled 32.0 2.8 2.5 Multi-walled 32.0 6.8 6.7
125
(a) (b) (c) 1 µm 1 µm 200 nm
Figure 7.20. STEM micrographs of MWCNTs (grey lines) (a) before filtration/jetting; (b) after filtration; (c) after filtration/jetting. Images were obtained with the assistance of Ilya Gourevich of the Centre for Nanostructure Imaging, Department of Chemistry, University of Toronto.
Although more MWCNTs were able to pass through the filter, more were also retained in the jets. The likely explanation for these phenomena is the difference in shape between the nanotube types. The MWCNTs are generally more rigid and stiff due to their multilayered structure (Chang et al. 2005). The SWCNTs, because of their extremely small diameter, may bend or “crumple” into different shapes, such as balls or bundles. In both cases, passage through the filter was contingent upon orientation relative to the filter fibres; transverse orientation facilitated the penetration of the CNTs through the fibre pores, which were significantly larger than the CNTs’ diameter. The SWCNTs, being potentially bent and crumpled, were less likely to pass through the fine pores of the filter. Conversely, the rigid, straight MWCNTs passed through the filter when oriented correctly. However, it appeared from the UV-vis study that a larger proportion of MWCNTs passed when undergoing both filtration and printing stages. The reason for this phenomenon was hypothesized to be the presence of differing amounts of soot and metal catalyst in the two CNT samples. The MWCNT sample used contained only 7.5% CNTs, with the remainder being miscellaneous amorphous carbon soot or residual catalyst from CNT production. The SWCNT sample contained 50-70% CNTs. With soot particles sufficiently small to pass through the filter, the 500 nm absorbance of the carbon soot, being identical to that of CNTs themselves (Zeng et al. 1999) was stronger in the MWCNT samples. This suggests that this method of analyzing CNT passage may be somewhat limited by the CNT sample type, and that purification of the nanotubes may be a necessary precursor for future studies. Regardless of
126
the composition of the sample, these filtration and jetting steps removed >80% of the CNTs, greatly limiting the practicality of incorporation of CNTs into a composite with PEDOT.
The apparently improved passage of MWCNTs over SWCNTs was more directly examined using STEM imaging (Figure 7.20) of dried MWCNT solution. A drop of solution observed under STEM without any filtration indicated a relatively large amount of MWCNTs distributed across the sample grid (Figure 7.20a), along with SLS crystals and other particles, possibly of the aforementioned soot or catalyst. Upon filtration of the solution, only small clumps of MWCNTs were visible sporadically (Figure 7.20b). Finally, after filtration and passage through the printhead, almost no MWCNTs were located, and only clumps of surfactant crystals or metal catalyst/soot particles were visible (Figure 7.20c). X-ray fluorescence linescans of the visible particles indicated that they were primarily composed of carbon, transition metals (particularly nickel), and sodium from the SLS surfactant. Qualitatively, it appeared that the MWCNTs did not pass through the printhead, and the material observed using UV-vis spectrometry was soot.
0
-10 bare CNTs (no SLS)
-20 CNT threshold of stability -30 (a)
-40 (b) potential (mV) potential - (c)
Zeta -50
-60 PEDOT:PSS stock
-70
Figure 7.21. Zeta-potential of CNT/SLS solutions and 50/50 (v/v) PEDOT:PSS/CNT mixtures. (a) Unfiltered; (b) filtered at 0.2 µm; (c) filtered at 0.2 µm and jetted through DMP2831 printhead. Error bars represent standard deviation.
127
The SWCNTs were very small in diameter and therefore difficult to image using STEM; instead, their passage was correlated to of the solutions, according to the theory that increasingly negative values of corresponded to lower concentrations of the generally unstable (Kim & Ma 2011) aqueous CNT dispersions. Therefore, solutions containing larger numbers of CNTs were hypothesized to have lower absolute values of ζ. This behaviour was observed in both the neat CNT solutions and 50/50 mixtures of CNT solution and stock PEDOT:PSS suspension (Figure 7.21).
After filtration and jetting, | ζ | of the resulting solutions increased, indicating improved dispersion stability. In the case of MWCNTs jetted after filtration, ζ decreased to almost 0, suggesting that almost all of the MWCNTs had been removed, leaving behind only distilled water. This result corresponded to that observed using STEM. This approach still did not address the actual amount of SWCNTs removed by the filtration and jetting steps; in order to quantitatively establish the passage of SWCNTs through the printer, a purified SWCNT sample would be required.
Another related observation was the improved stability of PEDOT/CNT mixtures, likely due to the presence of PSS- anion as a secondary dispersant for the CNTs. The stability did not markedly differ for SWCNTs and MWCNTs in the PEDOT:PSS mixture; however, | ζ | was higher for pure PEDOT:PSS stock than for that mixed with SWCNTs, suggesting that dispersion was somewhat compromised by the presence of CNTs, as expected. Flocculation of the PEDOT particles in similar PEDOT:PSS/CNT inks over time was observed in our previous work (Angelo & Farnood, 2010a). Again, after filtration and jetting, the MWCNT-containing mixture approached the | ζ | value measured for pure PEDOT:PSS suspension, indicating the removal of nearly all of the MWCNTs during these processes. Considering these results, only SWCNTs were retained as potential additives for conductivity improvement in the inks. As discussed in the following sections, however, SWCNTs were similarly removed, although not to the same extent, and so the use of smaller carbon species was considered as a possible means of delivering more conductive material to the substrate during jetting.
128
7.2.3 Substrates
The use of non-porous, non-rough substrates for preparing electronic layers has been widely studied; their impermeability, smoothness, and chemical inertness makes them ideal for the deposition of thin layers of functional materials. Therefore, a study on the effects of a substrate like glass, ITO, PET, acetate, or so forth on the functionality of the layers it supported was considered redundant to this study. However, as was discussed in Section 3.5, porous substrates – specifically paper – have attracted some interest for certain electronics; and, of course, paper was used in the first stages of this work to support an ACPEL (Section 5.4). In this study, where PEDOT:PSS/SWCNTs form the electrodes, these layers would be the only ones in direct contact with the paper surface. When printing a PEDOT:PSS ink (whose exact formulation will be discussed in Section 7.3) onto paper, the effect of ink absorption into the sheets was immediately evident. On all of the sheets examined, 3 layers of PEDOT:PSS ink were printed at a drop spacing of 25 µm. With a drop volume of 12 pL (shown in Figure 7.34) at S.G. ~ 1, the grammage of wet ink deposited was 58 g/m2 on each substrate. The large volume of ink did result in dimensional instability due to fibre swelling on unsized sheets. The grammage of dried ink (containing 0.45% PEDOT:PSS) was therefore 0.25 g/m2.
100
10-1 Conductivity (S/cm) Conductivity 10-2
10-3 High-yield Inkjet Cardstock Glossy Photo- Cellulose paper paper paper paper acetate
Figure 7.22. Printed PEDOT:PSS conductivity differences between different commercial paper types.
129
10-2 Hardwood pulp Softwood pulp
10-3
(a)
10-4 Conductivity (S/cm) Conductivity
(b)
10-5 0 2 4 6 8 10 12 14 16
TiO2 filler content (w/w%)
Figure 7.23. Conductivity of printed PEDOT:PSS (single layer) as a function of added filler. Dashed lines (a) and (b) represent the conductivity of handsheets containing retention aid but no TiO2: the amount of retention aid is equal to that of 15% TiO2 handsheets for (a) and 30% TiO2 handsheets for (b). Error bars represent standard deviation.
Conductivity was affected greatly by both chemical and physical means, varying widely between sheets of commercial paper (Figure 7.22). Indeed, the thickness of printed ink layers and their bulk resistances were widely varied across all of the sheets, both commercially prepared and lab- made (see APPENDIX L). Printing of the liquid PEDOT:PSS-SWCNT ink onto the porous, absorbent sheets of paper naturally resulted in the penetration of the conductive species into the paper. Certain sheets absorbed the majority of the ink, drying quickly; others retained an ink film at the surface and dried slowly; and all manner of print effects (feathering, mottle) were visible in the prints. Also, the penetration of ink into the paper may have induced a “chromatographic” effect, where certain ink components were adsorbed or absorbed at different spatial locations than others, resulting in a concentration gradient of conductive material. The distribution of ink components in this fashion has been previous observed by Filenkova et al. (2010). Depending on the location of the PEDOT and any conductivity enhancers, apparent conductivity could have been improved or compromised by this distribution effect. According to the cross-sectional imaging of printed PEDOT:PSS/SWCNT layers discussed previously (Angelo
130
10-2
& Farnood, 2010c), PEDOT:PSS tended to be more concentrated near the surface of the sheets, potentially improving conductivity versus a completely uniform “film”.
Conductivity of these samples was four (4) orders of magnitude smaller than the bulk conductivity of PEDOT (550 S/cm as reported by Crispin et al. 2003), and an order of magnitude smaller than that on cellulose acetate. Despite slight differences in ink absorption rate and surface properties (Table 6.3), there was little if any difference between SW and HW handsheets in terms of conductivity. However, it was observed that the physical differences between unfurnished sheets (i.e. sheets containing no sizing or fixation agent) and furnished sheets (which contained either or both of these materials) affected conductivity. Figure 7.23 shows that with the addition of TiO2 and its accompanying filler retention aid (Polymin SK), printed conductivity on handsheets decreased by nearly two orders of magnitude. However, this decline was primarily due to the presence of retention aid, as the control sheets – containing appropriate amounts of Polymin SK but no TiO2 (dashed lines) – performed similarly to those with both filler and retention aid. It is also worth noting that conductivity was slightly higher in sheets containing filler over those with only retention aid because of the adsorption of Polymin SK to – TiO2, which reduced the amount of Polymin SK reacting or interacting with the PSS stabilizer, as discussed below.
The above observations suggest that an interaction between the Polymin SK and the PEDOT:PSS-SWCNT ink took place resulting in compromised electrical performance. The most likely interaction is the neutralization of the excess ionic charge provided by the PSS– counterion in the ink, a function for which polyethyleneimide (PEI), an active component of this retention aid, is known (Neimo 1999). Furthermore, due to this strong interaction with the PEDOT:PSS complex, non-conductive PEI was likely incorporated within the conductive layer, dramatically reducing conductivity of the layer as a whole. This type of interaction has been previously exploited by Lin et al. (2007), modulating PEDOT:PSS conductivity by five orders of magnitude at a ratio of PEDOT:PSS to PEI of 1:1. However, in the case of printed handsheets, PEI was dispersed throughout the sheet in a relatively low concentration, and hence had a less drastic effect on conductivity. These results suggest that the effect of the filler itself on conductivity was evidently negligible compared to PEDOT interaction with the PEI retention aid.
131
ToF-SIMS mapping of PEDOT, TiO2, and PEI (Figure 7.24) confirms that the presence of isolated clusters of TiO2 did not disturb the spatial distribution of PEDOT in the handsheets. This observation implies that these clusters of TiO2 were also saturated with the conductive ink, and the filler species did not interrupt the conductive path. Hence, in the absence of retention aid, finely divided filler particles are expected to have little or no effect on the conductivity of paper- based printed PEDOT-SWCNT films. However, with the addition of a strongly positively- charged retention aid, filler can increase conductivity by binding to the retention aid, thereby diminishing its interaction with the PSS- counterion. ToF-SIMS images of printed handsheets also revealed that PEDOT and PSS were located in exactly the same regions but were less concentrated where PEI was localized (Figure 7.24). The resulting non-uniformity of the conductive layer would have an adverse effect on the conductivity of filled handsheets. It appears that a very small amount of PEI is actually interacting with the TiO2, and the majority resides in or on the fibres, where it can readily interact with the PSS–. In fact, ToF-SIMS peaks 4+ for Ti -PEI (m/z = 91) and TiO2-PEI (m/z = 123) were very weak, suggesting that little PEI was adsorbed or bonded to the surface of the filler, leaving the rest to freely interact with PSS– in or on the fibres. These observations are consistent with the hypothesis that PEI-PSS– interaction was primarily responsible for increased electrical resistance.
PEDOT PSS PEI TiO2
20 µm
Figure 7.24. ToF-SIMS maps of relative distribution of PEDOT, PSS, PEI, and TiO2 on SW/30% TiO2 sheet.
132
-2 10 (a)
(b)
10-3
10-4 Conductivity (S/cm) Conductivity 10-5
10-6 0 15 30 TiO filler added during sheet forming (w/w%) 2
Figure 7.25. Estimated conductivity of PEDOT-SWCNT ink on SW fibres: (a) unfurnished (b) PDADMAC fixation agent. Error bars represent standard deviation.
The addition of a cationic ink fixation agent, PDADMAC, decreased conductivity of the printed handsheets, and this effect was more pronounced at higher filler addition levels (Figure 7.25). With a similar effect to the retention aid, this was likely caused by the interaction of the cationic PDADMAC with the PSS– counterion and the introduction of non-conductive PDADMAC into the PEDOT film. However, the lower concentration of PDADMAC versus PEI resulted in a smaller decrease in estimated conductivity. In addition, the highly-charged, cationic nature of PDADMAC may also have caused agglomeration of the PEDOT or SWCNTs after printing due to destabilization of these suspensions through interaction with the PSS- and lauryl sulfate dispersants. Such an interaction would create a non-uniformly distributed conductive ink and hence a lower sheet conductivity. The non-uniform distribution and the apparent agglomeration of both PEDOT and PSS in the ToF-SIMS maps of handsheets containing PDADMAC supported this theory (Figure 7.26). With increasing filler levels, the detrimental effect of PDADMAC also became more pronounced as the contact angle decreased. A lower contact angle implied better penetration of PEDOT:PSS into the sheet, and hence reduced conductivity.
133
PEDOT PSS PDADMAC Total ion
20 µm
Figure 7.26. Relative distribution of PEDOT:PSS and PDADMAC in HW sheet (30% TiO2, no sizing).
10-2
(b) (a)
(c) 10-3
10-4 Conductivity (S/cm) Conductivity
10-5 0 15 30
TiO2 filler added during sheet forming (w/w%)
Figure 7.27. Estimated conductivity of PEDOT-SWCNT printed ink on HW fibres: (a) unfurnished; (b) with internal AKD sizing; (c) with internal AKD sizing and fixation agent. Error bars represent standard deviation.
The most notable effect on conductivity aside from that of PEI was that of internal sizing (Figure 7.27). In every case, the addition of sizing agent resulted in an increase in conductivity. As shown in Table 6.3, internal sizing decreased ink spreading and absorption by increasing the contact angle of the ink. Therefore, a more uniform ink layer (containing fewer non-conductive fibres/filler particles) with higher connectivity and conductivity was obtained. Moreover, the ink absorption rate was reduced by several orders of magnitude in the sized sheets, allowing a longer time for the ink to rest on the surface during drying of the PEDOT-SWCNT ink. Cross-sectional
134
images confirmed the presence of a relatively thin PEDOT-SWCNT layer on the sized sheets. Figure 7.28 shows the cross-sections of printed SW handsheets (30% filler) for an internally AKD sized handsheet and an unsized handsheet. In the case of the sized sheet, the PEDOT- SWCNT ink (blue-coloured) is concentrated near the surface of the sample while in the latter case ink is distributed throughout the sheet thickness.
AKD sized unsized
inked region
inked region
50 µm
Figure 7.28. Cross-sections of printed SW handsheets (30% filler) showing PEDOT:PSS ink penetration.
The positive effect of AKD sizing appeared to be more pronounced for handsheets containing retention aid and/or fixation agent. In other words, AKD sizing appeared to largely eliminate the adverse effects of PDADMAC and PEI that are distributed throughout the sheet, by reducing the amount of contact of the conductive ink with these molecules. Because of this correlation between ink “holdout” in the printed sheets and the performance of the conductive layers, paper with minimal porosity and high hydrophobicity (for aqueous inks, at least) should be considered ideal for electrically conductive paper production. The performance of such materials might be improved by the deposition of larger amounts of ink, as the initial print passes would fill the pores and become absorbed into the fibres and filler clusters, providing a less absorbent surface for subsequent print passes. However, this approach would of course consume more ink and require more processing time to achieve similar performance to appropriately sized sheets. The significant improvement in conductivity across all the sheets resulting from internal sizing is likely more easily and economically achieved by sheet pretreatment.
135
(a) (b) (c) (d)
20 µm
% covered by PEDOT (a) sized HW: 45.9% (b) unsized HW: 25.6% (c) sized SW: 40.2% (d) unsized SW: 25.1%
Figure 7.29. Greycale ToF-SIMS images of PEDOT distribution on unfilled (0% TiO2) sheets with no PDADMAC. % coverage by PEDOT was estimated using grey-level thresholding.
ToF-SIMS mapping of the ink components on the sized handsheets further confirmed the improved ink retention on the sheet surface. It is evident in Figure 7.29 that internal sizing allowed the PEDOT ink to coat the surface of the sheets to a greater degree, rather than being absorbed into them, resulting in a larger proportion of interconnected PEDOT-SWCNT regions. However, there was still not a completely contiguous PEDOT/PSS/SWCNT layer on the surface of the sized sheets, and fragments of cellulosic materials and filler were clearly visible in the spectral maps. As is also shown in Figure 7.28, even sized sheets absorbed the ink to a certain degree. However, it is worth noting that the intensity of the PEDOT signal in the ToF-SIMS maps of sized sheets was significantly higher indicating a higher concentration of conductive ink on the surface of the sized samples.
Another variable that might assist in improving conductivity, in the same vein, is sheet smoothness. In fact, a brief examination of printed PEDOT-SWCNT layers on rough, uncalendered handsheets revealed that their conductivity was close to those of the base sheets themselves. The universal calendering of the handsheets used in these experiments provided a high degree of smoothness, improving the ink holdout and conductivity of printed samples.
So, if paper were to be considered as a substrate for electronics deposition, the type of paper itself would play a major role in the functionality of the device. Because many planar electronic devices involve the deposition of interconnects and electrodes as a base layer, the conductive ink’s interaction with the surface would determine its conductivity and the device’s performance.
136
The use of paper with good ink holdout (i.e. low absorption), chemically compatible additives, and similarly-charged species – avoiding cationic materials – is critical to maximizing conductivity in an electrode ink.
7.3 Ink formulation
7.3.1 Conductive ink
The initial ink formulation was based on a PEDOT:PSS solution described by Bronzyck (2003) used for spin-coating onto paper, containing a mixture of glycerol (20%), isopropanol (12%), and PEDOT:PSS suspension (68%). Because of DMSO’s beneficial effect on conductivity in PEDOT:PSS films, this formulation was refined to contain DMSO, yielding a similar ink to that presented by Garnett & Ginley (2005). Given that the viscosity of the DMSO-containing ink was >12 cP at room temperature, water was used to dilute the PEDOT:PSS to a suitable viscosity before treating it with other additives (IPA, DMSO, etc). To clarify the following discussion, “water content” in the inks refers to the amount of water added to the PEDOT:PSS suspension, not the total amount of water in the ink (which included the water contained in the original PEDOT:PSS suspension).
It was observed during preliminary testing that moderate-viscosity (i.e. below 10 cP) PEDOT:PSS inks could be jetted for extended periods of time without clogging, whereas jetting of PEDOT:PSS inks with viscosity greater than 10 cP, while stable for a short period, resulted in rapid clogging of the printer nozzles. Also, the higher viscosity inks displayed an increased amount of splattering onto the substrate due to the formation of long “tails” on the droplets that formed satellite droplets, due to increased viscoelastic forces. This issue could not be resolved by increasing the drive voltage to its maximum (40 V) nor by increasing the printhead temperature (which aggravated the issue by boiling off the solvent prematurely and causing curing of the ink in the nozzle and on the nozzle plate). Therefore, considering these factors, a more moderate viscosity in the range of 5-6 cP was targeted. This range also corresponded to the suggested viscosity (5 cP) of piezoelectric inkjet inks found in literature (Di Risio & Yan 2007, Dong et al. 2006, Calvert 2001).
137
18
16
14
12
10
8 Viscosity (cP) Viscosity 6
4
2
0 0 5 10 15 20 25 30 35 40 45 50 Added water content (w/w%)
Figure 7.30. Effect of added water on PEDOT:PSS suspension viscosity. Error bars represent standard deviation.
Figure 7.30 shows the effect of added water on viscosity, indicating that a water content of approximately 35 w/w% provided a suitable viscosity (~4.75 cP). The choice of a slightly smaller value of viscosity was prompted by the necessity for the subsequent addition of the more highly viscous DMSO and glycerol co-solvents. Adding glycerol as a co-solvent/humectant, while expected to change viscosity, did not cause any significant variability, even at the highest typical humectant loading of 20 w/w% (Magdassi 2010). The viscosity of the undiluted ink containing 20 w/w% glycerol was 12.1 cP, versus 11.4 cP for 0% glycerol. At 17 w/w% glycerol, the viscosity was still 11.9 cP (i.e. in the jettable range). In order to avoid clogging of the piezoelectric nozzles with dried and crusted ink, 17 w/w% glycerol was added to the ink. The ink containing both 35 w/w% water and 17 w/w% glycerol (balance PEDOT:PSS suspension) had a viscosity of 4.9 cP.
Using the glycerol-treated ink, DMSO addition was studied. According to Figure 7.31, increasing the DMSO content did not demonstrate any trend in terms of viscosity variation. The effect of DMSO addition on the electrical resistance was also examined by cast coating 0.25 mL of the ink on cellulose acetate with a calibrated pipette. It was noticed that adding DMSO up to
138
10 w/w% reduced electrical resistance in cast-coated conductive films PEDOT:PSS while beyond 10 w/w% DMSO, resistance remained nearly unchanged (Figure 7.32). Therefore, a DMSO content of 10 w/w% was chosen, resulting in a mixture viscosity of 4.8 cP.
14 0 w/w% DMSO 1 w/w% DMSO 12 5 w/w% DMSO 10 w/w% DMSO 10 15 w/w% DMSO 20 w/w% DMSO 8
6 Viscosity (cP) Viscosity 4
2
0 0 10 20 30 40 50 Added water (w/w%)
Figure 7.31. Effect of added DMSO on PEDOT:PSS/glycerol mixture’s viscosity, at different water loadings
Ω) Resistance ( Resistance
0 2 4 6 8 10 12 14 16 18 20 DMSO content (w/w%)
Figure 7.32. 2-point resistance of cast-coated PEDOT:PSS/glycerol/water (48%/17%/35%) films with different DMSO contents (displacing water in the solution).
139
To optimize the surface tension, the surfactants discussed in Table 6.4 were then added to the ink (displacing water) to reduce its surface tension to a jettable level. Because of the addition of DMSO/glycerol, even the ink containing no surfactant had acceptable surface tension for jetting (33.1 mN/m). Nevertheless, as the PEDOT:PSS ink was needed to wet the surface of the hydrophobic polymer PVK in some of the device structures, further reduction of surface tension to 30 mN/m was desirable. Surface tension generally decreased up to a certain critical loading of surfactant, and then either plateaued (Zonyl FS-300, ZetaSperse 3700) or began to increase again (Triton X-100, Igepal CA-720, SLS) as the surfactant likely detached from the particles’ surface and formed double-layer micelles of pure surfactant in solution (Spasic & Hsu 2006). Each surfactant except ZetaSperse 3700 was capable of reducing the surface tension to 30 mN/m (or significantly lower), as is shown in Figure 7.33. The addition of surfactant also affected viscosity. Certain surfactants were observed to decrease viscosity markedly, SLS being the best example. The effect on viscosity was expected, as each surfactant was likely responsible for a different thinning effect (Jansen et al. 2001). In no case did the viscosity fall so low that the ink was no longer jettable, although the surface tension dropped below the jettable range, particularly for Zonyl FS-300 (expected for a fluorosurfactant).
The measured maximum particle sizes of each ink varied with the surfactant type as well, but were generally close to the cutoff of 200 nm. This was simply by virtue of selecting PEDOT:PSS as the starting material. Also, the addition of surfactants served to improve dispersion stability (| | > 40 mV) in most cases, or at least maintain it above the threshold of stability. Igepal CA-720 appeared to reduce | |, by a mechanism that was not well-understood, as it is a non-ionic molecule with no positively-charged regions. It is possible that the PSS- was displaced by the OH-bearing end of the molecule to some degree, and the Igepal served as a poorer stabilizer for the PEDOT micelles. A summary of the dispersion and particle size properties, as well as the rheological properties of the inks (with surfactants loaded to their CMCs), is given in Table 7.3. Each ink, again, contained 17% glycerol, 35% water (minus the amount replaced by surfactant), 10% DMSO, and the balance as PEDOT:PSS suspension.
140
40 Triton X-100 Zonyl FS-300 Igepal CA-720 Zetasperse 3700 SLS
35
30
Surface tension (mN/m) tension Surface 25
20 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 Surfactant content (w/w%)
8
7
6
5
4
Viscosity (cP) Viscosity 3
2
1
0 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 Surfactant content (w/w%)
Figure 7.33. Surface tension and viscosity in PEDOT:PSS inks with different surfactant types. Error bars represent standard deviation.
141
Table 7.3. PEDOT:PSS inks’ fluid properties.
-1 Ink (cP) (mN/m) dp, max (nm) Z En ζ (mV)
No surfactant 4.1 33.1 212 7.0 ~0 -68.8 Igepal CA-720 (0.2%) 4.9 26.3 235 5.2 ~0 -55.3 SLS (0.5%) 2.2 30.2 200 12.5 ~0 -67.5 Triton X-100 (0.1%) 5.7 30.3 236 4.8 ~0 -67.9 Zonyl FS-300 (0.02%) 4.3 22.5 82 5.5 ~0 -63.8 ZetaSperse 3700 (0.2%) 3.7 31.4 178 7.6 ~0 -65.8
Because of the wide variation in surface tension between inks, inks with identical formulations except for the surfactant type exhibited markedly different jettability and film formation. Images of the droplet and film formation can be seen in Figure 7.34 and Figure 7.35, respectively. For reference, an example of good drop formation is that of the SLS-bearing ink; poor drop formation, with nozzle plate splattering, is visible with Zonyl-bearing ink. Good film formation is visible in the Triton and Igepal-bearing inks, whereas the ink containing no surfactant formed a poor film.
200 µm
No surfactant Triton X-100
Igepal CA-720 Zonyl FS-300
SLS ZetaSperse 3700
Figure 7.34. PEDOT:PSS ink droplet formation during ejection from DMP2831 cartridge nozzles. (b) (b) Figure 7.34 shows that the drop size was similar(a) for various inks, with an average diameter of about(a) 28 µm (or droplet volume of 12 pL), close to the(d) expected 10 pL delivered by the
142
(b) (e) DMP2831 cartridge. Within the range of this study, viscosity had less of a bearing on drop formation and jettability, although it did affect Z-1, by definition (which still remained in the jettable range). Surface tension had much greater bearing on jetting performance. The inks with the lowest surface tension, containing Zonyl FS-300 and Igepal CA-720, demonstrated poorer drop formation than the other inks. With these two inks, the formation of small droplets which detached from the nozzle rapidly and dropped satellites onto the nozzle plate is clearly visible.
No surfactant Igepal CA-720 SLS 2 mm
2 mm
Triton X-100 Zonyl FS-300 ZetaSperse 3700
Figure 7.35. Printed patterns of PEDOT:PSS inks on acetate. Films consist of a single printed layer of PEDOT:PSS ink (25 µm drop spacing). Images were adjusted for colour and contrast.
Each of these inks was jetted onto cellulose acetate. The quality of the printed films varied widely among samples. Triton- and SLS-bearing inks, both having a surface tension around 30 mN/m, formed contiguous films while Igepal CA-720, with a slightly lower surface tension, created a more uniform but “striated” film, with visible printed lines and unprinted or only lightly inked areas. These results suggest that surface tension plays a major role in printed film quality with a higher surface tension better tolerated than a lower one. Poor quality films in
143
general exhibited either such striations or “island” formation. Striations in the film, characteristic of the failure to deliver ink to the entire printed area either due to clogging or poor drop formation, were visible in the Zonyl and Igepal-containing ink films. The lower surface tension may have produced smaller droplets which were not spaced closely enough. In the case of Zonyl FS-300, the entire printed pattern failed to be delivered to the substrate, which is typical when nozzles clog during printing, or, more likely, when the nozzle plate becomes wet with fluid through which drops cannot penetrate. These results also suggest that the use of potent surfactants such as fluorosurfactants is unsuited to this application. A similar effect (poor drop formation and print uniformity) was observed with a ZnS:Mn ink containing Zonyl FS-300 (Angelo & Farnood 2011). However, a higher surface tension produced isolated “islands” and unwetted regions on the substrate. Without the capability to wet the substrate, drops tended to merge into thick puddles, resulting in their characteristic “islands” upon drying.
The print quality issues resulting from the use of different surfactants had a major bearing on conductivity of printed film. Regardless of surfactant type, the conductivity of spin-coated – not printed – films on glass were fairly similar at just over 1000 S/cm, with all values within the same statistical envelope, indicating that the surfactants themselves did not have any significant effect on the conductivity. Compared to spin- coated samples, conductivity of inkjet-printed films was 3-5 orders of magnitude lower and varied widely from 0.016 S/cm for Zonyl-bearing ink to 1.03 S/cm for SLS-bearing ink, when resistance was measured parallel to the print direction (Figure 7.36). As would be expected, printed films that exhibited striated structures had lower conductivity than those with more uniform print quality. The inks containing no surfactant and ZetaSperse, which both were non-uniform in terms of print quality as well (the “islands” previously mentioned due to poor wetting), still had higher conductivity than the striated films, where definite gaps between printed lines were visible. The presence of large patches or blobs of PEDOT:PSS in these inks likely served as bridges between adjacent printed lines. Measuring conductivity in the direction normal to print (i.e. across rather than along the print lines) confirmed this hypothesis, as the striated inks (Igepal and Zonyl) effectively were not conductive in this direction. In every case, conductivity was reduced by measuring across the samples, suggesting that the printed lines merged poorly regardless of the surfactant. It was also
144
thought that viscosity may play an important role in determining the conductivity of inkjet- printed films, and in the case of conductive films, a lower viscosity might actually be beneficial. The lowest-viscosity ink containing SLS, where scalloping and spreading across the printed lines occurred (Figure 7.36), achieved the highest conductivity. The spread of the ink more rapidly due to less viscoelastic resistance to flow was expected to help fill in some of the inter-line gaps and produce better conductivity. The use of a strong anionic surfactant has also been linked to improved conductivity in PEDOT:PSS films due to “unzipping” of the PEDOT-PSS chain through surfactant exchange with PSS- and lengthening of the PEDOT chain as a result (Fan et al. 2008). This might explain the improved conductivity seen in both SLS and ZetaSperse- bearing inks, when printed, but there was no similar improvement observed in spin-coated films, casting doubt on the likelihood of this mechanism. Furthermore, the large improvement in conductivity provided by DMSO likely overshadowed any benefit provided by SLS or ZetaSperse.
104
103
102
101
100
10-1
10-2
Estimated conductivity conductivity (S/cm)Estimated
10-3
10-4
No surfactant Igepal CA720 SLS Triton X-100 Zonyl FS-300 ZetaSperse 3700 Printed (parallel direction) Printed (normal direction) Spin-coated
Figure 7.36. 2-point estimated conductivity of printed PEDOT:PSS inks on cellulose acetate (single printed layer, 25 µm drop spacing).
145
Nevertheless, the unexpectedly high conductivity of the anionic surfactant-bearing films was examined more closely, to determine if a chemical or conformational change was occurring in the films. Raman spectrometry indicated that the chemical structure remained the same in the presence of most of the surfactants except ZetaSperse (Figure 7.37). All spectra were red-shifted from that of stock PEDOT:PSS, due to the presence of DMSO as a conductivity enhancer, representing the benzoid-to-quinoid transition of the PEDOT structure (Figure 7.37 inset). The quinoid structure has higher electron mobility (Ouyang et al. 2004, 2005), due to the loss of aromaticity (and stabilization energy) in the ring, which in turn reduces the energetic bandgap (Cheng et al. 2009) and forms a conjugated structure in the polymer chain. A smaller bandgap is synonymous with improved carrier mobility. A peak at ~1460 cm-1 of Raman shift is representative of the benzoid thiophene ring, whereas the red-shifted peak at 1425 cm-1 represents the conjugated quinoid structure. The distinct shoulder peak at 1460 cm-1 (most easily visible in the PI spectrum), as well as the broadness of the other peaks (besides PZS, which will be discussed further below), suggests that the addition of 10% DMSO resulted in the transition of a certain fraction of the ethylenedioxythiophene repeating units. The shoulder represents remaining benzoid units. As the intensity of the red-shifted peak at 1425 cm-1 increased, this shoulder became less pronounced.
It was notable that the slope of the ZetaSperse 3700-treated film was the greatest, with a very minor shoulder peak present. The shape of this peak, being very similar to that of untreated PEDOT:PSS, suggested that all of the benzoid structures had been converted to quinoid structures in this ink. Either the functional molecule in ZetaSperse 3700 itself (which was proprietary) or the solvent system bearing it may have furthered the complete transition to the quinoid structure. Since ZetaSperse is water/propylene glycol-borne, it may have simply been the glycol which caused the transition, as has been reported previously (Kim et al. 2003). Another notable point was the further red shift in the Raman spectrum for this ink compared to the other inks. This result suggested that the addition of ZetaSperse 3700 may have caused a further conformational change in the PEDOT molecule. Examination of this surfactant by FTIR indicated the presence of primary and secondary amine groups in the ZetaSperse 3700 formulation. Therefore, it is possible that the peak in the PZS spectrum at 1415 cm-1 could
146
represent the deformational vibtrations attributed to primary methylated amine groups, masking the PEDOT:PSS peak.
1525 1475 1425 1375 1325 1275 Raman shift (cm-1) No surfactant Igepal CA-720 SLS Triton X-100 Zonyl FS-300 ZetaSperse 3700 PEDOT:PSS
Figure 7.37. Raman spectra (excitation wavelength = 785 nm) of PEDOT:PSS inks. The inset shows the benzoid-to- quinoid structural transition.
Based on the above consideration, the addition of either 0.5% SLS or 0.2% ZetaSperse 3700 was found to result in optimum conductivity of the pinted PEDOT film. The much superior conductivity and print quality offered by the SLS-bearing ink versus the ZetaSperse-bearing ink made it the more attractive option. Foaming was an issue with the use of SLS, which produced a large volume of foam upon agitation – a problematic situation for piezoelectric jets. However, a small amount (0.05%) of defoamer (Surfynol DF-110D) was sufficient to reduce foam in the PEDOT:PSS ink.
The consideration of CNT addition also was made to further improve printed conductivity to a value useful for electronics fabrication. Because the CNTs were suspended in water using SLS,
147
the use of the same surfactant in the ink was ideal. At this point, treatment of the ink with CNT solution (replacing water) was considered. Because MWCNTs appeared not to pass through the filter/printhead in any significant amount, SWCNTs were used. Figure 7.38 shows that there was a gradual increase in conductivity with SWCNT inclusion in printed PEDOT:PSS ink, although this increase was not consistent at higher loadings. Conductivity increased significantly with the addition of a smaller fraction of SWCNTs (i.e. from 0-2 w/w%) and then varied irregularly, although it remained at ~1 S/cm at 2% CNTs. It was evident from the CNT passage studies discussed in the previous section that only a very small fraction of CNTs passed through the printer to the substrate. However, as more nanotubes were added, the nanotubes were delivered to the substrate in larger amounts, indicated by increasing conductivity. Even in pure SWCNT stock, however, less than 8% of the CNTs (i.e. 3 × 10-2 w/w%) remained after filtration, which likely included some soot and catalyst materials. This implied that the highly CNT-loaded PEDOT-SWCNT inks were limited to approximately this number of nanotubes passing, so even if all of the SWCNTs in the 8% CNT ink passed through the printer, the 9 w/w% and 10 w/w% inks would be limited to 3 × 10-2 w/w% CNTs passing to the substrate. This maximum number of CNTs passing through the printer provided a limit above which the addition of CNTs served no further purpose, hence the plateau effect observed at 5-6% SWCNT solution. Also, at higher CNT concentrations, the likelihood of interaction and agglomeration of the CNTs is increased, reducing the passage of CNTs through the nozzles and causing nozzle clogging. This may offer another explanation for the increased variability of the conductivity data at higher CNT loadings. The shape of the curve describing conductivity dependence on SWCNT content on acetate also suggested a logarithmic or power-law relationship between CNT concentration and probability of passing through the printer before the plateau conductivity value.
The mechanism of conductivity enhancement might be attributed to the presence of more-highly conductive SWCNTs within the PEDOT:PSS matrix . As was described by Kerner (1956), the incorporation of a highly conductive species in a composite matrix may enhance the conductivity of the composite. However, the extremely small proportion of CNTs relative to PEDOT could not reasonably account for the measured increase in the overall conductivity of the composite, where conductive “filler” materials generally have to be in contact with one another to allow
148
charge transport (Kerner 1956, Kim & Ma 2011). As an alternative hypothesis, the physical conformation of the CNTs may assist in “bridging” non-conductive regions – i.e. that their length would connect nearby or adjacent printed regions of PEDOT and thereby decrease resistive losses. However, because the probability of any SWCNTs passing through the printer was extremely low after filtration, the “bridging” hypothesis was unlikely. A more realistic mechanism was increased doping of the PEDOT backbone by anionic species present in the CNT soot/catalyst. The presence of additional dopant (supplementing the effect of PSS-) would improve carrier mobility in the LUMO of the conjugated PEDOT, enhancing conductivity.
101 Acetate
Photo paper
100
10-1 Conductivity (S/cm) Conductivity
-2 10 0 1 2 3 4 5 6 7 8 9 10 SWCNT solution content (w/w%)
Figure 7.38. Conductivity of printed PEDOT:PSS-SWCNT ink (SLS surfactant) at varying SWCNT loadings (single printed layer, 25 µm drop spacing).
Firstly, C60 and S-SWCNTs, which did not have a high enough aspect ratio to physically bridge gaps between conductive regions, were jetted in the inks. Both C60, with a diameter of 0.7 nm (Yeo et al. 2009) and chemically-shortened SWCNTs, with an average length <60 nm (Chen et al. 2006), were sufficiently small to pass through the filters and printheads. Therefore, it was expected that if the higher conductivity of carbon relative to PEDOT was improving the composite’s conductivity, not only would inks containing these materials demonstrate similar trends in conductivity to those containing SWCNT solution, but that these trends would be more
149
pronounced due to their wholesale passage through the printer. However, this was not the case, and neither C60 nor S-SWCNTs had any significant effect on conductivity when printed at 25 µm drop spacing for a single print pass (Figure 7.39). The x-axis of this figure shows the amount of C-species solution that was added to the ink, where the solutions contained 0.04 w/w% of their respective C-species. S-SWCNTs, in particular, which were chemically shortened from the same batch of SWCNTs used in the ink, did not improve conductivity at all. The S-SWCNTs had identical properties to the SWCNTs, with the only difference being length; this directly supported the idea that SWCNTs were “bridging” materials. There was a small improvement in conductivity due to C60 addition, and a similar trend of irregular increase above 5 w/w% C60 solution. This suggested that perhaps a component of the phenomenon of increased conductivity in SWCNT-containing PEDOT layers was simply the presence of conductive carbon. The lack of a similar effect in S-SWCNTs was attributed to the chemical process by which they were shortened, which involved the use of strong mineral acids; dopant species may have been consumed during this reaction or removed by rinsing.
1.6 SWCNTsSWCNTs 1.4 USS-SWCNTs
1.2 C60C60
1
0.8
0.6 Conductivity (S/cm) Conductivity 0.4
0.2
0 0 1 2 3 4 5 6 7 8 9 10 C species sol'n (w/w%)
Figure 7.39. Conductivity of inkjet-printed PEDOT:PSS-carbon composites on acetate (single printed layer, 25 µm drop spacing).
150
Spatial location of the SWCNTs was not feasible using SEM, partially due to their small size, but more importantly because of their incorporation into an organic matrix and resulting lack of contrast. Again, ToF-SIMS mapping confirmed the physical behaviour of the conductive ink on the various substrates in terms of ink penetration and retention (Figure 7.40). The shallow penetration depth of the ToF-SIMS primary ions indicated significant retention on the acetate and photo paper substrates even at the very surface of their respective coating layers.
Photo-paper 40 µm
3+ Al PSS PEDOT Total ion
Cellulose acetate
Acetate PSS PEDOT Total ion
Figure 7.40. ToF-SIMS maps of PEDOT and substrate component distribution, single printed layer. Al3+ comprises the majority of the coating layer of the photo-paper (as alumina).
ToF-SIMS images show that on both acetate and photo-paper the printed PEDOT-SWCNT films were not uniform. In the case of the photo-paper, the PEDOT-SWCNTs and surface coatings formed composite layers with micro- or nano-scale non-uniformities in the form of finely divided coating pigments and binder. In the cases where non-uniformity was most pronounced on a spatial level, with permeable substrates, conductivity was the lowest and the least improved by the addition of SWCNTs. As PEDOT-laden regions moved closer together, in the coated sheets, conductivity increased, as did the effect of CNT addition. However, there was no visible means by which the “bridging” of conductive regions might have improved conductivity, and the practical invisibility of CNTs using ToF-SIMS examination meant that this mechanism could not
151
be reliably confirmed. Therefore, increased PEDOT doping by impurities in the CNTs remained the most sensible mechanism of conductivity enhancement. Further examination of the means of CNT enhancement of conductivity was discussed in Paper 4; in short, however, the introduction of CNTs into inkjet inks is of limited utility, due to their nearly wholesale removal from the ink during the printing process.
In any case, the slight beneficial effect of SWCNT/soot addition on conductivity was evident, to a point. Referring again to Figure 7.38, as well as work described in Paper 1, the addition of 3-4 w/w% SWCNT solution increased conductivity of the printed PEDOT:PSS films, and further addition did not yield any noticable improvement, as conductivity plateaued at about 1 S/cm at this point (for a single layer printed at 25 µm drop spacing) on cellulose acetate.
The detailed and rigorous formulation procedure outlined above for the PEDOT:PSS/SWCNT ink encompassed a large part of the work. This procedure was adapted to each respective ink depending on the mechanism of that ink’s function. In this case, physical connectedness was vital to proper ink performance. However, other key variables might affect the performance of various inks, such as topography, thickness, roughness, and so forth. The basic tenets of the formulation procedure remained universally the same, however; modification of the fluid properties of a dispersion in a linear fashion, establishing one and moving on to the next, while always making sure that each new additive did not have any adverse effects on the previously optimized properties. Again, because PEDOT:PSS/SWCNTs served as a “model” ink for the formulation procedure, all of the details of the formulations of the other inks will not be included here, for the sake of brevity – but unique challenges encountered with the formulation of each will be addressed. In each case, however, at least ten different ink formulations were tested before a suitable mixture was found (all of which are listed in APPENDIX I).
7.3.2 ZnS inks
Each of the aqueous and organic ZnS inks had one primary component – the nanoparticle suspension. A solids concentration of 1 wt.% ZnS was targeted. An initial attempt (described in Paper 7) at making an ink using AA-capped ZnS:Mn in an ink composed primarily of AA and other aprotic solvents yielded a very dilute suspension (< 0.8% in an ink almost completely
152
composed of AA) of weakly photoluminescent nanoparticles. The almost pure acid comprising the ink meant that it damaged the printhead, cartridge, substrate, and underlying layers very rapidly, and the slow heat-polymerization of the AA monomer allowed the solids to move to the edges of the printed pattern (Figure 7.41). Also, the capping agent (AA) did not quench surface states as well as 3-MPA, resulting in red-shifted emission and blue emission from S2- vacancies. The reformulation of ZnS:Mn and ZnS:Cu inks capped with 3-MPA are described below.
Visible 302 nm UV
Figure 7.41. Printed ZnS:Mn/AA ink on cellulose acetate, 10 printed layers.
Table 7.4. Ink components incompatible with ZnS nanoparticle suspensions (aqueous and organic).
Component Purpose Issue Ink
Mercaptosuccinic acid Dispersant, viscosity modifier Immediate precipitation Aqueous ZnS:Mn, Cu (Small et al. 2010)
PVP (>1wt%) Binder Gradual precipitation Aqueous ZnS:Mn, Cu
Air Products DF-110D Defoamer (req’d in Triton- Gradual precipitation All defoamer containing inks) Gradual precipitation, Carboxymethylcellulose Viscosity modifier insoluble in organic All solvents Immediate precipitation Isopropanol Surface tension modifier Aqueous ZnS:Cu (SHMP insoluble) Immediate precipitation Butoxyethanol Viscosity modifier Aqueous ZnS:Cu (SHMP insoluble) Viscosity modifier 3-amino-1-propanol Reaction with TGA Aqueous ZnS:Mn PL enhancer (Wang et al. 2009)
Dimethyl sulfoxide Co-solvent Gradual precipitation All
As was mentioned in Section 6.2.3, the a mixture of isopropanol, butoxyethanol, water, and PVP 1 300 000 were suitable to prepare an aqueous ink containing ZnS:Mn. The high viscosity and
153
low volatility but moderate boiling point of butoxyethanol made it a good co-solvent and humectant, and isopropanol served to reduce surface tension without having to introduce surfactants – the potential issues which have been elaborated upon in Section 7.3.1. However, the addition of certain materials during formulation caused precipitation of the nanoparticles for a variety of reasons and imposed limitations on ink formulation (Table 7.4). The inability to add any solvents other than water to the ZnS:Cu aqueous suspension due to the insolubility of SHMP, for example, meant that it was not suitable to use in an aqueous inkjet ink.
While ZnS:Cu was precluded from use in an aqueous inkjet ink, ZnS:Mn encountered problems with application in an organic ink. The first issue that arose was a practical one: the phase transfer process failed to produce as stable a suspension of ZnS:Mn in toluene, whereas in ZnS:Cu, no such problem was encountered. More importantly, however, the organic-dispersible ZnS nanoparticles were to be bound with PVK in a DC-LED structure, where charge transfer was vital between the matrix and the nanoparticles. The semi-conductive CuS which forms in DC-driven devices is integral to this charge transfer (Ono 1995); MnS does not similarly conduct. Therefore, ZnS:Cu was the only choice for the emitter in the PVK layer. So, one ink was made containing ZnS:Mn as an aqueous suspension, and one ink was made containing ZnS:Cu as an organic suspension, meaning that the DC-LEDs were to be built using ZnS:Cu and the PELs to be built using ZnS:Mn.
(a)
ZnS:Mn (aqueous) 200 µm
(c)(b)
ZnS:Cu (organic)
Figure 7.42. Droplet formation of ZnS inks at 10 µs intervals.
Jetting waveforms producing stable, spherical droplets were created for both inks (Figure 7.42); see APPENDIX J for the waveforms. No significant bead-on-a-string effects were observed even with the loading of polymers into the inks. Even with the low viscosity observed in the
154
toluene-based ZnS:Cu ink, drop formation was uniform and stable, although with some splattering.
There was no means by which to increase viscosity in the ZnS:Cu layer except by adding more PVK (typical viscosity modifiers were insoluble). However, at higher PVK loadings, film structure was observed to be poor; however, inks with a lower PVK loading actually formed superior films due to their lower viscosity. Although other viscosity modifiers were considered, the desire for a semiconducting film free of high-boiling solvents in DC-LED structure precluded their use. This was the only ink which did not meet the jetting criteria specified for the printer, with µ = 0.97 cP; however, inks containing toluene at low µ values have occasionally been printed on the DMP2831 (Sumerel et al. 2007). Regardless, film formation on the surface was successful with both inks, producing gap-free layers (Figure 7.43). The optimization of the layers’ structure will be discussed in Section 7.4. PL emission was immediately visible after deposition, indicating that not only had the ink jetted properly, but the formulation allowed for the retention of functionality and had no chemical compatibility issues. 302 nm UV excitation
500 µm
100 µm
Figure 7.43. Left: optical microscope imaging of a printed ZnS:Mn/PVP (aqueous) film on glass (single layer); centre: PL on glass, 1-5 printed layers; right: 10 printed layers on photo-paper. PL was induced by excitation with a 302 nm UV source
As described in Section 6.2.2, the BaTiO3 ink utilized an ethanol/MMA solvent mixture and PEG
300 to reduce volatility and increase viscosity. 5% BaTiO3 was added to the ink, as this was the limit of stability (see Section 7.2.1). No surfactant was required due to the low surface tension values of the ethanol and MMA. Drop formation was ideal with this ink (Figure 7.44).
155
200 µm
Figure 7.44. Droplet formation of BaTiO3 ink at 5 µs intervals.
The relatively high viscosity of the ink contributed to stable, uniform drop ejection from the piezoelectric nozzles. There was no evidence of the formation of satellite droplets, wetting of the nozzle plate, nozzle clogging, or other undesirable print effects. The drops were approximately 10 pL in volume, estimating from the drop formation images, indicating ideal operation of the printhead and successful voltage waveform design. The uniformity of the jetted drops allowed moderately high resolution lines of the ink to be printed, with drop sizes as small as 25 µm (Figure 7.45).
(a) (b) 200 µm (a) 100 µm (b)
Figure 7.45. (a) Jetted drops of BaTiO3 ink on ITO PET; (b) edge of BaTiO3/PMMA film on ITO glass.
7.3.3 Optimized ink formulations
A summary of all the finalized ink formulations is given in Table 7.5. Summaries of these formulations and the iterations attempted to reach them are given in APPENDIX H and APPENDIX I, respectively.
156
Table 7.5. Finalized ink formulations.
Fluid properties Solvent Ink Composition (w/w) system dp, max -1 µ (cP) Z En (mN/m) (nm)
. 34% PEDOT:PSS susp’n (1.3% in water) . 17% glycerol ~180 (no . 10% DMSO CNTs) PEDOT:PSS Aqueous . 10% SWCNT sol’n 2.2 30.2 12.5 ~0 /SWCNTs (0.4 % in ~400 (due water/SLS) to CNTs) . 0.5% SLS . 0.5% DF-110D . 28% H2O
. 5% BaTiO3 . 33% ethanol . 28% MMA BaTiO . 0.5% Surfynol CT- 3 Organic 9.6 31.7 ~30 2.9 ~0 PMMA 324 . 0.5% Disperbyk 111 . 33% PEG 300
. 40% ZnS:Mn (2.5% in water/TGA/NaOH) . 15% ZnS:Mn Aqueous butoxyethanol 12.1 32.1 ~16 2.3 ~0 PVP . 10% isopropanol . 0.1% PVP 1,300,00 . 34.9% H2O
. 40% ZnS:Cu (2.5% in toluene/olelyamine ZnS:Cu /3-MPA) Organic 0.97 28.4 ~33 27.5 ~0 PVK . 10% cyclohexane . 0.075% PVK . 49.925% chlorobenzene
157
7.4 Ink film formation
Following ink formulation, the procedure for determining ideal drop spacing, film topography, and film thickness for inkjet printing of the inks is presented in this section. In this case,
BaTiO3/PMMA ink was used as the model ink. Again, it must be noted that each ink had several substrates upon which it was deposited. These included an impermeable substrate (glass, in this case) for topography, functionality, and thickness observation, and any layers lying below it in the actual device structure. The surfaces upon which each ink was deposited are summarized in Table 7.6.
7.4.1 Drop spacing
BaTiO3 was to be deposited on top of the ZnS:Mn/PVP layer in the ACPEL stack, as well as on
ITO for capacitance testing and on glass for thickness testing. The drop sizes of BaTiO3 on each of these surfaces varied widely, as can be seen in Figure 7.46. On the surface of the ZnS:Mn/PVP layer, some interlayer mixing appeared to occur, and the droplet naturally widened as it dissolved into this layer. This was problematic for the construction of the ACPEL devices (see Section 6.7.1). Also, the different surface energies of each substrate meant that drops dried different ways; the ITO-based drop, for example, produced a pronounced coffee-ring effect, while the glass-based drop spread evenly. The different surface energies and roughnesses of these underlying layers were the determinants of drop spreading.
Slide glass ITO glass ZnS:Mn/PVP
55 µm 40 µm 65 µm
50 µm
Figure 7.46. Drop sizes of BaTiO3/PMMA ink on various substrates.
158
The size of the isolated droplets determined the line spacing in a printed layer, where line overlap contributed to film roughness, and too-wide spacing produced holes in the conductive layer. However, spacing the drops at the diameter of the individual drops did not necessarily result in the formation of a smooth film (Figure 7.47). The BaTiO3 ink on ITO, for example, did not form (c) (d) a uniform film at 40 µm drop spacing, and began to overlap excessively at 30 µm drop spacing – suggesting an ideal drop spacing of 35 µm. However, it formed a smooth film on slide glass at a drop spacing which was the same as its individual drop size (55 µm). The way the solids distributed in the drying droplet, as shown in the previous figure, determined the degree of overlap required to produce a film containing no holes.
On ITO glass
50 µm
Drop spacing: 40 µm 35 µm 30 µm 25 µm
On slide glass
Drop spacing: 60 µm 55 µm 50 µm
Figure 7.47. Printed lines of BaTiO3 ink (single jet, single layer) at different drop spacings.
As the lines are brought closer together, more and more material is deposited in a ridge. This exact problem was encountered by Haverinen et al. (2010), where it caused localized dimming of the devices. In some cases, the wetting of the surface was poor (as with ITO), so that close spacing of the lines was necessary to prevent pinholes. However, excessively closely-spaced lines were also expected to cause such an increase in film thickness as well as roughness. Ideal
159
drop spacing was established by jetting several lines of ink at different drop spacings in increments of 5 µm, ranging from the measured single drop diameter down to 50% of the single drop diameter (i.e. for a drop size of 60 µm, drop spacing ranging from 30-60 µm was tested. The resulting films were characterized using optical microscopy for uniformity (see APPENDIX (c) (d) (e) D). Ideal drop spacing was considered(b) to be when lines were fully merged (no holes) but not overlapping. Because the printer was limited to 5 µm increments in drop spacing, a compromise between overlap and merging of lines was often necessary. The drop sizes and spacings for various inks are given in Table 7.6.
Table 7.6. Summary of drop sizes and line spacing for all inks.
Ink Underlying layer Drop size (µm) Correct line spacing (µm) Glass 45 25 PEDOT:PSS BaTiO /PMMA 30 25 SWCNTs 3 ZnS:Mn/PVP 45 25 (aqueous) ZnS:Cu/PVK 50 45
BaTiO3 Glass 55 55 PMMA ITO 40 35 (organic) ZnS:Mn/PVP 65 65*
ZnS:Mn Glass 65 55 PVP ITO 50 45 (aqueous) PEDOT/SWCNTs 60 50
ZnS:Cu Glass 65 30 PVK ITO 45 30 (organic) PEDOT/SWCNTs 50 45
*the BaTiO3 ink damaged the ZnS:Mn/PVP layer extensively.
There was no consistent trend for any substrate in terms of drop size versus drop spacing, other than that drop spacing was generally smaller than the average drop size. Perhaps the most notable observation was that PEDOT:PSS wet most surfaces worse than the other inks (smaller drop sizes), and was not able to form smooth films except at narrower drop spacings. This may have been a result of the use of a surfactant rather than a fluid with low surface tension – wetting became a dynamic process rather than an instantaneous one as the surfactant molecules adhered to the substrate. As the droplets rapidly dried, they would not have yet had a chance to spread to their fullest diameter. Also, the (relatively) low solids content likely meant that capillary flow of solvent/particles to the droplet edges was not a dominant process, and that drops dried more in a
160
“dome” shape than a coffee-ring or flat morphology. Indeed, in inks with higher solids contents, the Marangoni effect was much more pronounced, as will be seen in the following section, whereas PEDOT:PSS/SWCNT ink formed smoother films.
7.4.2 Film topography
When a 3-D examination of film topography was made using optical profilometry, the problems with solute movement during drying became self-evident. These were observed on both a micro1.5- µm and macro-scale. In the latter case, the BaTiO3 ink’s chemistry was problematic upon film curing. The component that contributed primarily to viscosity and assisted with jetting, PEG 300, also made the jetted films difficult to rapidly cure and therefore affected their topography; indeed, the images of the BaTiO3 films shown above represent films which were not fully cured. PEG 300 has a very high boiling point compared to the volatile ethanol solvent, and took much longer to evaporate; also, the polymerization of MMA took some time. It is possible, in fact, that it did not evaporate entirely, and remained entrapped within the printed films even after they appeared to be dry. In any case, the slow curing caused a visible shift in film topography, inducing a macroscale version of the troublesome coffee-ring effect (Figure 7.48). A portion of the ink migrated to the edges of the printed pattern, forming a ridge of substantial height, compared to the average film thickness of the samples. Because of the long drying time of the PEG 300, the large surface tension gradient between PEG 300 and the other components (43.5 mN/m versus 28 mN/m and 23 mN/m for MMA and ethanol, respectively) caused these ridges to form (Tracton 2005). The non-ridged regions of the film had very high smoothness and uniformity; reduction of the effects of surface tension gradients by reformulation or immediate high-temperature curing would likely alleviate this ridging effect. . The ridges observed between printed lines in the drop spacing tests were not visible in the fully cured films. Therefore, the slow drying time and viscous co-solvent did assist in film leveling to a degree, but produced the ridging problem at the edge of the film, regardless, due to surface tension gradients. Film thickness averaged at 150 nm (not including the ridged edge of the sample).
161
nm 0.4 700 (b)
0.3
500 0.2
300 (µm) height Film 0.1
600 µm 450 µm 0 100 0 200 400 600 x-dimension (µm) 0 6.3 µm Figure 7.48. 3-D profile (left) and 2-D linescan (right) of a single printed BaTiO3 on slide glass. The arrow on the scale bar on the 3-D plot refers to the level of the glass surface.
(a) (b)
BaTiO3 cluster
10 µm 5 µm
Figure 7.49. SEM micrographs of printed BaTiO3 ink (one layer). (a) Surface at 70° tilt; (b) 0° tilt. Yellow circles on the high-magnification micrographs highlight some of the localized clusters of BaTiO3 where solvent concentrated.
The surface of the BaTiO3 film’s non-ridged regions was very smooth, and apparently pinhole- free. SEM imaging (Figure 7.49) shows the relatively smooth surface of the BaTiO3 films. More importantly, deep pinholes were not observed. If the films were to be used as gate dielectrics or insulators in a display, pinholes could cause catastrophic breakdown of the entire device. Localized thinness of the insulator might also be problematic, and some shallow pits were observed in the SEM images. These were likely a result of solvent vapour bubbles escaping from the film as it cured. The use of lower temperature vacuum-drying might alleviate this problem to a degree; however, it is unavoidable that bubbling of solvent vapours will occur in
162
any wet-processed film. The relatively shallow depth of the pits in the surface suggested that the presence of high-boiling PEG 300 assisted(a) in the leveling of such voids, as it flowed in(b) to fill them after the volatile solvents were removed. In regions where volatile solvents were more concentrated during curing, clusters of BaTiO3 were left behind – resulting in a non-uniform distribution of insulating material. Some of these clusters can be clearly seen in Figure 7.49b. The insulator was generally well dispersed and assembled, but small holes and some non- uniformity had occurred as a result of rapid ethanol evaporation.
1600
1400
1200
1000
800
600
Film thickness (nm)thicknessFilm 400
200
0 0 1 2 3 4 5 6 7 8 # of printed layers
Figure 7.50. BaTiO3/PMMA average dried ink film thicknesses on slide glass.
The thickness of the BaTiO3 layer was examined during over-printing of multiple layers, as well. Because a thicker layer (~1 µm) was needed for the deposition of an ACPEL, several layers of
BaTiO3 would be required. The thickness was estimated at the centre of the films (not at the raised edges) using the procedure outlined in APPENDIX C. As the number of layers increased, the thickness increased in a linear fashion; at 8 printed layers, the film was thick enough at ~1 µm (Figure 7.50) to apply in the ACPEL structure, according to the desired layer thicknesses specified earlier in Section 6.7. It was expected at this thickness that the film would be sufficienctly insulating to function in an ACPEL device. The other inks, which did not contain the high-boiling PEG 300, formed films with varying topographies. In each case, some degree of “coffee-ring” formation was observed, with the edges of printed lines forming raised edges,
163
resulting in a “peak-and-valley” structure. The peak-and valley structure was particularly rough and uneven, with large variations between the highest and lowest points, listed in Table 7.7.
Table 7.7. Printed film roughnesses and peak-to-valley differences in ZnS and PEDOT:PSS/SWCNT inks.
Ink Underlying layer Valley thickness (nm) Peak thickness (nm) RMS roughness (nm) Glass 130 400 40 PEDOT:PSS/ BaTiO3/PMMA 145 500 30 SWCNTs ZnS:Mn/PVP 25 300 70 ZnS:Cu/PVK 110 620 80
ZnS:Mn/ ITO 55 680 40 PVP PEDOT/SWCNTs 30 300 10
ZnS:Cu/ ITO 65 1000 300 PVK PEDOT/SWCNTs 65 1100 230
This problem was the most pronounced in the ZnS:Mn/PVP ink, which contained a higher loading of large-Mw polymer than either of the other inks. It was noticed in the ZnS:Cu ink and another trial ink that was prepared using only PVK that these had the worst topography of any – suggesting that the presence of dissolved polymers caused problems with coffee-ring formation. PEDOT:PSS/SWCNT ink, which contained no dissolved polymer, had the smoothest topography of any of the materials used, and no peak-and-valley formation was visible in the BaTiO3 inks either, supporting the hypothesis that dissolved polymers were responsible for this rough topography. Again, Haverinen (2010) discussed this precise issue during the printing of QDs; the approach in that work was to remove a fraction of the solids in the ink, resulting in smoother film topography. The hypothesis presented was that the larger amount of solvent present in the ink as a result of reducing solids content would prolong drying time and thereby aid in surface wetting. This is possible, considering that the worst-offending ink was ZnS:Mn/PVP, which had the lowest boiling-point solvents of any of the inks. However, by Haverinen’s explanation, only a very small amount of solvent was added (reducing solids concentration from 0.7 wt.% to 0.5 wt.% and replacing it with an additional 0.2 wt.% of solvent). This trivial amount of added solvent did not seem to support the claim of enough of an increased drying time to affect film topography. It was more likely that better leveling occurred in inks with the proper fluid characteristics, where a lower viscosity allowed more rapid flow of solvent and better film formation – as evidenced by the large drop sizes observed for ZnS:Cu/PVK (µ = 0.97 cP) (Table 7.6) and the superior film formation of PEDOT:PSS/SWCNTs (µ = 2.2 cP). The high-viscosity
164
inks, containing BaTiO3 and ZnS:Mn/PVP, experienced the most drastic problems with comparatively rougher surfaces. Examples of ink topography on glass slides are shown in Figure 7.51.
PEDOT:PSS/SWCNTs ZnS:Cu/PVK
ZnS:Mn/PVP BaTiO3/PMMA
600 µm 450 µm
Figure 7.51. Topography of single printed layers of all inks on slide glass.
Besides PEDOT:PSS/SWCNTs, most of the other inks demonstrated a relatively high degree of roughness, and in some cases, isolated ridges or features of several hundred nanometres in height. The edges of printed lines observed during optical microscopy to determine drop spacing were clearly visible as repeated ridges, the extreme example of which is seen in the ZnS:Mn/PVP ink. In this case, the finished film was to be relatively thick (~20 µm), meaning that topography was not as much of a concern – so the ink was not reformulated to improve its leveling ability. The relatively high smoothness of the ZnS:Cu/PVK and PEDOT:PSS/SWCNT films, which were to be used in the thin (sub-µm) film DC-LEDs was suitable.
165
The film thickness was found to vary almost linearly with the number of printed layers on every substrate, meaning that films of controlled thickness were readily deposited. Where a single ink film was thicker than the layer’s targeted thickness, only a single layer was used. Thus, the parameters for inkjet-printing of all of the device structures outlined in Chapter 6 were finalized and summarized in Table 7.8. It was noted that the ink layer thicknesses were best-suited to thin- film devices, as many print passes were required to provide the thick films characteristic of PELs.
Table 7.8. Ink layer thicknesses and layers required for device construction. Single layer Layer & desired Number of printed Ink Underlying layer Device type thickness (nm) thickness (nm) layers required Glass All 130 Anode (100) 1 PEDOT:PSS BaTiO /PMMA ACPEL 145 Cathode (1000) 7 SWCNTs 3 ZnS:Mn/PVP DCPEL 25 Cathode (1000) 40 (aqueous) ZnS:Cu/PVK DC-LED 110 Cathode (1000) 9
BaTiO3 PMMA ZnS:Mn/PVP ACPEL Dissolved (n/a) Insulator (1000) n/a (organic)
ZnS:Mn ITO PEL 55 Emitter (20,000) 363 PVP PEDOT/SWCNTs PEL 30 Emitter (20,000) 667 (aqueous)
ZnS:Cu ITO DC-LED 65 Emitter (200) 3 PVK PEDOT/SWCNTs DC-LED 65 Emitter (200) 3 (organic)
7.4.3 Interlayer interactions
When printing ACPEL structures several issues became apparent. Firstly, the BaTiO3 ink completely dissolved through the ZnS:Mn ink below it, regardless of the degree of cross-linking. Rub testing further confirmed that heat- or UV-treated films of ZnS:Mn/PVP were completely pervious to the BaTiO3 ink’s solvents. A second major issue was with the number of printed layers necessary to produce the thick film necessary for PEL function, which was exacerbated by the small average thickness of the ZnS:Mn films (due to the peak-and-valley formation). Time consumption aside, overprinting with a large number of layers also presented difficulties in maintaining print resolution. When a large number of ZnS:Mn/PVP layers were jetted onto glass, the edges of the pattern began to lose sharpness. As droplets impacted the wet ink on the
166
surface during subsequent print passes, ink was ejected from the previous films and landed randomly on the substrate surface (Figure 7.52).
Visible # layers UV 1 Visible UV 200 µm
2
2 3
4
5
10
20
50 (a) (b)
Figure 7.52. Multiple layers of ZnS:Mn/PVP ink printed on (a) aluminum foil (1-50 layers) and (b) ITO PET (50 layers), showing edge splattering with excessive overprinting.
Because of these issues associated with the production of ACPELs, this structure was dropped.
The BaTiO3 ink developed in this study may find future use as a dielectric in an AC-driven device using non-printed micron-sized phosphors, or as a gate dielectric in a transistor structure.
The rub-test was carried out on the ZnS:Mn/PVP layer as well with PEDOT:PSS/SWCNT ink – resulting in its complete dissolution, even after heat- and UV-crosslinking. Overprinting of the ZnS:Mn/PVP layer with even a single layer of PEDOT:PSS/SWCNT ink caused it to dissolve. So, with the inks developed in this work, a fully inkjet-printed DCPEL seemed unlikely to be realistically constructed. However, this issue could be addressed by vacuum-deposition of Al as top conductive layer.
167
ZnS:Cu/PVK (3-MPA) ZnS:Mn/PVP (TGA)
ITO
Ink droplet 600 µm 450 µm
0.15 0.06
0.1 0.04
0.05 0.02
0 0
-0.05 -0.02
Topographical height (µm) height Topographical -0.1 -0.04
-0.15 -0.06 0 200 400 600 0 200 400 600 x-dimension (µm)
Figure 7.53. ITO dissolution by ZnS:Cu/PVK ink, containing 3-MPA, and ZnS:Mn/PVP ink, containing TGA. Top: 3-D profiles of the surfaces; bottom: x-direction linescans across the droplets.
Another issue that arose with device deposition was the sensitivity of ITO to the acid (3-MPA or TGA)-bearing inks – i.e. the emitter species. As is shown in Figure 7.53, the 3-MPA in particular etched into the ITO’s surface and penetrated through its entire thickness, the ITO only being ~100 nm thick, as specified by the supplier. The etching effect of the TGA was not as severe, but it was still present. It is possible that the bases (olelylamine and NaOH) present in the inks were responsible for the etching effect as well. There was no way to further treat the ITO to prevent this from occurring; however, the dried PEDOT:PSS/SWCNT layers survived the rub tests from both of the ZnS inks with no apparent damage. Therefore, PEDOT:PSS was used as the anode in the place of ITO, further narrowing down the number of device structures that
168
were actually built and tested. While this was initially considered a problematic issue, the usefulness of PEDOT:PSS as a readily-patterned flexible material (as described in Paper 1) versus the brittle, photolithography-patterned ITO made it a better-suited candidate for all- printed devices. Also, the HOMO value of PEDOT:PSS, reported as 5.2 eV (Mihailetchi et al. 2003), is higher than that of plasma-treated ITO, which is typically around 4.7 eV (Shlaf et al. 2001). For the purpose of charge injection, this is energetically favourable, allowing holes to flow easily into the emissive material.
Based on the above considerations among the device structures described in Table 6.2, only three structures remained to be constructed and examined for their functionality. To facilitate charge injection into the anode and to provide a connection point for the power source, patterned ITO glass slides were used as the substrate to demonstrate the feasibility of the above fully inkjet printed EL devices. The device structures were:
DC-LED: PEDOT:PSS/SWCNTs – ZnS:Cu/PVK – Al DC-LED: PEDOT:PSS/SWCNTs – ZnS:Cu/PVK – PEDOT:PSS/SWCNTs DCPEL: PEDOT:PSS/SWCNTs – ZnS:Mn/PVP – Al
169
PEDOT:PSS/SWCNTs on glass (a) (b)
600 µm 450 µm
ZnS:Cu/PVK ZnS:Mn/PVP on spin-coated PEDOT:PSS/SWCNT ink on spin-coated PEDOT:PSS/SWCNT ink
PEDOT:PSS/SWCNTs on spin-coated ZnS:Cu/PVK ink
Al cathode (PEDOT:PSS/SWCNT ink dissolved ZnS:Mn/PVP layer)
Figure 7.54. Optical profilometry of printed ink layers, on the surfaces they would cover in printed devices. The underlying(c) surfaces were prepared by spin-coating of the respective(d) inks.
The film structures of the individual layers did not change much from those deposited on glass, when deposited on their respective underlying layers in the three device structures described above. The topography of the films remained smooth for PEDOT:PSS/SWCNTs (being
170
deposited on glass); while the topography remained as the “peak-and-valley” type for ZnS ( Figure 7.54) on PEDOT:PSS/SWCNTs. The high variability in the thickness of ZnS film presented the greatest concern for device functionality. Any regions where the film was thicker (peaks) would present localized points through which current could tunnel and short the device, or alternatively, would be too resistive to allow the passage of current and prevent device function. Haverinen (2010) observed that the raised regions in a single printed layer of CdS QDs would not light in a finished device. In the DCPEL, the thin regions would provide pathways through which hot electrons could arc between the electrodes, causing device burnout in those regions (Ono 1995). The pronounced peak-and-valley topography in the ZnS:Mn/PVP films made this latter concern a major issue. The necessity for overprinting each of the layers to reach a desired thickness alleviated this concern somewhat, as continued overprinting appeared to reduce the peak-and-valley formation (Figure 7.55). The usefulness of overprinting was limited by issues of edge resolution, as described above, and by the desired thickness of the printed film. However, in driving a printed device, it would most likely be more practical to have a thicker layer that required more voltage to achieve EL than to have a rough layer which could fail catastrophically. The eventual application of such a device might be difficult with a high driving voltage, but as a model for an all-printed device, some of the disadvantages of the inkjet printing method could be overcome by overprinting. In the following section, the relationship of overprinting to functionality is illustrated for PEDOT:PSS/SWCNTs and BaTiO3.
5 printed layers 10 printed layers
600 µm 450 µm
Figure 7.55. Reduction of “peak-and-valley” topography in ZnS:Cu/PVK films with successive overprints on slide glass. The topography of a single film of ZnS:Cu/PVK was presented in Figure 7.51.
171
7.5 Functional testing of individual layers
7.5.1 PEDOT:PSS/SWCNTs
From the examination of the PEDOT:PSS/SWCNT ink during formulation and substrate interaction studies, functionality for this layer has been already established. Because it was to be applied to glass, printed layers of PEDOT:PSS/SWCNTs on glass were tested for conductivity, demonstrating that an acceptable conductivity could be achieved even with a single printed layer (Figure 7.56). Successive overprinting yielded improvement in conductivity, as any gaps in the printed pattern or localized topography were smoothed out – this smoothening behaviour was universally seen for the printed films, as discussed in the previous section. Also, the introduction of successively more conductive “pigment” (i.e. SWCNT) into the films and the overlaying of a network of PEDOT:PSS chains and SWCNTs improved the connectivity of the conductive regions. The single print-pass conductivity was 4.6 S/cm, which is significantly lower than ITO (averaging at about 100 S/cm – a conductivity not achieved until 7 print passes with PEDOT:PSS/SWCNTs had been made). However, this was more than sufficient for application as an anode in an electronic device (Hsiao et al. 2008).
1000
100
10 Conductivity (S/cm) Conductivity
1 0 1 2 3 4 5 6 7 8 9 10 # of printed layers
Figure 7.56. PEDOT:PSS/SWCNT ink conductivity when printed on slide glass.
172
PEDOT:PSS/SWCNTs were also comparable with ITO in terms of their optical transmittance. ITO’s optical transmittance is generally 80% or lower, depending on the film thickness (Wei et al. 2001) as is the case with PEDOT:PSS, where more layers of PEDOT:PSS/SWCNTs produce a darker blue colour. A single printed film of PEDOT:PSS/SWCNT ink on cellulose acetate had a transmittance (measured by UV-visible spectrometry) in the visible range of ~83%.
Of course, the PEDOT:PSS/SWCNT ink could not match the conductivity of a metallic layer. When used as a cathode (10 printed layers), the conductivity across the plane of the film was 200 S/cm, several orders of magnitude lower than most metals. Furthermore, the direction of current flow in the functioning device was both across the plane of the film as well as across the film thickness. Depending on the alignment of the CNTs and PEDOT chains, conductivity might be anisotropic. It was expected as a result that a higher drive voltage would be required to observe uniform EL from the devices, overcoming resistive losses across the PEDOT:PSS/SWCNT electrode that would be negligible with a metal electrode.
7.5.2 ZnS
The spin-coated devices were prepared to qualify the phosphor materials and to ensure that they were actually electroluminescent. In these devices, printed-line topography and other pitfalls associated with inkjet printing were not expected to be relevant. However, because there is no way to pattern to spin-coated layers, the entire ITO slide was coated with the device, and the cathode (Al) was deposited directly onto the top of the device, with no point away from the emissive area to which to connect an electrode. Therefore, when electrodes were contacted to the device, they tended to penetrate the Al (and, indeed, the rest of the device) very rapidly, causing shorting. However, short-lived blue-green electroluminescence was observed from films of ZnS:Cu/PVK deposited on PEDOT:PSS-coated ITO. Momentary orange electroluminescence was also observed from ZnS:Mn/PVP films. Drive voltage in all cases was >30 VDC. Because of the poor device architecture, shorting occurred within a few seconds of successful illumination. However, the experiment did successfully demonstrate that the ZnS quantum dots were capable of electroluminescence under DC drive. For the sake of completeness, AC-driven devices were also tested (without the BaTiO3 layer, as it would have dissolved the underlying
173
ZnS:Mn/PVP film. No EL was observed from these devices. Unfortunately, the nature of the EL material, being nanoparticulate and unencapsulated, as well as the absence of an insulating layer, prevented these devices from functioning (Bredol & Dieckhoff 2010). Since the electroluminescent properties of DC-driven ZnS nanoparticles had been demonstrated, testing of the fully printed devices was undertaken.
7.5.3 BaTiO3
180 1600
160 1400 140 1200 120 1000 100 800 80 600 60
400 (nm)thicknessFilm
40 Dielectric constant(unitless) Dielectric 20 200
0 0 0 1 2 3 4 5 6 7 8 # of printed layers
Figure 7.57. Estimated relative dielectric constants of printed BaTiO3 films. Film thickness at the centre of the sample (in the non-ridged region) is also indicated.
Even though the BaTiO3 film was not used in the LEDs, its functionality was established in order to demonstrate its usefulness as an insulating layer in other electronic devices. In general, the
BaTiO3/PMMA films were sufficiently thick, uniform, and pinhole-free to demonstrate moderate dielectric constants (Figure 7.57). Dielectric constant () scaled with film thickness, which naturally scaled with the number of deposited layers, as per the definition of . In the case of a single printed film, which had some small voids and surface pinholes (Figure 6), current was able to tunnel through the film and no capacitance was measurable. However, as the pinholes were filled by subsequent printed layers and tighter packing of the larger amounts of BaTiO3 on the surface, the dielectric constant also increased. The occasional irregularities seen in the optical
174
profiles of the single BaTiO3/PMMA film were not present in the films with more layers, likely resulting in not only the increase in at higher film thicknesses, but also the nearly linear relationship between and film thickness above 5 printed layers. Above a certain number of layers, pinholes were completely covered by repeated print passes and breakdown became less and less likely. Therefore, deposition became a compromise between the pronounced ridging effect at the edge of the films and increased dielectric performance. The estimated value of and the film smoothness (in the non-ridged region) would be sufficient for application of the printed BaTiO3 films in a number of electronic devices, including emissive displays (Ono 1995).
7.6 EL device testing
Upon printing of the full LEDs, EL emission was achieved in the DC-LED device builds, with current-voltage (I-V) characteristics of these devices showing typical diode behaviour (Figure 7.58, Figure 7.59). The I-V curve for the fully-printed (PEDOT:PSS/SWCNT cathode) device was more linear than that for the Al-cathode device, suggesting that the relatively high resistance across the cathode was causing resistor-like behaviour. This deviation from diode behaviour might explain the weaker EL from this device. Weak blue-green emission characteristic of ZnS:Cu was visible over the 2 mm2 emissive area of the devices (Figure 7.60). In both cases, sudden shorting ended the functioning, which was visible during testing by their “sparkle”, representing multiple shorts during testing. Unfortunately, this shorting prevented further testing of the devices, and the weak emission strength made it difficult to obtain a meaningful EL spectrum. It was thought that the roughness of the ink layers was sufficient to cause localized channeling of current and prevent any further radiative recombination at the Cu2+ luminescent centres during the shorting events. However, the brief functioning of these devices did establish the feasibility of an all-printed LED using simple processing conditions, even if the EL was weak.
175
4 0.3
3.5 Al 0.25
3 ZnS:Cu/PVK
) PEDOT:PSS 0.2 2 2.5 substrate
2 0.15
Current (mA) Current 1.5
0.1 Luminance (cd/m Luminance 1 0.05 0.5
0 0 0 5 10 15 20 25 30 35 Voltage (VDC)
Figure 7.58. Current-voltage (I-V) and luminance-voltage (L-V) characteristics of PEDOT:PSS/SWCNT – ZnS:Cu/PVK – Al LED. Blue diamonds: I-V curve; red circles: L-V curve.
12.0 0.16
0.14 10.0
0.12
) 8.0 2 0.1
6.0 0.08
Current (mA) Current 0.06
4.0 PEDOT:PSS Luminance (cd/m Luminance 0.04 ZnS:Cu/PVK 2.0 PEDOT:PSS 0.02 substrate
0.0 0 0 5 10 15 20 Voltage (VDC)
Figure 7.59. Current-voltage (I-V) and luminance-voltage (L-V) characteristics of PEDOT:PSS/SWCNT – ZnS:Cu/PVK – PEDOT:PSS/SWCNT LED. Blue diamonds: I-V curve; red circles: L-V curve.
176
0 eV quantized quantization Ec = 1.5 E = 1.8
c LUMO = 2.1
LUMO = 1.9
Al = 4.1
PVK PEDOT:PSS Al = 5.1 HOMO HOMO = 5.3 ZnS:Cu
Ev = 5.5
nm
nm nm
nm nm
300 300
100 100 200
100 100 200 1µm 8 eV
PEDOT:PSS Al ZnS:Cu/PVK ZnS:Cu/PVK PEDOT:PSS PEDOT:PSS substrate substrate
Figure 7.60. Electronic band structures (top), device architecture (middle) and EL emission (bottom) of printed ZnS:Cu DC-LEDs. Left : with Al electrode, Right: with PEDOT electrode.
Another reason for the weak EL, even at relatively high drive voltages (> 30 V), was charge transfer into the emissive matrix. The bandgaps of the PVK and ZnS were similar (both emitting light in the blue region, when undoped), so charge injection from the semiconducting matrix into the nanoparticles themselves was not expected to be a problem. The HOMO value of PEDOT:PSS (5.1 eV) was also close to that of the PVK (5.3 eV), which in turn was close to that of the valence band energy of ZnS (5.5 eV), meaning that hole transfer was expected to be energetically favourable after charge injection into the PEDOT:PSS/SWCNT layer. However, electron injection through the cathode, be it Al or PEDOT:PSS/SWCNTs, had to overcome a
177
large energy barrier due to the wide bandgaps of both PVK and ZnS:Cu. Furthermore, the quantization of the ZnS:Cu nanoparticles significantly widened their bandgap, increasing the energetic difference between the Al work function or PEDOT:PSS LUMO and the ZnS:Cu conduction band. The quantization effect may also have made charge transfer from the PVK to the ZnS:Cu more difficult, as the quantized ZnS:Cu was a very wide-gap (4.0 eV) semiconductor – almost an insulator. The energetically unfavourable situation meant that higher voltages were required to drive the devices, and with their topographical irregularities, high voltage drive inevitably lead to device breakdown and shorting. Compared to typical LEDs, the drive voltages of >30 VDC were notably quite high.
With these problems in mind, there exist several possibilities for device improvement. The first of these concerns the cathode. Even when using Al, the work function is relatively high – and therefore provides a large barrier to charge injection. Other metallic cathodes with lower work functions, such as calcium (2.87 eV) or magnesium (3.66 eV) might alleviate this problem. However, the production of well-dispersed colloids of either of these materials has not been widely reported; colloidal Mg in THF was prepared by Kalidindi & Jagirdar (2009), and polydisperse colloidal Ca in THF was similarly prepared by Sanyal et al. (2012). The high reactivity of both of these metals might preclude them from incorporation into an ink; however, it is possible that they might be suitable candidates for a cathode material. Once again, the likely high sintering temperature of these metals would be an issue for any polymeric components/substrates. A more likely method to improve charge injection from the cathode would be to include an electron-transport layer of a polymer with a narrower bandgap than PVK. Of course, this approach necessitates the formulation of another ink, with orthogonal solvents once again being an added challenge. The presence of wide-bandgap quantized ZnS was the primary cause for energy band misalignment. The use of CdS, almost ubiquitous with solution- processed LEDs, avoids this issue (with certain colours, in any case) by controlling bandgap via particle size. However, for reasons discussed in Chapter 3, CdS was not considered ideal for this application; moreover, the stability of CdS or CdSe colloids was observed to be compromised when incorporating them into a multicomponent ink. A sample of orange-emitting (585 nm) CdSe colloids in water was mixed with the ink components containing in the aqueous ZnS:Mn
178
ink; the result was immediate flocculation of the colloids upon the addition of either isopropanol or butoxyethanol. PEG and glycerol had the same effect. Addition of the polymer binders (PVP and PVK) to both water- and toluene-soluble CdSe colloids resulted in immediate flocculation as well. So, Cd-based nanoparticles, likely due to their surface/cap chemistry, were not suitable for jetting, unlike the extremely stable ZnS:Mn and ZnS:Cu inks. Because energy level misalignment was the primary reason for poorer device performance, and the misalignment was a function of the materials being used, alteration of the cathode material or inclusion of new charge transport layers was the only practical way to consider improving functionality.
It is also possible that testing apparatus for the devices limited their luminescent intensity somewhat. The apparatus was capable of delivering 10 mA of current maximum, and the resistive losses across the devices because of roughness and the use of PEDOT:PSS rather than ITO or Al meant that only a fraction of that current was able to flow through the layers and induce radiative recombination. In previous studies with DC-LEDs using ZnS:Cu nanoparticles – most notably, that of Schrage et al. (2010) – significantly more current was drawn across the devices. Looking at Figure 7.59, the full 10 mA were drawn at maximum luminance; if more current had been available at lower voltages, stronger EL may have been produced at a lower potential. However, the use of self-limiting testing apparati caused immediate shorting, as the sources delivered as much current as needed to drive the devices, and any topographical flaws caused a large amount of current to be drawn, breaking down the layers.
The use of PVP as the binder polymer in the DCPEL structure exacerbated the need for high- voltage drive and the resulting failure of the devices. No significant EL emission was visible from these devices (Figure 7.61). Again, the I-V curve appeared more linear, suggesting resistor rather than diode behaviour, due to the thick layer of non-conductive PVP. Although non- conductive binders such as PMMA have been previously used (Schrage et al. 2010), the amount of current required to drive such devices far exceeds that output by the testing apparatus (>40 mA), and the devices also required high potential to drive. A different power source may be sufficient to drive these devices successfully. It would be ideal to be able to use the water- dispersible ZnS:Mn quantum dots as an emissive layer for ease of processing, and because of the difficulties mentioned above with phase transferring the ZnS:Mn quantum dots to the organic
179
phase. However, the interlayer dissolution issue which arose with overprinting of PEDOT:PSS onto the PVP-bound layer remains to be addressed.
0.02 0.16
Al 0.018 ZnS:dopant/PVP 0.14 0.016 PEDOT:PSS substrate 0.12
0.014
) 0.1 2 0.012
0.01 0.08
Current (µA) Current 0.008
0.06 Luminance (cd/m Luminance 0.006 0.04 0.004
0.02 0.002
0 0 0 5 10 15 20 25 Voltage (VDC)
Figure 7.61. Current-voltage (I-V) and luminance-voltage (L-V) characteristics of PEDOT:PSS/SWCNT – ZnS:Mn/PVP – Al DCPEL. Blue diamonds: I-V curve; red circles: L-V curve.
The use of inkjet printing as the preparation method for these devices limited their functionality in several ways. The first, and most obvious, is the topography of the resulting films, which in no way approached the smoothness of typical LED layers. The thickness of the films was well- controlled, but the presence of ridges between printed lines was not. When stacking layers, any topographical effects between them are amplified, particularly if the droplet spacing is the same or close to the same for those layers. A slight offset in overprinted layers (i.e. by half of the drop spacing) might partially alleviate this issue. However, judging by the work presented by Haverinen et al. (2010) and Wood et al. (2009), some degree of ridging in printed nanoparticle layers is almost unavoidable, so a better approach to improving functionality is to consider
180
energetic efficiency. Overprinting to improve surface properties resulted in devices through which no current was able to flow; the I-V plots are not presented here because they simply show a minimal current flow, no diode behaviour and no luminance. The added thickness of these layers further reduced the energetic favourability of carrier recombination due to increased resistance (Schrage et al. 2010).
Energetic efficiency can be improved primarily by clever materials selection. Materials selection, however, is limited by what is jettable (and orthogonally jettable), and it is here that the major stumbling block in the way of printed LEDs and PVs is encountered. In this study, the use of ZnS meant that a material was suitable for jetting which was readily prepared by wet methods, highly luminescent, and nearly monodisperse. However, it also meant that a wide bandgap material was the luminescent centre, requiring a large amount of energy to overcome barriers to charge transfer. Furthermore, the binder matrix was also a wide-bandgap material. The preparation of a practically applicable fully printed device presents a challenge for these reasons. The inkjet printing process may be limited or even completely crippled in its capacity to produce these devices because of these definitive problems. It is likely that single-layer devices with improved layer smoothness are the most realistic approach to optimizing these devices for future application.
181
Chapter 8 8 Conclusions
A model electronic device, a DC-driven LED, was prepared using inkjet printing to produce each of its layers. The device established a proof-of-concept; i.e. that an entire electronic device could be built using a single unit operation at ambient conditions with solution-based materials. The device also served to illustrate the many challenges involved with transferring conventionally vacuum-processed materials to the liquid phase, and further, to jettable inks.
The fundamental contribution of the work can be stated as the establishment of a detailed sample procedure for the development of a desired functional, printed material from chemical precursors. Of course, the demonstration of a fully-printed LED was also a major contribution, but this device requires further optimization before it becomes practically applicable. The many considerations of materials selection, dispersion, ink formulation, deposition, and layer formation comprise the greater part of the difficulty of inkjet manufacturing. The functionality of the resulting devices is entirely dependent on each of these prior steps, and so, this work has sought to comprehensively outline the means by which each step may be successfully undertaken.
The advantages of having such an outline to inkjet processing are manifold. Firstly, this work attempted to illustrate the difficulty of moving a material from the bulk phase to the dispersed liquid phase – particularly with traditionally-used inorganic semiconductors and insulators. This stage is likely the limiting factor in the success or failure of ink development. Secondly, the complexity and time-consuming nature of ink formulation is often overlooked in previous reports on inkjet deposition. Particularly in the case of commercial application and production of inks, repeatability and reliability of print is absolutely vital; therefore, the stability of ink dispersions and the maintenance of their fluid properties must be established. As is often the case with research endeavours, the practical application and scale-up of the technologies being researched may be overlooked, but in this case, the establishment of industrially viable ink formulations was considered an important aspect of the work. Also, the use of multiple ink layers pressed home the necessity for early consideration of device structures; the many different layer stacks that might
182
use the same materials in different orders required the production of inks with different solvent resistances. Thirdly, having an established routine for determining ideal drop spacing to achieve a certain topography meant that inks could be rapidly tested and used or discarded depending on their jetted performance. Lastly, several rapid but effective means for determining layer functionality before device construction were outlined, including capacitance testing and conductivity testing, even on porous substrates. It is hoped that by following this procedural example, electronics printing will become a better-streamlined process and that opportunities for utilizing all of the advantages of the inkjet printer will be realized.
8.1 Major findings
The use of paper and conventional paper printing and coating techniques was suitable for the production of paper-based AC-driven powder ELDs, as is described in detail in Paper 6. By avoiding the use of ITO and Al, the devices were readily produced at ambient conditions. However, they still presented difficulties in fabrication, due to the need for masking and the wastefulness of most coating methods. This served as the motivation for the examination of the exact procedure needed to move from these bulk materials to inkjet-printable materials. There were numerous findings at each stage of the ink development. To summarize the absolute milestones of the work: the successful preparation and deposition of a functional conductor, semiconductor, and insulator, and their integration into a printed LED. However, the study of each of these inks as models for the stages of the LED’s development yielded several other findings.
In the case of the PEDOT:PSS/SWCNT ink, conductivity of inkjet-printed films was shown to not only be a function of the conductivity of its constituents but also depended on the jetting characteristics and ink-substrate interactions. In particular, ideal droplet formation and ink spreading were found to play a dominant role in determining the film resistance, with less pronounced effects resulting from chemical/conformational changes in the PEDOT:PSS molecule. Optimizing viscosity and surface tension are needed both for acceptable jetting and for maximizing the connectivity and hence conductivity of such films. However, it was first
183
observed with the PEDOT:PSS ink that the reduction of viscosity to a value significantly lower than that recommended for inkjet printing resulting in better spreading of the ink – improving both film topography and conductivity. The addition of SWCNTs assisted in improving conductivity, as well, by providing long conductive pathways between isolated conductive regions of PEDOT:PSS. However, the use of high-aspect-ratio materials such as CNTs necessitates closer examination of their successful delivery to the substrate, due to the filtration and jetting stages of the process, which remove most large particles.
If paper substrates are to be considered as realistic supports for electronic devices, the lessons learned by observing their effect on PEDOT:PSS/SWCNT film conductivity must be considered. Paper’s absorbent and non-conductive nature made it a major hindrance to the transport of charge through the PEDOT:PSS/SWCNT layers; similar behaviour could be expected with any conductive material. Internal sizing or surface coating improved the ink holdout and therefore resulted in a higher conductivity by forming a thin layer of PEDOT:PSS/SWCNTs on the surface of the paper. Care should be taken in the treatment of paper with charged additives which may chemically interact with the conductive species, such as PDADMAC and PEI, in this case. The usually negatively-charged colloids in many inks would encounter similar problems.
However, on paper, values of conductivity were estimated from the dimensions of the ink layers. It should be emphasized that the key obstacle to objectively quantify the conductivity of printed PEDOT:PSS on paper arises from the irregularity and unpredictability of the ink distribution. Therefore, ideally, to optimize the performance of conductive papers, the uniformity of the paper itself needs to be controlled as tightly as possible. It is likely that the best result that can be offered for a given paper type is a range of conductivity values, which cannot be compared in an absolute sense to the conductivity of typical PEDOT:PSS/SWCNT films on impermeable surfaces. The smooth films formed by PEDOT:PSS/SWCNT ink on such impermeable surfaces, as well as its good electrical conductivity – suitable for use as an anodic material – suggested that the details of formulation that were refined had a major bearing on ink function. The resulting PEDOT:PSS/SWCNT ink emphasized the need to make a careful study of ink properties in order to maximize its functionality, even in adverse conditions, such as on the surface of paper. It also emphasized that even minor components – in this case, surfactants – have a major bearing on ink
184
function, once again reiterating the necessity for careful formulation. The ink in its current incarnation could be widely applied as an anode, charge transport layer, or paper-based electrode in a wide variety of electronic devices.
The insulating ink, although not included in the final LEDs, performed its intended role as a standalone material. BaTiO3/PMMA composite films with high dielectric constants were successfully printed. The smoothness, uniformity, and resolution of the printed layers was sufficient to consider this method of dielectric film fabrication a realistic alternative to vacuum- deposition conventionally used. This ink avoided the pitfalls of sol-gel processing and illustrated a single-step means of depositing a crystalline insulating layer. It also served as an example of the difficulty of forming a smooth ink film while maintaining jettability. The detrimental effect of surface tension gradients, even on the macro-scale, was shown. These issues with film formation served as examples for refinements to ink formulation in the ZnS-bearing inks. Also, the positive effect of layer overprinting on functionality was shown here, and replicated in the PEDOT:PSS/SWCNT ink.
Aqueous synthesis has been successfully employed to produce brightly luminescent ZnS:Mn and ZnS:Cu nanoparticles. However, the use of the aqueous method requires careful tailoring of the synthesis procedures to ensure good performance – sufficient dopant and small particle size being the primary concerns. A simple and environmentally friendly synthesis method is practicable for this optoelectronic material, and can produce nanoparticles not only in an aqueous system, but also readily provide a means for their redispersion in a variety of solvents. It was found that the use of 3-mercaptopropionic acid (3-MPA) as a capping agent, at controlled pH, temperature, and reaction time, was suitable for forming ZnS nanoparticles of sufficiently small size to be redispersible in an aqueous inkjet ink. The particles produced using this method were so small and well-passivated that their energy states were quantized due to the confinement of charge carriers, resulting in bright PL emission that was greatly blue-shifted for ZnS:Cu (510 to 470 nm). Effective passivation was achieved by bonding of the sulphur atom in 3-MPA to the S2- vacancy sites, resulting in primarily characteristic dopant emission with minimal emission in the blue range. Their small size and dispersity were crucial in formulation of an inkjet ink containing a relatively high loading of solids. The small size of the particles in the ink would
185
allow for use of nozzles of significantly smaller diameter and resulting higher resolution, if needed. The particles remained so well-passivated that even loading the dispersions with polymer could not cause them to become unstable, making them ideal for incorporating into a binder polymer. Printing of these nanoparticles produced brightly photoluminescent films of controlled thickness, albeit with poor topography due to solids migration during drying. Again, the method of overprinting assisted in reducing this effect – however, this remained as a problem for device function. Also, the wide bandgap of the quantized particles made it difficult to transfer charge from the electrodes and binder matrix into the luminescent centres. If used outside of LEDs, the ZnS layers have many possible applications – a particular example of this that was considered was the idea of anti-counterfeiting. The strong PL of the ZnS particles, at a wavelength determined by the degree of quantization and a peak intensity determined by the wavelength of the interrogator, could provide a means of identity verification for any printed surface.
8.2 Recommendations & future work
In order to improve the poor functionality of the LEDs, which was the major limitation of results produced by the work, there are several approaches which might be taken. Firstly, the improvement of ink leveling would not only probably improve the electronic characteristics of the devices, but it would also normalize the layer thicknesses. Having tightly-controlled and consistent layer thicknesses would assist greatly in applying these films to different electronic devices, such as thick-film devices like ELDs. This might involve surface treatments of underlying layers, or minor adjustments to the ink formulations.
Leveling was particularly poor in the aqueous ZnS:Mn/PVP layers. Ideally, aqueous inks would be used to build the entire device, due to their ease of processing and handling; PEDOT:PSS/SWNCT layers demonstrated that they were not soluble when overprinted with aqueous inks, as PEDOT is not water soluble. However, this implies that the emitter layer must somehow be rendered insoluble. The use of in-place polymerization with a printed initiator and cross-linker of a suitable binder polymer might serve to alleviate this problem.
186
Regardless of the film topography, however, charge transfer still remains the major hurdle to better LED function. The use of another narrow-bandgap emitter material such as CdS is a possibility for encouraging better charge transfer into the luminescent centres. If ZnS is to be used, however, due to its avoidance of heavy metals and ease of synthesis, a power source capable of delivering higher current is likely the best way to improve emission intensity. Also, an examination of the problems with phase-transfer of ZnS:Mn would assist in getting this material into the printed LED structure as well. To counter the effects of resistive losses, a printed metal cathode might be used. On glass, a sintered metallic ink would be suitable; on polymers, a novel means of low-temperature sintering would have to be applied, with the assumption that it would yield films of greater conductivity than the currently-used PEDOT:PSS/SWCNTs. Also, careful selection of a material with a lower work function to assist in electron injection might also bring down the operating voltage. It was concluded that the physical structure of the device was not so much of an issue to observing function as was the absence of a well-suited testing platform, and possibly a metallic cathode.
A structure that has been considered recently as an alternative – which should be tested to confirm the theories presented in this thesis about printed LED function – involves the use of a silver ink which “self-sinters” in the presence of a cationic polyelectrolyte. The formulation of a silver ink was undertaken but not included in this thesis, as no low-temperature-curing ink was successfully produced. However, a stable ink which was jettable was prepared. Using this ink, and another ink containing a cationic species (such as PDADMAC), a sinter-free Ag layer might be deposited as a cathode (Magdassi et al. 2010). The remaining layers would remain the same – ZnS:Cu/PVK and PEDOT:PSS. This would entail the formulation of a PDADMAC ink. The interesting aspect of this study lies in the use of PDADMAC as an underlying layer for the printed Ag. If a smooth surface – such as photo-paper – was used as the support for a thick PDADMAC film, which itself was smooth, the possibility of realizing a paper-based printed LED would be conceivable.
With an optimized printed LED, other inks could be considered to better its function, to construct other layers mentioned in this document. These might include charge transport layers, blocking layers, and an encapsulant. Because the focus of this work was simply to prove that an LED
187
could be printed by following a rigorous ink-formulation scheme, optimization was not considered. Not only might this involve the production of new inks, it might also involve changing deposition conditions, to avoid dust, humidity, and excessive handling. At this point, the fabrication process was very rough and rudimentary, but collaborative efforts with expertise in physics, materials science, and electrical engineering might assist in elaborating upon the opportunities and limitations associated with this extremely young technology.
Regardless of the printed LEDs’ mediocre performance, this technology still presents many unique possibilities and advancements in the fields of ink development and printed electronics. The limitations on resolution caused by drop size mean that high-resolution electronics manufacturing, on a scale similar to that of photolithography/shadowmask techniques, is not practical. Also, the ever-present issue of orthogonality means that overlaying of many layers – such as, in a display, the emissive material, transistor layer, colour filter, and so forth – becomes progressively less feasible. Therefore, a good possibility for application of the LEDs prepared in this work is for low-information-content displays, or for lighting. The flexibility of the printed materials, and the infinite patterning possibilities, mean that printed passive-matrix, single-pixel, or lamp-type LEDs are realistic applications. As demands for low-energy lighting and lighted signage increase, simple, rapid fabrication and especially patterning techniques such as those outlined here may become dominant. Even in the context of more complex non-emissive LCDs, low-energy, large-area, durable backlighting is an area of intense focus; most LCDs now contain LED backlights. With LCDs currently ubiquitous with consumer displays, the rapid fabrication of simple backplanes represents a unique opportunity to utilize printed LED technology. In many consumer electronics, also, backlit keyboards, accent lighting, and indicators currently use filtered fluorescent, incandescent, or LED lighting, behind a patterned stencil. Direct printing of such accents avoids material and energy wastage. The use of environmentally inert and non- cytotoxic ZnS as the emissive material also presents a unique advantage over Cd and Pb-based quantum dots. With a heavy focus on recycling and reuse of electronic components, LEDs based solely on polymers and inert inorganics are of particular commercial interest. Finally, with the groundwork laid for novel substrates incorporating these materials – paper being the most notable – the idea of biodegradeable and disposable electronics may be realized.
188
This work has ideally served as a stepping-stone for the development of fully-printed electronics in outlining the challenges and opportunities associated with their production. While not idealizing either materials or structure, the LEDs produced demonstrated the process of materials and ink development, and the bearing these stages have on the eventual production of a printed device. It is anticipated that such incremental advancements in the field, from the perspective of printing science and technology, will contribute to the refinement and optimization of future work on such devices, and that the detailed and rigorous map from raw materials to printed electronic layers will assist others in their respective studies in this fascinating field.
189
References
Adachi, D.; Hama, T.; Morimoto, T.; Toyama, T; Okamoto, H. (2008): Orange electroluminescence from chemically synthesized zinc sulfide nanocrystals doped with manganese, J. Non-Crys. Sol. 354, 2740. Adachi, D.; Hama, T.; Toyama, T; Okamoto, H. (2007): Electroluminescent properties of chemically synthesized zinc sulfide nanocrystals doped with manganese, J. Mater. Sci.: Mater. Elec. 10, 1. Adamovich, V.; Cordero, S.; Djurovich, P.; Tamayo, A.; Thompson, M.; d’Andrade, B.; Forrest, S. (2003): New charge-carrier blocking materials for high-efficiency OLEDs, Org. Electron. 4(2-3), 77. Agarwal, M.; Lvov, Y.; Varahramyan, K. (2006): Conductive wood microfibers for smart paper through layer-by-layer nanocoating, Nanotech.17, 5319. Aguey-Zinsou, K.; Ares-Fernandez, J. (2008): Synthesis of colloidal magnesium: a near room temperature store for hydrogen, Chem. Mater. 20(2), 376.
Al-Dahoudi, N.; Aegerter, M. (2006): Comparative study of transparent conductive In2O3:Sn (ITO) coatings made using a sol and a nanoparticle suspension, Thin Solid Films 502, 193. Allen, M.; Aronniemi, M.; Mattila, T.; Alastalo, A.; Ojanpera, K.; Suhonen, M.; Seppa, H. (2008): Electrical sintering of nanoparticle structures, Nanotechnology 19(17), 1. Althues, H.; Palkovits, R.; Rumplecker, A.; Simon, P.; Sigle, W.; Bredol, M.; Kynast, U.; Kaskel, S. (2006): Synthesis and characterization of transparent luminescent ZnS:Mn/PMMA nanocomposites, Chem. Mater. 18(4), 1068. American Ceramic Society (2010): A day made of glass – made possible by Corning.
190
Angelo, P.; Farnood, R. (2011): Photoluminescent inkjet ink containing ZnS:Mn nanoparticles as pigment, J. Exper. Nanosci. 6(5), 473. Angelo, P.; Cole, G.; Sodhi, R.; Farnood, R. (2012): Conductivity of inkjet-printed PEDOT:PSS- SWCNTs on uncoated papers, Nordic Pulp & Paper Res. J. (in press). Angelo, P.; Farnood, R.; Chlebowski, M (2012): Mn incorporation into ZnS nanoparticles prepared in aqueous solution, J. Nanoparticle Res. (submitted). Ardebili, H.; Pecht, M. (2009): Encapsulation technologies for electronic applications, William Andrew Publishing, Norwich, New York. Asberg, P.; Inganas, O. (2003): Hydrogels of a conducting conjugated polymer as 3-D enzyme electrode, Biosensors & Bioelectronics 19, 199. ASTM Standard D150 – 98 (2004): Standard Test Methods for AC Loss Characteristics and Permittivity (Dielectric Constant) of Solid Electrical Insulation, ASTM International, Subcommittee D09.12, Volume 10.01, West Conshohocken, Pennsylvania. ASTM Standard D5402-06 (2006): Standard Practice for Assessing the Solvent Resistance of Organic Coatings Using Solvent Rubs, ASTM International, Subcommittee D01.46, Volume 06.02, West Conshohocken, Pennsylvania. Atkinson, A.; Doorbar, J.; Hudd, A.; Segal, D.; White, P. (1997): Continuous inkjet printing using sol-gel ceramic inks, J. Sol-Gel Sci. Tech. 8, 1093. Aubry, N.; Singh, P.; Janjua, M.; Nudurpati, S. (2008): Micro- and nanoparticles self-assembly for virtually defect-free, adjustable monolayers, Proc. Nat. Acad. Sci. 105(10), 3711. Aytouna, M.; Bartolo, D.; Wegdam, G.; Bonn, D.; Rafai, S. (2010): Impact dynamics of surfactant laden drops: dynamic surface tension effects, Exp. Fluids 48, 49. Bakasybramanian, N.; Subrahmanyam, A. (1991): Studies on evaporated indium tin oxide (ITO)/silicon junctions and an estimation of ITO work function, J. Electrochem. Soc. 138, 322. Bakueva, L.; Musikhin, S.; Hines, M.; Chang, T.; Tzolov, M.; Scholes, G.; Sargent, E. (2003): Size-tunable infrared electroluminescence from PbS quantum dot nanocrystals in semiconductive polymer, App. Phys. Lett. 82(17), 2895. Berger, L. (1996): Semiconductor materials, CRC Press, Boca Raton. Bergeron, V.; Bonn, D.; Martin, J. (2000): Controlling droplet deposition with polymer additives, Nature 405, 772. Bhandari, S.; Deepa, M.; Srivasta, A.; Joshi, A.; Kant, R (2009) : Electrochemical and optical properties of the poly(3,4-ethylenedioxythiophene) film electropolymerized in an aqueous sodium dodecyl sulfate and lithium tetrafluoroborate medium, J. Phys. Chem. B 113, 9416. Bhargava, R.; Gallagher, D.; Hong, X.; Nurmikko, A. (1994): Optical properties of Mn-doped nanocrystals of ZnS, Phys. Rev. Lett. 72(3), 416.
191
Biswas, P.; De, A.; Dua, L.; Chkoda, L. (2006): Work function of sol–gel indium tin oxide (ITO) films on glass, Appl. Surf. Sci. 253, 1953. Blackburn, J.; Barnes, T.; Beard, M.; Kim, Y.; Tenent, R.; McDonald, T.; To, B.; Coutts, T.; Heben, M. (2008): Transparent conductive single-walled carbon nanotube networks with precisely tunable ratios of semiconducting and metallic nanotubes, ACS Nano 2(6), 1266. Blazdell, P.; Evans, J.; Edirishinghe, M.; Shaw, P.; Binstead, M. (1995): The computer aided manufacture of ceramics using inkjet printing, J. Mater. Sci. Lett. 14(22), 1562. Borovitskaya, E.; Shur, M. (eds.) (2002): Quantum dots, World Scientific, Hackensack, New Jersey. Bowers, R.; Melamed, N. (1955): Luminescent centres in ZnS:Cu, Cl phosphors, Phys. Rev. Lett. 99(6), 1781. Brandrup, J.; Immergut, E. (eds.) (1989): Polymer handbook, 3rd ed., Wiley-Interscience, New York, New York. Bredol, M.; Dieckoff, H (2010): Materials for powder-based AC electroluminescence, Materials 3, 1353. Brinker, C.; Scherer, G. (1990): Sol-gel science: the physics and chemistry of sol-gel processing, Academic Press, New York, New York. Bronzcyk, A. (2003): Fabrication of an organic light-emitting diode on a paper substrate, NSF/SRC Engineering Research Center for Environmentally Benign Semiconductor Manufacturing, Thrust C. Brus, L. (1986): Zero-dimensional excitons in semiconductor clusters, J. Quantum Electron. 22, 1909. Bube, R. (1953): Electronic transitions in the luminescence of zinc sulfide phosphors, Phys. Rev. 90(1), 70. Bucknall, D. (ed.) (2005): Nanolithography and patterning techniques in microelectronics, Woodhead Publishing, Cambridge. Bulanyi, M.; Polezhaev, B.; Prokofev, T. (1998): The nature of Mn luminescence centers in ZnS single crystals, Semiconductors 32(6), 603. Bulanyi, M.; Kovalenko, A.; Polezhaev, B. (2002): Spectra of manganese emission centers in zinc sulphide, J. Appl. Spectroscopy 69(5), 747. Caglar, U. (2009): Studies of inkjet printing technology with focus on electronic materials, doctoral thesis, Tampere University of Technology, Finland. Calvert, P. (2001): Inkjet printing for materials and devices, Chem. Mater. 13, 3299. Campbell, S. (2001): The science and engineering of microelectronic fabrication, Oxford University Press, Oxford. Capek, I. Nanocomposite structures and dispersions. Elsevier (2006): Amsterdam
192
Carter, E.; Gardiner, F. (eds.) (2009): Polymer electronics: a flexible technology, Smithers, London, England. Caruso, F. (ed.) (2004): Colloids & colloid assemblies, Wiley, Weinheim, Germany. Chander, H. (2005): Development of nanophosphors – a review, Mater. Sci. Eng. R49, 113. Chang, T.; Guo, W.; Guo, X. (2005): Buckling of multiwalled carbon nanotubes under axial compression and bending via a molecular mechanics model, Phys. Rev. B 72(6), 64101. Chang, P. (2005): Improving performance in oligothiophene based semiconductor devices via functionality, PhD thesis, University of California, Berkeley. Changneng, Z.; Mingtai, W.; Fang, L.; Mingguang, K.; Li, G.; Weiwei, X.; Xiaoguang, Z.; Kongjia, W. (2005): A polymeric/inorganic nanocomposite for solid-state dye-sensitized solar cells, Plasma Sci. Tech. 7(4), 2692. Chen, Z.; Cotterell, C.; Wang, W. (2002): The fracture of brittle thin films on compliant substrates in flexible displays, Eng. Fracture Mech. 69(5), 597. Chen, Z.; Kobashi, K.; Rauwald, U.; Booker, R.; Fan, H.; Hwang, W.; Tour, J. (2006): Soluble ultra-short single-walled carbon nanotubes, J. Amer. Chem. Soc. 128, 10568. Chen, J.; Cranton, W.; and Fihn, M. (2011): Handbook of visual display technology, Springer, New York. Cheng, K.; Yang, M.; Chiu, W.; Huang, C.; Chang, J.; Ying, T.; Yang,Y. (2005): Ink-jet printing, self-assembled polyelectrolytes, and electroless plating: low cost fabrication of circuits on a flexible substrate at room temperature, Macromol. Rapid Comm. 26(4), 247. Cheng, Y.; Yang, S.; Hsu, C. (2009): Synthesis of conjugated polymers for organic solar cell applications, Chem. Rev. 109(11), 5868. Chien, A.; Speck, J.; Lange, F.; Daykin, A.; Levi, C. (1995): Low temperature/low pressure hydrothermal synthesis of barium titanate: powder and heteroepitaxial thin films, Journal of Materials Research 10(7), 1784. Cho, K.; Cha, T. (2009): Quantum dot ink composition for inkjet printing and electronic device using the same, US Patent #20090314991. Cho, Y.; Kim, H.; Yi, G.; Hong, J.; Jang, S.; Yang, S. (2009): Dispersion stabilization of conductive transparent oxide nanoparticles, Coll. & Surf. A: Physicochem. Eng. Aspects 336, 88. Cho, Y.; Yi, G.; Hong, J.; Jang, S.; Yang, S. (2006): Colloidal indium tin oxide nanoparticles for transparent and conductive films, Thin Sol. Films 515, 1864. Chow, E.; Herrmann, J.; Barton, C.; Raguse, B.; Wieczorek, L. (2009): Inkjet-printed gold nanoparticle chemiresistors: Influence of film morphology and ionic strength on the detection of organics dissolved in aqueous solution, Anal. Chem. Acta. 632(1), 135. Chung, S.; Jang, J.; Cho, J. (2010): All-inkjet-printed organic thin-film transistors with silver gate, source/drain electrodes, Jap. J. Appl. Phys. 50(3), S1.
193
Chung, S.; Kim, S.; Kwon, S. (2011): All-inkjet-printed organic thin-film transistor inverter on flexible plastic substrate, IEEE Electron Dev. Lett. 32(8), 1134.
Cibis, D.; Krueger, K. (2005): DOD printing of conductive silver tracks, 1st Int. Conf. Ceram. Interconnect Ceram. Microsys. Techn., Baltimore, Maryland. Clymer, N.; Asaba, S. (2004): A new approach for understanding dominant design: the case of the ink-jet printer, J. Eng. Tech. Management 25, 137. Coe-Sullivan, S. (2005): Hybrid organic-quantum dot thin film structures and devices, PhD thesis, Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Boston, Massachusetts. Coe-Sullivan, S.; Woo, W.; Steckel, J.; Bawendi, M.; Bulović, V. (2003): Tuning the performance of hybrid organic/inorganic quantum dot light-emitting devices, Org. Electron. 4, 123. Colclough, W. (1968): The chemistry of solvents used for ultrasonic cleaning, Ultrasonics 6(1), 21. Collins, P.; Avouris, P. (2000): Nanotubes for electronics, Scientific American 283(6), 62. Colvin, V.; Schlamp, M.; Alivisatos, A. (1994): Light-emitting diodes made from cadmium selenide nanocrystals and a semiconducting polymer, Nature 370(4), 354. Cranton, W.; Stevens, R. Thomas, C. (1998): AC thin film electroluminescent device, US Patent 5742322. Craster, R.; Matar, O.; Sefiane, K. (2009): Pinning, retraction and terracing of evaporating droplets containing nanoparticles, Langmuir 25(6), 3601. Crispin, X.; Jakobsson, F.; Crispin, A.; Grim, P.; Andersson, P.; Volodin, A.; van Haesendonck, C.; Van der Auweraer, M; Salaneck, W.; and Berggren, M. (2006): The origin of the high conductivity of poly(3,4-ethylene dioxythiophene)-poly(styrene sulfonate) (PEDOT-PSS) plastic electrodes, Chem. Mater. 18, 4354. Crispin, X.; Marciniak, S.; Osikowicz, W.; Zotti, G.; Denier van der Gon, A.; Louwet, F.; Fahlman, M.; Groenedaal, L.; De Schryver, F.; and Salaneck, W. (2003): Conductivity, morphology, interfacial chemistry, and stability of poly(3,4-ethylene dioxythiophene)- poly(styrene sulfonate): a photoelectron spectroscopy study, J. Poly. Sci. B: Poly. Phys. 41, 2561. Crone, B.; Campbell, I.; Davids, P.; Smith, D. (1998): Charge injection and transport in single- layer organic light-emitting diodes, Appl. Phys. Lett. 73, 3162. Croucher M.; Hair, M. (1989): Design criteria and future directions in inkjet ink technology, Indus. Eng. Chem. Res., 28(11), 1712. Cui, W.; Lu, W.; Zhang, Y.; Lin, G.; Wei, T.; Jiang, L. (2010): Gold nanoparticle ink suitable for electric-conductive pattern fabrication using in ink-jet printing technology, Coll. & Surf. A: Physicochem. Eng. Aspects 358(1-3), 35.
194
Cullity, B.; and Stock, S. (2001): Elements of X-ray diffraction, 3rd edition, Prentice-Hall Inc., New Jersey. Curling, C. (2006): Prototype line for manufacturing flexible active matrix, USDC 5th Flex. Disp. Microelec. Conf. Cutnell, J.; Johnson, K. (2008): Physics, 8th edition, John Wiley & Sons, Hoboken, New Jersey. D’Errico, G.; Lellis, M.; Mangiapia, G.; Tedeschi, A.; Ortona, O.; Fusco, S.; Borzacchiello, A.; Ambrosio, L. (2008): Structural and mechanical properties of UV-photo-cross-linked poly(N-vinyl-2-pyrrolidone) hydrogels, Biomacromolecules 9, 231. Dabbousi, B.; Rodrigues-Viejo, J.; Mikulec, F.; Heine, J.; Mattoussi, H.; Ober, R.; Jensen, K.; Bawendi, M. (1997): CdSe-ZnS core-shell quantum dots: synthesis and characterization of a size series of highly luminescent nanocrystallites, J.Phys. Chem. B 101, 9463. Daniel, A.; and Fotheringham, A. (2007): Microstructured conducting polymer coatings on plastic strip for contact electrodes: Formulation and tuning of sheet resistance, J. Appl. Polym. Sci. 104, 234. Daniel, R.; Berg, J. (2006): Spreading on and penetration into thin, permeable print media: application to ink-jet printing, Adv. Coll. Interface Sci. 123-126, 439. Davison, G.; Lane, B. (eds.) (2003): Additives in water-borne coatings, Royal Society of Chemistry, Cambridge. de Gans, B.; Duineveld, P.; and Schubert, U. (2004): Inkjet printing of polymers: state of the art and future developments, Adv. Mater. 16(3), 203. Dearden, A.; Smith, P.; Shin, D.; Reis, N.; Derby, B.; O’Brien, P. (2005): A low curing temperature silver ink for use in ink-jet printing and subsequent production of conductive tracks, Macromol. Rapid Comm. 26, 315. Delgado, A.; Gonzalez-Caballero, F.; Hunter, R.; Koopal, L.; Lyklema, J. (2005): Measurement and interpretation of electrokinetic phenomena, Pure Appl. Chem. 77(10), 1753. Denneulin, A.; Blayo, A.; Bras, J.; and Neuman, C. (2008): PEDOT:PSS coating on specialty papers: process optimization and effects of surface properties on electrical performances, Prog. Org. Coatings 63, 87. Denneulin, A.; Blayo, A.; Bras, J.; and Neuman, C. (2011): Infra-red assisted sintering of inkjet printed silver tracks on paper substrates, J. Nanoparticle Res. 13(9), 3815. Derby, B.; Reis, N. (2003): Inkjet printing of highly loaded particulate suspensions, MRS Bull. 28, 816. Deslandes, Y.; Pleizier, G.; Poire, E.; Sapieha, S.; Wertheimer, M.; and Sacher, E. (1998) : The surface modification of pure cellulose paper induced by low-pressure nitrogen plasma treatment, Plasmas & Polym. 3(2), 61. Destriau, G. (1936): Research on the scintillations of zinc sulphide to rays, J. Phys. Chem. 33, 587.
195
Di Risio, S.; Yan, N. (2007): Piezoelectric ink-jet printing of horseradish peroxidase: effect of ink viscosity modifiers on activity, Macromol. Rapid Comm. 28, 1934. Di Risio, S.; Yan, N. (2009): Adsorption and inactivation behaviour of horseradish peroxidise on cellulosic fiber surfaces, J. Colloid & Interface Sci. 338, 410. Dimitrakopoulos, C.;Mascaro, M. (2001): Organic thin-film transistors: a review of recent advances, IBM J. Res. Dev., 45,11. Dimitrijev, S. (2005): Principles of semiconductor devices, Oxford University Press, Oxford. Ding, X.; Li, Y.;Wang, D.;Yin, Q. (2004): Fabrication of BaTiO3 dielectric films by direct inkjet printing, Ceram. Inter. 30, 1885. Dobbelin, M.; Marcilla, R.; Salsamendi, M.; Pozo-Gonzalo, C.; Carrasco, P.; Pomposo, J.; and Mecerreyes, D. (2007): Influence of ionic liquids on the electrical conductivity and morphology of PEDOT:PSS films, Chem. Mater. 19, 2147. Dong, H.; Carr, W.; and Morris, J. (2006): An experimental study of drop-on-demand drop formation, Phys. Fluids 18, 1. Dresselhaus, M.; Dresselhaus, G.; Avouris, P. (eds.) (2001): Carbon nanotubes, Springer, New York, New York. Eastoe, J.; Dalton, J. (2000): Dynamic surface tension and adsorption mechanisms of surfactants at the air-water interface, Adv. Coll. Interf. Sci. 85, 103. Eftekhari, A. (2010): Nanostructured conductive polymers, John Wiley & Sons, Ltd, Hoboken, New Jersey. Eisenstein, E. (1997): The printing press as an agent of change, Cambridge University Press, United Kingdom. Electronics.ca Publications (2011): Flat panel displays: global strategic business report. Fakhri, N.; Tsyboulski, D.; Cognet, L.; Weisman, R.; Pasquali, M. (2009): Diameter-dependent bending dynamics of single-walled carbon nanotubes in liquids, Proc. National Academy of Sci., 106(34), 14219. Fan, B.; Mei, X.; and Ouyang, J. (2008) : Significant conductivity enhancement of poly(3,4- ethylenedioxythiophene):poly(styrenesulfonate) films by adding anionic surfactants into polymer solution, Macromolecules 41, 5971. Fan, Z.; Wei, T.; Luo, G.; and Wei, F. (2005): Fabrication and characterization of multi-walled carbon nanotubes-based ink, J. Mater. Sci. 40, 5075. Fardim, P.; Holmbom, B. (2005): ToF-SIMS imaging: a valuable chemical microscopy technique for paper and paper coatings, Appl. Surf. Sci. 249, 393. Fehse, K.; Walzer, K.; Leo, K.; Lovenvich, W. And Elschner, A. (2007): Highly conductive polymer anodes as replacements for inorganic materials in high-efficiency organic light- emitting diodes, Adv. Mater. 19, 441.
196
Filenkova, A.; Acosta, E.; Brodersen, P.; Sodhi, R.; Farnood, R. (2010): Distribution of inkjet ink componenets via ToF-SIMS imaging, Surface & Interface Anal. 43(1-2), 576. Forest Products Laboratory (FPL) (1964): Ignition and charring temperatures of wood, Forest Service, U. S. Department of Agriculture. Washington D.C. Fujifilm-Dimatix (2006a): Dimatix Materials Printer DMP-2800 Series User Manual. Fujifilm-Dimatix (2006b): Fujifilm-Dimatix ink tutorial. Fukai, J.; Shiiba, Y.; Yamamoto, T.; Miyatake, O.; Poulikakos, D.; Megaridis, C.; Zhao, Z. (1995): Wetting effects on the spreading of a liquid droplet colliding with a flat surface: experiment and modeling, Phys. Fluids 7, 236. Fukai, J.; Zhao, Z.; Poulikakos, D.; Megaridis, C.; Miyatake, O. (1993): Modeling of the deformation of a liquid droplet impinging upon a flat surface, Phys. Fluids A, 5, 2588. Gan, Y.; Liu, J.; Zeng, S. (2006): Transparent conductive indium tin oxide film fabricated by dip-coating technique from colloid precursor, Surf. Coat. Tech. 201(1-2), 25. Gane, P. (2004): Absorption properties of coatings: a selected overview of absorption criteria derived from recent pore network modeling, J. Disper. Sci. Tech., 25, 389. Gaponenko, S. (1998): Optical properties of semiconductor nanocrystals, Cambridge University Press, Cambridge, England. Gaponik, N.; Talapin, D.; Rogach, A. (1999): A light-emitting device based on a CdTe nanocrystal/polyaniline composite, Phys. Chem. Chem. Phys. 1, 1787. Garnett, E.; Ginley, D. (2005): Electrical and morphological properties of inkjet-printed PEDOT:PSS films, J. Undergraduate Res. 15, 24. Geim, A.; Novoselov, K. (2007): The rise of graphene, Nature Mater. 6(3), 183. German Flat Panel Display Forum (DFF) (2008): European technology: flat panel displays, 6th ed. Gracia-Espino, E.; Sala, G.; Pino, F.; Halonen, N.; Luomahaara, J.; Maklin, J.; Toth, G.; Kordas, K.; Jantunen, H.; Terrones, M.; Helisto, P.; Seppa, H.; Ajayan, P.; Vajtai, R. (2010): Electrical transport and field-effect transistors using inkjet-printed SWCNT films having different functional side groups, ACS Nano 4(6), 3318. Gullichsen, J. and Paulapuro, H (1999): Papermaking science and technology 6A: chemical pulping, TAPPI & Fapet Oy, Helsinki. Hadley, A. (n.d.) CombineZ software homepage. http://www.hadleyweb.pwp.blueyonder.co.uk/ Haider, M.; Pitters, J.; DiLabio, G.; Livadaru, L.; Mutus, J.; Wolkow, R. (2009): Controlled coupling and occupation of silicon atomic quantum dots at room temperature, Phys. Rev. Lett. 102, 046805. Ham, H.; Choi, Y.; Chee, M.; Cha, M.; and Chung. I. (2008): PEDOT-PSS/singlewall carbon nanotubes composites, Poly. Eng. Sci. 48, 1.
197
Hamaker, H. (1937): The London-van der Waals attraction between spherical particles, Physica A 10, 1058. Hamberg, I.; Granqvist, C. J. Appl. Phys. 60 (1986): R123-R159. Haranath, D.; Chander, H.; Bhalla, N.; Sharma, P.; Sood, K. (2005): Surface distribution studies and improved photoluminescence characteristics of silica coated ZnS:Mn nanophosphor layers, Appl. Phys. Lett., 86(20), 201904. Harizanov, O.; Harizanova, A.; and Ivanona, T. (2004): Formation and characterization of sol– gel barium titanate, Mater. Sci. Eng. B106, 191. Harrison, P. (2000): Quantum wells, wires, and dots: theoretical and computational physics, J.Wiley & Sons, Hoboken, New Jersey. Haverinen, H.; Myllyla, R.; Jabbour, G. (2009): Inkjet printing of light-emitting quantum dots, App. Phys. Lett. 94, 073108. Haverinen, H.; Myllyla, R.; Jabbour, G. (2010): Inkjet printed RGB quantum dot-hybrid LED, J. Display Tech. 6(3), 87. Haverinen, H. (2010): Inkjet-printed quantum dot light-emitting devices – towards display applications, PhD thesis, Faculty of Technology, Department of Electrical and Information Engineering, University of Oulu. He, G.; Pfeiffer, M.; Leo, K.; Hofmann, M.; Birnstock, J.; Pudzich, R.; Salbeck, J. (2004): High- efficiency and low-voltage p-i-n electrophosphorescent organic light-emitting diodes with double-emission layers, Appl. Phys. Lett. 85(17), 3911. Heaney, M. (2000): Electrical conductivity and resistivity, In: The measurement, instrumentation, and sensors handbook, CRC Press, Oxford, 43-1. Hebner, T.; Wu, C.; Marcy, D.; Lu, M.; Sturm, J. (1998): Ink-jet printing of doped polymers for organic light emitting devices, Appl. Phys. Lett. 72, 519. Hensema, E. (1997): Sizing and ink jet printing of European multi-purpose office paper, TAPPI Proc.: Eng. Papermakers Conf. 851. Hieronymas A. (2002): Thin film inorganic light-emitting diode, US Patent 20020153830. Higashimura, T.; Teranishi, H.; and Sawamoto, M. (1980): Living cationic polymerization of n- vinylcarbazole with iodine, Poly. J. 12(6), 393. Hines, M.; Guyot-Sionnest, P. (1996): Synthesis and characterization of strongly luminescing ZnS-capped CdSe nanocrystals, J. Phys. Chem. 100, 468. Holman, R.; Cima, M.; Uhland, S.; Sachs, E. (2002): Spreading and infiltration of inkjet-printed polymer solution droplets on a porous substrate, J. Coll. Interface Sci. 249, 432. Hong, C. (2000): Inkjet printed copper source/drain metallization for amorphous silicon thin-film transistors, IEEE Elec. Dev. Lett.21(8), 384. Hostetler, M.; Wingate, J.; Zhong, C.; Harris, J.; Vachet, R.; Clark, M.; Londono, J.; Green, S.; Stokes, J.; Wignall, G.; Glish, G.; Porter, M.; Evans, N.; Murray, R. (1998):
198
Alkanethiolate gold cluster molecules with core diameters from 1.5 to 5.2 nm: core and monolayerproperties as a function of core size, Langmuir 14, 17. Hoth, C.; Choulis, S.; Schilinksky, P.; Brabec, C. (2007): High photovoltaic performance of inkjet-printed polymer:fullerene blends, Adv. Mater. 19, 3973. Hsiao, Y.; Whang, W.; Chen, C.; and Chen, Y. (2008): High-conductivity poly(3,4- ethylenedioxythiophene): poly(styrenesulfonate) film for use in ITO-free polymer solar cells, J. Mater. Chem. 18, 5948. Huang, A. (2009): Recent advances in dielectric materials, Nova Science Publishers, Commack, New York. Huang, D.; Liao, F.; Molesa, S.; Redinger, D.; Subramanian, V. (2003): Plastic-compatible low- resistance printable gold nanoparticle conductors for flexible electronics, J. Electrochem. Soc.150, 412. Huang, L.; Huang, Y.; Liang, J.; Wan, X.; Chen, Y. (2011): Graphene-based conducting inks for direct inkjet printing of flexible conductive patterns and their applications in electric circuits and chemical sensors, Nano Res. 4(7), 675. Hutchison, D.; Morril, N.; Aten, Q.; Turner, B.; Jensen, B.; Howell, L.; Vanfleet, R.; Davis, R. (2010): Carbon nanotubes as a framework for high aspect ratio MEMS fabrication, J. MEMS, 19(1), 75. Hwang, J.; Oh, M.; Kim, I.; Lee, J.; and Ha, C. (2005): Preparation and characterization of ZnS based nanocrystalline particles for polymer light emitting diodes, Hwang, Y.; Son, Y.; Lee, Y. (2007): Printed circuit patterns of conductive polymer, Mol. Cryst. Liq. Cryst. 472, 113. Igarashi, T.; Isobe, T.; Senna, M. (1997): EPR study of Mn2+ electronic states for the nanosized ZnS:Mn powder modified by acrylic acid, Phys. Rev. B 56(11), 6444. Jaeger, R. (2002): Introduction to microelectronic fabrication, Prentice-Hall, Upper Saddle River. Jaeger, R.; Balock, T. (2003): Microelectronic circuit design, McGraw-Hill, New York. Jang, D.; Kim, D.; Moon, J. (2009): Influence of fluid physical properties on ink-jet printability, Langmuir 25, 2629. Jang, S.; Seo, Y.; Choi, J. Kim, T.; Cho, J.; Kim, S.; Kim, D. (2010): Sintering of inkjet printed copper nanoparticles for flexible electronics, Scripta Materialia 62(5), 258. Jansen, K.; Agterof, W.; and Mellema, J. (2001): Viscosity of surfactant-stabilized emulsions, J. Rheol. 45(6), 1359. Jensen, G.; Krause, C.; Sotzing, G.; Rusling, J. (2011): Inkjet-printed gold nanoparticle electrochemical arrays on plastic: application to immunodetection of a cancer biomarker protein, Phys. Chem. Chem. Phys. 13(11), 4888.
199
Jeong, M.; Byeon, C.; Cha, O.; Jeong, H.; Han, J.; Choi, Y.; An, K.; Oh, K.; Kim, K.; Lee, Y. (2008): Purity measurement of single-walled carbon nanotubes by UV-vis-nIR absorption spectroscopy and thermogravimetric analysis, Nano Brief Rep. Rev. 3(2), 101. Jeong, S.; Song, H.; Lee, W.; Choi, Y.; Ryu, B. (2010): Preparation of aqueous Ag Ink with long-term dispersion stability and its inkjet printing for fabricating conductive tracks on a polyimide film, J. App. Phys. 108, 1. Jia, Q.; Miranda, F.; Oates, D.; Xi, X. (2000): Materials issues for tunable RF and microwave devices, Proc. Mater. Res. Soc. 603, 181. Jiang, D.; Cao, L.; Liu, W.; Su, G.; Qu, H.; Sun, Y.; and Dong, B. (2009): Synthesis and luminescence properties of core/shell ZnS:Mn/ZnO nanoparticles, Nanoscale Res. Lett. 4, 78. Jindal, Z.; Verma, N. (2008): Photoluminescent properties of ZnS:Mn nanoparticles with in-built surfactant, J. of Mater. Sci. 43, 6539. Jonsson, S.; Birgerson, J.; Crispin, X.; Greczynski, G.; Osikowicz, W.; van der Gon, A.; Salaneck, W.; Fahlman, M. (2003): The effects of solvents on the morphology and sheet resistance in poly(3,4-ethylenedioxythiophene)-polystyrenesulfonic acid (PEDOT-PSS) films, Synth. Metals 139(1), 1. Jung, H.; Cho, S.; Joung, J.; Oh, Y. (2007): Studies on inkjet-printed conducting lines for electronic devices, J. Electron. Mater. 36(9), 1211. Jung, S.; Baeg, K.; Yoon, S.; You, I.; Lee, J.; Kim, Y.; Noh, Y. (2010): Low-voltage-operated top-gate polymer thin-film transistors with high capacitance poly(vinylidene fluoride- trifluoroethylene)/poly(methyl methacrylate) dielectrics, J. Appl. Phys. 108, 102810. Kalidindi, S.; Jagirdar, B. (2009): Highly monodisperse colloidal magnesium nanoparticles by room temperature digestive ripening, Inorg. Chem. 48(10), 4524. Kanungo, M.; Lu, H.; Malliaras, G.; Blanchet, G. (2009): Suppression of metallic conductivity of single-walled carbon nanotubes by cycloaddition reactions, Science 323, 234. Karar, N.; Chander, H.; Shivaprasad, S. (2004a): Enhancement of luminescent properties of ZnS:Mn nanophosphors by controlled ZnO capping, Appl. Phys. Lett. 85(21), 5058. Karar, N.; Raj, S.; Singh, F. (2004b): Properties of nanocrystalline ZnS:Mn, J. Crys. Growth 268, 585. Karlin, K. (2005): Progress in inorganic chemistry v. 54, Wiley-Interscience, Hoboken, New Jersey. Karsa, D. (ed.) (2003): Surfactants in polymers, coatings, inks and adhesives, Blackwell, Oxford. Kavoosi, B. (2001): Active imaging and visual information on paper and paper-based products, Printed Microsystems on Paper, 1-106. Kelly, M. (1995): Low-dimensional semiconductors, Clarendon Press, Gloustershire, England. Kerner, E. (1956): The electrical conductivity of composite media, Proc. Phys. Soc. B 69, 802.
200
Kim, I.; Kim, J. (2010): The effect of reduction atmospheres on the sintering behaviors of inkjet printed Cu interconnectors, J. App. Phys. 108, 102807. Kim, J.; Cho, J.; Chung, S.; Kwak, J.; Lee, C.; Hong, Y.; Kim, J. (2009): Inkjet-printed silver gate electrode and organic dielectric materials for bottom-gate pentacene thin-film transistors, J. Korean Phys. Soc. 54(1), 518. Kim, J.; Ma, P. (2011): Carbon nanotubes for polymer reinforcement, CRC Press, Boca Raton, Florida. Kim, J.; Park, S.; Jeong, T.; Bae, M.; Song, S.; Lee, J.; Han, I.; Jung, D.; Yu, S. (2010): Paper as a substrate for inorganic powder electroluminescent devices, IEEE Trans. Electron. Dev. 57(6), 1470. Kim, J.; Jung, J.; Lee, D.; and Joo, J. (2002): Enhancement of electrical conductivity of poly(3,4- ethylenedioxythiophene)/poly(styrenesulfonate) by a change of solvents, Synth. Metals 126, 311. Kim, N.; Han, K. (2010): Future direction of direct writing, J. Appl. Phys. 108, 102801. Kim, S.; McKean, D. (1998): Aqueous TiO2 suspension preparation and novel application of ink-jet printing technique for ceramics patterning, J. Mater. Sci. Lett. 17, 141. Kim, Y.; Moon, D.; Han, J. (2004): Organic TFT array on a paper substrate, IEEE Electron Device Lett. 25(10), 702. King, C. (1998): Electroluminescent displays, Mater. Res. Soc. Symp. Proc. 508. King, C. (2000): Electroluminescent displays, Planar Systems Inc. King, C.; Coovert, R. (1990): Thin film electroluminescent display with improved contrast, US Patent 4963788. Kitai, A. (ed). (2008): Luminescent materials and applications, Wiley, Hoboken, New Jersey. Klausch, A.; Althues, H.; Schrage, C.; Simon, P.; Szatkowski, A.; Bredol, M.; Adam, D.; Kaskel, S. (2010): Preparation of luminescent ZnS:Cu nanoparticles for the functionalization of transparent acrylate polymers, J. Luminescence 130, 692. Ko, S.; Pan, H.; Grigoropolous, C.; Luscombe, C.; Frechet, J.; Poulikakos, D. (2007): All-inkjet- printed flexible electronics fabrication on a polymer substrate by low-temperature high- resolution selective laser sintering of metal nanoparticles, Nanotech. 18, 1. Konishi, M.; Isobe, T.; and Senna, M. (2001): Enhancement of photoluminescence of ZnS:Mn nanocrystals by hybridizing with polymerized acrylic acid, J. Lumin. 93, 1. Kroto, H.; Heath, J.; O'Brien, S.; Curl, R.; Smalley, R. (1985): C60: buckminsterfullerene, Nature 318,162. Kumar, N.; Joshi, M.; Friend, C.; Prasad, P., Burzynski, R. (1997): Organic-inorganic heterojunction light-emitting diodes based on poly(p-phenylene vinylene)/cadmium sulfide thin films, App. Phys. Lett. 71(10), 1388.
201
Kwong, C.; Djurisic, A.; Choy, W.; Li, D.; Xie, M.; Chan, W.; Cheah, K.; Lai, P.; Chui, P. (2005): Efficiency and stability of different tris(8-hydroxyquinoline) aluminum (Alq3) derivatives in OLED applications, Mater. Sci. Eng. B 116, 75. Kymakis, E.; Klapsis, G.; Koudoumas, E.; Stratakis, E.; Kornilios, N.; Vidakis, N.; and Franghiadakis, Y. (2007): Carbon nanotube/PEDOT:PSS electrodes for organic photovoltaics, Eur. Phys. J. Appl. Phys. 36, 257. Le, H. (1998): Progress and trends in ink-jet printing technology, J. Imaging Sci. Tech., 42(1), 49. Lee, E. (2003): Microdrop generation, CRC Press, Boca Raton. Lee, I.; Pennathur, S.; Goodnick, S.; Wager, J. (1997): band structure and high-field electron transport of a ZnS phosphor in AC thin-film electroluminescent devices, J. Korean Phys. Soc., 31(3), 517. Lee, C.; Joo, J.; Han, S.; and Koh, S. (2005a): An approach to durable PVDF cantilevers with highly conducting PEDOT/PSS (DMSO) electrodes, Sens. & Actuators A: Phys., 121, 373. Lee, K.; Blanchet, G.; Gao, F.; Loo, Y. (2005b): Direct patterning of conductive water-soluble polyaniline for thin-film organic electronics, Appl. Phys. Lett., 86, 0741021. Lee, F.; Mills, R.; Talke, F. (1981): Drop-on-demand inkjet printing at high print rates and high resolution, IBM Research Report 3154, IBM Corporation, New York. Lee, Y.; Choi, J.; Lee, K.; Stott, N.; Kim, D. (2008a): Large-scale synthesis of copper nanoparticles by chemically controlled reduction for applications of inkjet printing electronics, Nanotech. 19(41), 415604. Lee, S.; Yoon, S.; Shin, H.; Joo, W.; Yi, D.; Choi, J.; Amarnath, C.; Paik, U. (2008b): Hierarchical organization of Au nanoparticles in a poly(vinyl carbazole) matrix for hybrid electronic devices, Nanotech. 19, 1. Leising, G.; Stefan, H.; Meghdadi, F.; Tasch, S. (1999): Electroluminescent display device with continuous base electrode, US Patent 5949188. Li, D.; Kaner, R. (2006): Shape and aggregation control of nanoparticles: not shaken, not stirred, J. Am. Chem. Soc. 128(3), 968. Li, D.; Sutton, D.; Burgess, A.; Graham, D.; Calvert, P. (2009): Conductive copper and nickel lines via reactive inkjet printing, J. Mater. Chem. 19, 3719. Lia, S. M. Chan, M. Fung, C. Lee, and S. Lee. (2003): Concentration effect of glycerol on the conductivity of PEDOT film and the device performance, Mater. Sci. Eng. B104, 26. Lim, W.; Douglas, E.; Kim, S.; Norton, D.; Pearton, S.; Ren, F.; Shen, H.; Chang, W. (2009): High mobility InGaZnO4 thin-film transistors on paper, Appl. Phys. Lett. 94, 072103. Lim, J.; Kim, J.; Yoon, Y. (2010): All-inkjet-printed metal-insulator-metal (MIM) capacitor, Curr. Appl. Phys. 12(1), E14.
202
Lima, S.; Cremona, M.; Davolos, M.; Legnani, C.; and Quirino, W. (2007): Electroluminescence of zinc oxide thin-films prepared via polymeric precursor and via sol–gel methods, Thin Solid Films 516, 165. Lin, D.; Liu, N.; Yang, K.; Xing, B.; Wu, F. (2010): Different stabilities of multiwalled carbon nanotubes in fresh surface water samples, Enviro. Pollution 158, 1270. Lin, Y.; Li, Y.; Yeh, H.; Wang, Y.; Wen, T.; Huang, L.; and Chen, Y. (2007): Blending of poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) with poly(ethyleneimine) as an active layer in depletion-mode organic thin film transistors, Appl. Phys. Lett. 91, 253501. Liu, X.; Ni, X.; Wang, J.; Yu, X. (2008): A novel route to photoluminescent water-soluble Mn- doped ZnS quantum dots via photopolymerization initiated by the quantum dots, Nanotech. 19, 1. Liu, Z.; Su, Y.; Varahramyan, K. (2005): Inkjet-printed silver conductors using silver nitrate ink and their electrical contacts with conducting polymers, Thin Solid Films, 478, 275. Lopez, M.; Sanchez, J.; and Estrada, M. (2008) : Characterization of PEDOT :PSS dilutions for inkjet printing applied to OLED fabrication, Proc. 7th Int. Carribean Conf. Dev. Circ. Sys. Lua, Y.; Pew, C.; Zhang, F.; Fillmore, W.; Bronson, R.; Sathyapalan, A.; Savage, P.; Whittaker, J.; Davis, R.; and Linford, M. (2005): Polyelectrolytes as new matrices for secondary ion mass spectrometry, J. Am. Soc. Mass Spectrom. 16(10), 1575. Magdassi, S. (ed.) (2010): The chemistry of inkjet inks, World Scientific Publishing, Singapore. Magdassi, S.; Grouchko, M.; Berezin, O.; Kamyshny, A. (2010): Triggering the sintering of silver nanoparticles at room temperature, ACS Nano 4(4), 1943. Mahalik, N. (2006): Micromanufacturing and nanotechnology, Springer, Berlin. Manglik, R.; Wasekar, V.; Zhang, J. (2001): Dynamic and equilibrium surface tension of aqueous surfactant and polymeric solutions, Exper. Thermal & Fluid Sci. 25, 55. Manzoor, K.; Vadera, S.; Kumar, N.; Kutty, T. (2003): Synthesis and photoluminescent properties of ZnS nanocrystals doped with copper and halogen, Mater. Chem. Phys. 82, 718. Matsuda, H.; Kobayashi, N.; Kobayashi, T.; Miyazawa, K.; Kuwabara, M. (2000): Room- temperature synthesis of crystalline barium titanate thin films by high-concentration sol- gel method, J. Non-Crystalline Solids 271, 162. Matsumoto, S. (1984): Electronic display devices, Ohmsha Ltd., Tokyo. Mattoussi, H.; Radzilowski, L.; Dabbousi, B.; Thomas, E.; Bawendi, M.; Rubner, M. (1998): Electroluminescence from heterostructures of poly(phenylene vinylene) and inorganic CdSe nanocrystals, J. App. Phys. 83(12), 7965.
203
Mauthner, G.; Landfester, K.; Kock, A.; Bruckl, H.; Kast, M.; Stepper, C.; List, E. (2008): Inkjet printed surface cell light-emitting devices from a water-based polymer dispersion, Org. Electron. 9, 164. McGinness, John (1972): Mobility gaps: a mechanism for band gaps in melanins, Science 177, 896. McGown, D.; Parfitt, G. (1967): Stability of non-aqueous dispersions, part 5: theoretical predictions for dispersions in hydrocarbon media, Colloid & Poly. Sci. 219(1), 48. Mei, J.; Lovell, M.; Mickle, M. (2005): Formulation and processing of novel conductive solution inks in continuous inkjet printing of 3-d electric circuits, IEEE Trans. Electron. Packaging Manuf. 28(3), 265. Meixner, R.; Cibis, D.; Krueger, K.; and Goebel, H. (2008): Characterization of polymer inks for drop-on-demand printing systems, Microsys. Techn. 14, 1137. Menawat, A.; Joseph, H., Siriwardane, R. (1984): Control of surface energy of glass by surface reactions: contact angle and stability, J. Colloid & Interface Sci., 101(1), 110. Mentley, D. (2002): State of flat panel display technology and future trends, Proc. IEEE 90(4), 453. Meziani, M.; Bunker, C.; Lu, F.; Li, H.; Wang, W.; Guliants, E.; Quinn, R.; Sun, Y. (2009): Formation and properties of stabilized aluminum nanoparticles, ACS Appl. Mater. Interfaces 1(3), 703. Mihailetchi, V.; Blom, P.; Hummelen, J.; Rispens, M. (2003): Cathode dependence of the open- circuit voltage of polymer:fullerene bulk heterojunction solar cells, J. App. Phys. 94(10), 6489. Miki, T.; Ogawa, T.; Isobe, T.; and Sfihi, H. (2004): Low-temperature synthesis of nanosize 2+ Zn2SiO4:Mn phosphor from the precursor sol in supercritical ethanol and its photoluminescent properties, J. Sol-Gel Sci. & Tech. 31, 73. Mire, C.; Agrawal, A.; Wallace, G.; Calvert, P.; Panhuis, M. (2011): Inkjet and extrusion printing of conduction poly(3,4-ethylenedioxythiophene) tracks on and embedded in biopolymer materials, J. Mater. Chem. 21, 2671. Miura, N. (2008): Proc. 14th Int. Workshop Inorg. Org. Electroluminescence, 427. Molesa, S; Redinger, D.; Huang, D.; Subramanian, V. (2003): High-quality inkjet-printed multilevel interconnects and inductive components on plastic for ultra-low-cost RFID applications, Proc. Mater. Res. Soc. 769, Symp. H8.3. Molesa, S. (2006): Ultra-low-cost printed electronics, PhD thesis, Department of Electrical Engineering, University of California, Berkeley. Molina-Lopez, F.; Briand, D.; De Rooij, N. (2012): All additive inkjet printed humidity sensors on plastic substrate, Sensors & Actuators B 166, 212.
204
Montibon, E.; Jarnstrom, L.; and Lestelius, M. (2009): Characterization of poly(3,4- ethylenedioxythiophene): poly(styrenesulfonate) (PEDOT:PSS) adsorption on cellulosic materials, Cellulose 16, 807. Montibon, E.; Jarnstrom, L.; Lestelius, M. (2010): Electroconductive paper prepared by coating with blends of poly(3,4-ethylenedioxythiophene)/poly(4-styrenesulfonate) and organic solvents, J. Appl. Poly. Sci. 117(6), 3524. Moon, J.; Park, J.; Lee, T.; Kim, Y.; Yoo, J.; Park, C.; Kim, J.; and Jin, K. (2005): Transparent conductive film based on carbon nanotubes and PEDOT composites, Diamond & Related Mater. 14, 1882. Moore, G. (1965): Cramming more components onto integrated circuits, Electronics 38. Mu, J.; Gu, D.; and Xu, Z. (2005): Effect of annealing on the structural and optical properties of nano-crystalline particles for polymer light-emitting diodes, Mater. Sci. Forum 475, 1795. Mullen, K.; Scherf, U. (eds.) (2006): Organic light emitting devices: synthesis, properties, and applications, Wiley-VCH, Berlin, Germany. Mustonen, T.; Kordas, K.; Saukko, S.; Toth, G.; Penttila, J.; Heliston, P.; Seppa, H.; and Jantunen, H. (2007): Inkjet printing of transparent and conductive patterns of single- walled carbon nanotubes and PEDOT-PSS composites, Physica Status Solidi B 244(11), 4336. Nalwa, H. (2000): Handbook of nanostructured materials and nanotechnology, Academic Press, New York, New York. Nalwa, H. (ed.) (2003): Handbook of organic-inorganic hybrid materials and nanocomposites 2nd ed., American Scientific Publishers, Valencia, California. Naumov, A.; Kuznetsov, O.; Harutyunyan, A.; Green, A.; Hersam, M.; Resasco, D.; Nikolaev, P.; Weisman, R. (2009): Quantifying the semiconducting fraction in single-walled carbon nanotube samples through comparative atomic force and photoluminescence microscopies, Nano Lett. 9(9), 3203. Neamen, D. (2002): Semiconductor physics and devices, 3rd ed., McGraw-Hill, New York, New York. Neimo, L. (1999): Papermaking science and technology 4: papermaking chemistry. TAPPI & Fapet Oy, Helsinki. Nguyen, B.; Gautrot, J.; Nguyen, M.; Zhu, X. (2007): Nitrocellulose-stabilized silver nanoparticles as low conversion temperature precursors useful for inkjet printed electronics, J. Mater. Chem. 17, 1725. Nguyen, T.; Yee, R.; Schwartz, B. (2001): Solution processing of conjugated polymers: the effects of polymer solubility on the morphology and electronics properties of semiconducting polymer films, J. Photochem. Photobio. A: Chem. 144, 21.
205
Nien, Y.; Hsu, H.; Chen, I.; Hwang, C.; and Chu, S. (2006): Photoluminescence and microstructure of Mn-doped ZnS phosphor powder co-fired with ZnS nanocrystallites, J. Non-Crystal. Sol. 352, 2539. Nobusa, Y.; Yomogida, Y.; Matsuzaki, S.; Yanagi, K.; Katatura, H. (2011): Inkjet printing of single-walled carbon nanotube thin-film transistors patterned by surface modification, Appl. Phys. Lett. 99, 183106. Noguchi, Y.; Sekitani, T.; Someya, T. (2006): Organic-transistor-based flexible pressure sensors using ink-jet-printed electrodes and gate dielectric layers, App. Phys. Lett. 89, 1. O’Mara, W.; Herring, R.; and Hunt, L. (1990): Handbook of semiconductor silicon technology, William Andrew Publishing, Norwich. Ohnesorge, W. (1936): Formation of drops by nozzles and the breakup of liquid jets, J. Appl. Math. Mech. 16, 355. Ohring, M. (1991): The materials science of thin films, Academic Press, Walthen, Massachusetts. Oittinen, P.; Hannu, S. (1998): Printing, papermaking science and technology, Finnish Paper Engineers Association/TAPPI, Helsinki, Finland. Okamoto, K.; Nasu, Y.; Hamakawa, Y. (1981): Low threshold-voltage thin-film electroluminescent devices, IEEE Trans. Electron Dev. 28(6), 698. Ono, Y. (1995): Electroluminescent displays, World Scientific Publishing Co., Singapore. Ould-Ely, T.; Luger, M.; Kaplan-Reinig, L.; Niesz, K.; Doherty, M.; Morse, D. (2011): Large- scale engineered synthesis of BaTiO3 nanoparticles using low-temperature bioinspired principles, Nature Protocols 6(1), 97. Ouyang, J.; Chu, C.; Chen, F.; Xu, Q.; and Yang, Y. (2005) : High conductivity poly(3,4- ethylene-dioxythiophene): poly(styrenesulfonate) film and its application in polymer optoelectronic devices, Adv. Func. Mater. 15(2), 203. Ouyang, J.; Xu, Q.; Chu, C.-W.; Yang, Y.; Li, G.; and Shinar, J. (2004): On the mechanism of conductivity enhancement in poly(3,4-ethylenedioxythiophene): poly(styrenesulfonate) film through solvent treatment, Polymer 45, 8443. Papageorgiou, D. (1995): On the breakup of viscous liquid threads, Phys. Fluids 7(7), 1529. Park, J.; Lee, S.; Kim, J.; Kwon, A.; Park, H.; Han, S. (2007): White electroluminescent device with ZnS:Mn, Cu, Cl phosphor, J. Luminescence 126, 566. Park, S.;Kuo, C.; Anthony, J.; Jackson, T. (2005): High mobility solution-processed OTFTs, Tech. Digest – Int. Elec. Devices Meet. 113. Parker, I. (1994): Carrier tunneling and device characteristics in polymer light emitting diode, J. Appl. Phys. 75, 1656. Peng, W.; Qu, S.; Cong, G.; Wang, Z. (2005) Structural and optical investigation of Mn-doped ZnS nanocrystals , Curr. Appl. Phys. 5, 31.
206
Perelaer, J.; de Laat, A.; Hendriks, C.; Schubert, U. (2008): Inkjet-printed silver tracks: low temperature curing and thermal stability investigation, J. Mater. Chem. 18, 3209. Phillips, Julia; Cava, R.J.; Thomas, G.A.; Carter, S.A.; Kwo, J.; Siegrist, T.; Krajewski, J.; Marshall, J.; Peck, W. Jr.; Rapkine, D. (1995): Zinc-indium-oxide: a high-conductivity transparent conducting oxide, Appl. Phys. Lett. 67(15), 2246. Polasik, J.; Schmidt, H. (2005): Conductive polymer PEDOT/PSS electrodes on the piezoelectric polymer PVDF, Proc. SPIE, 5759, 114. Porambo, M.; Marsh. A. (2009): Synthesis and photoluminescent properties of doped ZnS nanocrystals capped by poly(vinylpyrrolidone), Optical Mater. 31(11), 1631. Pushparaj, V.; Shaijumon, M.; Kumar, A.; Murugesan, S.; Ci, L.; Vajtai, R.; Linhardt, R.; Nalamasu, O.; Ajayan, P. (2007): Flexible energy storage devices based on nanocomposite paper, Proc. Nat. Acad. Sci. 104(34), 13574. Rack P, P. Holloway. (1998): The structure, device physics, and material properties of thin film electroluminescent displays, Mater. Sci. Eng. R21, 171. Rae, A.; and Hammer-Fritzinger, D. (2006): Creating metal and nonmetal nanosystems using conductive jettable inks, Sol. St. Tech. 49(4), 53. Reddish, W. (1962): Chemical structure and dielectric properties of high polymers, Pure App. Chem. 5, 723. Rembe, C.; Wiesche, S.; Hofer, E. (1999): Dynamics in a thermal ink jet: a model for identification and simulation, Tech. Proc. 1999 Int. Conf. Modeling & Simulation Microsys., 553. Ridley, B.; Nivi, B.; Jacobson, J. (1999): All-inorganic field effect transistors fabricated by printing, Science 286, 746. Robel, L.; Kuno, M.; Kamat, P. (2007): Size-dependent electron injection from excited CdSe quantum dots into TiO2 nanoparticles, J. Amer. Chem. Soc. 129, 4136. Ryu, B.; Choim, Y.; Park, H.; Byun, J.; Kong, K.; Lee, J.; Chang, H. (2005): Synthesis of highly concentrated silver nanosol and its application to inkjet printing, Colloids & Surfaces A 270, 345. Saad, G. (1994): Dielectric behaviour of cyanoethylated cellulose, Polym. Int. 34, 411.
Sakai, Y.; Futakuchi, T.; Ijima, T.; Adachi, M. (2006): Preparation of (Ba,Sr)TiO3 thick film on ZrO2 substrates by inkjet printing, Jap. J. Appl. Phys. 45(2A), 846. Sanchez-Romaguera, V.; Madec, M.; Yeates, S. (2008): Inkjet printing of 3D metal-insulator- metal crossovers, Reactive & Func. Poly. 68(6), 1052. Sangoi, R.; Smith, C.; Seymour, M.; Venkataraman, J.; Clark, D.; Kleper, L.; Kahn, B. (2005): Printing radio frequency identification (RFID) tag antennas using inks containing silver dispersions, J. Dispers. Sci. Tech. 25(4), 515 (2005).
207
Sankir, N. (2008): Selective deposition of PEDOT/PSS on to flexible substrates and tailoring the electrical resistivity by post treatment, Circuit World 34(4), 32. Sanyal, U.; Datta, R.; Jagirdar, B. (2012): Colloidal calcium nanoparticles: digestive ripening in the presence of a capping agent and coalescence of particles under an electron beam, RSC Adv. 2, 259. Sartale, S.; Sankapal, B.; Lux-Steiner, M.; and Ennaoui, A. (2005): Preparation of nanocrystalline ZnS by a new chemical bath deposition route, Thin Solid Films 480, 168. Satoh, T.; Nakasuta, T.; Tsuruya, K.; Tabata, Y.; Tamura, T.; Ichikawa, Y.; Tango, H. (2007): Electrical properties of two-sided luminescence powder-distributed inorganic electroluminescence panels J. Mater. Sci. Mater. Electron. 18, S239. Savvatimskiy, A. (2005): Measurements of the melting point of graphite and the properties of liquid carbon (a review for 1963–2003), Carbon 43(6), 1115. Schlaf, R.; Murata, H.; Kafafi, Z. (2001): Work function measurements on indium tin oxide films, J. Electron. Spectroscopy & Related Phenomena 120, 149. Schrage, C.; Althues, H.; Klausch, A.; Adam, D.; Kaskel, S. (2010): ZnS:Cu polymer nanocomposites for thin film electroluminescent devices, J. Nanosci. Nanotech. 10, 4335. Sears, F.; Zemansky, M.; Young, H. (1982): University physics, 6th ed., Addison-Wesley, Boston, Massachusetts. Sedra, A. and Smith, K. (1997): Microelectronic circuits, Oxford University Press, Oxford. Selwitz, C. (1988): Cellulose nitrate in conservation, J. Paul Getty Trust, Los Angeles, California. Seshan, K. (2002): Handbook of thin film deposition, William Andrew, Norwich, New York. Setti, L.; Fraleoni-Morgera, A.; Mencarelli, I.; Fillipini, A.; Ballerin, B.; Di Biase, M. (2007): An HRP-based amperometric biosensor fabricated by thermal inkjet printing, Sensors and Actuators B 126, 252. Sharma, P.; Varadan, V.; Varadan, V. (2000): Porous behavior and dielectric properties of barium strontium titanate synthesized by sol-gel method in the presence of triethanolamine, Chem. Mater. 12, 2590. Sharma, H.; Mansingh, A. (1998): Sol-gel processed barium titanate ceramics and thin films, J. Mater. Sci. 33, 4455. Sharma, H.; Sarma, H. (1998): Electrical properties of sol-gel-processed barium titanate films, Thin Solid Films, 330, 178. Shi, L.; Wang, R.; Cao, Y.; Feng, C.; Liang, D.; Tay, J. (2007): Fabrication of poly(vinylidene fluoride-co-hexafluropropylene) (PVDF-HFP) asymmetric microporous hollow fiber membranes, J. Membrane Sci. 305, 215. Shield, T.; Bogy, D.; Talke, F. (1987): Drop formation by DOD ink-jet nozzles: a comparison of experiment and numerical simulation, IBM J. Res. Dev. 31(1), 96.
208
Shik, A. (1997): Quantum wells, World Scientific, Hackensack, New Jersey. Shin, D.; Smith, P. (2008): Theoretical investigation of the influence of nozzle diameter variation of the fabrication of thin film transistor liquid crystal color filter, J. Appl. Phys. 103, 114905. Singh, M.; Ohji, T.; Asthana, R.; Mathur, S. (eds). (2011): Ceramic integration and joining technologies, John Wiley & Sons, Hoboken, New Jersey. Singh, V.; Morton, D. (1992): A model for electroluminescence in SrS:Ce ACTFEL display devices, IEEE Trans. Electron Devices, 39(6), 1331. Singh, M.; Haverinen, H.; Dhagat, P.; Jabbour, G. (2010): Inkjet printing: process and its applications, Adv. Mater. 22, 673. Sirringhaus, H.; and Shimoda, T. (2003): Inkjet printing of functional materials, MRS Bulletin 28(11), 802. Skolnick, M. (1981): Time resolved DC electroluminescence studies in ZnS:Mn,Cu powder phosphors, J. Phys. D: Appl. Phys. 14, 301. Skotheim, T. (1997): Handbook of conducting polymers, 2nd ed., CRC Press, Boca Raton, Florida. Skotheim, T.; Reynolds, J. (eds.) (2006): Conjugated polymers: theory, synthesis, properties, and characterization, CRC Press, Boca Raton, Florida. Small, A.; Johnston, J.; Clark, N. (2010): Inkjet printing of water “soluble” doped ZnS quantum dots, Eur. J. Inorg. Chem. 242. Smook, G. (1992): Handbook for pulp and paper technologists, 2nd ed., TAPPI, Angus Wilde Publications, Vancouver, British Columbia. Sodhi, R.; Sun, L.; Sain, M.; and Farnood, R. (2008): Analysis of ink/coating penetration of paper surfaces by time-of-flight secondary ion mass spectrometry (ToF-SIMS) in conjunction with principal component analysis (PCA), J. Adhes. 84, 277. Solomon, S.; Bahadory, M.; Jayarajasingham, V.; Rutkowsky, S.; Boritz, C. (2007): Synthesis and study of silver nanoparticles, J. Chem. Edu. 84(2), 322. Soltman, D.; Subramanian, V. (2008): Inkjet-printed line morphologies and temperature control of the coffee-ring effect, Langmuir 24(5), 2224. Song, J.; Kim, Y.; Yoon, Y. (2009): The production of transparent carbon nanotube field emitters using inkjet printing, Physica E41, 1513. Spasic, A.; and Hsu, J. (2006): Finely dispersed materials. CRC Press, Boca Raton. Srivastava, B.; Jana, S.; Pradhan, N. (2011): Doping Cu in semiconductor nanocrystals: some old and some new physical insights, J. Amer. Chem. Soc. 133(4), 1007. Sumerel, J.; Nguyen, C.; Sudau, K.; Fabris, D. (2006): Deposition of MWCNTs using the Dimatix Materials Printer, Fujifilm-Dimatix, Santa Clara, California.
209
Sumerel, J.; Staton, J.; Sudau, K. (2007): Inkjet printing toluene using the Dimatix Materials Printer, Fujifilm-Dimatix, Santa Clara, California. Sun, C. (2005): Size and confinement effects on nanostructures, School of Elec. & Electron. Eng., Singapore. Tadros, T. (ed). (1987): Solid-liquid dispersions, Academic Press, London. Takahashi, H.; Tetsuhiko, I. (2005): Photoluminescence enhancement of ZnS:Mn nanocrystal phosphors: comparison of organic and inorganic surface modifications, Jap. J. Appl. Phys. 44(2), 922. Talapin, D.; Rogach, A.; Kornowski, A.; Haase, M.; Weller, H. (2001): Highly luminescent monodisperse CdSe and CdSe/ZnS nanocrystals synthesized in a hexadecylamine −trioctylphosphine oxide−trioctylphospine mixture, Nano Lett. 1(4), 207. Taylor, R.; Church, K.; Sluch, M. (2007): Red light emission from hybrid organic/inorganic quantum dot AC light emitting displays, Displays 28, 92. Te, L.; Ervin, M.; Qiu, H.; Fuchs, B.; Lee, W. (2011): Graphene supercapacitor electrodes fabricated by inkjet printing and thermal reduction of graphene oxide, Electrochem. Comm. 13, 355. Tekin, E.; de Gans, B.;Schubert, U. (2004): Inkjet printing of polymers – from single dots to thin-film libraries, J. Mater. Chem. 14, 2627. Tekin, E.; Smith, P.; and Schubert, U. (2008): Inkjet printing as a deposition and patterning tool for polymers and inorganic particles, Soft Matter 4, 703. Titcomb, S. (1981): Dynamic surface tension of jet inks, Ind. Eng. Chem. Prod. Res. Dev. 20, 680. Tobjork, D.; Aarnio, H.; Pulkkinen, P.; Bollstrom, R.; Maattanen, A.; Ihalainen, P.; Makela, T.; Peltonen, J.; Toivakka, M.; Tenhu, H.; Osterbacka, R. (2012): IR-sintering of ink-jet printed metal-nanoparticles on paper, Thin Solid Films 520(7), 2949. Tobjork, D.; Osterbacka, R. (2011): Paper electronics, Adv. Mater. 23(17), 1935. Torrisi, F.; Hasan, T.; Wu, W.; Sun, Z.; Lombardo, A.; Kulmala, T.; Hshieh, G.; Jung, S.; Bonaccorso, F.; Paul, P.; Chu, D.; Ferrari, A. (2012): Ink-jet printed graphene electronics, ACS Nano 6(4), 2992. Tracton, A. (2005): Coatings technology handbook, CRC Press, Boca Raton, Florida. Tseng, W.; Lin, S.; Wang, S. (2006): Particulate dispersion and freeform fabrication of BaTiO3 thick films via direct inkjet printing, J. Electroceram. 16, 537. Tseng, H.; Subramanian, V. (2011): All inkjet-printed, fully self-aligned transistors for low-cost circuit applications, Org. Electron. 12(2), 249.
210
Tsyboulski, D.; Bachilo, S.; Kolomeisky, A.; Weisman, R. (2008): Translational and rotational dynamics of individual single-walled carbon nanotubes in aqueous suspension, ACS Nano 2(9), 1770. Valero, J.; Jarom, C.; Comas, E. (2007): Optimized automatic recovery of nozzle health in inkjet systems, Int. Conf. Digital Printing Tech. Digital Fab. NIP23, 766. van der Waarden, M. (1950): Stabilization of carbon-black dispersions in hydrocarbons, J. Colloid Sci. 5(4), 317. van Osch, T.; Perelaer, J.; de Laat, A.; Schubert, U. (2008): Inkjet printing of narrow conductive tracks on untreated polymeric substrates, Adv. Mater. 20, 343. Vazquez, G.; Alvarez, E.; Navaza, J. (1995): Surface tension of alcohol + water from 20 to 50°C, J. Chem. Eng. Data 40(3), 611. Vetanen, W.; Aguilera, M. (1997): Electroluminescent device construction employing polymer derivative coating, US Patent 5650692. Vilms, J.; Spicer, W. (1965): Quantum efficiency and radiative lifetime in p-type gallium arsenide, J. Appl. Phys. 36 , 2815. Vissenberg, M.; Matters, M. (1998): Theory of the field-effect mobility in amorphous organic transistors, Phys. Rev. B 57,12964. Vlasenko, N.; Popkov, I. (1960): Study of the electroluminescence of a sublimed ZnS-Mn phosphor, Optics & Spectroscopy 8, 39. Vogel, W.; Borse, P.; Deskmukh, N.; and Kulkarni, S. (2000): Structure and stability of monodisperse 1.4-nm ZnS particles stabilized by mercaptoethanol, Langmuir 16(4), 2032. Wang, G.; Tau, X.; and Wang. R. (2008a): Flexible organic light-emitting diodes with a polymeric nanocomposite anode, Nanotech. 19, 145201. Wang, J. (2002): Systematic consideration for the thermodynamic stability of pigmented inkjet dispersions and inks, Int. Conf. Digital Printing Tech. NIP18, 176. Wang, M.; Zhang, M.; Qian, J.; Zhao, F.; Shen, L.; Scholes, G.; Winnik, M. (2009): Enhancing the photoluminescence of polymer-stabilizer CdSe/CdS/ZnS core/shell/shell and CdSe/ZnS core/shell quantum dots in water through a chemical-activation approach, Langmuir 25(19), 11732. Wang, W.; Fernando, K.; Lin, Y.; Meziani, M.; Veca, L.; Cao, L.; Zhang, P.; Kimani, M.; Sun, Y. (2008b): Metallic single-walled carbon nanotubes for conductive nanocomposites, J. Am. Chem. Soc. 130, 1415. Wang, Y.; Gebler, D.; Lin, L.; Blatchford, J.; Jessen, S.; Wang, H., Epstein, A. (1996): Alternating-current light-emitting devices based on conjugated polymers, Appl. Phys. Lett. 68, 894.
211
Warad, H.; Ghosh, S.; Hemtanon, B.; Thanachayanont, C.; Dutta, J. (2005): Luminescent nanoparticles of Mn doped ZnS passivated with sodium hexametaphosphate, Sci. Tech. Adv. Mater. 6, 296. Wei, Q.; Zheng, H.; Huang, Y. (2001): Direct patterning ITO transparent conductive coatings, Solar Energy Mater. & Solar Cells, 68(3-4), 383. Wei, T.; Ruan, J.; Fan, Z. (2007): Preparation of a carbon nanotube film by ink-jet printing, Carbon 45, 2712. Whitmarsh, J. (2005): Flexible electronics: silicon meets paper and beyond, Microelec. Int. 22(3), 16. Wijshoff, H. (2008): Structure and fluid dynamics in piezo inkjet printheads, Océ Technologies, Venlo, TheNetherlands. Wood, C. (ed.) (2000): Handbook of thin film devices, Academic Press, Missouri. Wood, V.; Panzer, M.; Chen, J.; Bradley, M.; Halpert, J.; Bawendi, M.; Bulovic, V. (2009): Inkjet-printed quantum dot–polymer composites for full-color AC-driven displays, Adv. Mater. 21, 2151. Wu, Y.; Li, Y.; Ong, B. (2007): A simple and efficient approach to a printable silver conductor for printed electronics, J. Amer. Chem. Soc. 129, 1862. Xia, Y. and Ouyang, J. (2009): Salt-induced charge screening and significant conductivity enhancement of conducting poly(3,4-ethylenedioxythiophene): poly(styrenesulfonate), Macromol. 42, 4141. Xie, X.; Samsudeen, F.; Farnood, R.; Kortschot, M. (2008): Roughening due to ink jet rewetting: effect of paper treatment and composition, J. Imag. Sci. Tech. 52(1), 010506-1. Xing, Y.; Xu, X.; Zhang, P.; Tian, W.; Yu, G.; Lu, P.; Liu, Y.; Zhu, D. (2005): Carbazole-pyrene based organic emitters for electroluminescent device. Chem. Phys. Lett.408(1–3), 169. Xu, J.; Zhai, J.; Yao, X. (2006): Structure and dielectric properties of barium titanate films grown by sol-gel-hydrothermal process, Appl. Phys. Lett. 89, 252902. Xue, F. and Su, Y. (2005): Modified PEDOT-PSS conducting polymer as S/D electrodes for device performance enhancement of P3HT TFTs, IEEE Trans. Electron. Devices 52(9), 1982. Yakovlev, S. (2004): Lanthanide ions doping effects on structural, electrophysical and functional properties of sol-gel fabricated PbTiO3 thin films, PhD thesis, University of Kiel. Yang, A.; Cheng, C.; Lin, C. (2006): Investigation of droplet-ejection characteristics for a piezoelectric inkjet printhead, Proc. IMechE C: J. Mech. Eng. Sci. 220, 435. Yang, H.; Holloway, P.; Ratna, B. (2003): Photoluminescent and electroluminescent properties of Mn-doped ZnS nanocrystals, J. Appl. Phys. 93(1), 586. Yang, H.; Santra, S.; Holloway, P. (2005): Syntheses and applications of Mn-doped II-VI semiconductor nanocrystals, J. Nanosci. Nanotech 5, 1364.
212
Yang, F.; Zhang, X.; Wu, X.; Tian, F.; Gan, F. (2007): Preparation of highly dispersed antimony- doped tin oxide nanopowders by azeotropic drying with isoamyl acetate, Trans. Nonferrous Met. Soc. China 17, 626. Yang, H.; Holloway, P. (2003): Electroluminescence from hybrid conjugated polymer- CdS:Mn/ZnS core/shell nanocrystals devices, J. Phys. Chem. B 107, 9705. Yang, H.; Holloway, P. (2004): Efficient and photostable ZnS-passivated CdS:Mn luminescent nanocrystals, Adv. Func. Mater., 14(2), 152. Yang, P.; and Bredol, M. (2008): Surface passivation and photoluminescence of Mn-doped nanocrystals Res. Lett. Mater. Sci. 1. Yang, Y.; Huang, J.; Liu, S.; Shen, J. (1997): Preparation, characterization, and electroluminescence of ZnS nanocrystals in a polymer matrix, J. Mater. Chem. 7(1), 131. Yarema, M.; Pichler, S.; Kriegner, D.; Stang, J.; Yarema, O.; Kirchsclager, R.; Tollabimazraehno, S.; Hume, M.; Haringer, D.; Kohl, M.; Chen, G.; Heiss, W. (2012): From highly monodisperse indium and indium tin colloidal nanocrystals to sel-assembled indium tin oxide nanoelectrodes, ACS Nano (online only). Yarin, A.; Szczech, J.; Megaridis, C.; Zhang, J.; and Gamota, D. (2006): Lines of dense nanoparticle colloidal suspensions evaporating on a flat surface: formation of non- uniform dried deposits, J. Coll. & Interf. Sci. 294, 343. Yen, W. (2004): Inorganic phosphors: compositions, preparation and optical properties, CRC Press, Boca Raton. Yen, W. (2006): Phosphor handbook, 2nd edition, CRC Press, Boca Raton.
Yeo, S.; Pode, R.; Ahn, J. (2009): Study on the size of fullerene (C60) aggregates in solution by photoluminescence and HRTEM measurements, J. Korean Phys. Soc. 55(1), 322. Yoshioka, Y.; Jabbour, G. (2006): Simple modification of sheet resistivity of conducting polymeric anodes via combinatorial ink-jet printing techniques, Synth. Metals 156, 779. Yoshioka, Y.; Calvert, P.; Jabbour, G. (2005): Simple modification of sheet resistivity of conducting polymeric anodes via combinatorial ink-jet printing techniques, Macromol. Rapid. Comm. 26, 238. Yu, I.; Isobe, T.; Senna, M. (1996): Optical properties and characteristics of ZnS nanoparticles with homogeneous Mn distribution, J. Phys. Chem.: Solids 57(4), 373. Yuan, Z.; Dan-Li, X. (2008): Porous PVDF/TPU blends asymmetric hollow fiber membranes prepared with the use of hydrophilic additive PVP (K30), Desalination 223, 438. Zeng, J.; Lin, C.; Li, J.; Li, K. (1999): Low-temperature preparation of barium titanate thin films by a novel sol-gel-hydrothermal method, Materials Letters 38, 112. Zhang, W.; Lee, H. (2010): Synthesis and optical property of water-soluble ZnS:Cu quantum dots by use of thioglycolic acid, Appl. Optics 49(14), 2566.
213
Zhou, Z.; Yang, Z.; Yuan, Q. (2008): Barium titanate ceramic inks for continuous ink-jet printing synthesized by mechanical mixing and sol-gel methods, Trans. Nonferrous Met. Soc. China 18, 150. Zhouping, Y.; Yongan, H.; Ningbin, B.; Xiaomei, W.; Youlun, X. (2010): Inkjet printing for flexible electronics: materials, processes, and equipment, Chinese Sci. Bulletin 55(30), 3383. Zhuang, J.; Zhang, X.; Wang, G.; Li, D.; Yang, W.; Li, T. (2003): Synthesis of water-soluble ZnS : Mn2+ nanocrystals by using mercaptopropionic acid as stabilizer, J. Mater. Chem. 13, 1853.
214
APPENDIX A. ZnS nanoparticle synthesis & dispersion Synthesis
All ZnS nanoparticles were synthesized using a competitive precipitation method (Figure A.1). Synthesis procedures for ZnS:Mn were based on those suggested by Adachi et al. (2007, 2008) and Althues et al. (2006) using caps other than 3-MPA, and Zhuang et al. (2003) using 3-MPA.
Zn2+
- 2+ - Zn(Ac ) Zn Ac 2+ 2+ 2 Zn Mn Ac- Zn2+ H O - 2 2+ 2+ 2- Ac Zn Mn S S2- - - - Ac Ac - Mn(Ac )2 Ac - 2+ ZnS Ac 2+ Mn Zn ZnS ZnS 2- 2- + S ZnS S MnS Ac- Ac- 2- 2- S S S2- 2- 2- S source H2O S 2- 2+ 2- 2- S Zn 2+ S S ZnS:Mn Zn Zn2+
Appendix Figure A.1. Competitive precipitation process for producing doped ZnS nanoparticles.
Synthesis procedures for ZnS:Cu were based on those described by Manzoor et al. (2003) and Small et al. (2011) using caps other than 3-MPA, and Schrage et al. (2010) using 3-MPA. Phase transfer of ZnS nanoparticles to make them organic-soluble was achieved using the method described by Klausch et al. (2010). Figure A.2 shows a schematic of the synthesis procedure of ZnS:Mn using 3-MPA, which was typical of all the syntheses, with a few minor differences. Figure A.3 shows a schematic of the phase-transfer procedure.
215
Zn(Ac)2 2+ Zn[SCH2CH2COOH]2 Zn pH = 10.3
Mn2+ H2O Mn[SCH2CH2COOH]2 NaOH Mn(Ac)2 - CH3COO (1) - CH3COO + - 3-MPA (H )(SCH2CH2COOH )
Na2S pH = 10 S2- S2- H2O pH = 10 Δ + + (2) pH 10 Na Na buffer T = 70°C
2+ Zn O
2+ Zn2+ Mn S OH O O Ac- S2- S2- OH - OH Ac - 2+ ZnS Ac 2+ Mn Zn Δ ZnS ZnS 16 h ZnS:Mn (1) + (2) S2- 2- O ZnS S OH MnS - - OH T = 70°C Ac Ac S2- O HO Zn2+ Zn2+ 2+ O Zn
Appendix Figure A.2. Competitive precipitation process for producing doped ZnS nanoparticles.
216
ZnS:Mn/Cu, SHMP/PVP/citrate/chitosan caps
1) Zn2+/dopant/cap solution:
50 mmol Zn(Ac)2 2H2O
12.5 mmol Mn(Ac)2 4H2O (50 at. % - only 1.5 at. % incorporated) OR
0.5 mmol Cu(Ac)2 H2O (1 at. %)
150 mL H2O
w:w ratio of SHMP OR PVP OR citrate OR chitosan*:Zn(Ac)2 2H2O = 1:20 (SHMP), 1:100 (PVP), 1:1.4 (citrate), 1:20 (chitosan)
*chitosan solution displaced 10 mL of H2O with 10 mL of HAc.
2) S2- solution:
25 mmol Na2S 9H2O (Mn-doped) OR
25 mmol Na2S2O3 5H2O (Cu-doped)
50 mL H2O
3) Heat the Zn2+/dopant/cap solution to 70°C with gentle stirring.
4) Using a burette, add the S2- solution dropwise to the Zn2+/dopant/cap solution. A white precipitate forms immediately. Continue stirring at 70°C for 16 hours, using an air condenser to provide reflux.
5) Remove from heat. Fill 50 mL centrifuge tubes with reaction solution (use enough tubes to use all the solution). Centrifuge at 5100 RPM for 30 minutes. Repeat twice, rinsing the particles with 30 mL acetone the first time, and 30 mL water the second time.
6) Dry particles in air overnight.
217
ZnS:Mn/Cu, 3-MPA cap
1) Zn2+/dopant/cap solution:
6 mmol Zn(Ac)2 2H2O
0.09 mmol Mn(Ac)2 4H2O (1.5 at. %) OR
0.06 mmol Cu(Ac)2 H2O (1 at. %)
30 mL H2O
24 mmol (Mn-doped) or 18 mmol (Cu-doped) 3-MPA
2) S2-/buffer solution:
3 mmol Na2S 9H2O (Mn-doped) OR 18 mmol thiourea (Cu-doped)
120 mL H2O
4 mL pH buffer (pH 10 for Mn-doped, pH 8 for Cu-coped)
3) Add 2 M NaOH dropwise, with stirring, to Zn2+/dopant/cap solution until pH = 10 (Mn- doped) or 8 (Cu-doped). Solution clarifies and becomes transparent.
4) Meanwhile, heat S2-/buffer solution to 70°C (Mn-doped) or 90°C (Cu-doped).
5) Add Zn2+/dopant/cap solution to S2-/buffer solution, with stirring, and maintain temperature (70°C for Mn-doped, 90°C for Cu-doped). Heat and stir for 8 hours (Mn- doped) or 20 hours (Cu-doped), refluxing with an air-condenser in a round-bottomed flask. Remove from heat. Solution is still transparent. Proceed to next step if organic- soluble particles are desired; for aqueous particles, skip to step 9.
218
6) With stirring, add 6 g octylamine to the reaction solution. Phase separation should immediately occur. With vigorous stirring (>1000 RPM), allow the separation to take place for 30 minutes.
7) Add 75 mL pentane to the reaction mixture, and stir at 750 RPM for 30 minutes.
8) With gentle heating (50°C max) and gentle stirring (100 RPM max) to encourage the organic phase to the top of the mixture, use a pipette to remove the organic phase to a separate flask. When all of the organic phase is removed, discard the aqueous phase.
9) Fill 50 mL centrifuge tubes with 15 mL reaction solution (use enough tubes to use all the solution). Add 30 mL acetone to each tube. Solution should become cloudy as nanoparticles precipitate out. Centrifuge at 5100 RPM for 30 minutes. Repeat twice, rinsing the particles with 30 mL fresh acetone each time.
10) Redisperse particles in water or toluene (2.5-5 w/w%).
O O O OH OH - + O H3N + ZnS ZnS Lipophilic tail (Mn, Cu) O stirring OH (Mn, Cu) OH O HO Emulsion Hydrophilic tail NH2 O
O
- + O H3N ZnS + Δ (Mn, Cu)
Dispersion
Appendix Figure A.3. Competitive precipitation process for producing doped ZnS nanoparticles.
219
Shelling
Shelling procedures for adding ZnO shells were based on work by Karar et al. and Jiang et al. (2004, 2009), the work of Althues et al. (2006) was used for AA caps, and that of Konishi et al. (2001) was used for PAA caps. Where particles were capped with both PAA/AA and ZnO, the ZnO cap was applied first.
ZnO shelling
1) Add 200 mg of dried water-soluble ZnS:Mn or :Cu nanoparticles to 200 mL of water, dispersing them by ultrasonication for 20 minutes.
2) Add 10 mL of 0.05 M Zn(NO3)2 6H2O solution to the suspension.
3) Add 0.1 M NaOH, dropwise, to the solution, until pH = 10. Allow the solution to mix vigorously for 20 minutes. Isolate/dry the particles by centrifugation, as described in the procedures above.
AA capping (polymerized in situ)
1) Add 200 mg of dried water-soluble ZnS:Mn or :Cu nanoparticles to 22.5 mL of water, dispersing them by ultrasonication for 20 minutes.
2) Add 5.486 g of AA to the solution. Cover and stir for 24 hours at 80°C to induce polymerization of the AA.
3) Remove from heat; isolate/dry the particles by centrifugation, as described in the procedures above.
PAA capping (polymerized before addition)
1) Add 200 mg of dried water-soluble ZnS:Mn or :Cu nanoparticles to 22.5 mL of water, dispersing them by ultrasonication for 20 minutes.
220
2) Add 0.315 g of dissolved PAA (35 w/w% in water) to 5 mL of water. Add to the ZnS dispersion and ultrasonicate for 30 minutes
3) Isolate/dry the particles by centrifugation, as described in the procedures above.
Dispersion
Dispersion of synthesized nanoparticles was accomplished by adding several different reagents, depending on the dopant and the solvent (the cap was irrelevant). In water, dispersion of ZnS:Mn was accomplished using a method described by Yang and Bredol (2008), and dispersion of ZnS:Cu was accomplished using a method based on that suggested by Hieronymas (2002). ZnS:Cu was reactive with the TGA used to stabilize the ZnS:Mn, necessitating a different dispersion method. Dispersion of either ZnS:Mn or Cu in toluene was accomplished using a method described by Klausch et al. (2010).
ZnS:Mn (aqueous)
1) Prepare a dispersion of 2.5 w/w% ZnS:Mn powder in water by ultrasonication for 15 minutes.
2) Adjust the pH of the dispersion to 9 using 2M NaOH.
3) Add enough TGA to bring the TGA concentration to 0.023 M.
4) Ultrasonicate in an ice bath for 1 hour; a transparent dispersion should result.
ZnS:Cu (aqueous)
1) Prepare a dispersion of 17.5 w/w% ZnS:Cu powder in water by ultrasonication for 15 minutes. The dispersion will be cloudy and unstable.
2) Separately, prepare a solution of 1.17 w/w% SHMP in water.
3) While ultrasonicating the SHMP solution in an ice bath, add the ZnS:Cu dispersion to it dropwise. The weight ratio of SHMP solution to ZnS:Cu dispersion should be 6:1, yielding a 2.5 w/w% ZnS:Cu dispersion, which should be transparent.
221
ZnS:Mn/Cu (organic)
1) Prepare a solution of 1 w/w% 3-MPA and 1 w/w% oleylamine in toluene.
2) While ultrasonicating this solution in an ice bath, add enough ZnS:Mn/Cu to achieve a concentration of 2.5 w/w%. The solution should remain transparent.
222
APPENDIX B. Procedure for conductivity estimation in PEDOT/SWCNT films
Samples of conductive ink were tested using a modified 2-point method. The 2-point method, while not as commonly used as the 4-point method (Heaney, 2000), was found to give more consistent results on porous substrates, such as the different paper types that were tested. The 2 points refer to two electrodes connected to an ohmmeter which were then contacted to a printed film. In order to measure an average resistance across the entire width of the film, bus bars were applied using a carbon paint (Luxprint 7102, Dupont Microelectronics), to which silver contact points were applied with a silver pen. Figure B.1 shows the experimental setup for testing resistance in a typical square sample of printed PEDOT/SWCNT ink.
(a) R (b) 2 mm
base sheet C C (b)
Ag Ag PEDOT:PSS-SWCNT layer
Appendix Figure B.1. (a) Schematic of resistance testing setup, showing bus bars and silver contacts. (b) Actual resistance sample, on paper, with bus bars and contacts applied.
Measurement across the sample yielded bulk resistance (R). This value is independent of sample geometry. Therefore, to obtain a more meaningful figure – conductivity – resistance was used to calculate resistivity (), the inverse of which is conductivity (). Resistance is related to conductivity by the following expressions (Heaney 2000):
223
푅푤ℎ 휌 휎 푙 휌 where w, h, and l are the width, thickness, and length of the sample, respectively. With a square sample, w = l and = (Rh)-1. Therefore, sample thickness was required to estimate conductivity.
In the case of impermeable substrates, such as glass, the thickness was easily obtained using optical profilometry (see Appendix E). In the case of permeable paper (or even the absorbent coating on the surface of inkjet acetate), thickness was not uniform throughout the sample and depended on ink penetration depth. Therefore, a cross-section of the sample and the average thickness of the ink layer in the exposed section were used to estimate conductivity. The conventional method of preparing cross-sections, by embedding samples in epoxy resin and sectioning with a microtome, was not used because of PEDOT dissolution and migration in the epoxy, observed under SEM. Instead, samples were fastened between two glass slides and a cross-section was cut along the edges of the slides using a surgical scalpel. In order to maintain as flat a section as possible, normal to the surface of the sheet, a sample was fastened to a glass slide with double-sided tape, overhanging the edge. The slide was then placed, sample side down, on a flat surface, and the scalpel was drawn along the edge of the slide to produce a smooth section parallel to the edge of the slide. Scalpels were replaced after preparing 50 sections to preserve sharpness. A second slide, again with double-sided tape, was applied to the other side of the sample, forming a “sandwich”.
The noticeable blue-black colour of the PEDOT/SWCNT ink allowed it to be readily observed under an optical microscope (Leica DM-LA) and attached camera at 100x magnification. In order to compensate for any roughness in the section, images were captured at several focal depths and stacked to generate an in-focus image. The section was then converted to a greyscale image – the dark-coloured ink appeared as a darker grey. The image was then colour-inverted, and the grey levels at or above that of the ink colour were highlighted using thresholding in ImageJ software, which highlighted only these brighter areas. The highlighted area therefore corresponded to the inked region, which was then measured as a percent-area of the entire image.
224
Multiplying the percent-area by the image dimension parallel to the layer thickness (“length”) gave an average layer thickness (“width”) for the sample:
푎푟푒푎 푡ℎ푟푒푠ℎ표푙푑푒푑 푎푟푒푎 표 𝑖푚푎𝑔푒 푎푟푒푎 표 푡ℎ푟푒푠ℎ표푙푑푒푑 푟푒𝑔𝑖표푛 푎푟푒푎 표 푡ℎ푟푒푠ℎ표푙푑푒푑 푟푒𝑔𝑖표푛 푎푟푒푎 표 푡ℎ푟푒푠ℎ표푙푑푒푑 푟푒𝑔𝑖표푛 푡ℎ푟푒푠ℎ표푙푑푒푑 푟푒𝑔𝑖표푛 푤𝑖푑푡ℎ 𝑖푚푎𝑔푒 푙푒푛𝑔푡ℎ 푡ℎ푟푒푠ℎ표푙푑푒푑 푟푒𝑔𝑖표푛 푙푒푛𝑔푡ℎ
A sequential representation of this procedure is shown in Figure B2. This procedure was repeated on several sections, yielding an accurate estimate of average ink layer thickness.
1) colour glass slide 2) greyscale tape inked region
non-inked region tape glass slide 3) inverted 4) cropped grey-level threshold
5) threshold- ing applied
100 µm
Appendix Figure B.2. Estimation of cross-sectional ink layer thickness using thresholding, showing stages 1-5 of obtaining a thresholded inked region.
On non-permeable substrates, such as glass, ITO, or cleaned acetate (i.e. with the absorbent layer removed), resistance was measured using the same method, but thickness was measured using optical profilometry (Veeco WYKO NT-1100), as described in Appendix C, following.
225
APPENDIX C. Film thickness estimation
For estimation of conductivity and dielectric constant, as well as for determining the number of print passes required to achieve a desired layer thickness, film thickness had to be measured. Optical profilometry was used for this purpose – specifically, using the Veeco WYKO NT1100 profiler. A film was deposited on the desired (impermeable) substrate and an optical profile was obtained at the edge of the film, where it met the substrate surface. After obtaining a profile, an x-direction linescan profile was extracted at several points (5-10) along the sample using the
WYKO software (Figure C.1).
glass
200 µm 0.2
(µm)
0.1 thickness
0
-0.1 0 200 400 600 PEDOT:PSS film x-dimension (µm)
Appendix Figure C.1. Extraction of a linescan profile from optical profile (1 layer PEDOT:PSS ink on glass).
During measurement, the stage upon which the slides rested was moved and tilted to bring the desired location into focus and to align the optical fringes upon that location. As a result, the linescan profiles were not always level and the substrate surface was not located at thickness = 0 (see Figure C.1). Measurement of thickness took place not only at several linescan locations, but at several locations along each linescan itself, to obtain a representative value. Therefore, profiles were zeroed and leveled before measurements took place.
The zeroing procedure simply involved finding the value of the measured thickness at the very edge of the substrate (not the film). The difference between this value and 0 was then added to all data points in the linescan (Figures C.2a, b). For leveling the linescan, the slope of the substrate was estimated by plotting only the substrate region and fitting it to a linear regression
226
(Figure C.2c). The data points were then indexed by number, starting with point 1 as the edge of the substrate. Levelled data was obtained by the following formula:
푙푒푣푒푙푙푒푑 푣푎푙푢푒 표푟𝑖𝑔𝑖푛푎푙 푣푎푙푢푒 𝑖푛푑푒 푛푢푚푏푒푟 푠푢푏푠푡푟푎푡푒 푠푙표푝푒
0.2 (a) 605.88327607.12995608.37662609.62329610.86997612.11664613.36331614.60999615.85666617.10333618.35001619.59668620.84335622.09003604.6366623.3367 (b) 573.46976574.71644575.96311577.20978578.45646579.70313582.19648583.44315584.68982587.18317588.42984589.67652590.92319592.16987593.41654594.66321595.90989597.15656598.40323599.64991600.89658602.14325603.38993580.9498585.9365 549.78297551.02964552.27632553.52299554.76966556.01634557.26301558.50968559.75636561.00303563.49638564.74305565.98972568.48307569.72974570.97642572.22309548.5363562.2497567.2364 527.34285528.58952531.08287532.32954533.57622534.82289536.06956537.31624538.56291539.80958541.05626542.30293544.79628546.04295547.28962529.8362543.5496 496.17601497.42269498.66936499.91603501.16271502.40938503.65605504.90273507.39607508.64275509.88942511.13609512.38277513.62944514.87611516.12279517.36946518.61613519.86281521.10948522.35615523.60283526.09617506.1494524.8495 zeroing 465.00918466.25585467.50253469.99587471.24255472.48922473.73589474.98257476.22924477.47591478.72259479.96926481.21593482.46261483.70928484.95595486.20263488.69597489.94265491.18932492.43599493.68267494.92934468.7492487.4493 433.84234435.08902436.33569437.58236438.82904440.07571441.32238442.56906443.81573446.30908447.55575448.80242451.29577452.54244453.78912455.03579456.28246457.52914458.77581460.02248461.26916462.51583445.0624450.0491463.7625 0.1 402.67551403.92218405.16886406.41553408.90888410.15555411.40222413.89557415.14224416.38892417.63559418.88226420.12894421.37561422.62228423.86896425.11563427.60898428.85565430.10232432.59567407.6622412.6489426.3623431.349 value 371.50867372.75535374.00202375.24869376.49537377.74204378.98871380.23539381.48206382.72873383.97541385.22208386.46875387.71543390.20877391.45545392.70212393.94879395.19547396.44214397.68881398.93549400.18216401.42883388.9621 340.34184341.58851342.83519344.08186345.32853346.57521347.82188349.06855350.31523352.80857354.05525355.30192356.54859357.79527359.04194360.28861361.53529362.78196364.02863365.27531366.52198367.76865369.01533351.5619370.262 310.42168311.66835312.91502315.40837316.65504317.90172319.14839320.39506321.64174322.88841324.13508325.38176326.62843329.12178330.36845331.61512334.10847335.35514336.60182337.84849339.09516314.1617327.8751332.8618309.175 278.00817279.25484280.50152281.74819282.99486284.24154285.48821286.73488287.98156289.22823291.72158292.96825294.21492296.70827297.95494299.20162300.44829301.69496302.94164304.18831305.43498306.68166307.92833290.4749295.4616 246.84133248.08801249.33468250.58135251.82803254.32137255.56805256.81472258.06139259.30807260.55474261.80141263.04809264.29476265.54143266.78811268.03478269.28145270.52813273.02147274.26815275.51482276.76149253.0747271.7748 216.92117218.16785219.41452220.66119221.90787223.15454224.40121225.64789226.89456228.14123229.38791230.63458231.88125233.12793235.62127236.86795238.11462239.36129240.60797241.85464243.10131244.34799245.59466215.6745234.3746 0 184.50766185.75434187.00101188.24768189.49436190.74103193.23438194.48105195.72772198.22107199.46774200.71442201.96109203.20776204.45444205.70111206.94778208.19446209.44113211.93448213.18115214.42782191.9877196.9744210.6878 153.34083155.83418157.08085158.32752160.82087162.06754163.31422164.56089165.80756167.05424168.30091169.54758170.79426172.04093174.53428175.78095177.02762179.52097180.76764182.01432183.26099154.5875159.5742173.2876178.2743 122.17399123.42067124.66734125.91401127.16069128.40736129.65403130.90071132.14738133.39405134.64073137.13407138.38075139.62742140.87409142.12077143.36744144.61411145.86079147.10746148.35413149.60081150.84748152.09415135.8874 Thickness (µm) Thickness 97.24052598.48719999.733872100.98055102.22722103.47389104.72057105.96724107.21391108.46059109.70726110.95393112.20061113.44728114.69395115.94063118.43397119.68065120.92732117.1873 93.50050594.74717895.993852(µm) Thickness 63.58034364.82701767.32036468.56703771.06038472.30705773.55373174.80040476.04707777.29375178.54042479.78709881.03377182.28044483.52711884.77379186.02046587.26713888.51381189.76048591.00715892.25383262.3336766.0736969.81371 31.16683532.41350833.66018234.90685536.15352937.40020238.64687539.89354941.14022242.38689643.63356944.88024246.12691647.37358948.62026349.86693651.11360952.36028353.60695656.10030357.34697659.84032361.08699754.8536358.59365 -0.1 1.24667342.49334683.74002024.98669367.48004048.72671389.973387211.22006112.46673413.71340714.96008116.20675417.45342818.70010119.94677421.19344822.44012123.68679524.93346826.18014127.42681528.67348829.9201626.2333670 0 200 400 600 0 200 400 600 x-dimension (µm) x-dimension (µm)
y = x (c) index # (d) 600.89658602.14325603.38993605.88327607.12995608.37662609.62329610.86997612.11664613.36331614.60999615.85666617.10333618.35001619.59668620.84335622.09003604.6366623.3367 597.15656598.40323599.64991600.89658602.14325603.38993605.88327607.12995608.37662609.62329610.86997612.11664613.36331614.60999615.85666617.10333618.35001619.59668620.84335622.09003604.6366623.3367 569.72974570.97642572.22309573.46976574.71644575.96311577.20978578.45646579.70313582.19648583.44315584.68982587.18317588.42984589.67652590.92319592.16987593.41654594.66321595.90989597.15656598.40323599.64991580.9498585.9365 565.98972568.48307569.72974570.97642572.22309573.46976574.71644575.96311577.20978578.45646579.70313582.19648583.44315584.68982587.18317588.42984589.67652590.92319592.16987593.41654594.66321595.90989567.2364580.9498585.9365 561.00303563.49638564.74305565.98972568.48307562.2497567.2364 557.26301558.50968559.75636561.00303563.49638564.74305562.2497 n 531.08287532.32954533.57622534.82289536.06956537.31624538.56291539.80958541.05626542.30293544.79628546.04295547.28962549.78297551.02964552.27632553.52299554.76966556.01634557.26301558.50968559.75636529.8362543.5496548.5363 526.09617527.34285528.58952531.08287532.32954533.57622534.82289536.06956537.31624538.56291539.80958541.05626542.30293544.79628546.04295547.28962549.78297551.02964552.27632553.52299554.76966556.01634529.8362543.5496548.5363 0 498.66936499.91603501.16271502.40938503.65605504.90273507.39607508.64275509.88942511.13609512.38277513.62944514.87611516.12279517.36946518.61613519.86281521.10948522.35615523.60283526.09617527.34285528.58952506.1494524.8495 494.92934496.17601497.42269498.66936499.91603501.16271502.40938503.65605504.90273507.39607508.64275509.88942511.13609512.38277513.62944514.87611516.12279517.36946518.61613519.86281521.10948522.35615523.60283506.1494524.8495 496.17601497.42269 492.43599493.68267 465.00918466.25585467.50253469.99587471.24255472.48922473.73589474.98257476.22924477.47591478.72259479.96926481.21593482.46261483.70928484.95595486.20263488.69597489.94265491.18932492.43599493.68267494.92934468.7492487.4493 461.26916462.51583465.00918466.25585467.50253469.99587471.24255472.48922473.73589474.98257476.22924477.47591478.72259479.96926481.21593482.46261483.70928484.95595486.20263488.69597489.94265491.18932463.7625468.7492487.4493 433.84234435.08902436.33569437.58236438.82904440.07571441.32238442.56906443.81573446.30908447.55575448.80242451.29577452.54244453.78912455.03579456.28246457.52914458.77581460.02248461.26916462.51583445.0624450.0491463.7625 430.10232432.59567433.84234435.08902436.33569437.58236438.82904440.07571441.32238442.56906443.81573446.30908447.55575448.80242451.29577452.54244453.78912455.03579456.28246457.52914458.77581460.02248445.0624450.0491431.349 402.67551403.92218405.16886406.41553408.90888410.15555411.40222413.89557415.14224416.38892417.63559418.88226420.12894421.37561422.62228423.86896425.11563427.60898428.85565430.10232432.59567407.6622412.6489426.3623431.349 398.93549400.18216401.42883402.67551403.92218405.16886406.41553408.90888410.15555411.40222413.89557415.14224416.38892417.63559418.88226420.12894421.37561422.62228423.86896425.11563427.60898428.85565407.6622412.6489426.3623 371.50867372.75535374.00202375.24869376.49537377.74204378.98871380.23539381.48206382.72873383.97541385.22208386.46875387.71543390.20877391.45545392.70212393.94879395.19547396.44214397.68881398.93549400.18216401.42883388.9621 367.76865369.01533371.50867372.75535374.00202375.24869376.49537377.74204378.98871380.23539381.48206382.72873383.97541385.22208386.46875387.71543390.20877391.45545392.70212393.94879395.19547396.44214397.68881388.9621370.262 340.34184341.58851342.83519344.08186345.32853346.57521347.82188349.06855350.31523352.80857354.05525355.30192356.54859357.79527359.04194360.28861361.53529362.78196364.02863365.27531366.52198367.76865369.01533351.5619370.262 337.84849339.09516340.34184341.58851342.83519344.08186345.32853346.57521347.82188349.06855350.31523352.80857354.05525355.30192356.54859357.79527359.04194360.28861361.53529362.78196364.02863365.27531366.52198351.5619 310.42168311.66835312.91502315.40837316.65504317.90172319.14839320.39506321.64174322.88841324.13508325.38176326.62843329.12178330.36845331.61512334.10847335.35514336.60182337.84849339.09516314.1617327.8751332.8618 306.68166307.92833310.42168311.66835312.91502315.40837316.65504317.90172319.14839320.39506321.64174322.88841324.13508325.38176326.62843329.12178330.36845331.61512334.10847335.35514336.60182314.1617327.8751332.8618309.175 279.25484280.50152281.74819282.99486284.24154285.48821286.73488287.98156289.22823291.72158292.96825294.21492296.70827297.95494299.20162300.44829301.69496302.94164304.18831305.43498306.68166307.92833290.4749295.4616309.175 276.76149278.00817279.25484280.50152281.74819282.99486284.24154285.48821286.73488287.98156289.22823291.72158292.96825294.21492296.70827297.95494299.20162300.44829301.69496302.94164304.18831305.43498290.4749295.4616 249.33468250.58135251.82803254.32137255.56805256.81472258.06139259.30807260.55474261.80141263.04809264.29476265.54143266.78811268.03478269.28145270.52813273.02147274.26815275.51482276.76149278.00817253.0747271.7748 245.59466246.84133248.08801249.33468250.58135251.82803254.32137255.56805256.81472258.06139259.30807260.55474261.80141263.04809264.29476265.54143266.78811268.03478269.28145270.52813273.02147274.26815275.51482253.0747271.7748 218.16785219.41452220.66119221.90787223.15454224.40121225.64789226.89456228.14123229.38791230.63458231.88125233.12793235.62127236.86795238.11462239.36129240.60797241.85464243.10131244.34799245.59466246.84133248.08801234.3746 216.92117218.16785219.41452220.66119221.90787223.15454224.40121225.64789226.89456228.14123229.38791230.63458231.88125233.12793235.62127236.86795238.11462239.36129240.60797241.85464243.10131244.34799215.6745234.3746 188.24768189.49436190.74103193.23438194.48105195.72772198.22107199.46774200.71442201.96109203.20776204.45444205.70111206.94778208.19446209.44113211.93448213.18115214.42782216.92117191.9877196.9744210.6878215.6745 184.50766185.75434187.00101188.24768189.49436190.74103193.23438194.48105195.72772198.22107199.46774200.71442201.96109203.20776204.45444205.70111206.94778208.19446209.44113211.93448213.18115214.42782191.9877196.9744210.6878 157.08085158.32752160.82087162.06754163.31422164.56089165.80756167.05424168.30091169.54758170.79426172.04093174.53428175.78095177.02762179.52097180.76764182.01432183.26099184.50766185.75434187.00101159.5742173.2876178.2743 155.83418157.08085158.32752160.82087162.06754163.31422164.56089165.80756167.05424168.30091169.54758170.79426172.04093174.53428175.78095177.02762179.52097180.76764182.01432183.26099154.5875159.5742173.2876178.2743 127.16069128.40736129.65403130.90071132.14738133.39405134.64073137.13407138.38075139.62742140.87409142.12077143.36744144.61411145.86079147.10746148.35413149.60081150.84748152.09415153.34083155.83418135.8874154.5875 123.42067124.66734125.91401127.16069128.40736129.65403130.90071132.14738133.39405134.64073137.13407138.38075139.62742140.87409142.12077143.36744144.61411145.86079147.10746148.35413149.60081150.84748152.09415153.34083135.8874 118.43397119.68065120.92732122.17399123.42067124.66734125.91401117.1873 114.69395115.94063118.43397119.68065120.92732122.17399117.1873(µm) Thickness 87.26713888.51381189.76048591.00715892.25383293.50050594.74717895.99385297.24052598.48719999.733872100.98055102.22722103.47389104.72057105.96724107.21391108.46059109.70726110.95393112.20061113.44728114.69395115.94063 83.52711884.77379186.02046587.26713888.51381189.76048591.00715892.25383293.50050594.74717895.99385297.24052598.48719999.733872100.98055102.22722103.47389104.72057105.96724107.21391108.46059109.70726110.95393112.20061113.44728 56.10030357.34697659.84032361.08699763.58034364.82701767.32036468.56703771.06038472.30705773.55373174.80040476.04707777.29375178.54042479.78709881.03377182.28044483.52711884.77379186.02046558.5936562.3336766.0736969.81371 53.60695656.10030357.34697659.84032361.08699763.58034364.82701767.32036468.56703771.06038472.30705773.55373174.80040476.04707777.29375178.54042479.78709881.03377182.28044454.8536358.5936562.3336766.0736969.81371 31.16683532.41350833.66018234.90685536.15352937.40020238.64687539.89354941.14022242.38689643.63356944.88024246.12691647.37358948.62026349.86693651.11360952.36028353.60695654.85363(µm) Thickness 27.42681528.67348829.92016231.16683532.41350833.66018234.90685536.15352937.40020238.64687539.89354941.14022242.38689643.63356944.88024246.12691647.37358948.62026349.86693651.11360952.360283 1.24667342.49334683.74002024.98669367.48004048.72671389.973387211.22006112.46673413.71340714.96008116.20675417.45342818.70010119.94677421.19344822.44012123.68679524.93346826.18014127.42681528.67348829.9201626.2333670 1.24667342.49334683.74002024.98669367.48004048.72671389.973387211.22006112.46673413.71340714.96008116.20675417.45342818.70010119.94677421.19344822.44012123.68679524.93346826.1801416.2333670 450 500 550 600 0 200 400 600 x-dimension (µm) x-dimension (µm)
2 (e) 1.9 1.8
1.7 1.6 1.5 1.4 1.3 1.2 1.1 1 thickness 0.9 0.8 0.7 0.6 0.5 0.4 Thickness (µm) Thickness 0.3 0.2 0.1 0 0 200 400 600 x-dimension (µm)
Appendix Figure C.2. Zeroing and levelling of x-direction linescan for thickness estimation (1 layer PEDOT:PSS ink on glass). (a) Location of zero-point; (b) zeroing of linescan; (c) obtaining substrate slope; (d) levelling of entire linescan; (e) levelled, zeroed linescan.
227
APPENDIX D. Drop and line spacing optimization
Drop spacing was an important variable in the formation of smooth, thin films for depositing the various functional materials. Drop spacing refers to the distance between the centre of adjacent jetted drops, and was controlled by changing the angle of the nozzle array on the DMP Materials Cartridges. The DMP2831 printer was capable of depositing drops at a spacing as low as 5 µm, and as high as 250 µm, in 5 µm intervals. To print as smooth a film as possible, the space between individual jetted drops was controlled. Individual drops spread to an average radius and coalesced into films when they were close enough together; however, overly close drops overlapped, resulting in rough topography (Figure D.1). Also, the different surface properties of different substrates had an effect on the wetting behaviour of the various inks, so each ink needed to be tested on each surface it might contact during LED construction. Therefore, a 3-stage approach was used to determine optimum drop spacing.
90 µm 80 µm 60 µm 40 µm
drop spacing (90 µm)
Appendix Figure D.1. Drop spacing of ZnS:Mn-Aq ink on glass.
Preparation of jetting surfaces/substrates
Bare substrates, such as glass, ITO glass, or ITO PET, were cleaned as usual (see Appendix F). However, because the LED structure required stacking of different inks, surfaces covered in each of the inks had to be prepared. This was achieved by spin-coating of the inks onto glass slides (which had themselves been previously cleaned). All spin-coated slides were dried at 150°C for 30 min in air. The spin speeds used are listed below.
228
PEDOT:PSS/SWCNT ink: 3000 RPM for 20 s; 500 RPM for 10 s.
ZnS:Cu/PVK and ZnS:Mn/PVP inks: 2500 RPM for 15 s; 200 RPM for 45 s.
Determination of drop size
Each ink was jetted onto each substrate/surface using a pattern designed to deliver an array of drops at 250 µm spacing, so that individual drops would remain distinct. Without drying the drops, the pattern was examined under the optical microscope, photographed, and the drop diameter measured for several drops in ImageJ, and averaged (Figure D.2). The purpose of this stage was to estimate the likely drop spacing that would result in the coalescence of adjacent drops and the formation of printed lines or films. The drop diameter represented the largest drop spacing that could realistically be expected to form a film.
100 µm
Appendix Figure D.2. 250 µm-spaced drops of ZnS:Mn-Aq ink on ITO glass (circled with dotted lines).
Determination of optimum drop spacing
Ideal drop spacing was established by jetting several lines of ink at different drop spacings in increments of 5 µm, ranging from the measured single drop diameter down to 50% of the single drop diameter (i.e. for a drop size of 60 µm, drop spacing ranging from 30-60 µm was tested. The resulting films were characterized using optical microscopy for uniformity (Figure D.3). Ideal drop spacing was considered to be when lines were fully merged (no holes) but not overlapping. Because the drop spacing was only tunable in multiples of 5 µm, precise spacing to avoid holes and overlapping was usually not possible, so a minimal amount of overlap without any holes was targeted. In Figure D.3, this would be at 55 µm, for example. Once ideal drop
229
spacing was established, the ideally-spaced printed films were optically profiled to confirm the absence of holes and large overlap.
25 µm 35 µm 45 µm 55 µm 65 µm
100 µm
230
APPENDIX E. Paper substrate preparation
Northern bleached SW and HW kraft pulps were used as supplied. 40 g of oven-dry (OD) pulp was soaked in 500 mL of deionized water for 12 h. The excess water was pressed out of the pulp in order to calculate consistency, which was 23-25% for both pulp types. Pulp was then refined in a Noram refiner for 5100 revolutions after which 24 g (dry pulp weight) of refined pulp was added to 2 L of water, and was then mixed in a Durant disintegrator for 15000 revolutions. The resulting slurry was made into handsheets containing 1.5 g dry pulp per sheet, using a non- recirculating Noram sheet former according to TAPPI Standard Method T-205. TiO2 filler and alkylketene dimer (AKD) internal sizing agent (Hercon 115, Hercules) were added to the pulp slurry in filled sheets. Internal sizing agent was added at 0.8 w/w% on an OD pulp basis. A PEI-based retention aid (Polymin SK, 30 w/w% active ingredient in water, BASF) was added to sheets containing filler during sheet forming, with a ratio of retention aid:filler of 4:1. An ink fixation agent, PDADMAC, was added at 2 w/w% on an OD pulp basis during sheet forming. Handsheets were air dried in a conditioning room at 25°C and 75% relative humidity for 24 h. All sheets were calendared at 80oC, 100 kPa and 3 nips using a Beloit Wheeler laboratory calendar, couch side up.
Commercial sheets were used as received, without additional cleaning. Sheets were stored in low-oil vacuum foil in the same conditioning room as the handsheets until use.
231
APPENDIX F. Impermeable substrate preparation
Several non-paper substrates were used in this work: slide glass, ITO glass (patterned and unpatterned), ITO PET, and cellulose acetate. Patterned ITO substrates were purchased from Kintech Limited. Unpatterned ITO glass and PET were purchased from Sigma-Aldrich. Cellulose acetate was purchased from Avery. Finally, patterned slide glass and cellulose acetate were prepared by printing them with 2 layers of the PEDOT:PSS/SWCNT ink, and curing for 2 hours at 150°C on a hotplate in air. All of the substrates listed above, including the PEDOT:PSS/SWCNT-printed ones, were cleaned before overprinting with functional layers. The cleaning process was carried out in an ultrasonic bath, as follows:
(i) Soap and water, 30 minutes, followed by rinsing with water;
(ii) Acetone, 15 minutes;
(iii) 50:50 vol:vol methanol and ethanol, 15 minutes.
Substrates were air-dried between each step with a jet of compressed air. Washed ITO substrates were then also plasma-treated with oxygen plasma for 15 minutes to remove residual contaminants.
232
APPENDIX G. Paper substrate characterization
Several variables had an effect on ink performance on a given paper type. Of these, the most important that were noted were filler type and content, surface energy, and surface absorbency.
Sheets were characterized for thickness using a TMI micrometer at 10 different points on the sheet. Average sheet thickness (tavg) was also used to estimate sheet density () by weighing (m) a sample of a fixed area (A).
푚 휌 푡 퐴
Filler/pigment content was determined by burning a portion of each sheet of a fixed weight in an oven at 500oC for 1 h, at which point the fibres were completely ashed. The actual weight percent of filler (plus a small amount of ash) retained in the sheet was estimated by dividing filler weight by the original sample weight (before burning).
Contact angle was estimated using an aqueous solution of crystal violet dye (test ink) prepared according to TAPPI Standard Method T431 for measuring ink absorbency into paper. The test ink had a surface tension of 62 mN/m. 30 µL of this ink was dropped with a calibrated micropipette onto a handsheet, and the resulting drop was immediately photographed from the side using a Canon Rebel XT-ME DSLR camera with a MP-E 65 mm macro lens. Ink absorbency of the sheets was observed by measuring the time for complete absorption of a 30 µL sample of the same test ink into the surface. During this test, the samples were placed directly under a 60 W incandescent lamp elevated 30 cm from the test specimen’s surface. Complete absorbency was defined according to Test Method T431 as the point at which light reflection from the droplet on the surface was no longer visible. Micrographs of paper sheets were used to observe filler, coating, and surface pore distribution. A JEOL-7001 JSM scanning electron microscope (SEM), employing the backscattered electron (BSE) detector on the SEM, was used to improve contrast between the filler and fibres.
233
APPENDIX H. Detailed ink formulations
All formulations are given on a w/w% basis. All reagents, except where otherwise specified, were supplied by Sigma-Aldrich Canada. One reagent in particular, poly(n-vinylcarbazole) or PVK, was synthesized by fellow doctoral candidate Graham Morse using the method described by Higashimura et al. (1980). The solutions used in several of the inks are composed as follows (again, on a w/w% basis):
PEDOT:PSS: 1.3% PEDOT particles in water, 2:1 PSS:PEDOT ratio
SWCNTs: 0.04% SWCNTs (1.2-5 nm diameter, 2.5 µm length) in water, 0.5% SLS
ZnS (aqueous): 2.5% 3-MPA-capped ZnS NPs in pH 9 aqueous NaOH solution, 0.25% TGA
ZnS (organic): 2.5% octylamine-capped ZnS NPs in toluene, 1.25% 3-MPA and olelyamine
ZnS (acid): 1% AA-capped ZnS NPs in AA
PEDOT:PSS/SWCNTs (PS) ZnS/PVP (ZnS-Aq)
34% PEDOT:PSS 40% ZnS (aqueous)
10% SWCNTs 10% IPA
10% DMSO 15% butoxyethanol
17% glycerol 34.9% water
0.5% SLS 0.1% PVP 1 300 000
0.5% Surfynol DF-110D defoamer
28% water
234
ZnS/PVK (ZnS-Org) BaTiO3/MMA (BT)
40% ZnS (organic) 5% BaTiO3 NPs, <25 nm diameter
10% cyclohexane 33% PEG 300
0.075% PVK 50 000 33% EtOH
49.025% chlorobenzene 0.5% Surfynol CT-324 nonionic surfactant
0.5% BYK Chemie Disperbyk 111 dispersant ZnS/AA (ZnS-AA) 28% MMA 80% ZnS (acid)
10% sulfolane
8% diethylene glycol diacrylate
2% Zonyl FS-300 fluorosurfactant
235
ZnS/PVDF/PVP Meyer-coating paste BaTiO3/PVDF Meyer-coating paste
26.7% Sylvania GlacierGlo ZnS phosphor 31.7% BaTiO3 <3 µm diameter powder 47.3% DMAc 48% DMAc 1.5% BYK Chemie Disperbyk 111 2.7% Cytec Modaflow 2100 dispersant
13.4% PVDF 19.5% PVDF
1.4% PM(MA-co-EA)
7.7% PVP 10 000
236
APPENDIX I. Ink iterations
There were many different iterations of each ink that was tested. This appendix is intended to list the major iterations of each ink type that was prepared for this study and notes on why it was or was not used in the final device construction. The appendix has been broken down into several tables, each one describing a particular ink type.
Appendix Table I.1. Organic solvent-based ZnS inks.
Ink name Component Amount (w/w%) Notes CB-1 ZnS:Mn@SHMP 0.25 No PL on printing. Triton X-100 7 Filtration removes all Chlorobenzene 72 ZnS:Mn. Poor solubility in Cyclohexane 20 chlorobenzene. Polyvinylcarbazole 0.75
AA-1 ZnS:Mn@AA 0.25 No PL in solution. Acrylic acid 10 Weak PL on deposition Methyl methacrylate 20 Damaging to printhead. Damaging to substrate. Sulfolane 20 Red-shifted emission. DEG-diacrylate 1.5 N,N-dimethylformamide 48.25
CB-2 ZnS:Mn@SHMP 0.25 No PL on printing Triton X-100 7 Again, poor solubility in Chlorobenzene 72 chlorobenzene.
Cyclohexane 19.75 Polyvinylcarbazole 1
CB-3 ZnS:Mn@SHMP 2.5 No PL on printing. Triton X-100 7 Printhead clogs almost Chlorobenzene 70 immediately.
Cyclohexane 20
Polyvinylcarbazole 0.5
CB-4 ZnS:Mn@SHMP 0.1 No PL on printing. Chlorobenzene 77 Cyclohexane 21.9 Polyvinylcarbazole 1
237
CT-1 ZnS stock solution, in 20 No PL in solution, chlorobenzene or when printed. Chlorobenzene 56.375 Stock solution very poorly Cyclohexane 22.125 dispersed. Polyvinylcarbazole 1.5
CT-2 ZnS stock solution, in 40 Same issues as CT-1. chlorobenzene Chlorobenzene 20 Cyclohexane 9.5 Toluene 30 Polyvinylcarbazole 0.5
CT-3 ZnS stock solution, in 40 Better dispersion with THF chlorobenzene added. Chlorobenzene 20 No PL. Tetrahydrofuran 19.5 Very long drying Toluene 20 time. Polyvinylcarbazole 0.5
TT-1* ZnS stock solution, in 40 Prints well. toluene Well dispersed and Chlorobenzene 49.025 transparent. Cyclohexane 10 Polyvinylcarbazole 0.075
238
Appendix Table I.2. Aqueous ZnS inks.
Ink name Component Amount (w/w%) Notes AQ-1 ZnS:Mn@SHMP 0.25 Very long drying time. Beckman-Coulter Dispersant IIIA 2 No PL in printed film. Triton X-100 1
water 56.25
glycerol 20 PEG-300 20
AQ-2 ZnS:Mn@SHMP 0.25 Longest drying time. Beckman-Coulter Dispersant IIIA 2 No PL in printed film. Triton X-100 1
water 56.25 PEG-300 40
AQ-3 ZnS:Mn@PVP 0.25 Unstable. glycerol 20 Very viscous. polyvinyl alcohol 3 Non-jettable.
Triton X-100 1
Beckman-Coulter Dispersant IC 0.5 Beckman-Coulter Dispersant IIIA 0.5 water 74.75
AQ-4 ZnS:Mn@PVP 0.25 Very unstable. glycerol 20 Very viscous. polyvinyl alcohol 1.5 Non-jettable.
Triton X-100 1
Beckman-Coulter Dispersant IC 0.5 Beckman-Coulter Dispersant IIIA 0.5 water 76.25
AQ-5 ZnS:Mn@PVP 0.25 No PL on printing. glycerol 40 Poor resolution. Zonyl FS-300 0.1 Surface tension too low.
Beckman-Coulter Dispersant IIIA 1 water 58.65
AQ-6 ZnS:Mn@SHMP 2.5 Difficult to filter. glycerol 40 Bright PL, but red-shifted. Zonyl FS-300 0.1 Clogs printhead.
Beckman-Coulter Dispersant IIIA 1 water 56.4
239
AQ-7 ZnS:Mn@SHMP 0.1 Difficult to filter. glycerol 40 Dim PL. Zonyl FS-300 0.1 No PL on printing. Eventually unstable. Disperbyk 111 0.005 water 59.795
AQ-8 ZnS:Mn@SHMP 0.1 Precipitated out. glycerol 20 Unstable. Zonyl FS-300 0.1
Disperbyk 111 0.005 water 79.795
AQ-9 ZnS:Mn@SHMP 0.1 Similar to AQ-8; glycerol 20 Longer shelf life, but Zonyl FS-300 0.1 eventually unstable.
Disperbyk 111 0.005
water 69.795
dimethyl sulfoxide 10
AQ-10 ZnS:Mn@SHMP 0.1 Weak PL. glycerol 10 Difficult to filter. Zonyl FS-300 0.1 No PL when printed.
Disperbyk 111 0.005
water 69.795 dimethyl sulfoxide 20
AQ-11 ZnS:Mn@SHMP 0.1 Similar to AQ-10. glycerol 5 Poor dispersion. Zonyl FS-300 0.1 Unstable over time.
Disperbyk 111 0.005
water 69.795 dimethyl sulfoxide 25
AQ-12 ZnS:Mn@SHMP 0.1 Similar to AQ-10, AQ-11. Zonyl FS-300 0.1 Poor dispersion. Disperbyk 111 0.005 Unstable over time. DMSO = bad stability. water 69.795 dimethyl sulfoxide 30
AQ-13 ZnS:Mn@SHMP 0.1 Filters well. Zonyl FS-300 0.1 Unstable over time. Disperbyk 111 0.005 No PL when printed. water 99.795 Zonyl does not affect stability.
240
AQ-14 ZnS:Mn@SHMP 0.1 Disperbyk 111 is Disperbyk 111 10 Insoluble at high loadings. water 89.9 Unstable.
AQ-15 ZnS:Mn@SHMP 0.1 Jets extremely poorly. Disperbyk 111 0.005 Good dispersion. water 79.895 Unable to print. 3-amino-1-propanol 20 Bright PL.
AQ-16 ZnS:Mn@SHMP 0.1 No PL in printed film. Zonyl FS-300 0.1 Print quality very poor. Disperbyk 111 0.005 Droplet formation poor.
water 69.795 3-amino-1-propanol 30
AQ-17 ZnS:Mn@SHMP 0.1 No PL in printed film. Disperbyk 111 0.005 Droplet formation poor. water 69.895 3-amino-1-propanol 30
WA-1 ZnS:Mn stock solution 20 Based on Small et al's PVP10K 1.5 formulation. Igepal CA-720 0.25
Surfynol DF-110D 0.25 Bad mercaptan stench and extremely unstable. mercaptosuccinic acid 6.162 water 60 dimethyl sulfoxide 11.838
WA-2 ZnS:Mn stock solution 20 Good printing. PVP10K 1 Bright, but not extremely Igepal CA-720 0.25 bright.
glycerol 8 Drop formation ideal. water 60.75 dimethyl sulfoxide 10
WA-3 ZnS:Mn stock solution 40 Similar to WA-2. PVP10K 1 Bright PL in solution and Igepal CA-720 0.25 printed. Higher ZnS:Mn glycerol 8 concentration. water 40.75 dimethyl sulfoxide 10 Jetting issues - over time, jetting is compromised.
241
WA-4 ZnS:Mn stock solution 40 Not jettable (3-AP is a PVP10K 1 problem). Igepal CA-720 0.25
glycerol 8
water 36 dimethyl sulfoxide 10 3-amino-1-propanol 4.75
WA-5 ZnS:Mn stock solution 40 High viscosity - good for PVP10K 1 jetting. Igepal CA-720 0.75 glycerol 58.25 Drop formation deteriorated rapidly.
WA-6 ZnS:Mn stock solution 40 High viscosity (lots of PVP) PVP10K 40 Unstable. Igepal CA-720 0.75 PL quenched by PVP. dimethyl sulfoxide 19.25
WA-7 ZnS:Mn stock solution 40 Clogging. PVP10K 5 Igepal CA-720 0.75 Jetting is poor. glycerol 54.25
WA-8 ZnS:Mn stock solution 40 Clogging. PVP10K 0.5 Igepal CA-720 0.75 Jetting is still poor. glycerol 58.75
WA-9 ZnS:Mn stock solution 40 Jetting is poor. PVP10K 0.5 Excessive foaming. Triton X-100 1 DF-110D destabilizes glycerol 40 solution. water 18.5
AQ-11-Mn ZnS:Mn stock solution 40 Viscosity is very high with PVP10K 0.5 CMC. Triton X-100 0.75 Stability compromised. Films will contain CMC and glycerol 5 PVP. water 53.35 carboxymethylcellulose 0.4
242
AQ-12-Mn ZnS:Mn stock solution 40 Similar problems as AQ-12- PVP10K 0.5 Mn. Igepal CA-720 0.75
glycerol 5
water 53.55 carboxymethylcellulose 0.2
AQ-13-Mn ZnS:Mn stock solution 40 Removal of PVP/CMC helps Igepal CA-720 0.75 jetting. glycerol 10
water 48 Foaming is a problem.
pH 10 buffer 0.25 0.5 M NaOH 1
AQ-14-Mn ZnS:Mn stock solution 40 Addition of isopropanol helps isopropanol 10 jetting. butoxyethanol 15 No foaming.
water 33.188 Stable with TGA added. PVP10K 0.35 thioglycolic acid 0.212 pH 10 buffer 0.25 0.5 M NaOH 1
AQ-15-Mn ZnS:Mn stock solution 65.6 Higher concentration of isopropanol 10 ZnS:Mn. butoxyethanol 15 Addition of butoxyethanol caused destabilization. water 7.848
PVP1.3M 0.09 Too much stock solution. thioglycolic acid 0.212 pH 10 buffer 0.25 0.5 M NaOH 1
AQ-16-Mn ZnS:Mn stock solution 40 Stable and jettable. isopropanol 10 Not durable to overprinting butoxyethanol 15 with aqueous ink. water 33.483
PVP1.3M 0.055 Not enough PVP, doesn’t thioglycolic acid 0.212 form films. pH 10 buffer 0.25 0.5 M NaOH 1
243
AQ-17-Mn ZnS:Mn stock solution 40 Stable and jettable. isopropanol 10 butoxyethanol 15
water 33.188
PVP1.3M 0.35 thioglycolic acid 0.212 pH 10 buffer 0.25 0.5 M NaOH 1
AQ-15-Cu ZnS:Cu stock solution 99.945 Stable. PVP1.3M 0.055 Jets poorly.
AQ-16-Cu ZnS:Cu stock solution 74.945 Unstable. PVP1.3M 0.055 Ethanol desolvates SHMP. ethanol 10 butoxyethanol 15
AQ-17-Cu ZnS:Cu stock solution 99.65 PVP1.3M 0.35
AQ-18-Mn* ZnS:Mn stock solution 40 Stable and jettable. isopropanol 10 butoxyethanol 15
water 34.65 PVP1.3M 0.35
244
Appendix Table I.3. BaTiO3 ink formulations.
Ink name Component Amount (w/w%) Notes BT-old BaTiO3 NPs x PEG-200 30 EtOH 50 Zetasperse 1200 x/10 Methyl methacrylate balance
BT BaTiO3 NPs 5 PEG-200 33 EtOH 33 Zetasperse 1200 0.5 Methyl methacrylate 28.5
BT-1 BaTiO3 NPs 5 PEG-200 33 EtOH 33 Surfynol CT-324 0.5 Disperbyk 111 0.5 Methyl methacrylate 28
BT-2 BaTiO3 NPs 5 EtOH 60 Surfynol CT-324 0.5 Disperbyk 111 0.5 Methyl methacrylate 28.4 Polyvinyl acetate 5 Surfynol DF-110D 0.5 Sodium citrate 0.1 BT-3 BaTiO3 NPs 5 EtOH 60 Surfynol CT-324 0.5 Disperbyk 111 0.5 Methyl methacrylate 32.4 Polyvinyl acetate 1.5 Surfynol DF-110D 0.5 Sodium citrate 0.1
BT-4* BaTiO3 NPs 5 PEG-300 33 EtOH 33 Surfynol CT-324 0.5
Disperbyk 111 0.5 Methyl methacrylate 28
245
BT-5 BaTiO3 NPs 5 PEG-300 10 EtOH 33 Surfynol CT-324 0.5 Disperbyk 111 0.5 Methyl methacrylate 51
BT-6 BaTiO3 NPs 5 PEG-300 5 EtOH 33 Disperbyk 111 0.5 Methyl methacrylate 56 carboxymethylcellulose 0.5
BT-7 BaTiO3 NPs 5 PEG-300 5 EtOH 33 Surfynol CT-324 1 Disperbyk 181 1 Methyl methacrylate 55
BT-8 BaTiO3 NPs 5 PEG-300 10 EtOH 33 Surfynol CT-324 1 Disperbyk 181 0.5 Disperbyk 111 0.5 Methyl methacrylate 45 Diethylene glycol 5 diacrylate
BT-9 BaTiO3 NPs 5 Butoxyethanol 15 EtOH 33 Surfynol CT-324 1 Disperbyk 181 0.5 Disperbyk 111 0.5 Methyl methacrylate 40 Diethylene glycol 5 diacrylate
246
BT-10 BaTiO3 NPs 5 Butoxyethanol 10 PEG-200 5
EtOH 33
Surfynol CT-324 1 Disperbyk 181 0.5 Disperbyk 111 0.5 Methyl methacrylate 40 Diethylene glycol 5 diacrylate
247
Appendix Table I.4. PEDOT:PSS/SWCNT ink formulations.
Ink name Component Amount (w/w%) Notes Bronczyk Model PEDOT:PSS 68 isopropanol 12 glycerol 20
Garnett Model PEDOT:PSS 45 isopropanol 10 glycerol 15 dimethyl sulfoxide 30
Standard PEDOT:PSS 52.5 PEG-200 10 isopropanol 10 glycerol 25 Disperbyk 111 2.5
Standard + DMSO PEDOT:PSS 42 PEG-200 7 isopropanol 7 glycerol 14 dimethyl sulfoxide 28 Disperbyk 111 2
Standard + DMSO, PEDOT:PSS 41.7 MWCNTs PEG-200 7 isopropanol 7 glycerol 13.9 dimethyl sulfoxide 27.9 MWCNT solution 0.7 Disperbyk 111 1.8
Inkjet formulation PEDOT:PSS 32.7 water 24.5 isopropanol 5.4 glycerol 10.9 dimethyl sulfoxide 21.8 MWCNT solution 0.5 Disperbyk 111 1.4 sodium lauryl sulfate 2
248
PB PEDOT:PSS 34 water 35.5 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5
PT PEDOT:PSS 34 water 35.4 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 Triton X-100 0.1
PI PEDOT:PSS 34 water 35.3 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 Igepal CA-720 0.2
PZ PEDOT:PSS 34 water 35.3 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 Zonyl FS-300 0.2
PZS PEDOT:PSS 34 water 35.3 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 ZetaSperse 3700 0.2
PSB PEDOT:PSS 34 water 35 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 sodium dodecyl benzene 0.5 sulfonate
249
PLS PEDOT:PSS 34 water 35 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 sodium lignosulfonate 0.5
PBK PEDOT:PSS 34 water 35 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 benzalkonium chloride 0.5
PS PEDOT:PSS 34 water 35 glycerol 17 dimethyl sulfoxide 10 SWCNT solution 3 Surfynol DF-110D 0.5 sodium lauryl sulfate 0.5
PS-1 PEDOT:PSS 34 water 35 glycerol 17
dimethyl sulfoxide 10
SWCNT solution 3 Surfynol DF-110D 0.5 sodium lauryl sulfate 0.5
PS-2 PEDOT:PSS 34 glycerol 45 dimethyl sulfoxide 10 SWCNT solution 10 Surfynol DF-110D 0.5 sodium lauryl sulfate 0.5
PS-3 PEDOT:PSS 44 water 25 glycerol 10 dimethyl sulfoxide 10 SWCNT solution 10 Surfynol DF-110D 0.5 sodium lauryl sulfate 0.5
250
APPENDIX J. Jetting waveforms
35 PEDOT:PSS/SWCNT ink 30 Jetting frequency = 3 kHz
25
20
15 Voltage (V) Voltage 10
5
0 0 10 20 30 40 50 Time (µs)
30
BaTiO3 ink 25 Jetting frequency = 8.8 kHz
20
15
Voltage (V) Voltage 10
5
0 0 10 20 30 40 50 60 Time (µs)
251
35
ZnS/PVP/H2O ink 30 Jetting frequency = 2 kHz
25
20
15 Voltage (V) Voltage 10
5
0 0 5 10 15 20 Time (µs)
25 ZnS/PVK/chlorobenzene ink
20
Jetting frequency = 2 kHz
15
10 Voltage (V) Voltage
5
0 0 20 40 60 80 Time (µs)
252
APPENDIX K. ToF-SIMS fragments analyzed and construction of molecular maps
Time-of-flight secondary ion mass spectrometry (ToF-SIMS) was used to map the distribution of PEDOT and PSS ions and ion fragments relative to the paper fibres and filler. Samples were printed an immediately placed in high-vacuum foil to prevent them from contacting any oils or salts during handling. Before measurement, they were removed from the foil and placed with tweezers into the analysis chamber.
An ION-ToF ToF-SIMS IV apparatus was used to perform the measurements. A Bi3 primary ion gun was used to induce ion ejection and fragmentation. Images of the samples’ surfaces were obtained using the high-spatial-resolution detector, as the spatial location of the molecules in question was to be ascertained. The imaged areas had a size of 100 μm ×100 μm, with a spatial resolution of 390 nm. Both positive and negative ion spectral maps were obtained. A full peak list is shown in Table J.1; identification of the cellulosic fragments was based on the work of Fardim & Holmbom (2005), Sodhi et al. (2008) and Delandes et al. (1998). It is important to note that the secondary ion masses listed were variable by several mass units due to protonation or deprotonation of the fragments, a common occurrence in ToF-SIMS analysis of organic species (Lua et al. 2005).
Appendix Figure K.1.1. ToF-SIMS ion fragments analyzed (high spatial resolution). Negative fragments Component Source Chemical formula(e) Secondary ion masses (amu)
- PSS Ink (PSS) C8H7O3S 183
- C6H3O2S 139 2- C6H2O2S 138 PEDOT Ink (PEDOT) 3- C6HO2S 137 4- C6O2S 136
253
Positive fragments Component Source Chemical formula(e) Secondary ion masses (amu)
Base sheet (filler) Ti4+ 48 Metals Ink (surfactant) Na+ 23 Contaminants from handling Mg2+ 24
+ Ink (surfactant, defoamer) CnH(2n-1) (C=C) 27, 29, 43, 55, 69, 85, 83, 97, Hydrocarbons + Contaminants from handling CnH(2n+1) (C-C) 99, 113
+ C6H7O3 127 + Cellulose Base sheet (fibres) C6H9O4 145 + C6H9O5 161
254
APPENDIX L. Printed PEDOT:PSS/SWCNTs on paper: thickness and bulk resistance
All sheets were printed with 3 layers of ink at 25 µm drop spacing, which corresponded to a loading of 58 g/m2 of wet ink or 0.25 g/m2 of dried ink.
Commercial sheets
100
80
60
40
20 Printed layer (µm) thicknesslayer Printed
0 cellulose photo-paper cardstock inkjet paper high-yield glossy paper acetate paper
Appendix Figure L.1. Printed PEDOT:PSS/SWCNT layer thickness, commercial sheets.
25
20
) 15
10
5 Bulk resistance (k resistanceBulk
0 High-yield Inkjet Cardstock Glossy Photo- Cellulose paper paper paper paper acetate
Appendix Figure L.2. Printed PEDOT:PSS/SWCNT bulk resistance, commercial sheets.
255
Handsheets
70 SW fibres
60
50
40 unfurnished sizing 30 fixation agent sizing + fixation agent
20 Printed layer (µm) thicknesslayer Printed 10
0 0% 15% 30%
TiO2 content (wt%)
90 HW fibres 80
70
60
50 unfurnished sizing 40 fiaxtion agent 30 sizing + fixation agent
Printed layer (µm) thicknesslayer Printed 20
10
0 0% 15% 30%
TiO2 content (wt%)
Appendix Figure L.3. Printed PEDOT:PSS/SWCNT layer thickness, handsheets.
256
100000 SW fibres
10000
) 1000 unfurnished sizing 100 fixation agent
Bulk resistance (k resistanceBulk sizing + fixation agent
10
1 0 15 30
TiO2 content (wt%)
10000 HW fibres
1000
)
unfurnished 100 sizing fixation agent
Bulk resistance (k resistanceBulk sizing + fixation agent
10
1 0 15 30
TiO2 content (wt%)
Appendix Figure L.4. Printed PEDOT:PSS/SWCNT bulk resistance, handsheets.
257