Roadmap for optofluidics

Item Type Article

Authors Minzioni, Paolo; Osellame, Roberto; Sada, Cinzia; Zhao, S; Omenetto, F G; Gylfason, Kristinn B; Haraldsson, Tommy; Zhang, Yibo; Ozcan, Aydogan; Wax, Adam; Mugele, Frieder; Schmidt, Holger; Testa, Genni; Bernini, Romeo; Guck, Jochen; Liberale, Carlo; Berg-Sørensen, Kirstine; Chen, Jian; Pollnau, Markus; Xiong, Sha; Liu, Ai-Qun; Shiue, Chia-Chann; Fan, Shih-Kang; Erickson, David; Sinton, David

Citation Minzioni P, Osellame R, Sada C, Zhao S, Omenetto FG, et al. (2017) Roadmap for optofluidics. Journal of 19: 093003. Available: http://dx.doi.org/10.1088/2040-8986/aa783b.

Eprint version Post-print

DOI 10.1088/2040-8986/aa783b

Publisher IOP Publishing

Journal Journal of Optics

Rights This is an author-created, un-copyedited version of an article accepted for publication/published in Journal of Optics. IOP Publishing Ltd is not responsible for any errors or omissions in this version of the manuscript or any version derived from it. The Version of Record is available online at http:// doi.org/10.1088/2040-8986/aa783b

Download date 29/09/2021 05:10:27

Link to Item http://hdl.handle.net/10754/626071 Page 1 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 Roadmap on optofluidics 5 1, 2 3 4 4 5 6 Paolo Minzioni *, Roberto Osellame , Cinzia Sada , S. Zhao , F. G. Omenetto , Kristinn B. Gylfason , 5 6 6,7 8 9 10 7 Tommy Haraldsson , Yibo Zhang , Aydogan Ozcan , Adam Wax , Frieder Mugele , Holger Schmidt , 8 Genni Testa11, Romeo Bernini11, Jochen Guck12, Carlo Liberale13, Kirstine Berg-Sørensen14, Jian Chen15, 9 Markus Pollnau16, Sha Xiong17, Ai-Qun Liu17, Chia-Chann Shiue18, Shih-Kang Fan18, David Erickson19 and 10 David Sinton20 11 12 13 14 Affiliations 15 16 1Department of Electrical, Computer and Biomedical Engineering, University of Pavia, Via Ferrata 5A, 17 18 27100 Pavia, Italy

19 2 20 Istituto di Fotonica e Nanotecnologie - CNR, P.za Leonardo da Vinci 32, I-20133 Milano, Italy 21 3 22 Dipartimento di Fisica e Astronomia G. Galilei, Università di Padova, Via Marzolo 8, 35131, Padova, Italy 23 4 24 Department of Biomedical Engineering, Tufts University, Medford, Massachusetts 02155, USA 25 26 5Micro and Nanosystems, KTH Royal Institute of Technology, Osquldas väg 10, SE-100 44 Stockholm, 27 Sweden 28 29 6Electrical Engineering Department, Bioengineering Department, and California NanoSystems Institute 30 (CNSI), University of California, Los Angeles, California 90095, USA 31 32 7Department of Surgery, David Geffen School of Medicine, University of California, Los Angeles, 33 California 90095, USA 34 35 8 36 Department of Biomedical Engineering, Duke University, Durham, NC 27708, USA

37 9 38 Physics of Complex Fluids Group, MESA+ Institute, University of Twente, Enschede, The Netherlands 39 10 40 School of Engineering, University of California Santa Cruz, 1156 High Street, Santa Cruz, California 41 95064, USA 42 43 11Istituto per il Rilevamento Elettromagnetico de ll’Ambiente, National Research Council, Via Diocleziano 44 328, 81031 Naples, Italy 45 46 12Biotechnology Center, Technische Universität Dresden, Am Tatzberg 47-49, 01307 Dresden, Germany 47 48 13Biological and Environmental Science and Engineering Division, King Abdullah University of Science and 49 Technology (KAUST), Thuwal 23955-6900, Saudi Arabia 50 51 14 52 Department of Physics, Technical University of Denmark, DK-2800 Kongens Lyngby, Denmark 53 15 54 State Key Laboratory of Transducer Technology, Institute of Electronics, Chinese Academy of Sciences, 55 Beijing 100190, China 56 16 57 Department of Materials and Nano Physics, School of Information and Communication Technology, 58 KTH−Royal Institute of Technology, Electrum 229, Isafjordsgatan 22−24, 16440 Kista, Sweden 59 60 17School of Electrical and Electronic Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798

18Department of Mechanical Engineering, National Taiwan University, Taipei 10617, Taiwan AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 2 of 59

1 2 3 19Sibley School of Mechanical and Aerospace Engineering, Cornell Unversity, Ithaca, New York 14853, 4 USA 5 6 20 7 Department of Mechanical and Industrial Engineering, University of Toronto, 5 King’s College Road, 8 Toronto, Ontario M5S 3G8, Canada 9 10 11 12 *Guest editor of the roadmap 13 14 Email: [email protected] 15 16 17 18 19 Abstract 20 21 Optofluidics, nominally the research area where optics and fluidics merge, is a relatively new research field 22 23 and it is only in the last decade that there has been a large increase in the number of optofluidics applications 24 as well as in the number of research groups devoted to the topic. Nowadays optofluidics applications include, 25 without being limited to, lab-on-chip devices, fluid-based and controlled lenses, optical sensors for fluids and 26 for suspended particles, biosensors, imaging tools, etc. The long list of potential optofluidics applications, 27 28 which have been recently demonstrated, suggests that optofluidic technologies will become more and more 29 common in everyday life in the future, causing a significant impact on many aspects of our society. 30 31 A characteristic of this research field, deriving from both its inter-disciplinary origin and applications, is that 32 in order to develop suitable solutions it is often required to combine a deep knowledge in different fields, 33 ranging from to photonics, from to molecular biology and . As a 34 35 direct consequence, also being able to understand the long-term evolution of optofluidics research is not an 36 easy target. In this article we report several expert-contributions on different topics, so as to provide 37 guidance for young scientists. At the same time we hope that this document will also prove useful for 38 funding institutions and stake holders, to better understand the perspectives and opportunities offered by this 39 40 research field. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 3 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 5 6 Contents 7 8 9 10 1. Introduction 11 12 Materials and fabrication techniques 13 14 2. 3D optofluidic devices by femtosecond laser micromachining 15 3. Lithium niobate as an optofluidic platform 16 17 4. Silk fibroin films for biophotonic and microfluidic applications 18 5. Integration of microfluidics with silicon photonic sensors 19 20 Imaging 21 22 6. Holographic on-chip microscopy and tomography using lensless computational imaging 23 7. Live cell imaging with optofluidics 24 25 8. Optofluidic microlenses: from tunable focal length to aberration control 26 27 Strong light-fluid interaction 28 29 9. Integration of reconfigurable photonics and microfluidics 30 10. Optofluidic waveguides and resonators 31 32 Optics and acoustic forces for cell analysis 33 34 11. High-content physical phenotyping of biological objects with microfluidic dual-beam laser 35 36 traps 37 12. Integration of fiber-based tweezers and microfluidic systems 38 13. Acoustic prefocusing in optofluidic systems 39 40 DNA and molecular analysis 41 42 14. Optofluidics for single-cell protein analysis 43 15. Optofluidics for DNA analysis 44 45 16. Nanoscale optofluidics 46 17. Optofluidic immunoassays 47 48 Promising future applications 49 50 18. Optofluidics and point-of-need diagnostics for precision medicine and global health 51 19. Optofluidics in energy 52 53 54 55 56 57 58 59 60 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 4 of 59

1 2 3 researchers, scientists from other disciplines, and 4 1. Introduction – Paolo Minzioni optofluidics experts. 5 University of Pavia 6 Young researchers, starting their activity in 7 Optofluidics: an emerging and promising topic optofluidics could significantly benefit from this 8 The field of Optofluidics is a relatively new one in the article, as it represents a reference text, showing 9 possibilities and challenges of this specific research 10 scientific panorama. Although the idea of using fluids to control light, e.g. by spinning-mercury mirrors, field. Additionally the Roadmap also comes with a 11 th 12 dates back to the 18 century, no liquid-mirror- substantial bibliography making it easier to identify 13 telescope was practically realized before the end of the and access many helpful papers. th 14 20 century. In particular it is only during the last 15 The Roadmap can be useful as well for researchers 15 years that the term “optofluidics” has attracted a who have already developed a good expertise in a 16 significant attention, and that many groups have started contiguous field, such as chemical sensing, 17 devoting their research efforts to this field. microfluidic devices, high sensitivity molecule sensors, 18 19 The “young age” of this research field is demonstrated single-cell analysis, material-science and high- 20 by the fact that the first works indexed on the Web of precision diagnostics. For all of them optofluidics can 21 Science database and containing the word surely open new possibilities and scenarios, that this 22 “optofluidics” in the topic appears in 2005 [1,2]. Since Roadmap can help to unveil. 23 that moment the attention devoted to the realization of Finally, for experienced researchers in the optofluidic 24 systems exploiting the simultaneous control of fluidic field, the contributions included in this text by well- 25 conditions and optical beams roared. In the following 26 known experts allow having a clear reference point twelve years (2005-2016), the number of papers that about the current state of the art. The Roadmap offers 27 can be found using the same criteria increases from 0 28 them the possibility to get recent updates on the to 580. Recently about 70 new papers with the word 29 scientific activities and to analyze what other 30 “optofluidics” in the title are published every year, and experienced researchers see as future perspectives. 31 the number of citations paid to the “optofluidics” paper The aim of this document is that of showing not only 32 is continuously growing, almost linearly, and passed 33 from 20 in 2006 to about 1600 in 2016. the current technologies and applications but also showing promising approaches, still requiring research 34 Nowadays, what is generally included in the field of 35 and technological development. It is in any case 36 Optofluidics is, in the vast majority, based on the perfectly clear to the authors of this paper that what is 37 integration of optical technologies (for sensing, now perceived as a challenge may, in such a rapidly 38 actuation or imaging) within micro-fluidic (or nano- evolving field, already be achieved in one year, or even 39 fluidic) systems allowing the use of small samples less, or may not be considered any more as a relevant 40 volumes in a highly-controlled environment. As a target. Nevertheless, it is of great importance to define 41 result it is also observed that an increasing number of which research directions are now considered as the 42 short-schools for Ph.D. students are being scheduled emerging and promising, so that it will be possible in 43 by universities, and topical conferences expressly the future to re-assess the validity of this analysis. 44 dedicated to the field of Optofluidics are being 45 organized by important scientific associations (e.g. the Another fundamental point that this Roadmap wants to 46 EOS Conference on Optofluidics in Europe, the stress, thanks to the different contributions, is the 47 International Multidisciplinary Conference on strong connections linking applications, materials, and 48 methods. Exactly for this reason contributions on all 49 Optofluidics in Asia, etc.), so as to create meeting- points for researchers with different backgrounds and these aspects are included in the document. 50 Additionally, even if many topics are discussed in each 51 interests [3-5]. 52 section, the order of the different contributions has It is worth underlining that the field of Optofluidics is been selected to allow an easier sequential reading. 53 twice interdisciplinary: not only because of the 54 competences required for the design and realization of Current Status and Challenges 55 56 the devices, but also because of the countless This section wants to briefly provide young 57 possibilities offered by these devices in the fields of researchers, and all those needing some introduction to 58 chemical sensing, physical characterization and the basics of optofluidics, with a short list of the main 59 biomedical research. concepts useful to fully exploit the Roadmap content, 60 Intended Audience and Aim of the Roadmap and a list of suggested introductory readings. The intended audience of this document is constituted One of the basic ideas in optofluidics is that of taking by three different categories of possible readers: young advantage (at least in the majority of the cases) of the laminar-flow regime that is obtained as a consequence Page 5 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 of the typical dimensions and flow-speed used in smaller) obviously requires precision and attention. 4 Even more challenging is the collection of the output 5 microfluidic systems. The strong laminarity of these 6 systems is testified by their Reynolds number, defined samples in all those situations where sub-populations by the below equation, where ρ is the fluid density or cells (or even single cells) need to be extracted 7 3 3 8 (≈10 kg/m in case of room-temperature water), v is preserving their viability and without the risk of losing -5 -1 9 the fluid speed (generally in the range 10 – 10 m/s), them in microchip dead-volumes. D is the microchannel diameter (of the order of 10-4 m) 10 A second challenge is that of achieving high 11 and µ is the dynamic (which for water is performances and sensitivities while using low-cost, or 12 about 10 -3 Pa s): 13 disposable, systems. Such an achievement could allow the diffusion of optofluidic solutions in many 14 = 15 environments where lab equipment is not always 흆풗푫 available. 16 푹풆 17 If we compare the maximum Re value achievable with 흁 The third challenge that many microfluidic structures 18 the above reported numbers (10) with the reference 19 value of 3000, generally considered as a boundary are now trying to solve is that of creating easy-to-use 20 between laminar flow (Re < 2000) and turbulent flow devices. At the state-of-the-art microfluidic devices are 21 (Re > 4000), it is immediate to understand that unless generally tested on lab tables, requiring a lot of 22 turbulences are artificially created a laminar flow is additional instruments (laser sources, thermal 23 produced in these systems [6,7]. The flow laminarity, controllers, objectives, micropumps, cameras, etc.) and 24 which is a crucial point when well separated streams must be operated by experienced researchers. The 25 are needed, can become an issue when fluid mixing is integration of all these functionalities in a simple 26 required. In that case specific structures can be used to micro-system, would largely help the diffusion of these 27 break the system laminarity [8,9]. devices, making them user-friendly and more suitable 28 for point-of-care analysis. In particular a considerable 29 Another fundamental element in the optofluidics field attention has been paid in the last five years to the 30 is the good size-matching between biological objects development of compact and efficient imaging 31 (bacteria, cells, or organelles), microfluidic channels, systems, which will also be discussed in the following. 32 and the diameter of optical beams in the visible or 33 near-infrared spectral range. All these quantities are Acknowledgments – I want to thank Professor Ilaria 34 generally in the range between 0.1 µm and 500 µm, 35 Cristiani for her precious suggestions and Dr Jarlath and, while the size of biological objects is generally a 36 McKenna for the great help in organizing and given data, it is often possible to tailor the 37 coordinating the article. 38 microchannel width and optical beams diameter in order to obtain the desired values and fluidic/optical 39 40 effects [10]. References 41 Additionally, the forces that can be applied to cells by [1] C. Grillet, P. Domachuck, B. Eggleton, J. Cooper- 42 means of optical beams, to trap, push or deform them, White, “Optofluidics enables compact tunable 43 44 well match those generally involved in biological interferometer”, Laser Focus World 41 (2), 2005 45 processes [11-17]. [2] P. Domachuk, M. Cronin-Golomb, B.J. Eggleton, S. Mutzenich, G. Rosengarten, A. Mitchell, 46 Finally, a basic idea in the optofluidic field is that 47 “Application of optical trapping to beam thanks to the different response exhibited by the manipulation in optofluidics”, Optics Express 13 48 samples under analysis to optical or acoustic forces it is 49 (19), 7265-7275, 2005 also possible to use different actuation mechanisms as 50 [3] http://www.nanotech.dtu.dk/uddannelse/polynano- 51 “sensing mechanisms”, enabling for interesting summerschool 52 selection and sorting strategies [18-22]. [4] http://www.myeos.org/events/wpc2017/eosof 53 Some Trends & Challenges [5] http://www.optofluidics.sg/ 54 [6] O. Reynolds, “An experimental investigation of the 55 Without going now into the specific solutions that can circumstances which determine whether the motion 56 be envisaged, there are three main challenges that of water shall be direct or sinuous, and of the law of 57 future optofluidic devices should try to solve. resistance in parallel channels”, Philosophical 58 Transactions of the Royal Society. 174 935–982, 59 The first one is dealing with the sample insertion and 1883 60 collection from a microfluidic system. Although the [7] H. Bruus “Acoustofluidics 1: Governing equations use of extremely reduced sample volumes brings in microfluidics” Lab Chip 11, 3742-3751, 2011 several advantages it is nevertheless evident that properly managing volumes in the µL range (or AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 6 of 59

1 2 3 [8] C.Y. Lee, W.T. Wang, C.C. Liu, Lung-Ming Fu, 4 “Passive mixers in microfluidic systems: A review” 5 Chemical Engineering Journal 288, 146–160, 2016 6 [9] http://www.elveflow.com/microfluidic- 7 8 tutorials/microfluidic-reviews-and- 9 tutorials/microfluidic-mixers-short-review/ 10 [10] A.M. Smith, M.C. Mancini, S. Nie, “Bioimaging: 11 Second window for in vivo imaging” Nature 12 Nanotechnology 4 (11), 710-711, 2009. 13 [11] A.Ashkin, “Acceleration and trapping of particles by 14 radiation pressure” Phys. Rev. Lett. 24, 156–158 15 1970 16 [12] A. Ashkin, J.M. Dziezic, J.E. Bjorkholm, S. Chu, 17 “Observation of a single beam gradient force optical 18 trap for dielectric particles” Opt. Lett. 11, 288-290 19 1986 20 [13] A. Salandrino, S. Fardad, D.N. Christodoulides, 21 "Generalized Mie theory of optical forces," J. Opt. 22 Soc. Am. B 29, 855-866, 2012 23 [14] L. Ferrara, E. Baldini, P. Minzioni, F. Bragheri, C. 24 Liberale, E. Di Fabrizio, I. Cristiani “Experimental 25 study of the optical forces exerted by a Gaussian 26 beam within the Rayleigh range”, Journal of Optics, 27 13, 7, 075712, 2011 28 [15] M.J. Padgett, J. Molloy, D. McGloin Optical 29 Tweezers: Methods and Applications, CRC Press, 30 2010 31 [16] T. Yang, F. Bragheri, P. Minzioni “A 32 comprehensive review of optical stretcher for cell 33 mechanical characterization at single-cell level”, 34 MDPI Micromachines 7, 5, 2016. 35 [17] J. Guck, R. Ananthakrishnan, H. Mahmood, T.J. 36 Moon, C.C. Cunningham, J. Käs, “The Optical 37 38 Stretcher: A Novel Laser Tool to Micromanipulate 39 Cells” Biphysical Journal, 81 (2), 767–784, 2001 40 [18] H. Bruus “Acoustofluidics 2: Perturbation theory 41 and ultrasound resonance modes”, Lab Chip, 12, 20- 42 28, 2012 43 [19] J. Dual, T. Schwarz, “Acoustofluidics 3: Continuum 44 mechanics for ultrasonic particle manipulation” Lab 45 Chip 12, 244-252, 2012 46 [20] T. Yang, F. Bragheri, G. Nava, I. Chiodi, C. 47 Mondello, R. Osellame, K. Berg-Sørensen, I. 48 Cristiani, P. Minzioni “A comprehensive strategy 49 for the analysis of acoustic compressibility and 50 optical deformability on single cells” Scientific 51 Reports 6, 23946, 2016 52 [21] J. Shi, X. Mao, D. Ahmed, A. Colletti, T.J. Huang 53 “Focusing microparticles in a microfluidic channel 54 with standing surface acoustic waves (SSAW)” Lab 55 Chip 8, 221-223, 2008 56 [22] L.Y. Yeo, J.R. Friend, “Surface Acoustic Wave 57 Microfluidics”, Annual Review of Fluid Mechanics 58 Vol.46:379-406, Annual Reviews (2014) 59 60 Page 7 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 2. 3D optofluidic devices by femtosecond 5 laser micromachining – Roberto Osellame 6 Istituto di Fotonica e Nanotecnologie - CNR 7 8 Relevance of the topic 9 10 Femtosecond laser micromachining (FLM) of 11 transparent materials [1, 2] is a recent technique that 12 found wide application in optofluidic device 13 fabrication. FLM is based on a nonlinear absorption 14 process, selectively triggered in the focal volume by 15 ultrashort and tightly focused laser pulses (Fig. 1). 16 Tailoring of the irradiation parameters can produce 17 very different results in . In particular, one can 18 achieve a gentle and localized change of the refractive 19 20 index or a self-assembled nanostructuring of the material. The first modification is extremely relevant Figure 1. Femtosecond laser micromachining can create 21 optofluidic devices, encompassing optical waveguides and 22 as it allows drawing optical waveguides by suitably microfluidic channels, directly buried inside a glass sample. 23 moving the glass substrate with respect to the laser focus. The second modification is also important for 24 Current Status and Challenges 25 optofluidic applications since the nanostructured region 26 has an enhanced selectivity to wet chemical etching FLM has been widely exploited in the fabrication of 27 (e.g., acqueous solutions of hydrofluoric acid). This optofluidic devices with the main aim of integrating 28 means that an irradiated volume in this second regime optical detection and microfluidic handling of samples 29 will be preferentially etched away with respect to the [2 ]. The use of optical waveguides to deliver the 30 unirradiated material, thus creating cavities and probing light and to collect the output signal enables

31 microfluidic networks inside the glass. higher sensitivity, reproducible and alignment-free 32 measurements, and multipoint sensing. Taking 33 The first relevant advantage of this microfabrication advantage of the 3D capabilities of the technology, it 34 technology for optofluidics is the capability to produce, has been possible either to add specific components to 35 with the same tool, both optical waveguides and already existing devices or to fabricate the complete 36 microfluidic channels in glass. This puts the microsystem by FLM. An example of the former case 37 technology in a very favorable condition to combine 38 consists in adding optical waveguides to commercial these two elements in the same substrate and produce 39 microfluidic lab-on-chips to perform on-chip compact and complex optofluidic devices. The second 40 fluo rescence or label-free detection of relevant advantage of this technology is its unique capability to 41 biomolecules [2]. A different example of how FLM produce 3D structures taking advantage of the fact that 42 can be exploited to add specific components to already the induced modification is highly localized to the 43 existing devices is the concept of Lab-in-Fiber [5]. In focal volume and can thus be arbitrarily placed 44 this case, the starting device is an , where 45 anywhere in the glass volume. A third advantage is the the cladding is typically underused. In this approach, 46 possibility to combine very different fabrication FLM is exploited to add microchannels, cavities or 47 techniques with a single tool, from material property additional optical waveguides to produce highly 48 tuning (e.g. in waveguide writing), to selective material functionalized optical fibers that become self- 49 removal (e.g. for microchannel fabrication) and consistent and sophisticated optofluidic sensors (Fig. 50 additive manufacturing by two-photon polymerization 51 2a). The possibility to use FLM to fabricate the whole (e.g. for the creation of plastic microstructures 52 device further widens the design possibilities and fully embedded in the microfluidic channels [3,4]). A final 53 unleashes the 3D capabilities of the technology. 3D important advantage of FLM is the rapid prototyping 54 microfluidic layouts have been exploited to achieve capability; as any maskless and direct-write technique, 55 full-hydrodynamic-focusing and, combined with laser a software design can be transferred forthwith onto a 56 written optical waveguides, enabled the demonstration 57 working prototype, with the possibility to simply and 3 of a 1 mm cell counter, capable of counting up to 58 rapidly adjust the design or to produce highly 5000 cells/s [6]. Cell manipulation is another relevant 59 customized products. 60 application of FLM-produced devices (Fig. 2b). In fact, Further advances in this technology will provide an optical forces, exerted by the light coming from optical unprecedented level of integration of functionalities in waveguides, can be exploited to trap and deform cells a single microsystem, with very short time-to-market flowing in a microchannel, thus characterizing their of new products. mechanical response [7]. This analysis can AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 8 of 59

1 2 3 discriminate between different cells populations (e.g., cylindrical lens has been combined with a suitably 4 5 between healthy and cancer cells) without any marker shaped microchannel, allowing optical sectioning and 6 (see Section 11 for more details). Depending on the 3D reconstruction of cellular spheroids in suspension. 7 assay result, each cell can then be sorted in different The biological sample is driven through the light sheet, 8 output channels, again exploiting optical forces created by the cylindrical lens, with a uniform 9 produced by a dedicated waveguide. The combination microfluidic flow. of 3D microfluidics and optical waveguides has been 10 As already mentioned, additional functionalities can be 11 also exploited to produce a 3D-shaped nanoacquarium, integrated in the optofluidic device by two-photon 12 to study the behavior of cyanobacteria in controlled polymerization (2PP). In particular, the 3D capability 13 conditions [8]. In particular, optical waveguides of FLM enables the 2PP-fabrication of plastic 14 embedded in the chip were used to accurately monitor structures directly inside sealed microfluidic channels 15 the CO concentration in the nanoacquarium. Another 2 [3], with an approach that has been dubbed ‘ship in a 16 3D optofluidic device that allows sophisticated 17 bottle’ [4]. As an example, arrays of diffractive and fluorescence imaging of biological samples is the 18 refractive microlenses have been produced inside recent demonstration of selective plane illumination 19 FLM-fabricated microfluidic channels to achieve microscopy (SPIM) on-chip [9]. An optofluidic 20 white-light detection and counting of cells flowing in 21 the microchannel [4]. 22 23 A few technological challenges are still open to fully 24 exploit this technology. The first one is FLM 25 throughput; in fact, being a serial writing technique, it 26 requires the time to irradiate each device and does not 27 benefit from economy of scale. A second challenge is 28 the possibility to process different materials, other than 29 fused silica and Foturan glass. The third challenge is 30 further miniaturization to the nanoscale. 31 32 Advances in Science and Technology to Meet 33 Challenges 34 35 The challenges previously detailed have triggered a 36 significant effort in the community and, although still 37 open, there are already encouraging results towards 38 their solution. Regarding the scaling of fabrication 39 capability, fine optimization of the processing window 40 has already allowed writing velocities in the order of 41 cm/s for several processes. In addition, the use of 42 spatial light modulators has introduced the possibility 43 of multifoci writing, thus enabling the production of 44 45 several structures in parallel. These approaches have 46 still to be exploited fully to understand the ultimate 47 limits, but FLM will likely be competitive with 48 standard microfabrication technologies also for large 49 production batches and not just for a few prototypes. 50 Fused silica or Foturan are excellent substrates 51 for many optofluidic applications and in particular for 52 biophotonic ones. However, enlarging the portfolio of 53 materials amenable to FLM will be beneficial to target 54 new optofluidic applications. A better understanding of 55 Figure 2. Two examples of optofluidic devices fabricated by femtosecond laser micromachining. (a) Lab-in-fiber: Several the nanostructuring process, at the base of the 56 optical and fluidic components can be integrated in the microchannel formation, and its replication in other 57 cladding of a fiber to perform optofluidic sensing; SMF: single materials is an on-going activity that still requires some 58 mode fiber; FBG: fiber Bragg grating; FPI: Fabry-Perot 59 interferometer; TIR: total internal reflection. Reproduced from effort. Finally, further miniaturization to the nanoscale 60 [5] with permission of The Royal Society of Chemistry. (b) could be the next big evolution of the technology. The Optical cell stretcher: optical forces exerted by light delivered 2PP technique is already able to produce nanoscale by two optical waveguides allow characterizing the mechanical structures and recently nanofluidic channels have been properties of flowing cells [7]. Reproduced with permission © demonstrated by FLM [8]. This fabrication scale is still 2010 Optical Society of America.

Page 9 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 largely unexplored, however FLM has the unique [10] Hasegawa S, Ito H, Toyoda H and Hayasaki Y 2016 4 Massively parallel femtosecond laser processing, 5 capability of easily connecting nanoscale structures to Opt. Express 24 18513–18524 6 microscale ones, thus potentially improving the

7 usability of new devices at this very small scale. 8 Concluding Remarks and Perspectives 9 10 Femtosecond laser micromachining is a very powerful 11 technique for the fabrication of optofluidic devices, 12 allowing unique 3D layouts and an amazing level of 13 integration of different functionalities. Its versatility 14 opens the way to many different applications both for 15 research and for industry. In particular, in those fields 16 where standardization is not too rigid, the design 17 freedom allowed by FLM can produce revolutionary 18 19 microsystems that will set a new standard in the field. 20 References 21 22 [1] Gattass RR and Mazur E 2008 Femtosecond laser 23 micromachining in transparent materials Nature 24 Photonics 2 219–225 25 [2] Osellame R, Hoekstra HJWM, Cerullo G and 26 Pollnau M 2011 Femtosecond laser 27 microstructuring: an enabling tool for optofluidic 28 lab-on-chips Laser Photonics Rev. 5 442–463 29 [3] Amato L, Gu Y, Bellini N, Eaton SM, Cerullo G 30 and Osellame R 2012 Integrated three-dimensional 31 filter separates nanoscale from microscale elements 32 in a microfluidic chip Lab Chip 12 1135–1142 33 [4] Wu D, Xu J, Niu L-G, Wu S-Z, Midorikawa K and 34 Sugioka K 2015 In-channel integration of 35 designable microoptical devices using flat scaffold- 36 supported femtosecond-laser microfabrication for 37 coupling-free optofluidic cell counting Light: 38 Science & Applications 4 e228 39 [5] Haque M, Lee KKC, Ho S, Fernandes LA and 40 Herman PR 2014 Chemical-assisted femtosecond 41 laser writing of lab-in-fibers Lab Chip 14 3817– 42 3829 43 [6] Paiè P, Bragheri F, Martinez Vazquez R and 44 Osellame R 2014 Straightforward 3D hydrodynamic 45 focusing in femtosecond laser fabricated 46 microfluidic channels Lab Chip 14 1826–1833 47 48 [7] Bellini N, Vishnubhatla KC, Bragheri F, Ferrara L, 49 Minzioni P, Ramponi R, Cristiani I and Osellame R 50 2010 Femtosecond laser fabricated monolithic chip 51 for optical trapping and stretching of single cells 52 Opt. Express 18 4679–4688 53 [8] Sugioka K, Xu J, Wu D, Hanada Y, Wang Z, 54 Cheng Y and Midorikawa K 2014 Femtosecond 55 laser 3D micromachining: a powerful tool for the 56 fabrication of microfluidic, optofluidic, and 57 electrofluidic devices based on glass Lab Chip 14 58 3447–3458 59 [9] Paiè P, Bragheri F, Bassi A and Osellame R 2016 60 Selective plane illumination microscopy on a chip Lab Chip 16 1556–1560 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 10 of 59

1 2 3 4 3. Lithium niobate as an optofluidic 5 platform – Cinzia Sada 6 University of Padova, Italy 7 8 Relevance of the topic 9 10 Droplet-microfluidics technology has paved the way to 11 new approaches of chemical and biological analysis, 12 exploiting parallel processing and high throughput 13 response. Screening, rapid droplet sorting and 14 biochemical micro-reactors were in fact demonstrated 15 as well as bio-analytical assays, enzyme kinetics, cell 16 analysis, encapsulation and sorting. Thanks to the 17 exploitation of reduced volumes down to nano-litres to 18 femto-litres, as well as reaction times shortened to 19 20 seconds or less, chemical synthesis and complex 21 particle fabrication up to drug delivery and diagnostic 22 testing and biosensing were also reported [1]. Although 23 novel micro-fabrication techniques are continuously 24 being developed to improve the final performances, 25 lower the devices costs and differentiate functionalities 26 with appropriate fluidic interfacing scheme, fully Figure 1 – Lithium niobate (LiNbO3) main properties. 27 integrated opto-microfluidics devices are far from 28 being widely commercially available yet. The In lab-on-chip systems the integration of many stages 29 perspective of combining the optical tools versatility is a key point in portable compact devices. 30 with the potentiality of microfluidics has gained Consequently, flexible and easily-moulded materials 31 increasing interest. As already foreseen by Psaltis and (mainly polymers such as PDMS, PMMA) and 32 biocompatible substrates such as glass and silica have 33 co-workers in their pioneering work published in Nature on 2006, novel enabling technologies leading to usually been used leading to commercially available 34 disposable microfluidic systems. Opto-fluidics 35 true Lab-on-chip system have started to be under 36 debate where chemical, physical and biophysical platforms were therefore achieved by coupling both 37 sensors [2] could be integrated on multi-purpose laser beams (including, in best cases, fibres coupling) 38 portable devices. To this aim, several challenges and detection stages (often carried out by optical 39 should be faced both from a scientific and a microscopy) to microfluidics. To this direction, 40 technological point of view. An opto-microfluidic prototypes able to detect, count, sort and analyse the 41 portable device should in fact guarantee a trustworthy droplets have therefore been investigated such as: 42 response, i.e., a reproducible and repeatable output compact micro-flow-cytometers (typical throughput of 43 that, minimizing false responses and artefacts, is easy the order of several hundreds of cells per second and a 44 to read/store and fast to calibrate. Finally, to be really typical sample purity about 90%), velocity detectors 45 [4], biosensors based on fluorescence measures [5] just 46 competitive with full-optional bulky-lab apparatuses, it should be compact, low cost, easy to use and handle, to cite a few. In these cases, frequently polymers were 47 bonded to rigid materials: silicon, to provide for 48 and stable under different environmental conditions. At 49 first sight, all these features seem difficult to match. patterned substrate to anchor/align fibres to the 50 The demand of reducing costs has therefore prevailed, microfluidic circuitries; silica or glasses to provide for 51 leading towards the spread of disposable devices and a solid base; lithium niobate (LN) to add Surface 52 putting in second priority the possibility of checking, Acoustic Wave (SAW) stages working as micro-pumps 53 monitoring and exploiting the device response in a or mixers respectively. Quite surprisingly, only in the 54 longer time scale. However in the last decade, draining last decade significant advances were achieved by 55 the expertise in photonics, integrated optics and delivering monolithic, perfectly aligned, robust and 56 microelectronics MEMS processing into microfluidics, portable opto-fluidic prototypes. Both microfluidic 57 have conveyed advances to overcome and bypass these circuits and optical waveguides were synergistically 58 limitations by the suitable combination of materials, fabricated on the same substrate (fused silica) by 59 femtosecond laser micromachining (FLM) [6]. 60 preparation techniques and smart approaches respectively [3]. Optically controlled photonic prototypes to manipulate micron-scale dielectric objects by light-driven phenomena were therefore presented: optical tweezers Current Status and Challenges and stretchers were implemented with suitable focused Page 11 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 laser beams, either externally or fibre-coupled, at a 4 5 microfluidics level [7]. Similar results can be achieved 6 also in LN crystals substrates but are currently 7 underestimated (Fig.1). Thanks to its excellent optical 8 and electro-optical properties, in fact, high quality 9 optical waveguides, multiplexers, switches and optical 10 modulators are commercially available, therein 11 obtained mainly by Ti-in-diffusion technology. Its 12 ferroelectric property, instead, was exploited to 13 produce light sources by way of frequency conversion 14 in Periodically Poled LN structures. Other features, 15 instead, were less profit by: biocompatibility, 16 photorefractivity for holographic recording and 17 18 induced local space-charge-fields to exploit 19 dielectrophoretic and electrophoretic forces (either 20 generated by inhomogeneous illumination [8] or 21 heating [9]) as well as micromachining by high optical 22 quality dicing [10]. This material therefore presents all 23 the needed characteristics to host several different 24 functionalities on the same substrate in a monolithic

25 integrated opto-fluidic multi-functional platform. This Figure 2 – Lithium niobate based Opto-fluidic platform. 26 high level of integration represents a great challenge 27 but is required in portable Lab-on-chip systems that 28 LiNbO3–LiNbO3 bonding to get 3D configurations, could operate in a wide spectrum range, with tailored still almost unexplored. Despite of these technological 29 final response and real-time reconfigurability. Limiting 30 challenges, the real advance in science is factors mainly stem from either the tools to finely 31 straightforward: droplet counting, sorting depending on 32 control the fluids flux and the optimization of the light the size or content, droplet routing and storage all- 33 coupling and detection within the microfluidics integrated on the same substrate with micro-pumps 34 interface. Up to now expensive and bulky systems such fluid driving can be foreseen. This is possible by 35 as syringe or pressure based pumps are still used whilst implementing droplets optical transmission 36 optical functionalities are often granted by using high measurements and passage monitoring as well as 37 power or optical bench working lasers respectively. refractive index and/or fluorescence emission detection 38 When optical fibres are integrated, instead, optical (eventually spectral resolved by recording holographic 39 losses can represent a limiting factor for enhanced 40 gratings). Finally, waveguides can both confine light sensibilities. Hybrid configurations of polymers/ and allow for droplet routing and sorting, while storage 41 durable materials were therefore proposed to partially 42 can be achieved (Fig.2). This challenge can be faced overcome this issue. Less investigated, instead, were 43 only by exploiting light-induced phenomena in the the alternative solution of monolithically integration of 44 material, something possible in lithium niobate by way 45 several functionality on the same substrate. Both these of its photorefractive and photovoltaic properties, even 46 aspects somehow have hindered the full exploitation of enhanced by suitable local doping. Consequently 47 opto-microfluidics benefits and potentialities so far. tailored and reconfigurable set-ups could be designed 48 even in real-time by way of feedback systems triggered 49 Advances in Science and Technology to Meet by the optical waveguides response as well. 50 Challenges 51 52 In order to get a true portable device some issues must Concluding Remarks and Perspectives 53 be solved from a technological point of view. In Light-reconfigurable devices with multifunctional 54 integrated opto-microfluidics platforms, embedded or stages have not been proposed yet on the same 55 buried optical waveguides are in fact needed. Up to substrate, as well as systems that activate a function in 56 now only the FLM technique allowed one to reach real time depending on the other stages feedback. 57 such a result, but in fused silica and in a defined range 58 Together with the miniaturization advantages, a fully of possible functionalities limited to this materials integrated opto-fluidic platform promises to provide 59 properties. Although some attempts were made in 60 higher sensitivity and dynamic range, a finer control of LiNbO3, FLM is still far from being truly exploited. As the light power focused on the biological target and a an alternative, LN could be also micromachined and dynamical reconfigurability tailored depending on the bonded to silica and to polymers, allowing hybrid final purpose. Integrated optics can provide a configurations as well. Less hunted, instead, seems the perspective and a systematic analysis on the role of AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 12 of 59

1 2 3 different conditions of illumination and the relative [9] Ferraro P, Coppola S, Grilli S, PAturzo M, Vespini 4 V 2010 Dispensing nano–pico droplets and liquid 5 real-time monitoring even of cumulative light induced patterning by pyro-electrodynamic shooting Nature 6 effects still unknown or not investigated. Lithium Nanophotonics 5 429-435 7 niobate promises to be a valid alternative, as it was 8 demonstrated to host: efficient integrated optics stages; [10] Chauvet M, Al Fares L, Guichardaz B, Devaux F, 9 micro-pumps by Surface Acoustic Waves (SAW); Ballandras S 2012 Integrated optofluidic index 10 SHG light generation by PPLN; particle manipulation sensor based on self-trapped beams in LiNbO3 11 stages and droplet routing by way of photovoltaic Letters 101 (18) 181104-6 12 properties; engraved microfluidic circuitries by Femto- 1. 13 laser technique and high quality dicing. Consequently 14 it represents an ideal gym-floor to integrate all these 2. 15 functionalities together and achieve a portable full 16 optional Lab-on-chip device, tailored and 17 18 reconfigurable by way of optical induced phenomena. 19 20 Acknowledgments – The author kindly acknowledge 21 G. Bettella, G. Pozza, C. Montevecchi, E. Baggio, R. 22 Zamboni and A. Zaltron, co-workers at University 23 Padova. Fondazione Ca.Ri.Pa.Ro and the University of 24 25 Padova are acknowledged for funding the Excellence 26 Project-2012 and PRAT-2016 project respectively. 27 28 29 References 30 [1] Seeman R, Brinkmann M, Pfohl T, Herminghaus 31 S 2012 Rep. Prog. Phys. 75 016601-42 32 [2] Matos Pires NM, Dong T, Hanke U, Hoivik N 2014 33 Recent Developments in Optical Detection 34 Technologies in Lab-on-a-Chip Devices for 35 Biosensing Applications Sensors 14(8) 15458- 36 15479 37 [3] Pang L, Chen MH, Freeman LM, Fainman Y 2012 38 Optofluidic devices and applications in photonics, 39 sensing and imaging Lab on a Chip 12(19):3543-51 40 [4] Hsieh YW, Wang AB, Lu XY, Wang LA 2016 41 High-throughput online multidetection for refractive 42 index, velocity, size and concentration 43 measurements of micro-two-phase flow using 44 optical microfibers Sensors and Actuators B 237 45 841-848 46 47 [5] Testa G, Persichetti G, Sarro PM, Bernini R 2014 A 48 Hybrid silicon-PDMS optofluidic Platform for 49 sensing application Biomedical Optics Express 14 50 (2) 417-426 51 [6] Osellame R, Hoekstra HJWM, Cerullo G, Pollnau 52 M 2011 Femtosecond laser microstructuring: an 53 enabling tool for optofluidic lab-on-chips Laser & 54 Photon Rev 5 442–463 55 [7] Bragheri F, Ferrara L, Bellini N, Vishnubhatla KC, 56 Minzioni P, Ramponi R, Osellame R, Cristiani C 57 2010 Optofluidic chip for single cell trapping and 58 stretching fabricated by a femtosecond laser Journal 59 of Biophotonics 3 (4) 234-243 60 [8] Esseling M. Zaltron A, Horn W, Denz C, Optofluidic droplet router 2015 Laser and Photinics reviews 9 (1) 98-104 Page 13 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 4. Silk fibroin films for biophotonic and 5 microfluidic applications − S. Zhao, F. G. 6 Omenetto 7 Tufts University 8 9 Relevance of the topic 10 11 Microfluidics is a versatile technological platform for 12 precise manipulation of fluids of extremely small 13 volumes and has a broad range of biochemical and 14 medical applications. Polydimethylsiloxane (PDMS), 15 as one of the most commonly used microfluidic 16 materials, has revolutionized the field by enabling out- 17 of-cleanroom and inexpensive fabrication and 18 providing robust material properties to support a 19 20 variety of applications from large-scale chemical 21 synthesis to high throughput bimolecular assays. 22 However, the non-biofriendly fabrication process and 23 the limitation on bulk functionalization have limited 24 the utility of PDMS devices in the fields of tissue

25 engineering and regenerative medicine. More recently, Figure 1. (A) A silk hydrogel microfluidic device with 3D 26 biologically relevant materials have been utilized to microchannels. (B) Human endothelial cells cultured along 27 construct microfluidic devices, such as gelatin, microchannel surface. (C) Human fibroblast cells cultured in 28 collagen, alginate and silk. These natural biomaterials the bulk of silk hydrogel device. Scale bar shows 1cm in (A) 29 and 20µm in (B) and (C). Reprinted with permission from allow easy bulk and surface functionalization and Elsevier [2]. 30 exhibit fully compatible bio-interfaces. Among those 31 materials, silk fibroin has unique physicochemical been implemented thanks to the elastic nature of the 32 silk hydrogel, which would allow automatic control of 33 properties including highly tunable mechanical stiffness and in vivo degradability as well as long term silk microfluidics for high throughput large scale 34 devices. Human endothelial cells and fibroblast cells 35 stability in various medium environments. These 36 attributes are essentially important for advanced were successfully cultured along the channel surface 37 applications of microfluidics in tissue-related research and in the bulk of the device, respectively, which 38 and in vivo implantation. demonstrated the utility of the silk hydrogel devices in 39 potential tissue engineering applications. Current Status and Challenges 40 Silk microfluidics has made tremendous progress in 41 The first microfluidic device completely made of silk the last decade, but challenges still remain to be 42 fibroin material was introduced in 2007 [1]. The 43 addressed to bring this platform from research labs to microchannels were fabricated by aqueous casting of hospitals and clinical market. One of the key 44 silk fibroin solution on PDMS mold followed by 45 challenges is that the current 3D silk microfluidic water-stable treatment. The micromolded silk film was 46 fabrication requires time-consuming layer-by-layer 47 then laminated onto a flat silk film so closed assembly and significant human intervention, which 48 microchannels were formed. Human hepatocytes were largely limits the throughput. Ideally, complex and 49 successfully cultured in the microchannels and multilayer silk microfluidics will be fabricated with 50 perfused with cell culture medium for a prolonged automatic systems using either bottom-up (e.g. 3D 51 period of time to demonstrate the biocompatibility of printing) or top-down (e.g. 3D laser micromachining) 52 the devices. A new enzymatically crosslinked silk approaches. Moreover, fabricating ultrafine features 53 hydrogel material was later introduced and has also (from micrometer to nanometer) within the silk 54 been utilized for microfluidics fabrication (Figure 1). hydrogel is very challenging, which if addressed could 55 This material provides better tunability on mechanical 56 allow the construction of high density microvascular properties to match different human tissues, bulk networks and highly integrated microfluidic devices. 57 functionalization and cell encapsulation for artificial 58 tissue/organ constructions [2]. A gelatin sacrificial Advances in Science and Technology to Meet 59 Challenges 60 molding method was used to create microchannels in silk hydrogel and a layer-by-layer assembly approach 3D printing is a promising technology that allows allowed the fabrication of multilayered 3D facile fabrication of complex 3D structures with microfluidic devices. Pneumatic fluidic controls minimal human intervention. We have recently previously used in the PDMS counterparts have also AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 14 of 59

1 2 3 developed a 3D bio-printer that is able to print room- Proceedings of the National Academy of Sciences of 4 the United States of America, 112(39), 12052-12057 5 temperature curable silk hydrogels into various 2D and [5] Applegate MA, Partlow B, Coburn J, Marelli B, 6 3D formats [3]. Ultrafast laser pulses of relatively low Pirie, Pineda R, Kaplan DL, and Omenetto FG, 7 energy have also been utilized to directly carve high 8 resolution channels of micrometer dimension within 2016, Photocrosslinking of Silk Fibroin Using 9 bulk silk hydrogels, which provides a promising Riboflavin for Ocular Prostheses. Advanced 10 solution to automatic fabrication of high density silk Materials, 28(12), 2417-2420 11 microfluidics [4]. In addition to enzyme-induced [6] Kim S, Marelli B, Brenckle MA, Mitropoulos AN, 12 crosslinking, light and electron beams have also been Gil ES, Tsioris K, Tao H, Kaplan DL, and Omenetto 13 utilized to cure and pattern silk hydrogel materials FG, 2014, All-water-based electron-beam lithography using silk as a resist, Nature 14 under aqueous and bio-friendly conditions, which Nanotechnology, 9(4): p. 306-310 15 could be potentially used to fabricate features of 16 micrometer to nanometer resolution [5,6]. 17 18 19 20 Concluding Remarks and Perspectives 21 The next evolution of the field of microfluidics will be 22 based on the integration of new classes of active 23 materials as the bulk constituent of the devices. That is, 24 the availability of materials that are biologically active 25 26 and that can be integrated into living tissue. Together 27 with the possibility of incorporating dopant molecules, 28 which can surround the volumes of fluid flowing 29 through microchannels, with active environments that 30 can define diffusive profiles into or out of the 31 channnels or can provide externally addressable 32 functions. Future applications envisioned are 33 biologically integrated fluidic devices as smart tissue 34 engineering platforms, active bulk doped microfluidics 35 that embed optical and/or electronic function, and 36 dynamically reconfigurable, chemically active devices, 37 extending the utility of the platform. 38 39 40 Acknowledgments 41 The authors would like to acknowledge support from 42 the ONR, AFOSR, and NIH. 43 44 45 46 References 47 [1] Bettinger CJ, Cyr KM, Matsumoto A, Langer R, 48 Borenstein JT and Kaplan DL, 2007, Fibroin 49 Microfluidic Devices, Advanced Materials, 19(5): 50 2847–2850 51 [2] Zhao S, Chen Y, Partlow BP, Golding AS, Tseng P, 52 Coburn J, Applegate MB, Moreau JE, Omenetto FG, 53 Kaplan DL. 2016, Bio-functionalized silk hydrogel 54 microfluidic systems. Biomaterials, 93:60-70 55 [3] Jose RR, Brown JE, Polido L, Omenetto FG, Kaplan 56 DL, 2015, Polyol-Silk Bioink Formulations as Two- 57 Part Room-Temperature Curable Materials for 3D 58 Printing, ACS Biomater. Sci. Eng., 1 (9), 780–788 59 [4] Applegate MA, Coburn J, Partlow BP, Moreau JE, 60 Mondia JP, Marelli B, Kaplan DL, and Omenetto FG, 2015, Laser-based three-dimensional multiscale micropatterning of biocompatible hydrogels for customized tissue engineering scaffolds, Page 15 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 detected molecule as the analyte, and the technique as 4 5. Integration of microfluidics with silicon a whole as optical biosensing. Since the transparency 5 photonic sensors – Kristinn B. Gylfason and 6 Tommy Haraldsson window of silicon lies in the near- and mid-infrared, 7 KTH Royal Institute of Technology detection schemes based on refractive index changes 8 upon biomolecule binding are most commonly applied. 9 Relevance of the topic Fluorescence based detection is uncommon, since very 10 few fluorophores operate in the near- and mid-infrared. 11 With our rapidly advancing understanding of 12 molecular biology, the interest in analysing the The main benefit of silicon photonics is that by 13 molecular composition of various biological samples leveraging the investments made by the silicon-based 14 has increased dramatically in recent years. Medical microelectronics industry, silicon photonics can be 15 diagnostics are, for example, increasingly based on the efficiently manufactured on large diameter wafers 16 quantitative measurement of the concentration of using highly automated processes. Furthermore, silicon 17 certain biomolecules, referred to as biomarkers, in based optical circuits can be made very compact on the 18 patient samples such as blood and urine. Currently, this SOI platform, due to the high refractive index contrast, 19 which enables small waveguide bending radii. This 20 type of analysis is predominately performed in a laboratory setting, due to the stringent environmental permits efficient use of the wafer area, thereby keeping 21 device cost down. For bioanalysis, this means that 22 control and the high skill level of operators required for reliable results. However, there is a growing need for whole optical detector arrays can be integrated into a 23 single microfluidic channel. Another benefit of silicon 24 reliable mobile bioanalysis, e.g. to tackle epidemics 25 and biohazard monitoring on site. photonics is that electronics for readout and control can 26 be integrated into the same substrate, yielding an 27 The great success of current bioanalytical techniques is unprecedented integration density. The main limitation 28 largely built on the exploitation of highly selective of silicon photonics is that there are currently no 29 binding reactions of certain biomolecule pairs, e.g. monolithically integrated lasers available, but recent 30 certain proteins and nucleic acids, to perform advances on germanium lasers on silicon [2] hold great 31 biomolecule detection with high selectivity. In this promise for low cost silicon photonic systems in the 32 way, we take advantage of the work already done by near future. 33 evolution on the fine tuning of the control systems of 34 life. Alas, with this benefit comes a strict condition: we By combining the tools of silicon photonics and 35 need to handle liquid samples, since biological binding microfluidics, the basic toolset is in place for 36 reactions invariably take place in aqueous solutions. implementing robust mobile bioanalysis. A successful 37 merger, however, requires overcoming a number of 38 Microfluidics refers to the technology of fluid challenges outlined below. 39 manipulation on the microscale. From its inception 40 almost 30 years ago [1], this field has benefited greatly Current Status and Challenges 41 from the microfabrication techniques of the To target mobile bioanalysis, the chosen integration 42 microelectronics industry, but has recently developed technology should preserve the key benefits of silicon 43 its own toolset based on polymer micro-structuring, photonics mentioned above. This implies that both 44 often by casting or moulding. The rapid development 45 recognition element attachment and microfluidics of microfluidics has been driven by the desire to 46 integration should be done at wafer level, and without 47 automate and parallelize liquid sample handling for adding significant wafer footprint. So far, such an 48 applications such as medical diagnostics, biological approach has not been demonstrated. research, and drug development. 49 The second fundamental challenge is the volumes 50 The result of a bioanalysis must be quantified, and mismatch. For many important medical conditions, the 51 optical detection is a particularly popular quantification 52 concentration of the indicative biomarker analytes is approach, due to its large dynamic range, high spatial 53 very low in the sample liquid. Thus, for reliable 54 and temporal resolution, and strong resistance to detection, a certain minimum sample volume must be 55 electromagnetic interference. Silicon photonics is the analysed. For common cases, this results in sample 56 study and application of integrated optical systems volumes of 0.5 to 1 ml. Furthermore, most biosensing 57 which use silicon as an optical medium, usually by protocols call for washing with at least 5 times the 58 confining light in optical waveguides etched into the sample volume of clean buffer liquid. For comparison, 59 surface of silicon-on-insulator (SOI) wafers. By fixing a silicon photonic chip of 10×10 mm2 area covered 60 one part of the biological pairs mentioned above onto with a uniform 0.1 mm thick liquid film holds only the surface of such a waveguide, the selective binding 10 µl of liquid. Any microfluidic channel network of the other part can be optically detected. The fixed structured on top of such a chip will hold much less. molecule is referred to as a recognition element, the Hence, a mobile silicon photonics biosensing AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 16 of 59

1 2 3 instrument must be able to handle and store much more 4 5 liquid than can be fitted on the silicon chip surface. 6 The third fundamental challenge is the sensitivity of 7 the biological recognition elements to temperature and 8 solvents. This is particularly true for the more 9 interesting protein-based class of recognition elements 10 that generally neither tolerate heating significantly 11 12 above body temperature nor non-physiological 13 solvents, which severely limits fabrication options. 14 Nucleic acids, on the other hand, are more tolerant, but 15 then there already exist many more alternative analysis 16 methods for nucleic acids. 17 Figure 1 provides a schematic overview of the 18 Figure 1 – A schematic overview of the microfluidic 19 integration approaches that have been reported so far, integration approaches that have been reported, as well 20 as well as a few possibilities that have not been tried as a few new possibilities. 21 yet. The liquid handling methods can be classified into 22 two major groups, based on analyte transport modality, allow for low-temperature dry bonding both to silicon 23 as advective or diffusive. The advective class is and to other polymers. Using such a polymer, the 24 dominated by microfluidic channel based devices, alignment-critical first microfluidic channel layer could 25 usually employing pressure-driven flow in channels be made directly on the silicon photonics wafer, using 26 moulded into polymers such as polydimethylsiloxane spin coating, lithography, and development. Next, the 27 (PDMS) [3,4] and thiol-enes [5], and transfer-bonded 28 biological recognition elements could be deposited in a onto the silicon photonics. So far, only chip level serial process using a spotting robot, or preferably 29 processing, has been demonstrated. The critical aspect 30 using a wafer-scale printing scheme with a patterned of this approach is how to seal the channels after 31 stamp. Then, a second, lower resolution lid layer could 32 recognition element attachment, without affecting the be transferred and dry-bonded to form closed channels 33 biomolecules. PDMS is difficult to dry-bond to silicon with open vias aligned to the channel structure below. 34 after biomolecule attachment, since the required The lid layer could be injection moulded into a 35 oxygen plasma activation would destroy the thermoplastic, and thus also contain the large volume 36 biomolecules. Thiol-enes are more promising in this reservoirs necessary for ml-scale sample handling. 37 regard [6]. Another approach is to etch channels into a Another alternative would be to add a third layer 38 spin-coated perflouropolymer and then seal them with containing the large liquid reservoirs. 39 clamped gaskets [7]. The drawback of this approach is 40 that the poor alignment accuracy of the clamped gasket Currently, there exist polymers that have the potential 41 severely limits the integration density. Furthermore, a to allow such a processing scheme. In particular, the 42 leak tight seal will be difficult to achieve in a low-cost class of photo-curable thiolene-epoxy polymers [10] 43 consumer operated device. Pressure has also been used seems promising, due to its capability for high- 44 to flow samples through orifices in silicon photonic resolution photo-patterning and versatile low- 45 temperature dry bonding. 46 chips glued to the bottom of open ended plastic test 47 tubes [8]. Advective droplet transport (digital Another potentially rewarding development of silicon 48 microfluidics), which employs electro-wetting, has also photonics based biosensing would be to use IR 49 been reported [9], but the sample volumes that can be fluorescence in combination with silicon photonics. 50 handled in this way are very small. This would enable the combination of the high signal 51 Due to their simplicity, diffusive transport devices are to background possible with fluorescence based 52 sensing with the low-cost, high integration density 53 potentially very interesting for mobile biosensing and optics possible with silicon photonics. There are 54 high throughput screening applications. However, they 55 will struggle with the low analyte concentrations of already recent developments on IR fluorophores, but 56 most medical diagnostic applications. combining them with silicon photonics is uncharted 57 territory. Advances in Science and Technology to Meet 58 Concluding Remarks and Perspectives 59 Challenges 60 To be able to structure microfluidic circuits directly on Silicon photonics has already been commercially top of silicon photonic wafers, some advances in employed to biosensing in a laboratory setting. polymer science are needed. The ideal polymer should However, the big potential of the technology is to be patternable using standard photo-lithography, and enable quantitative and reliable mobile bioanalysis, for Page 17 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 example in a smart-phone attachment. To get there, Gunn L C 2010 Label-Free Biosensor Arrays Based 4 on Silicon Ring Resonators and High-Speed Optical 5 advances are necessary in wafer-scale polymer Scanning Instrumentation IEEE J. Sel. Top. 6 processing, as well as in wafer-scale biomolecule Quantum Electron. 16 654–661 7 attachment. Furthermore, a monolithically integrated 8 light source is also needed to realize this vision. 9 Finally, even though optical sensing, and biosensing in [8] Arce C L, Van Put S, Goes A, Hallynck E, Dubruel 10 particular, has been the driving force behind much of P, Komorowska K and Bienstman P 2014 Reaction 11 the recent work, the control and manipulation of light tubes: A new platform for silicon nanophotonic ring 12 by fluids is a potentially rewarding application of the resonator sensors J. Appl. Phys. 115044702+ 13 microfluidic integration technology discussed here. 14 [9] Lerma Arce C, Witters D, Puers R, Lammertyn J 15 and Bienstman P 2012 Silicon photonic sensors incorporated in a digital microfluidic system. Anal. 16 Acknowledgments – This work was partially 17 Bioanal. Chem. 404 2887–2894 supported by the Swedish Research Council (Grants 18 19 nr. B0460801 and 621-2012-5364). [10] Carlborg C F, Vastesson A, Liu Y, van der 20 Wijngaart W, Johansson M and Haraldsson T 2014 21 References Functional off-stoichiometry thiol-ene-epoxy 22 [1] Manz A, Graber N and Widmer H M 1990 thermosets featuring temporally controlled curing 23 Miniaturized total chemical analysis systems: A stages via an UV/UV dual cure process J. Polym. 24 novel concept for chemical sensing Sens. Actuators Sci. Part Polym. Chem. 52 2604–15 25 B Chem. 1 244–248 26 27 [2] Camacho-Aguilera R E, Cai Y, Patel N, Bessette J 28 T, Romagnoli M, Kimerling L C and Michel J 2012 29 An electrically pumped germanium laser Opt 30 Express 20 11316–11320 31 32 [3] Carlborg C F, Gylfason K B, Kazmierczak A, 33 Dortu F, Polo M J B, Catala A M, Kresbach G M, 34 Sohlström H, Moh T, Vivien L, Popplewell J, 35 Ronan G, Barrios C A, Stemme G and Wijngaart 36 2010 A packaged optical slot-waveguide ring 37 resonator sensor array for multiplex label-free 38 assays in labs-on-chips Lab Chip 10 281–290 39 40 [4] De Vos K, Girones J, Claes T, De Koninck Y, 41 Popelka S, Schacht E, Baets R and Bienstman P 42 2009 Multiplexed Antibody Detection With an 43 Array of Silicon-on-Insulator Microring Resonators 44 Photonics J. IEEE 1 225–235 45 46 47 [5] Errando-Herranz C, Saharil F, Romero A M, 48 Sandström N, Shafagh R Z, van der Wijngaart W, 49 Haraldsson T and Gylfason K B 2013 Integration of 50 microfluidics with grating coupled silicon photonic 51 sensors by one-step combined photopatterning and 52 molding of OSTE Opt Express 21 21293–21298 53 54 [6] Carlborg C F, Cretich M, Haraldsson T, Sola L, 55 Bagnati M, Chiari M and van der Wijngaart W 56 2011 Biosticker: Patterned microfluidic stickers for 57 rapid integration with microarrays 15th 58 International Conference on Miniaturized Systems 59 for Chemistry and Life Sciences (Seattle, 60 Washington, USA) pp 311–313

[7] Iqbal M, Gleeson M A, Spaugh B, Tybor F, Gunn W G, Hochberg M, Baehr-Jones T, Bailey R C and AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 18 of 59

1 2 3 4 6. Holographic on-chip microscopy and 5 tomography using lensless computational 6 imaging – Yibo Zhang and Aydogan Ozcan 7 University of California, Los Angeles 8 9 Relevance of the topic 10 11 The recent advances in micro- and nano-technologies 12 have greatly enhanced our capability to build cost- Figure1. (Left) Schematic setup of Holographic 13 effective and powerful chip-size microsystems that can Optofluidic Microscopy (HOM) [1]. (Right) Schematic 14 handle, e.g., cells and bio-fluids. These lab-on-a-chip setup of Holographic Optofluidic Tomography (HOT) 15 systems have wide applications in bioengineering and [10]. 16 medicine. The integration of optical imaging and 17 18 microscopy techniques with microfluidics can enable Lensless on-chip microscopy based on digital in-line 19 high-throughput detection and sensing of biological holographic imaging offers a wide FOV and a large 20 and/or chemical signals. To complement such DOF within a low cost and compact device platform 21 microfluidics and lab-on-a-chip systems, optical [1]–[10]. These features make it a promising solution 22 microscopy tools need to undergo a transformation in for optofluidic microscopy applications [1], [6], [8]. In 23 terms of their size, cost and throughput, and a holographic optofluidic microscopy (HOM) setup, 24 computational on-chip imaging provides unique the sample flows in a microfluidic channel, which is 25 opportunities for this broad goal. driven either by electro-kinetic force or pressure. A 26 partially coherent light source with a large illumination Current Status and Challenges 27 aperture (e.g., ~ 50-100 μm, filtered by a large pinhole 28 The optical interrogation of current microfluidic or butt-coupled to an optical fiber section) is placed 50- 29 samples is by and large conducted by conventional 30 100 mm (z1 distance) above the sample microfluidic light microscopy tools that use lenses for image 31 chip, while the microfluidic chip is placed directly on 32 formation, and have significantly smaller fields of view top of a CMOS image sensor. As a result, the sample 33 (FOV) than the size scale of a typical microfluidic chip of interest inside the micro-channel is typically located 34 and shallower depths of field (DOF) than the height of at ~ 1 mm (z2 distance) away from the imager chip. 35 typical microfluidic channels. Moreover, the FOV as The illumination light that can be from a low-cost 36 well as DOF of conventional microscopy scale down light-emitting diode (LED) acquires spatial coherence 37 when switched to higher-magnification objective after free-space propagation over a distance of z1, and 38 lenses (when higher resolution is needed), further the hologram recorded by the image sensor is formed 39 reducing the available imaging area and volume within by the interference of the light scattered by the sample 40 the sample. Besides, the bulkiness and high cost of within the fluid with the unscattered light that is 41 these conventional microscopes present another directly transmitted. Compared to laser sources, the use 42 obstacle to merge optical imaging with microfluidics 43 of a partially coherent light source (e.g., a simple LED) for building lab-on-a-chip systems, especially for use 44 reduces speckle and multiple-interference noise, 45 at low-resource and field settings. This mismatch resulting in cleaner images without loss of resolution 46 caused by the limitations of traditional optical imaging owing to on-chip imaging geometry. Moreover, the 47 modalities prompts advances in microscopy. cost of the setup is reduced by using low-cost LEDs as 48 Meanwhile, tomographic imaging techniques that light sources. 49 enable sectional imaging have been of great interest to 50 An important advantage of using the holographic 51 reveal 3D structures of various organisms. Sectional imaging principle is the capability to do single-shot 52 imaging modalities such as optical projection depth-resolved amplitude and phase imaging of a large 53 tomography, optical diffraction tomography, confocal sample volume. The holographic patterns can be 54 microscopy and light sheet microscopy, among others, digitally refocused to different z distances, with 55 can achieve high-resolution 3D optical imaging. particles/samples residing at different depths 56 However, these modalities are also limited by the respectively in focus. 57 aforementioned restrictions including low throughput, 58 high cost and bulkiness. When only one in-line hologram is used for 59 reconstruction, the major limiting factor for the 60 Therefore, it is of great importance to develop high- resolution is the pixel size, which is a disadvantage of resolution microscopic and tomographic imaging using unit fringe magnification – a characteristic techniques that have small footprints and large sample signature of an on-chip holographic microscope. imaging areas/volumes. However, the flow of the sample in a microfluidic Page 19 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 scattering object, and that the majority of the detected 4 5 photons should have experienced at most a single 6 scattering event. 7 A variant of the HOM and HOT methods was used for 8 the dynamic 3D tracking of sperm cells over large 9 volumes [4], [9]. In this system, two LED sources of 10 different wavelengths (red and blue) simultaneously 11 12 illuminate the microfluidic chamber from different 13 perspectives (vertical and 45° tilted, respectively). This 14 dual-wavelength illumination design enables digital 15 separation of the two holograms of the same sperm 16 Figure 2. 3D imaging of C. elegans using HOT.[10] cell, and the dual-angle design improves the cell 17 localization accuracy in the z direction via 18 channel (Fig. 1) can be leveraged to circumvent this triangulation. High frame rate lensless image 19 limit imposed by the pixel size by capturing multiple sequences were taken to resolve the rapid motion of the 20 holograms that are acquired at different time instances sperms within a very large volume of ~10 µL. Using 21 [1], [6], [8]. Due to the flow, these holograms generally this platform, rare trajectories of human sperms such as 22 have non-integer (sub-pixel) shifts with respect to each helical patterns were detected and quantified. A similar 23 other. A high-resolution hologram that is compatible platform can be used to detect and track a wide variety 24 25 with all the sub-pixel-shifted low-resolution holograms of other micro-swimmers that exhibit 3D locomotion 26 can be calculated by using a pixel super-resolution in fluids. (PSR) algorithm. 27 Concluding Remarks and Perspectives 28 Using lens-free holographic on-chip imaging together Lensless holographic microscopy and tomography are 29 with computed tomography principles, portable 30 promising techniques to complement microfluidic and tomographic microscopes can also be built, to achieve 31 lab-on-a-chip platforms as they present important 32 3D sectional imaging capability [3], [6]–[8], [10]. In advantages of e.g., high throughput, low cost, 33 this method, the sample is illuminated from multiple simplicity and small footprints, making them ideally- 34 angular positions ranging between ±50°, which suited to be used in low-resource and field settings. 35 provides different perspectives of the samples. These Manipulation of liquid samples and rapid screening of 36 images are holographically reconstructed, and a large volumes of fluids are some of the applications 37 computed tomogram is generated by applying a filtered that can benefit from these computational imaging 38 back-projection algorithm. Although the multi-angle techniques. 39 illumination increases the system complexity, it can 40 still be implemented in a cost-effective and portable Acknowledgments – The Ozcan Research Group at 41 device by using multiple LEDs that are butt-coupled to UCLA gratefully acknowledges the support of the 42 optical fiber sections at different angular positions. Presidential Early Career Award for Scientists and 43 Engineers (PECASE), the Army Research Office 44 PSR can be integrated onto the same platform by 45 electromagnetically actuation of the optical fibers (ARO; W911NF-13-1-0419 and W911NF-13-1-0197), 46 using coils and magnets [3]. This induced motion in the ARO Life Sciences Division, the National Science 47 the optical fibers results in sub-pixel shifts of the Foundation (NSF) CBET Division Biophotonics 48 acquired holograms, and can be used to synthesize Program, the NSF Emerging Frontiers in Research and 49 higher-resolution holographic images. With this Innovation (EFRI) Award, the NSF EAGER Award, 50 approach, tomographic imaging of approximately 15 NSF INSPIRE Award, NSF Partnerships for 51 mm3 has been achieved, with a spatial resolution of Innovation: Building Innovation Capacity (PFI:BIC) 52 <1μm × <1μm × <3μm in x, y and z direction. Program, Office of Naval Research (ONR), the 53 Tomographic imaging of an entire C. elegans has been National Institutes of Health (NIH), the Howard 54 demonstrated using the same technique [7]. Hughes Medical Institute (HHMI), Vodafone Americas 55 Foundation, the Mary Kay Foundation, Steven & 56 A unique advantage of the holographic optofluidic Alexandra Cohen Foundation, and KAUST. 57 tomography (HOT) method (Figs. 1-2) is its extended 58 DOF that is inherent to the holographic imaging References 59 principle. Unlike high-NA objective lenses that have a 60 very shallow DOF, a tomogram of an object can be [1] W. Bishara, H. Zhu, and A. Ozcan, “Holographic computed at any position within the imaged 3D opto-fluidic microscopy,” Opt. Express, vol. 18, no. volume of the micro-channel. However, notice that 26, pp. 27499–27510, Dec. 2010. HOT technique assumes that the sample is a weak AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 20 of 59

1 2 3 [2] S. O. Isikman, I. Sencan, O. Mudanyali, W. Bishara, 4 C. Oztoprak, and A. Ozcan, “Color and 5 monochrome lensless on-chip imaging of 6 Caenorhabditis elegans over a wide field-of-view,” 7 8 Lab Chip, vol. 10, no. 9, pp. 1109–1112, May 2010. 9 [3] S. O. Isikman, W. Bishara, U. Sikora, O. Yaglidere, 10 J. Yeah, and A. Ozcan, “Field-portable lensfree 11 tomographic microscope,” Lab. Chip, vol. 11, no. 12 13, pp. 2222–2230, 2011. 13 [4] T.-W. Su, L. Xue, and A. Ozcan, “High-throughput 14 lensfree 3D tracking of human sperms reveals rare 15 statistics of helical trajectories,” Proc. Natl. Acad. 16 Sci., vol. 109, no. 40, pp. 16018–16022, Oct. 2012. 17 [5] S. Seo et al., “High-Throughput Lens-Free Blood 18 Analysis on a Chip,” Anal. Chem., vol. 82, no. 11, 19 pp. 4621–4627, Jun. 2010. 20 [6] S. O. Isikman et al., “Lensfree On-Chip Microscopy 21 and Tomography for Biomedical Applications,” 22 IEEE J. Sel. Top. Quantum Electron., vol. 18, no. 3, 23 pp. 1059–1072, May 2012. 24 [7] S. O. Isikman et al., “Lens-free optical tomographic 25 microscope with a large imaging volume on a 26 chip,” Proc. Natl. Acad. Sci., vol. 108, no. 18, pp. 27 7296–7301, May 2011. 28 [8] W. Bishara, S. O. Isikman, and A. Ozcan, “Lensfree 29 Optofluidic Microscopy and Tomography,” Ann. 30 Biomed. Eng., vol. 40, no. 2, pp. 251–262, Feb. 31 2012. 32 [9] T.-W. Su, S. O. Isikman, W. Bishara, D. Tseng, A. 33 Erlinger, and A. Ozcan, “Multi-angle lensless 34 digital holography for depth resolved imaging on a 35 chip,” Opt. Express, vol. 18, no. 9, pp. 9690–9711, 36 Apr. 2010. 37 38 [10] S. O. Isikman, W. Bishara, H. Zhu, and A. Ozcan, 39 “Optofluidic Tomography on a Chip,” Appl. Phys. 40 Lett., vol. 98, no. 16, p. 161109, Apr. 2011. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 21 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 7. Live cell imaging with optofluidics – Adam 5 Wax 6 Duke University 7 8 Relevance of the topic 9 10 Cell biologists and microscopists will often image 11 individual live cells to establish the properties of a 12 population of interest. However, these studies can be 13 limited by the number of cells that can be practically 14 examined due to the need to reposition, refocus and 15 image each cell. As an alternative, flow cytometry can 16 enable large throughput studies but typically requires a 17 degree of instrument complexity. Perhaps a greater 18 limitation is that flow cytometry makes it difficult to Figure 1 – Surface enhanced Raman spectroscopy tags 19 follow the fate of individual cells. As an alternative allow rapid discrimination of cells in flow. Reprinted with permission from [6]. Copyright 2015 American 20 approach, optofluidics offer a unique opportunity to 21 Chemical Society examine live cells and easily manipulate them in a high 22 23 throughput scheme. pressure. A calibrated syringe allows the pressure to be varied across the array and thus implement 24 Many microfluidic platforms have been developed for simultaneous micropipette aspiration assays on each 25 examining individual live cells. The earliest trapped cell. This is a common technique in cell 26 implementations often included an optical scheme for biology for assessing cell biomechanical properties by 27 interrogating a cell, usually in the form of a fluorescent 28 noting the degree to which a cell is drawn into a reporter [1]. As the technology matured, optics were 29 narrow capillary due to a given force. This technique also adapted for implementing cell sorting [2]. The 30 is effective but can be quite laborious due to the need development of optics integrated into the microfluidic 31 to manually trap and image each cell. In comparison, platform birthed the field of optofluidics. The first key 32 the optofluidic platform allows highly parallel studies advances in this field were to enable high resolution 33 using brightfield microscopy to assess cell 34 imaging within the microfluidic platform. These low- deformation. In its initial demonstration, it was applied 35 cost implementations offered fit for purpose to distinguish breast cancer cells based on mechanical 36 alternatives to using a traditional microscope [3]. This properties. One particular challenge of this approach is 37 approach combines microfluidics for sample to keep the sample in focus during experiments. Here 38 manipulation with low cost sensors for imaging, and this is solved by integrating the platform with the 39 employ reconstruction algorithms to obtain high- microscope but other approaches are discussed below. 40 resolution images. Further development of these 41 42 approaches for cell imaging in a compact, low cost Raman spectroscopy is a powerful technique for 43 platform has enabled applications in new settings such distinguishing cells by their biochemical properties. 44 as resource limited locations and point of care medical However, in general the Raman cross section is low 45 diagnosis. Below, recent advances in live cell imaging resulting in poor efficiency and a need for long 46 using optofluidics are discussed alongside a brief look integration times. Analysis of the Raman signal from 47 at challenges in this field and suggestions on how cells inside an optofluidic platform usually must 48 future efforts may approach them. include a mechanism for trapping the cells using 49 mechanical or electrophoresis approaches. Using laser Current Status and Challenges 50 tweezers to trap cells for Raman measurements has 51 The true advantage of optofluidics for live cell imaging also been shown to be effective [5]. As an alternative 52 is the ability to apply a range of well controlled, approach surface-enhanced Raman spectroscopy 53 calibrated forces both to manipulate and characterize (SERS) has been used to discriminate cells in a 54 cells. These forces can be applied mechanically, 55 microfluidic platform [6]. SERS increases the strength 56 electrically or optically to isolate and interrogate of a Raman signal by orders of magnitude, enabling 57 individual cells for study. Cells under examination in more rapid detection. In this approach a SERS biotag 58 these devices are then typically profiled using an is created which is taken up by cells which are then 59 optical modality ranging from bright field imaging to introduced to the microfluidic. The flowing cells can 60 Raman spectroscopy to interferometry. then be rapidly profiled using the biotag for sorting. A recent example of this approach can be found in Lee Several efforts seek to increase the rate of cell imaging and Liu [4] where individual cells are trapped at as they flow through a device. While the use of designated sites in a micropipette array using flow today’s high pixel count sensors can allow for highly AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 22 of 59

1 2 3 parallel imaging approaches, these typically are limited 4 5 by the low numerical aperture optics required to have a 6 wide field of view. Light sheet imaging approaches 7 have shown promise for wide field and rapid 8 throughput but are limited to detecting fluorescent tags. 9 Novel imaging approaches, such as optical time stretch 10 imaging allows for more rapid imaging of cells [7], 11 achieving rates comparable to flow cytometers. In this 12 approach the imaging information is spectrally 13 encoded using a dispersed broadband pulse. 14 Extensions of this approach have shown the ability to 15 obtain phase information using an interferometry 16 scheme but can have difficulty with maintaining cells 17 18 in a focal plane. In addition, the complex scheme may 19 not be compatible with typical applications of 20 microfluidics as point of care devices. 21 Advances in Science and Technology to Meet 22 Challenges Figure 2 – Digital refocusing of a single red blood cell 23 image subjected to high flow rates. Top-Left: Optical 24 The continuing challenge in imaging of live cells in volume of RBC with flow on and off (rates indicated on 25 optofluidics is to achieve high throughput without plot). Blue, red are before, after digital refocusing. 26 sacrificing information or greatly increasing instrument Bottom-Left: Defocus distance. Reprinted from [10] with 27 complexity. Optofluidic devices become less efficient permission. 28 when there is a need to first trap each individual cell 29 before analyzing it. Instead, developing methods in regular image. These data are then inverse transformed 30 optofluidics seek to provide richer information while to deduce the 3D distribution. This advance was 31 preserving throughput. significant since it first demonstrated 3D tomographic 32 imaging of cells in flow. However, it had some 33 Holographic imaging offers a compelling way to obtain limitations, including requiring many exposures and 34 more information about cells in that it provides both 35 significant computation time, indicating that further phase and amplitude information on the optical field development is needed for high throughput application. 36 that has interacted with a cell. Unlike Raman 37 Further, this platform avoided the refocusing problem spectroscopy or fluorescence, interferometric signals 38 by confining cells to a channel just barely larger than 39 are quite robust yet the method does present some their size, which could become easily clogged if debris 40 challenges. For example, cell samples in a were present. 41 microfluidic may present some turbidity, due to 42 presence of other cells or fouling biomolecules. To As an alternative, our research group has shown that 43 implement holographic imaging in this situation, a refocusing of holographic cell images can avoid the 44 clean background image is needed for subtraction from need for restrictive channel dimensions. In this 45 later images. A recent study pointed the way to approach holographic cell images are acquired and 46 conquering turbidity through careful analysis of the then digitally refocused by leveraging the fact that the 47 holographic signal components [8]. The approach, complex optical field is obtained [10]. In this initial 48 termed multi-look, allowed for successful imaging of demonstration, it was shown that quantitative 49 cells in the face of a turbid sample in the microfluidic comparison of red blood cells was much more accurate 50 channel. after refocusing. We further employed this refocusing 51 algorithm in two studies of red blood cells infected 52 Another compelling aspect of holography is its ability with P. falciparum, a parasite responsible for malaria. 53 to image the three dimensional distribution of The first study used our quantitative phase 54 refractive index, an indicator of the cell’s mass 55 spectroscopy technique [11] to analyze the hemoglobin distribution. However, this approach for refractive content of individual red blood cells, revealing its 56 index cell tomography can require many camera 57 consumption by the parasite [12]. A second study used 58 exposures, typically from different angles, during machine learning to discriminate the presence of 59 which the cell must remain stationary. A recent infection by using a set of cell descriptors, producing 60 advance has pointed the way to 3D holographic high enough sensitivity and specificity to suggest imaging in flow [9]. In this approach a cell flows clinical utility[13]. Both of these studies employed the slowly past a high speed camera which captures the refocusing approach which will be essential in angular distribution of scattered light rather than a Page 23 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 development of holographic techniques for studying Measurement of Continuously Flowing Cells in a 4 cells using optofluidics. Microfluidic Channel. Physical Review Applied, 5 2014. 1(1): p. 014002. 6 Concluding Remarks and Perspectives [10] Rinehart, M.T., H.S. Park, and A. Wax, Influence of 7 In summary, optofluidics offers unique capabilities for defocus on quantitative analysis of microscopic 8 objects and individual cells with digital high throughput cell studies. Many research efforts are 9 holography. Biomedical optics express, 2015. 6(6): 10 underway to mine the richness of the optical spectrum p. 2067-2075. 11 to characterize cells while maintaining sufficient [11] Rinehart, M., Y. Zhu, and A. Wax, Quantitative 12 throughput to implement viable assays. New phase spectroscopy. Biomedical optics express, 13 techniques such as holographic imaging can provide 2012. 3(5): p. 958-965. 14 new capabilities but successful application can pose [12] Rinehart, M.T., H.S. Park, K.A. Walzer, J.-T.A. 15 new challenges as well. Chi, and A. Wax, Hemoglobin consumption by P. 16 falciparum in individual erythrocytes imaged via 17 Acknowledgments – Grant support from the National quantitative phase spectroscopy. Scientific reports, 18 Science Foundation CBET 1604562. 2016. 6. 19 [13] Park, H.S., M.T. Rinehart, K.A. Walzer, J.-T.A. 20 References Chi, and A. Wax, Automated Detection of P. 21 [1] Wheeler, A.R., W.R. Throndset, R.J. Whelan, A.M. falciparum Using Machine Learning Algorithms 22 Leach, R.N. Zare, Y.H. Liao, K. Farrell, I.D. with Quantitative Phase Images of Unstained Cells. 23 Manger, and A. Daridon, Microfluidic Device for PloS one, 2016. 11(9): p. e0163045. 24 Single-Cell Analysis. Analytical Chemistry, 2003. 25 75(14): p. 3581-3586. 26 [2] Wang, M.M., E. Tu, D.E. Raymond, J.M. Yang, H. 27 Zhang, N. Hagen, B. Dees, E.M. Mercer, A.H. 28 Forster, and I. Kariv, Microfluidic sorting of 29 mammalian cells by optical force switching. Nature 30 biotechnology, 2005. 23(1): p. 83-87. 31 [3] Cui, X., L.M. Lee, X. Heng, W. Zhong, P.W. 32 Sternberg, D. Psaltis, and C. Yang, Lensless high- 33 resolution on-chip optofluidic microscopes for 34 Caenorhabditis elegans and cell imaging. 35 Proceedings of the National Academy of Sciences, 36 2008. 105(31): p. 10670-10675. 37 [4] Lee, L.M. and A.P. Liu, A microfluidic pipette 38 array for mechanophenotyping of cancer cells and 39 mechanical gating of mechanosensitive channels. 40 Lab on a Chip, 2015. 15(1): p. 264-273. 41 [5] Lau, A.Y., L.P. Lee, and J.W. Chan, An integrated 42 optofluidic platform for Raman-activated cell 43 sorting. Lab on a Chip, 2008. 8(7): p. 1116-1120. 44 [6] Pallaoro, A., M.R. Hoonejani, G.B. Braun, C.D. 45 Meinhart, and M. Moskovits, Rapid Identification 46 by Surface-Enhanced Raman Spectroscopy of 47 Cancer Cells at Low Concentrations Flowing in a 48 Microfluidic Channel. ACS Nano, 2015. 9(4): p. 49 4328-4336. 50 [7] Yang, T., P. Paiè, G. Nava, F. Bragheri, R.M. 51 Vazquez, P. Minzioni, M. Veglione, M. Di Tano, 52 C. Mondello, and R. Osellame, An integrated 53 optofluidic device for single-cell sorting driven by 54 mechanical properties. Lab on a Chip, 2015. 15(5): 55 p. 1262-1266. 56 [8] Bianco, V., F. Merola, L. Miccio, P. Memmolo, O. 57 Gennari, M. Paturzo, P.A. Netti, and P. Ferraro, 58 Imaging adherent cells in the microfluidic channel 59 hidden by flowing RBCs as occluding objects by a 60 holographic method. Lab Chip, 2014. 14(14): p. 2499-504. [9] Sung, Y., N. Lue, B. Hamza, J. Martel, D. Irimia, R.R. Dasari, W. Choi, Z. Yaqoob, and P. So, Three- Dimensional Holographic Refractive-Index AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 24 of 59

1 2 3 4 8. Optofluidic microlenses: from tunable 5 focal length to aberration control – Frieder 6 Mugele 7 University of Twente 8 9 Relevance of the topic 10 11 Generating optical images by refractive lenses is still 12 the most important practical use of optical systems. In 13 principle only one single lens is needed to form an 14 optical image and two to generate a zoom system. In 15 practice, however, optical systems require a lot more 16 than two lenses in order to guarantee a decent – 17 ideally: diffraction limited – imaging quality. 18 Combinations of several lenses are typically used to 19 20 compensate imaging faults that arise from the various

21 aberrations of each individual lens. If zoom or focus Fig. 1: a) pressure-controlled optofluidics lens with 22 settings are to be changed, many of these lenses have pinned contact line. b) EW-controlled optofluidics lens 23 to be displaced with respect to each other in a well- as sessile drop (top) and in cylindrical tube (bottom). c) 24 defined manner. This typically requires fairly precise MTF for spherical lens with fixed diameter for 25 and bulky mechanical displacement stages that are increasing and curvature (see inset). d) MFT for 26 often not compatible with, e.g., the spatial constraints fixed pressure and increasing top voltage 퐷 that 27 퐿 in high tech devices such as mobile phones or induced increasingΔ푝 asphericity. Inset: electric field- 28 푇 endoscopes. induced surface deformation (adapted from푈 [2]). 29 30 Optofluidic lenses provide a way out of this dilemma. typically matched in order to avoid gravity-induced 31 Instead of translating lenses they adapt their imaging distortions.) Pressure and/or volume can be changed by 32 properties by adapting the shape of the lenses. This can a variety of different actuation schemes including 33 be achieved by a variety of external control parameters piezo-electric transducers, thermal expansion, and 34 including pressure variations, mechanical stresses, electrowetting, see [3] for a recent review. In some 35 thermal expansion, and electrowetting [3]. While these approaches, the interface between the two fluids is 36 approaches provide a fair degree of tunability of the covered by an elastomeric membrane, e.g. of PDMS 37 focal length, most of them retain a spherical shape of (polydimethylsiloxane) with a thickness of a few tens 38 the lens, as dictated by the mechanical equilibrium of 39 of micrometers. For not too large deformations, the 40 free liquid surfaces in the absence of external forces. resulting shape of the interface in these ‘liquid-filled’ 41 As a consequence, optical systems built of simple lenses is also approximately spherical cap-shaped with 42 optofluidics lenses often display mediocre imaging the elastic tension of the membrane taking over the role 43 quality due to uncompensated imaging faults such as of . 44 spherical and other optical aberrations. High quality 45 adaptive imaging systems require new strategies for In 휸the second variant of optofluidics lenses, the three 46 aberration correction in liquid microlenses while phase contact line is free to move along the solid 47 preserving the compactness of the design, i.e. without surface, but the contact angle is changed, Fig. 1b. 48 introducing mechanical stages to displace the lenses. Variations of in combination with the fixed radius 49 and/or volume of the lens fluid determine휽 the curvature 50 Current Status and Challenges of the lens. Reversible,휽 fast, and robust tuning of the 51 Standard optofluidic lens designs come in two basic contact angle over a wide range is best achieved using 52 variants as illustrated in Fig. 1a and b. In the first type, electrowetting (EW; see [5] for a review). EW requires 53 the edge of the lens, i.e. the three phase contact line of a conductive lens fluid and a dielectric ambient 54 fluid. It allows to vary ( ) by more than 90° within 55 between the solid, the lens fluid (LF) and the ambient fluid (AF) is pinned along the edge of a physical milliseconds with excellent reproducibility over 56 휽 푼 57 aperture, Fig. 1a. In this case, the curvature of the millions of actuation cycles. 58 lens is determined either by the pressure difference Yet, both approaches lead to spherical lenses for 59 = between the two fluids or by the휿 fixed cylindrically symmetric geometries. As a consequence, 60 volume of the lens fluid in combination with geometric such lenses always suffer from optical (in particular: 푳 constraints.횫풑 ퟐ휸휿 The interfacial tension between the two spherical) aberrations that compromise the resulting fluids guarantees the spherical cap shape of the image quality, as evidenced in plots of the modulus of interface. (The densities of lens and휸 ambient fluid are the optical transfer function (MTF) for increasing Page 25 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 curvature, Fig. 1c, and in the resulting images, Fig. 2a. 4 5 Such deteriorations are particularly dramatic for lenses 6 with high numerical aperture or small f-number, as 7 they are frequently needed in miniaturized devices. 8 In the case of liquid-filled lenses, combinations of two 9 differently curved surfaces in biconvex lenses were 10 demonstrated to reduce both spherical [6, 7] and 11 12 chromatic aberration. The different curvatures are 13 achieved by using two membranes with a different 14 stiffness resulting in different curvatures for the same 15 applied pressure, or more recently with elastomeric 16 membranes with a carefully designed radial thickness 17 profile [8]. While preserving the elegance of a single 18 actuator, the applied pressure, this fixed design of the 19 membranes, and their non-linear elastic and hysteretic 20 response, reduce to some extent the flexibility of this 21 approach. 22 23 Advances in Science and Technology to Meet 24 Challenges 25 26 To overcome these limitations, alternative actuation 27 schemes that enable not only a variation of the overall 28 average curvature of the lens, but also controlled 29 deviations from the spherical shape are needed. A few 30 recent studies suggest to address this problem using 31 electrowetting or a related electrical actuation in Figure 2: a) top: side view image (left) and image of a test grid (right) recorded with a spherical liquid lens with 32 combination with suitably patterned electrodes. One of finite longitudinal spherical aberration (LSA); bottom: 33 the first applications of this type was not intended for same as in a) for an aspherical lens for negligible LSA 34 imaging optics but for (wide angle) beam steering. deformed by optimized Maxwell stress. (adapted from 35 Smith et al.[9] used square or rectangular cuvettes with [1]) b) Zernike coefficient for astigmatism in vertical and 36 EW-functionalized sidewalls that could be addressed horizontal direction obtained by EW with segmented 37 individually. In this manner, they managed to generate electrodes in azimuthal direction. (reproduced from 38 [4]) 39 flat liquid-liquid interfaces with a widely tunable tilt 40 angle resulting in a ± ° beam steering capability. dependent electrical stress and the resulting position- 41 dependent local Laplace pressure and curvature of the Kopp and Zappe [4]ퟏퟓ recently presented a design in 42 which they segmented the electrode along the inner interface. For an unstructured electrode on the top, the 43 side of a cylinder. In this case, the liquid-liquid local curvature of the interface decreases 44 monotonically from the optical axis towards the edge 45 interface assumes in general complex shape that is of the lens, similar to a perfect aspherical lens. Indeed, 46 compatible with the local contact angles on electrode 47 and the requirement of a constant mean curvature in it turns out that the longitudinal spherical aberration 48 mechanical equilibrium. Using eight individually can be completely eliminated, resulting in diffraction- 49 addressable electrode segments they managed to limited MTF curves, Fig. 1d, as well as optical images 50 generate lenses with tunable focal length (7-12mm) (Fig. 2b). 51 and tunable astigmatism in 0, ± °, and 90° directions Extensions of the method using segmented top 52 with amplitudes up to 3µm (Zernike coefficient) while electrodes, a homogeneous one, allow to break the 53 ퟒퟓ keeping the amplitude of all other aberrations below cylindrical symmetry and the generate lenses with 54 100nm, Fig. 2b. 55 tunable astigmatism [10], coma, and various other 56 Mishra et al. [1] presented a different approach, in aberrations. 57 which they used another electrode placed in the Concluding Remarks and Perspectives 58 ambient dielectric above the liquid-liquid interface, see 59 inset Fig. 1d. Application of a high voltage to that EW and the related Maxwell stress-induced tuning of 60 electrode generates a position-dependent electric field liquid-liquid interfaces provide very fast and flexible that pulls on the liquid-liquid interface with a Maxwell tools to generate tunable lens profiles with variable stress ( ) = ( )/ . The equilibrium shape of local curvature. The design of optofluidics devices that the liquid is set byퟐ balance between this position- allow for correcting tilt, coma, astigmatism, and 횷 풓 흐ퟎ흐푬 풓 ퟐ AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 26 of 59

1 2 3 spherical aberration simultaneously with an all 4 5 electrical actuation seems within reach, possibly best 6 by combining the currently existing approaches 7 discussed above. The extent to which different 8 aberrations can be addressed completely independently 9 needs to be explored. Improved geometric designs 10 aided by numerical simulation and improved materials 11 (e.g. low interfacial tension fluids) will help to improve 12 the flexibility. Ideally, it is conceivable to achieve 13 almost arbitrary polarization-independent wavefront 14 shaping on top of a lensing effect by using arrays of 15 large numbers of individually addressable electrodes. 16 17 Acknowledgments – We thank the Dutch Science 18 Foundation NWO supporting our optofluidics 19 activities within their VICI program under grant 20 #11380. 21 22 References (separate from the two page limit) 23 [1] Mishra, K, Murade, C, Carreel, B, Roghair, I, Oh, JM, 24 Manukyan, G, van den Ende, D, and Mugele, F, 2014 25 Optofluidic lens with tunable focal length and 26 asphericity. Scientific Reports. 4:6378. 27 [2] Mishra, K and Mugele, F, 2016 Numerical analysis of 28 electrically tunable aspherical optofluidic lenses. Optics 29 Express. 24:14672-14681. 30 [3] Mishra, K, van den Ende, D, and Mugele, F, 2016 31 Recent Developments in Optofluidic Lens Technology. 32 Micromachines. 7:102. 33 [4] Kopp, D and Zappe, H, 2016 Tubular astigmatism- 34 tunable fluidic lens. Optics Letters. 41:2735-2738. 35 [5] Mugele, F and Baret, JC, 2005 Electrowetting: From 36 basics to applications. Journal of Physics-Condensed 37 Matter. 17:R705-R774. 38 [6] Fuh, YK, Lin, MX, and Lee, S, 2012 Characterizing 39 aberration of a pressure-actuated tunable biconvex 40 microlens with a simple spherically-corrected design. 41 Optics and Lasers in Engineering. 50:1677-1682. 42 [7] Yu, HB, Zhou, GY, Leung, HM, and Chau, FS, 2010 43 Tunable liquid-filled lens integrated with aspherical 44 surface for spherical aberration compensation. Optics 45 Express. 18:9945-9954. 46 [8] Zhao, PP, Ataman, C, and Zappe, H, 2016 Gravity- 47 immune liquid-filled tunable lens with reduced spherical 48 aberration. Applied Optics. 55:7816-7823. 49 [9] Smith, NR, Abeysinghe, DC, Haus, JW, and Heikenfeld, 50 J, 2006 Agile wide-angle beam steering with 51 electrowetting microprisms. Optics Express. 14:6557- 52 6563. 53 [10] Lima, NC, Cavalli, A, Mishra, K, and Mugele, F, 54 2016 Numerical simulation of astigmatic liquid lenses 55 tuned by a stripe electrode. Optics Express. 24:4210- 56 4220. 57 58 59 60 Page 27 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 9. Integration of reconfigurable photonics (a) 5 Holger Schmidt and microfluidics – (i) (ii) 6 University of California, Santa Cruz 7 a 8 Relevance of the topic 9 10 Ever since the emergence of optofluidics in the early 11 2000s, reconfigurability has been one of the field’s 12 defining characteristics, afforded by the potential to 13 replace and modify non-solid media to (rapidly) 14 change the optical properties of a photonic device 15 [1,2]. At the time, reconfigurable devices were inspired 16 by possible applications in optical communications. (b) 17 Thus, on-chip laser sources that could be tuned via 18 swapping the active (liquid) gain medium or 19 20 mechanically deforming the cavity were a natural early 21 demonstration of the power of this approach [3]. 22 Subsequently, reconfigurability was demonstrated in a 23 number of other photonic elements, including switches 24 [4], lenses [2], microring resonators [2], and spectral Figure 1: (a) Liquid-core waveguide optofluidic platform 25 filters [2]. (i) Schematic view of hybrid integration of microfluidic 26 and optofluidic chips [8], (ii) MMI waveguide creating 27 Current Status and Challenges multiple excitation spots in liquid-core channel [9], inset: 28 Reconfigurability remains one of the distinctive photo of silicon chip; (b) Left: Multiple ways for 29 dynamic reconfiguration of PDMS waveguides [10]; characteristics and a major thrust area for optofluidics. right: image of PDMS waveguide chip. 30 However, over the years, the main interest in terms of 31 real-world applications has shifted towards chemical enhance emerging optofluidic platforms for advanced 32 and biological analysis, in particular disease biochemical analysis. 33 34 diagnostics. Here, some of the opportunities lie in Advances in Science and Technology to Meet 35 integration of photonic functions with microfluidic Challenges 36 sample handling and the development of compact, Devices made from flexible materials such as PDMS 37 powerful point-of-care devices. For example, recently 38 single virus particles have been imaged on a offer the highest potential for dynamic reconfiguration 39 smartphone-based platform [5]. At the same time, as their optical properties can be modified in at least 40 hybrid integration of silicon-based optofluidic chips three ways – mechanical pressure, altering the fluid in 41 with glass or PDMS-based microfluidic chips has a channel, and by physical movement. These options 42 emerged as a powerful approach to create a single, are visualized in Fig. 1b (left). Recently, the first 43 optofluidic system that is optimized for both sample PDMS-based optofluidic platform that features all 44 preparation and analysis [6,7]. Using a combination of these options for dynamic reconfiguration was realized 45 microvalve-controlled microfluidics and liquid-core using a combination of fluid-filled microchannels 46 (liquid-core waveguides) and solid-core PDMS 47 waveguide-based fluorescence sensing as shown in Fig. 1a(i), amplification-free detection of single Ebola waveguides in which the core and cladding are defined 48 by PDMS layers with different precursor mixing ratios 49 nucleic acids was demonstrated with high specificity 50 over the entire clinically relevant viral load range [8]. and, thus, different refractive indices (Fig. 1b) [10]. 51 Moreover, multimode interference (MMI) waveguides When properly fabricated, this allows for guiding light 52 were integrated on the same silicon chip platform to through micron-scale solid-core waveguides and 53 add multiplexing capabilities by creating color- interacting with liquid channels, much like established 54 dependent spot patterns in a liquid channel as silicon-based devices [2,8,9]. A photograph of a 55 illustrated in Fig. 1a(ii) [9]. This approach was completed device that features both a microvalve- 56 successfully used to distinguish three influenza based sample preparation region and an optical 57 subtypes on the single virus particle level [9]. analysis region on the same chip is shown in Fig. 1b 58 (right). 59 These recent demonstrations are indicative of the 60 power of waveguide-based optofluidic devices. It is, Fig. 2 illustrates different aspects of this powerful, new therefore, a natural question whether the well- platform, highlighting the different reconfigurability established fluidic reconfigurability [1-4] can further options [10]. Fig. 2a summarizes the implementation of an MMI waveguide whose properties can be AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 28 of 59

1 2 3 (left). However, if the segment that seals the channel 4 (a) (b) (i) (dark gray) is made of high-index PDMS, the valve 5 (i) 6 also acts as an optical waveguide, symbolized by the 7 red confined mode in the figure. The structure now acts 8 as a “lightvalve” because lifting the central membrane 9 up by applying negative pressure in the pneumatic top 10 fixed movable layer results in the light passing through fluid that now (ii) SC-WG lightvalve 11 (ii) fills the space (Fig. 2b(i), right). This creates a new 12 type of optical waveguide that can be individually 13 controlled and dropped into fluidic channels at will. 14 Fig. 2b(ii) shows two examples for the vast number of 15 possibilities that this principle creates. The top image 16 shows a photograph of a straight lightvalve segment 17 18 that connects two fixed, solid-core waveguide sections. (iii) (iii) When resting on the substrate, light is passed with 19 0 w0=125µm, N=12 negligible loss, but upon pneumatic actuation (lifting 20 w0=50µm, N=2 3.4 w =200µm, N=33 -10 21 0 up or pushing down), the optical path across the 3.2 -20

22 L [mm] -30 lighvalve becomes extremely lossy. Fig. 2b(iii) shows 3 23 -40 how the transmission can be controlled by the (push-

2.8 [dB] Transmission 24 1.3 1.35 1.4 1.45 0 20 40 down) actuation pressure, and a tunable light switch n 25 c Pressure [PSI] with on-off ratios of up to 45dB is created. This switch 26 Figure 2: (a) Tunable MMI waveguide. (i) Schematic can be operated without degradation over 100,000 27 top-down view, (ii) image of light pattern in MMI times. 28 section (top) in excellent agreement with simulations 29 (bottom), (iii) tuning of spot pattern with core fluid index On the bottom of Fig. 2b(ii), another version of the 30 for different MMI widths [10]; (b) Actuatable lightvalve is shown. Here, the movable segment is 31 lightvalves. (i) schematic view of cross-section of lifting- shaped as a ring with short waveguide stubs that 32 gate valve with central high index layer for waveguiding, interface with fixed solid-core waveguide sections on 33 (ii) top-down images of straight waveguide on-off switch either end. This annulus was used as an actuatable trap 34 (top) and ring-shaped lightvalve for particle trapping and for particles that flow through the fluidic channel when analysis (bottom), (iii) pressure-dependent light 35 the ring is lifted. In this way, trapping and waveguide- 36 transmission across straight lightvalve switch from (ii) based optical analysis of a controlled number of 37 with high on-off ratio. fluorescent beads and E.coli bacteria was demonstrated 38 controlled using the core fluid or mechanical pressure. 39 [10]. This shows that the lightvalve concept has 40 Fig. 2a(i) shows a schematic top down view where tremendous flexibility both in geometrical design and 41 light from a single-mode solid-core PDMS waveguide functionality vis-a-vis transport and manipulation of 42 section (dark gray) enters a wide channel that can be both fluids and light. 43 filled with different liquids and acts as the MMI Concluding Remarks and Perspectives 44 waveguide. When filled with ethylene glycol, an index- 45 guiding MMI is formed. Fluorescent dye in the Dynamic reconfiguration can significantly expand the 46 solution reveals the multimode interference pattern functionality and performance of waveguide-based 47 (Fig. 2b(ii, top)) in near-perfect match with the optofluidic devices. This was illustrated with a PDMS- 48 simulated behavior of this structure (Fig. 2b(ii, based platform that combines liquid-core and solid- 49 bottom)). This spot pattern can be dynamically altered, core waveguides as well as fluidic sample handling in a 50 either by changing the core liquid or by applying single device. Light-guiding microvalves (lightvalves) 51 pressure in the air cavities surrounding the MMI, thus 52 that can be physically moved to alter fluid and light 53 changing its width. Fig. 2a(iii) shows how the location propagation further add to the optofluidic toolbox. It 54 of different spot numbers along the MMI waveguide can be expected that additional novel functionalities 55 can be tuned with the index of the liquid in the MMI will be demonstrated on this platform as well as using 56 section. other optofluidic waveguide types such as slot 57 A novel type of reconfiguration was demonstrated in waveguides, photonic crystal waveguides, or laser- 58 this platform by re-envisioning the design and function written waveguides. 59 60 of fluidic lifting-gate microvalves previously used for Acknowledgments – This work was supported by the carrying out distribution and preparation of biological NIH under grants 4R33AI100229 and 1R01AI116989- sample materials [8]. Fig. 2b(i) shows a schematic 01 as well as the W.M. Keck Center for Nanoscale cross section of such a valve where a thin PDMS Optofluidics at UC Santa Cruz. membrane sits on a substrate, thus blocking fluid flow Page 29 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 References 4 [1] Monat C, Domachuk P, and Eggleton BJ 2007 5 Integrated optofluidics: A new river of light. Nat. 6 Photonics 1, 106. 7 8 [2] Schmidt H., Hawkins AR 2011 The photonic 9 integration of non-solid media using optofluidics. 10 Nature Photonics 5, 598-604. 11 [3] Li Z, Zhang Z, Scherer A, Psaltis D 2006 12 Mechanically tunable optofluidic distributed 13 feedback dye laser. Opt Express 14, 10494-9. 14 [4] Campbell K, Groisman A, Levy U, Pang L, 15 Mookherjea, S,Psaltis D, and Fainman, Y 2004 A 16 microfluidic 2x2 optical switch. Appl. Phys. Lett. 17 85, 6119-6121. 18 [5] Wei Q, Qi H, Luo W, Tseng D, Ki SJ, Wan Z, 19 Göröcs Z, Bentolila LA, Wu TT, Sun R, and Ozcan 20 A 2013 Fluorescent Imaging of Single 21 Nanoparticles and Viruses on a Smart Phone ACS 22 Nano 7, 9147–55. 23 [6] Parks JW, Cai H, Zempoaltecatl L, Yuzvinsky TD, 24 Leake K, Hawkins AR 2013 Hybrid optofluidic 25 integration. Lab Chip 13, 4118-23. 26 [7] Testa G, Persichetti G, Sarro PM, Bernini R 2014 A 27 hybrid silicon-PDMS optofluidic platform for 28 sensing applications. Biomed Opt Express 5, 417-26. 29 [8] Cai H, Parks JW, Wall TA, Stott MA, Stambaugh 30 A, Alfson K, Carrion R, Patterson JL, Hawkins AR, 31 and Schmidt H 2015 Optofluidic analysis system for 32 amplification-free, direct detection of Ebola 33 infection. Scientific Reports 5, 14494. 34 [9] Ozcelik D, Parks JW, Wall TA, Stott MA, Cai H, 35 Parks JW, Hawkins AR, and Schmidt H 2015 36 Optofluidic wavelength division multiplexing for 37 38 single-virus detection. Proc. Nat. Acad. Sci. 112, 39 12933-12937. 40 [10] Parks JW, and Schmidt H 2016 Flexible optofluidic 41 waveguide platform with multi-dimensional 42 reconfigurability. Sci. Rep. 6, 33008. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 30 of 59

1 2 3 4 10. Optofluidic waveguides and resonators 5 – Genni Testa, Romeo Bernini 6 CNR-IREA 7 8 Relevance of the topic 9 10 Optofluidic waveguides offer the possibility of guiding 11 light through a fluid that could be a liquid, a gas or a 12 plasma. This peculiar property could be exploited in all 13 applications in which strong light−matter interaction 14 is required. In recent years, promising demonstration of 15 the potentiality of optofluidic waveguides have been 16 reported in sensing [1], non-linear photonics, atomic 17 spectroscopy and quantum interference [2]. Figure 1 – Schematic layout of a) Bragg, b) ARROW 18 and c) slot waveguide 19 Daniel Colladon gave the first demonstration of an 20 optofluidic waveguide in 1842, however, only in recent microsystems . SWs confine a substantial fraction of 21 years a great effort has been devoted to the the optical power into nanometer void thanks to TIR 22 development of reliable and high performance effect. Such a tight confinement in subwavelength 23 optofluidic waveguides and devices [3]. The main interaction area makes these waveguides very powerful 24 issue related to optofluidic waveguides is the difficulty 25 for non-linear photonics and sensing applications [3]. of fulfilling the total internal reflection (TIR) condition 26 Although SWs are extremely promising in detecting when low refractive index fluids are used as core 27 ultra-low concentrations, the fluidic transport scales material. However, optical confinement in a waveguide 28 unfavourably in nanofluidic channels, making the can be also based upon interference phenomena along 29 integration of fluidic system for sample manipulation the transverse direction [3]. The first Bragg fibre 30 still a challenging task. The implementation of 31 utilizing a 1D photonic crystal was proposed by Yeh et integrated waveguides with hollow core is also 32 al. in 1978. Russell successively extended this concept possible by exploiting interference effects. BWs and 33 in a 2D geometry. Hollow core photonic crystal ARROWs have been fabricated by creating 34 (HCPC) fibers have been successfully applied. multilayered dielectric cladding s that act as a high 35 However, HCPC fibers require complex fabrication performance reflector. This confinement mechanism 36 procedures and could not be easily integrated with causes propagation losses due to a not ideal boundary 37 other components. In order to overcome these 38 reflector, and thus requires an accurate design strategy. problems, many integrated architectures have been 2 39 Air guiding BWs with core sizes of about 60×2 μm proposed to develop integrated waveguides that can be 40 and propagation losses of few dB/cm have been considered for planar optofluidics, like slot waveguides 41 demonstrated and bent section investigated. However, (SWs), Bragg (BWs) or Antiresonant Reflection 42 the implementation of Bragg waveguides into complex Optical Waveguides (ARROWs) [3] (Fig.1). The 43 circuit layouts for sensing purposes has not been fully 44 advantage of these approaches relies on the possibility demonstrated. Single mode, linearly polarized, liquid 45 to use them as building block of more complex 2 core ARROW with mode area of 4.5 μm and low 46 devices. In particular, the opportunity to make attenuations of about 1dB/cm have been demonstrated 47 optofluidic resonators has attracted a lot of attention. by an optimized design of the waveguide structure. 48 These devices enable a strong increase of the Nonetheless, further improvements are attainable for 49 light−matter interaction, which is no more limited by 50 both BWs and ARROWs that could open-up new 51 the physical length but is related to the number of application scenarios, especially in the field of sensing. 52 revolutions of light in the resonator, characterised by An important concern in the waveguide performances 53 the quality factor Q. is to reduce the core size in order to further shrink the 54 Current Status and Challenges mode field volume. Small mode volume can 55 significantly enhance the light−matter interaction and Integrated optofluidic waveguides like SWs, BWs and 56 increase the sensitivity by reducing the background 57 ARROWs offer very attractive potential of low loss, optical noise. However, small core size poses 58 single mode operation, which is highly desirable to tremendous design and fabrication challenges since it 59 build photonic circuits and systems. Moreover, the typically leads to a rapid increase of the propagation 60 planar nature of these architectures lends itself to and bend losses. further integration of additional functionalities, like microfluidics, in a similar manner as integrated Until now, integrated optofluidic waveguides have microelectronics, leading to self-contained been mostly designed to operate in the visible/near-IR Page 31 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 spectral range with liquid core (n>1.33). Nonetheless, 4 5 they can be designed to operate with a gaseous core 6 (n≈1) and in a larger spectral window from the visible 7 to the far-IR, providing that the cladding materials and 8 the waveguide geometry are properly selected. In 9 particular, the mid-infrared (mid-IR) (2.5-25µm) is 10 very interesting as it supports strong molecular 11 absorption fingerprints, thus offering an interesting 12 spectral window for gas/vapour spectroscopy as the 13 hollow core can be used as micro chamber, enabling a 14 strong sensitivity enhancement. Promising results with Figure 2 – Scanning electron microscope picture of an 15 slot waveguides operating in the mid–IR for gas and 16 optofluidic ring resonator based on ARROWs. 17 liquid sensing have been proposed [4]. In the case of 18 ARROWs, when the core is filled with gases, the limited material choice of the integrated technologies 19 design optimization as far as the fabrication turns into a in the mid-IR. As an example, suspended silicon slot 20 nontrivial task, which is still being investigated. waveguides have been proposed [4]. 21 A promising strategy to further improve the 22 Concerning optofluidic resonators, there are two main − 23 light matter interaction is to increase the duration of objectives to address in order to improve their 24 the interaction by placing the sample instead of target performances and effectiveness. The first is to improve 25 inside an optical cavity performing a high quality the quality factor of the resonator by optimizing the 26 factor Q. Thus far, various types of optical resonators different optical elements (waveguide, bend regions, 27 have been demonstrated in optofluidic format. coupler) as also discussed above. The second is to 28 Optofluidic resonators in the form of planar ring have provide a fine manipulation of the sample fluid by 29 been obtained using SWs [5] and ARROWs [6] (Fig. integrating suitable microfluidic systems. From this 30 2), but also in the form of Fabry-Pérot (FP) [7]. Each point of view, polymer materials offer several 31 type of implementation has its own suitable advantages for microfluidics like reduced cost and 32 application. Optofluidic resonators based on slot and simplicity of fabrication. For this reason, hybrid 33 hollow core waveguides, despite having moderate Q- silicon/polymer approaches appear as an optimal 34 factors (up to 104), open up interesting perspectives for 35 design strategy [6],[10], requiring, however, still 36 planar integration. Nowadays the increase of the Q- considerable fabrication efforts to fully merge silicon 37 factor still remains a challenging task due to the leaky photonics and microfluidics. nature of the interference based optofluidic waveguides 38 Concluding Remarks and Perspectives 39 and the problem associated with fabrication roughness 40 into the nanometer-void of the SWs. The availability of In perspective, optofluidic waveguides have the 41 high Q-factor resonators could represent a paradigm potentiality to play the same role as solid-core 42 shift from passive to active sensors based on lasing waveguides do for conventional integrated photonic 43 action. Recent results have demonstrated that circuits. Promising performances have been 44 optofluidic lasers could result in a promising way demonstrated through the implementation of 45 towards ultrasensitive detection [7]. innovative waveguide layouts and high-yield 46 fabrication processes. Ongoing progress is also focused 47 Advances in Science and Technology to Meet to extend the range of applicability to the mid–IR range 48 Challenges 49 in order to address new applications including gas 50 The need to reduce the waveguides losses requires a sensing. twofold approach. First, the development and the 51 The tasks of improving their performance and the 52 optimization of new materials and refined fabrication processes. The proper selection of the cladding integration on a chip are nowadays the main challenges 53 that could be addressed in the near future by further 54 materials and deposition processes has proven to development of state-of-the art modelling and 55 effectively reduce waveguides losses. Second, the 56 development of novel waveguide configurations able manufacturing. By taking advantage of recent 57 to overcome the limitations of existing designs. Atomic innovative polymeric manufacturing techniques, hybrid 58 Layer Deposition (ALD) has been demonstrated to be a silicon/polymer approaches will pave the way to the 59 powerful tool to reduce loss in ARROWs [8], but also integration of additional functionalities on chip. 60 to realize novel slot waveguides [9]. New waveguide Moreover, polymer technologies could open the layouts are needed to extend the working range of prospect of high performance and low cost all-plastic optofluidic waveguides from vis-IR to mid-IR region optofluidic microsystems, also suited to disposable that can offer new integrated solutions to overcome the applications. AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 32 of 59

1 2 3 4 5 Acknowledgments – The authors acknowledge the 6 financial support from the Italian Minister of 7 University and Research (MIUR) through the Futuro in 8 Ricerca (FIR) Program under Grant RBFR122KL1. 9 10 References 11 [1] Testa G, Persichetti G and Bernini R 2015 12 Optofluidic approaches for enhanced microsensor 13 performances Sensors 15 465–484. 14 [2] Schmidt H and Hawkins A R 2010 Atomic 15 spectroscopy and quantum optics in hollow-core 16 waveguide Laser & Photon Rev 4 720–737. 17 [3] Schmidt H and Hawkins AR 2008 Optofluidic 18 waveguides: I. Concepts and implementations 19 Microfluid Nanofluid 4 3–16. 20 [4] Lin P T, Kwok S W, Lin H-Y G, Singh V, 21 Kimerling L C, Whitesides G M and Agarwal A 22 2014 Mid-Infrared Spectrometer Using Opto- 23 Nanofluidic Slot-Waveguide for Label-Free On- 24 Chip Chemical Sensing Nano Letters 14, 231–238 25 [5] Barrios C A, Bañuls M J, González-Pedro V, 26 Gylfason K B, Sánchez B, Griol A, Maquieira A, 27 Sohlström H, Holgado M and Casquel R 2008 28 Label-free optical biosensing with slot-waveguides 29 Opt Lett 33 708–710. 30 31 [6] Testa G, Collini C, Lorenzelli L and Bernini R 2016 32 Planar Silicon-Polydimethylsiloxane optofluidic 33 ring resonator sensors IEEE Photon Tech Lett 28 34 155–158. 35 [7] Fan X and Yun S-H 2014 The potential of 36 optofluidic biolasers Nature Methods 11 141–147. 37 [8] Testa G, Huang Y, Zeni L, Sarro P M and Bernini R 38 2010 Liquid Core ARROW Waveguides by Atomic 39 Layer Deposition IEEE Photon Tech Lett 22 616– 40 618. 41 [9] Häyrinen M, Roussey M, Säynätjoki A, Kuittinen M 42 and Honkanen S 2015 Titanium dioxide slot 43 waveguides for visible wavelength Appl Opt 54 44 2653–2657. 45 [10] Testa G, Persichetti G, Sarro P M and Bernini R 46 2014 A hybrid silicon-PDMS optofluidic platform 47 for sensing applications Biomed Opt Express 5 417– 48 426. 49 50 51 52 53 54 55 56 57 58 59 60 Page 33 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 11. High-content physical phenotyping of 5 biological objects with microfluidic dual- 6 beam laser traps – Jochen Guck 7 Technische Universität Dresden 8 9 Relevance of the topic 10 11 Dual-beam laser traps (DBLTs) were the first kind of 12 optical trap realized by Arthur Ashkin and predate 13 optical tweezers by two decades. Micron-sized 14 dielectric objects, such as biological cells, are trapped Figure 1 Principle of microfluidic dual-beam laser 15 by the combined scattering and gradient forces of two – 16 traps. Cells are serially trapped between counter- non-focused, counter-propagating Gaussian laser 17 propagating laser beams in a microfluidic channel. 18 beams (see Figure 1). The main advantages of DBLTs 19 are that (1) they can be simply realized by aligning two [1], or their local mass density, which could be 20 optical waveguides, such as single-mode optical fibers, important for phase separations and phase transitions in 21 (2) they can be combined with arbitrary imaging cells. The choice of laser wavelength and total power 22 optics, and (3) the lack of focusing permits the use of also determines water absorption and with that the 23 high laser powers without cell damage (‘opticution’). temperature in the volume of the trapped object [3]. By 24 intentional use of this laser heating, additional 25 The latter aspect enables the controlled and non- thermodynamic properties of the trapped object can be 26 destructive deformation of cells. When used for this 27 purpose, DBLTs are called optical stretchers [7]. The investigated [2]. DBLTs can also be combined with 28 cells’ mechanical properties measured in this way have other photonic techniques such as Raman and Brillouin 29 proven an important label-free marker of cell function, spectroscopy and microscopy for additional chemical 30 related to cancer progression, differentiation, cell and mechanical information. Ideally, all these 31 cycle, infection etc. [4, 6]. The simplicity of the trap properties should be resolved in 3D inside of cells with 32 allows the straight-forward integration into diffraction-limited resolution so that either fast point- 33 microfluidic lab-on-chip environments [5] (see also scanning, light-sheet or tomographic approaches 34 Section 2). Microfluidic optical stretchers (µOS), thus, become relevant. And finally, the investigation does 35 constitute a convenient technique to measure cell not need to remain limited to entire cells. Also other, 36 mechanics at rates up to 10 cells/min, which compares often rather delicate, biological objects can be gently 37 handled, manipulated and investigated by optical 38 favorably with other standard ways of measuring cell forces in combination with photonic techniques. In this 39 mechanics, such as AFM, micropipette aspiration, 40 where 10 cells/hour are more typical. context, vesicles, isolated and artificial cell nuclei, mitotic spindles, or protein droplets come to mind. 41 However, with the recent advent of really high- 42 throughput (1,000 cells/sec) microfluidic techniques Current Status and Challenges 43 for the characterization of mechanical phenotypes that 44 In order to realize the potential of DBLTs to deliver 45 approach the analysis rates of conventional, high-content physical information about cells, as 46 fluorescence-based flow cytometers [9], the high- outlined above, and considering the state-of-the-art in 47 throughput aspect of the µOS has lost some of its the field, there are a number of challenges to be met. 48 former appeal. Instead, the current focus is shifting to 49 the high-content investigation of cells, and also of The optical trapping itself [7], the extraction of detailed 50 other biological objects, which are temporarily, but viscoelastic properties [4], and the microfluidic 51 arbitrarily long immobilized until several different integration [5] has been well elaborated. The 52 physical parameters important for cell function have temperature increase in the trap was until recently 53 been determined in detail. For example, full coupled directly to the amount of power used, since 54 mechanical characterization, in particular steady-state DBLTs were conveniently realized with Ytterbium- 55 viscosity, which is particularly important to relate to doped fiber-lasers operating at 1064 nm. Here, 56 cell migratory processes [4], takes at least several ∆T = 13 K/W [3]. By the recent advent of alternative 57 high-power fiber-based lasers, such as frequency- 58 seconds per cell, and can inherently not be obtained in 59 high-throughput, where the measurement time per cell doubled Erbium lasers, also 780 nm sources are now 60 is in the milliseconds range. available, which reduces the temperature increase 8- fold and permits the decoupling of optically induced Other physical characteristics of interest are the stress and temperature [2]. refractive index (RI) of cells, which can be used to track cell differentiation and other functional changes AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 34 of 59

1 2 3 The combination of DBLTs with many different 4 5 conventional imaging techniques, such as phase 6 contrast microscopy, is straight-forward. For 7 fluorescence imaging, the limited usability of high-NA 8 objectives — due to the significant distance between 9 objective and trapped object, caused by the diameter of 10 the optical fiber and the mounting on thick microscope 11 slides — restricts the spatial resolution and light 12 gathering ability of weaker fluorescence signals. While 13 Ca2+ dyes and nuclear dyes can be readily imaged [2], 14 sparser fluorescent signals from GFP-proteins or 15 fainter signals from fluorescently labeled surface 16 markers pose considerable challenges at present. 17 Figure 2 – Cell rotation in an optical cell rotator can be 18 For quantitative phase imaging, often digital combined with any arbitrary microscopic or spectro- 19 holographic microscopes are used. Their integration scopic technique for full physical characterization. 20 with DBTLs, set up on a commercial microscope 21 stand, is less straight-forward especially when a translational and rotational stability of the object in the 22 separate reference beam needs to be implemented. 23 trap. From such quantitative phase information, assuming 24 spherical symmetry, which is often given for cells in Advances in Science and Technology to Meet 25 Challenges 26 suspension, a 2D integrated RI distribution and an 27 overall average refractive index can be calculated [10]. The challenge of high-quality fluorescence imaging in 28 In order to obtain the complete 3D distribution of RI, DBLTs can be addressed by an optimized integration 29 and mass density, tomographic reconstruction has to be into specific lab-on-chip setups with the specific 30 used. For this, phase images need to be obtained from consideration of the optical quality along the imaging 31 multiple angles. Here, the versatility of DBLTs offers axis and the minimization of the distance between 32 at least two attractive solutions: (1) either cells can be objective and trapped object [5] (see also Section 2). 33 trapped by a conventional DBLT with two counter- The use of more sensitive and faster fluorescence 34 propagating Gaussian laser beams with cylindrical cameras, and an improved translational stability of the 35 rotational symmetry and then rotated by fluid flow in trap (see also below) will improve detection of weak 36 the microfluidic channel or (2) at least one of the 37 signals and spatial resolution. The problem with 38 Gaussian laser beams is replaced by a laser beam with incorporating quantitative phase microscopy can be 39 higher-order modes that break the rotational symmetry, solved by in-line interferometry, such as quadrilateral 40 which leads to a preferential rotational alignment of the shearing interferometers. These are now commercially 41 cell in the trap. The rotation of the laser modes then available, and can be mounted onto microscope stands 42 leads to the rotation of the trapped cell around the just like other cameras. 43 optical axis of the laser trap – and through the focal The challenge of sufficient translational and rotational 44 plane of the imaging microscope. When used for this 45 latter purpose, DBLTs are also referred to as optical stability of DBLTs can be approached by fast feedback loops, involving real-time analysis of the position and 46 cell rotators [8] (see Figure 2). An important challenge 47 in this context is the translational and rotational 3D RI distribution of the trapped object and appropriate shaping of the light fields and their 48 stability of the trap, currently limiting the spatial 49 intensity using adaptive optics through the trapping resolution of any imaging. 50 fibers. The creation of complex light fields through 51 While Raman microscopy has been researched and multi-mode fibers using transfer-matrix operators has 52 described extensively, Brillouin microscopy of recently been demonstrated. This could also be used 53 biological cells has only recently been introduced. It for illuminating cells with a light-sheet through the 54 promises to yield additional mechanical information trapping fibers for improved fluorescence or even 55 via interrogation of the local phononic spectrum inside Brillouin microscopy. The problem with long 56 cells. The extraction of actual mechanical properties 57 integration times in Brillouin microscopy can 58 requires knowledge of the local density and RI, which potentially be ameliorated by stimulated Brillouin 59 can be provided by the simultaneous RI tomography as microscopy, and the use of line- or even light sheet- 60 described above. A significant challenge is the small imaging. frequency shift away from Rayleigh scattering and the very low intensity of the Brillouin signal, which leads Concluding Remarks and Perspectives to long integration times, amplifying the problem of Page 35 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 In summary, microfluidic DBLTs are ideal “optical [9] Otto O, Rosendahl Ph, Mietke A, Golfier S, Herold 4 Ch, Klaue D, Girardo S, Pagliara S, Ekpenyong A, 5 work benches” where cells can be temporarily held Jacobi A, Wobus M, Töpfner N, Keyser UF, Mansfeld 6 stationary, interrogated by a number of different J, Fischer-Friedrich E, and Guck J. 2015. Real-time 7 photonic techniques for their complete physical 8 phenotyping and then conveniently exchanged and deformability cytometry: on-the-fly cell mechanical 9 sorted by the microfluidic integration. Even though phenotyping. Nat. Methods 12 199–202. 10 their high-throughput mechanical phenotyping appeal [10] Schürmann M, Scholze J, Müller P, Chan CJ, 11 has faded with the recent advent of techniques that are Ekpenyong AE, Chalut KJ and Guck J. 2015. 12 100,000 times faster, their true potential lies in the Refractive index measurements of single, spherical 13 high-content characterization and is still to be realized. cells using digital holographic microscopy. Meth Cell Biol 125 143–159. 14 The best times of microfluidic DBLTs are yet to come. 15 16 Acknowledgments – The author would like to thank 17 R. Ananthakrishnan, L. Boyde, K. Chalut, C.J. Chan, 18 G. Cojoc, U. Delabre, A. Ekpenyong, Ch. Faigle, J. 19 Käs, M. Kreysing, F. Lautenschläger, B. Lincoln, P. 20 Müller, S. Rönicke, S. Schinkinger, M. Schürmann, K. 21 Travis, G. Whyte, and F. Wottawah for their essential 22 contributions to the development of microfluidic 23 DBLTs. Recent technological developments have been 24 25 funded by the ERC Starting Investigator Grant 26 “LightTouch” (#282060). The artistic illustrations are 27 provided by Effigos GmbH, Leipzig. 28 29 30 References 31 [1] Chalut KJ, Ekpenyong AE, Clegg WL, Melhuish IC 32 and Guck J. 2012. Quantifying cellular differentiation 33 by physical phenotype using digital holographic 34 microscopy. Integrative Biology. 4 280–284. 35 [2] Chan CJ, Whyte G, Boyde L, Salbreux G and Guck J. 36 2014. Impact of heating on passive and active 37 biomechanics of suspended cells. Interface Focus. 4 38 20130069–20130069. 39 [3] Ebert S, Travis K, Lincoln B and Guck J. 2007. 40 Fluorescence ratio thermometry in a microfluidic 41 dual-beam laser trap. Optics Express. 15 15493– 42 15499. 43 [4] Ekpenyong AE, Whyte G, Chalut KJ, Pagliara S, 44 Lautenschläger F, Fiddler C, Paschke S, Keyser UF, 45 Chilvers ER and Guck J. 2012. Viscoelastic Properties 46 of Differentiating blood cells are fate- and function- 47 dependent. PLoS ONE 7 e45237. 48 49 [5] Faigle C, Lautenschläger F, Whyte G, Homewood P, 50 Martin-Badosa E and Guck J. 2015. A monolithic 51 glass chip for active single-cell sorting based on 52 mechanical phenotyping. Lab Chip 15 1267–1275. 53 [6] Guck J and Chilvers ER. 2013. Mechanics meets 54 medicine. Sci Transl Med 5 212fs41. 55 [7] Guck J, Ananthakrishnan R, Mahmood H, Moon TJ, 56 Cunningham CC and Käs J. 2001. The optical 57 stretcher: a novel laser tool to micromanipulate cells. 58 Biophys J 81 767–784. 59 [8] Kreysing M, Ott D, Schmidberger MJ, Otto O, 60 Schürmann M, Martin-Badosa E, Whyte G and Guck J. 2014. Dynamic operation of optical fibres beyond the single-mode regime facilitates the orientation of biological cells. Nat Commun 5 5481–5486. AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 36 of 59

1 2 3 4 12. Integration of fiber-based tweezers and 5 microfluidic systems – Carlo Liberale 6 King Abdullah University of Science and Technology 7 (KAUST) 8 9 Relevance of the topic 10 11 Over the last decades, microfluidic technologies have 12 demonstrated their potential for the realization of lab- 13 on-chip (LoC) systems. LoC devices present 14 integration and miniaturization of various probes and 15 devices, along with parallelization of functions and 16 analyses, which allow cost-effectiveness (e.g. thanks to 17 low consumption of samples and chemicals) and 18 automated high throughput, that are particularly 19 beneficial to biomedical applications. In this context, 20 21 considerable interest is attracted by implementing Optical Tweezers (OT) in LoC devices, to support Figure 1 – (a) cross-section of the FOT probe presented 22 in [1]: beams propagating in two symmetrically 23 functions like precise positioning of samples within the micro-channel, single cell manipulation and sorting. positioned fibres are reflected at the interface between 24 microprisms and the outer medium; (b) isometric view of 25 Optical Tweezers are based on opto-mechanical forces, the FOT; (c) red blood cell trapped by the FOT in 26 originated by exchange of momentum between photons hypotonic solution; (d) picture of a microfluidic chip 27 and matter, that are exerted by a strongly focused where the FOT has been inserted. Adapted with 28 optical beam on micro- or nano-particles [2]. They permission from [1]. 29 have been used in many applications in biology and devices could integrate, besides optical trapping, more 30 physics, for a wide variety of experiments ranging advanced functions as imaging and/or enhanced 31 from trapping and manipulation of live bacteria and 32 spectroscopies with plasmonic probes. 33 cells, to cooling, trapping and manipulation of single 34 atoms [2]. Important results have been achieved also Current Status and Challenges 35 by combining OT with other optical techniques, such At the state of the art, three-dimensional optical traps 36 as fluorescence and Raman spectroscopy. Early OT are mainly implemented inside microfluidic devices by 37 setups were based on tight focusing of a laser beam by using three different approaches: 1) with high 38 means of high Numerical Aperture (NA) microscope Numerical Aperture (NA) microscope objective-based 39 objectives (usually NA>1)[2]. In order to overcome setups; 2) by two counter-propagating beams, delivered 40 issues that arise in this scheme when trapping particles 41 by two optical waveguides (which could be optical within thick samples or in turbid media, an effort to fibers) positioned on opposite sides of the microfluidic 42 implement OT based on optical fibers has been in place 43 channel; 3) by single optical fibers. Naturally only the since early 1990’s. However, the realization of a fiber- two latter methods bring capability for integration into 44 based OT (FOT) forming three-dimensional optical 45 microfluidic chips, and are then compatible with the 46 traps has proven to be a challenging task, mainly development of portable LoC devices. Optical 47 because of the low NA of optical fibers. Indeed three- Tweezers based on two opposed waveguides are also 48 dimensional optical trapping was initially demonstrated called “optical stretchers” and are mainly used to apply 49 only with configurations made of at least two separated forces that deform the trapped particle, to probe its 50 and opposing fibers. The early-proposed schemes mechanical properties [3]. This particular type of 51 based on a single fiber were essentially capable of only integrated OT is considered in Section 11 to this 52 two-dimensional trapping, therefore were unable to roadmap article. 53 stably trap particles in the middle of a fluid, or had 54 Integrated OT based on a single optical fiber, able to impractically small trapping distances, making it generate a three-dimensional optical trap far enough 55 difficult to trap a particle without physical contact. 56 from the fiber tip to avoid contact with the sample, are Only with more recent advances in technologies for currently possible only by the use of specialty fibers 57 microstructuring of the fiber tip it has become possible 58 (bundles, multicore or annular core fibers) along with 59 to design and fabricate OT based on single optical advanced methods to modify the fiber tip to overcome 60 fibers. Fiber-based OT have recently been its intrinsic low NA [4]. Remarkably, the continuous demonstrated, which present ready compatibility with progress made in the capability to structure the fiber tip microfluidic channels and are best suited towards has enabled ever more effective geometries for optical realization of fully integrated microfluidic chips. trapping. Indeed, different fiber tip modification Future developments of fiber-based OT for LoC Page 37 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 methods have been proposed so far. Initially, tapering envisioned, as an extremely powerful tool for rapid and 4 flexible prototyping of high optical quality elements, 5 of a standard single mode optical fiber to a very small 6 tip size with the heating and pulling method, resulted with dimensions that range from few to hundreds of 7 in a FOT with a very small trapping distance [5]. More microns. In particular, a development of a new design 8 recently, high-precision fabrication techniques were that further reduce the size of single fiber FOT will 9 used to implement an approach proposed in [6], where improve compatibility with small microfluidic 10 light beams having an annular arrangement (e.g. channels and will allow an easier integration of other 11 generated by an annular core fiber or carried by the fibers, e.g. a multimode fiber for imaging, and/or other 12 different fibers of a bundle) are deflected by Total structures, e.g. for enhanced spectroscopy, within the 13 Internal Reflection (TIR) towards a common crossing FOT probe, to obtain an ultra-compact multimodal tool 14 point, yielding the equivalent effect of a high-NA for LoC devices. Another important advance will be 15 microscope objective, as needed for OT (figure 1(a)- the combination with other technologies that could 16 (b)). The TIR deflecting structures have been enable new functions, as the use of Spatial Light 17 Modulators for fiber imaging [10] or towards creation 18 fabricated on the fiber tip by Focused Ion Beam 19 milling [6] or, more recently, by Two-Photon- of dynamical multiple traps, which will also allow 20 Lithography (TPL) (figure 1(c)) [1]. The latter manipulation of the trapped particle. fabrication method is significantly easier, more flexible 21 Concluding Remarks and Perspectives 22 and cost-effective than the former. The use of a fiber 23 bundle instead of an annular fiber allows specialization Integration of Optical Tweezers in microfluidic 24 of different subsets of fibers for optical trapping and channels will enable new functions in LoC devices. 25 for simultaneous probing of the trapped sample (e.g. Optical fibers offer an excellent platform for 26 with fluorescence). A similar scheme, based on fiber miniaturization of OT, so that they can be embedded in 27 annular core distribution and TIR-based deflection, has microfluidic circuits. Indeed single-fiber FOT have 28 been realized also by the grinding an polishing been recently reported, based on specialty fibers and 29 technique [7]. TIR structures, and inserted in a microfluidic channel. 30 Future developments will concern: further 31 As shown in [1], the single-fiber based OT can be 32 integrated into a microfluidic device (figure 1(d)), in a miniaturization of the FOT; addition of manipulation 33 vertical configuration that would allow to insert and multi-trap capabilities; integration of new 34 multiple FOT at different locations within the same functionalities, like imaging and enhanced 35 LoC device without interfering with the microfluidic spectroscopies, to the FOT probe. 36 channel network. Trapping within microfluidic 37 channels has also been lately demonstrated with a 38 tapered fiber creating a two-dimensional trap, under Acknowledgments – The author would like to thank 39 specific flow rate conditions, [8] and with a tapered the following current and former colleagues for their 40 fiber for three-dimensional trapping but with the valuable contributions in the joint work on the subject: 41 particle in contact with the tip [9]. Francesca Bragheri, Patrizio Candeloro, Gheorghe 42 A few challenges still remain towards the goal of Cojoc, Ilaria Cristiani, Enzo Di Fabrizio, Lorenzo 43 Ferrara, Paolo Minzioni, Gerardo Perozziello and 44 bringing full functionalities of well-developed “bulk” 45 OT in LoC devices. A first challenge is to allow all Vijayakumar P. Rajamanickam. 46 optical manipulation (translation in three dimensions, 47 rotation) of trapped samples without moving the FOT 48 probe. A second challenge is the creation of multiple References 49 optical traps with a single FOT within a microfluidic 50 channel. Further challenges will be the addition of [1] Liberale C, Cojoc G, Bragheri F, Minzioni P, Perozziello 51 imaging capabilities to the FOT probe, e.g. by G, La Rocca R, Ferrara L, Rajamanickam V, Di Fabrizio E 52 integrating multimode fibers [10], and of enhanced and Cristiani I 2013 Integrated microfluidic device for 53 single-cell trapping and spectroscopy Scientific Reports probing/sensing capabilities, e.g. with electrodes or 3:1258 54 plasmonic based structures. 55 [2] Ashkin A, Optical trapping and manipulation of neutral 56 Advances in Science and Technology to Meet particles using lasers. World Scientific: Singapore ; 57 Challenges Hackensack, NJ, 2006; p xxiv, 915 p. 58 [3] Yang T, Bragheri F and Minzioni P 2016 A 59 The addition of more advanced functionalities to single Comprehensive Review of Optical Stretcher for Cell 60 fiber FOT, for integrated LoC devices, will be possible Mechanical Characterization at Single-Cell Level with the progress on design and fabrication of complex Micromachines 7:5 photonic structures on the tip of optical fibers. In this [4] Ribeiro R S R, Soppera O, Oliva A G, Guerreiro A and regard, a key role of the TPL fabrication method is Jorge P A S 2015 New Trends on Optical Fiber Tweezers J Lightwave Technol 33 3394-3405 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 38 of 59

1 2 3 [5] Liu Z H, Guo C K, Yang J and Yuan L B 2006 Tapered 4 fiber optical tweezers for microscopic particle trapping: 5 fabrication and application Optics Express 14 12510-12516 6 [6] Liberale C, Minzioni P, Bragheri F, De Angelis F, Di 7 Fabrizio E and Cristiani I 2007 Miniaturized all-fibre probe 8 for three-dimensional optical trapping and manipulation 9 Nature Photonics 1 723-727 10 [7] Zhang Y, Liu Z H, Yang J and Yuan L B 2012 A non- 11 contact single optical fiber multi-optical tweezers probe: 12 Design and fabrication Optics Communications 285 4068- 13 4071 14 [8] Gong Y, Zhang C L, Liu Q F, Wu Y, Wu H J, Rao Y J 15 and Peng G D 2015 Optofluidic tunable manipulation of 16 microparticles by integrating graded-index fiber taper with a 17 microcavity Optics Express 23 3762-3769 18 [9] Xin H B, Li Y Y, Li L S, Xu R and Li B J 2013 19 Optofluidic manipulation of Escherichia coli in a 20 microfluidic channel using an abruptly tapered optical fiber 21 Applied Physics Letters 103:033703 22 [10] Morales-Delgado E E, Psaltis D and Moser C 2015 23 Two-photon imaging through a multimode fiber Optics 24 Express 23 32158-32170 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 39 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 13. Acoustic prefocusing in optofluidic 5 systems – Kirstine Berg-Sørensen 6 Technical University of Denmark 7 8 Relevance of the topic 9 10 The ability of acoustic fields to position small particles 11 was discussed already by Ernst Florens Friedrich 12 Chladni (1756-1827), and is familiar to many scholars 13 through the demonstration experiment named “The 14 Chladni plate”. When carried over to dimensions of 15 typical microfluidic systems, the resonance 16 phenomenon nicely illustrated by the Chladni plate 17 experiment, implies typical frequencies in the MHz 18 range. Many research papers have in recent years Figure 1 – Microfluidic all-glass device with a liquid 19 channel of a rectangular cross section, the channel is 100 20 demonstrated the ability of the effect, termed µm wide; details may be found in [3]. Shown are images 21 acoustophoresis, e.g. to sort cells according to their of 5 µm diameter polystyrene beads in water, visualized 22 acoustic properties; readers unfamiliar with the topic by standard bright field microscopy. Laser-written 23 may consult the tutorial series of papers in Lab-on- waveguides in the glass chip were in the focal plane; 24 Chip [1]. The second part of the topic, optofluidic bright spots indicate scattered laser light coupled into 25 systems, is the term currently used for microfluidic two opposing waveguides. (a) Acoustic vertical focusing 26 systems with embedded optical elements, be it a micro- turned off. (b) Acoustic vertical focusing turned on. 27 laser, or a passive or active optical element; the Scattering by the trapping laser light on the beads 28 relevance of optofluidic systems at large is illustrates nicely the acoustic prefocusing of the beads to 29 demonstrated by the collection of papers in the current the plane of the waveguides. 30 roadmap on Optofluidics. The benefits of a 31 combination of the effects of acoustic and optical ability to construct micro- and optofluidic systems in 32 manipulation has been illustrated by the ability for polymer materials since micro- and even nanofluidic 33 systems may be produced by large-scale production 34 active particle sorting [2], improved trapping efficiency in an optical stretcher setup [3] and access to techniques such as injection molding or roll-to-roll 35 imprinting rather than the expensive and time- 36 complementary mechanical characteristics for the same 37 single cancer cell [4]. Acoustic prefocusing in an consuming clean-room processes required for the 38 optical stretcher setup is illustrated in Figure 1. The production of a single microfluidic system in glass [7]. 39 results of references [2-4] suggest that further advances In addition, the polymer material can be chosen with 40 in the field may find both technological and biomedical properties that match well both optical and biomedical 41 applications. requests. Unfortunately, acoustofluidics with plastic 42 materials remain challenging due to the relatively low 43 Current Status and Challenges sound speed and acoustic impedance in most of the 44 The combination of acoustofluidics and optical transparent plastics. This implies that acoustic energy 45 manipulation is, to the best of the author’s knowledge, may be transferred to, and lost in, the device itself, and 46 that knowledge of the conditions for acoustic 47 only applied in very few labs [2-4], with promising recent advances involving high-resolution imaging [5]. resonance, leading to large acoustic forces, requires 48 detailed 3D modelling of the pressure fields in the 49 As discussed below, one may very well imagine 50 inclusion of other characterization tools based on entire device [8]. Such detailed models are in contrast 51 optics. to simple 1D models in which one assumes a large 52 degree of reflection of the sound wave at the interface On the side, an interesting development is the 53 between the liquid in the microfluidic channel and the 54 recent introduction of the concept of iso-acoustic hard walls of the glass or silicon microfluidic device. 55 properties, whereby the understanding and control in

56 acoustic focusing and sorting has been tremendously 57 widened [6]. A current challenge for the technology to 58 be more accessible for applications is the choice of 59 material for the construction of the microfluidic As discussed in section 3, lithium niobate offers an 60 systems. The successful demonstrations of combined attractive alternative choice of material.. In [5], a acousto- and optofluidics [2-5] are based on glass – a transparent transducer made out of LiNbO3 with hard and highly resistant material with excellent optical indium tin oxide (ITO) electrodes was applied, in a properties. Yet, much current research explores the layered device. The experimental setup was AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 40 of 59

1 2 3 demonstrated to be able to provide precise 4 5 determination of both optical and acoustic forces on 6 trapped silica microspheres. 7 Delivery or collection of light for optical manipulation 8 or characterization by optical means is another point to 9 decide on. One option is to have the light delivered by 10 embedded optical fibers [9], or waveguides directly 11 written into the microfluidic device [3-4], suitable for 12 13 optical traps based on two counter propagating laser 14 beams, or for collection of emitted light, e.g. for Figure 2 – Simple sketch of a future device for 15 Raman spectroscopy of the sample. Another option, simultaneous acoustic prefocusing, optical manipulation 16 relevant for single beam optical traps, is to rely on light and spectroscopic characterization. 17 delivery and light collection through an external 18 microscope objective [2, 5]. When this latter option is acoustic transducer, but capacitive micromachined 19 chosen, one should be aware that the acoustic force ultrasound transducers, which may even be controlled 20 field may be altered due to the mechanical coupling by light, may be an interesting alternative [10]. 21 between the immersion oil to the microscope objective Concluding Remarks and Perspectives 22 and thus the entire microscope [5]. 23 As discussed, manipulation by both sound and light 24 Advances in Science and Technology to Meet fields in microfluidic systems adds versatility to the 25 Challenges optofluidics toolbox. The text concentrated on merging 26 tools of manipulation applying both optical and sound 27 As alluded to above, on the materials side, the 28 combination of acoustophoresis and optofluidics would waves, yet the perspectives include other uses of light, 29 clearly benefit from the development of transparent e.g., for spectroscopic characterization. One may 30 polymer materials with larger acoustic contrast foresee a combination of optical manipulation and/or 31 between device material and liquid – in addition to spectroscopy on cells with particular iso-acoustic 32 requests of good optical quality and high chemical properties, with the perspective to learn many more 33 resistance. As this may not be realistic, the second best details, including molecular, about the properties of 34 is to develop and apply 3D modelling tools [8] that such individual cells. 35 enables prediction of the sound wave propagation in 36 the entire experimental device and thereby may suggest 37 a design of the microfluidic device optimized for the Acknowledgments – 38 simultaneous action of acoustic and optical forces. The author acknowledges discussions over the years 39 with G. Nava, T. Yang, P. Minzioni, I. Cristiani, as 40 The choice of a polymer material for the microfluidic well as with A. Kristensen and with H. Bruus and his 41 device couples well with a desire to develop single use 42 students. Financial support from the Carlsberg devices. Waveguides produced by post-processing Foundation and COST MP1205 is highly appreciated. 43 steps like DUV exposure and/or other means to embed 44 45 optical elements in a polymer chip produced by mass- References 46 production techniques would clearly enhance the [1] Tutorial series on acoustofluidics (2011): 47 employability of optofluidic devices, with or without Acoustofluidics-exploiting ultrasonic standing wave 48 acoustophoresis. A challenge to resolve is, however, forces and acoustic streaming in microfluidic 49 the typical loss of optical power of order 0.5-0.7 systems for cell and particle manipulation Lab Chip 50 dB/mm in waveguides written by DUV or direct laser 11 3579-3580 51 writing, in addition to the coupling losses into a square [2] G Thalhammer, R Steiger, M Meinschad, M Hill, S 52 or rectangular waveguide. Alternatively, light coupling Bernet, and M Ritsch-Marte (2011): Combined 53 through standard optical fibers suffers much less from acoustic and optical trapping Biomed Opt Exp 2 54 loss of optical power, and still offers a competitive 2860-2870 55 solution although manual assembly is required. With [3] G Nava, F Bragheri, T Yang, P Minzioni, R 56 either choice of solution for light propagation in the Osellame, I Cristiani, and K Berg‑Sørensen (2015): 57 All‑silica microfluidic optical stretcher with 58 microfluidic chip, the dream device would be a single acoustophoretic prefocusing Microfluid Nanofluid 59 use polymer device that is to be mounted in a system or 60 a holder that includes efficient and well-aligned 19 837–844 coupling to light sources as well as an efficient [4] T Yang, F Bragheri, G Nava, I Chiodi, C Mondello, coupling to a sound transducer. In this respect, piezo- R Osellame, K Berg-Sørensen, I Cristiani, and P ceramic crystals may still be a wise choice for the Minzioni (2016): A comprehensive strategy for the Page 41 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 analysis of acoustic compressibility and optical 4 deformability on single cells Sci Rep 6 23946 5 [5] G Thalhammer, C McDougall, MP MacDonald, and 6 M Ritsch-Marte (2016): Acoustic force mapping in a 7 8 hybrid acoustic-optical micromanipulation device 9 supporting high resolution optical imaging Lab Chip 10 16 1523-1532 11 [6] P Augustsson, JT Karlsen, H-W Su, H Bruus, and J 12 Voldman (2016): Iso-acoustic focusing of cells for 13 size-insensitive acousto-mechanical phenotyping 14 Nature Comm 7 11556 15 [7] H Becker and LE Locascio (2002): Polymer 16 microfluidic devices Talanta 56 267–287 17 [8] J Dual and D Möller (2012): Acoustofluidics 4: 18 Piezoelectricity and application in the excitation of 19 acoustic fields for ultrasonic particle manipulation 20 Lab Chip 12 506-514 21 [9] B Lyncoln, S Schinkinger, K Travis, F Wottawah, S 22 Ebert, F Sauer, and J Guck (2007): Reconfigurable 23 microfluidic integration of a dual-beam laser trap 24 with biomedical applications Biomed Microdevices 25 DOI 10.1007/s10544-007-9079-x 26 [10] B T Khuri-Yakub and Ö Oralkan (2011): Capacitive 27 micromachined ultrasonic transducers for medical 28 imaging and therapy J. Micromech Microeng. 21 29 054404 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 42 of 59

1 2 3 4 14. Optofluidics for single-cell protein 5 analysis – Jian Chen 6 Institute of Electronics, Chinese Academy of Sciences 7 8 Relevance of the topic 9 10 Proteins (i.e., macromolecules composed of chains of 11 12 amino acid residues) perform key functions within 13 organisms, including regulation of metabolic reactions, 14 DNA replication, stimuli responsiveness, and molecule transportations [1]. As key components of biological 15 Figure 1 – Variations in single-cell protein expressions 16 organisms, quantities and activities of proteins have are closely correlated with tumour biology, 17 been regarded as the most important indicators of microbiology, cellular differentiation and pharmaceutical 18 biological activities, which are closely related to 19 phenomena such as cellular differentiation, neuron types: (1) optofluidic fluorescent flow cytometry; (2) 20 transmission and disease progression [2]. droplet based optofluidic flow cytometry; (3) large- 21 array micro wells (microengraving) and (4) large-array 22 Within the last few decades, a large number of micro chambers (barcode micro chips). In this section, 23 characterization approaches (e.g., gel electrophoresis, we examine the advantages and limitations of each 24 technique and discuss future research opportunities by 25 immunoassay, chromatography and mass spectrometry) focusing on two key parameters: absolute 26 have been developed to estimate the quantities and 27 activities of proteins [3]. These results have quantification and throughput. 28 contributed to the rapid developments of biological and 29 medical sciences, which have significantly promoted Current Status and Challenges 30 the detection capabilities in clinical diagnosis [4]. 31 However, these approaches are only capable of In optofluidics, the fluorescent micro flow cytometry is 32 quantifying protein expressions from a large number of the first approach to quantify single-cell protein 33 cells, which neglects the fact that most of the expressions where single cells are mixed with 34 organisms are composed of cells with variations in fluorescence labeled antibodies are flushed into 35 their functionalities and thus protein expressions at the microfabricated flow channels with fluorescent 36 37 population level may mask key underlying intensities measured [9] (see Figure 2(a)). As a 38 mechanisms due to cellular heterogeneity. miniaturized version of conventional fluorescent flow 39 cytometry, the micro flow cytometry is featured with 40 Currently, fluorescent flow cytometry is the golden low sample requirement of cells and reagents. 41 standard in the field of single-cell protein analysis However, this approach cannot quantify the copy 42 where single cells mixed with fluorescence labeled number of intracellular proteins, due to the lack of 43 antibodies are flushed into a fluorescent measuring effective calibration approaches. 44 domain composed of laser excitation and 45 photomultiplier tube based fluorescent detection [5]. In the second approach, droplet based optofluidic flow 46 Based on this methodology, variations in single-cell cytometry was proposed where single cells, 47 48 protein expressions are closely correlated with tumor functionalized capture beads and fluorescence-labeled 49 biology, microbiology, cellular differentiation and drug secondary antibodies are encapsulated in droplets, 50 development (see Figure 1). However, fluorescent flow where micro beads bind cytokines secreted by single 51 cytometry can only provide an absolute quantitation of cells, which further bind fluorescence-labeled 52 surface proteins of single cells, while it cannot quantify secondary antibodies, leading to fluorescent signals 53 intracellular proteins due to the lack of calibration [10] (see Figure 2(b)). 54 approaches, and thus it cannot translate raw fluorescent 55 signals into copy numbers of intracellular proteins [6]. Compared to conventional fluorescent flow cytometry, 56 this approach can effectively assay secreted proteins at 57 Optofluidics is the science and technology on the the single cell line. However, due to the use of 58 processing and optical detection of small amounts of cytokine-capture beads, there is an uneven distribution 59 -9 -18 60 fluids (10 to 10 liters) in channels with dimensions of fluorescence within individual droplets (intensity of tens of micrometers [7]. Since the dimensions of peaks around individual beads), which poses obstacles optofluidics are comparable to biological cells, it has in calibration and thus the secreted proteins cannot be been used for the analysis of single-cell protein effectively quantified. expressions [8], which can be classified into four major Page 43 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 In the third approach, large-array micro wells 4 5 (“microengraving”) were proposed to isolate individual 6 cells with copy numbers of secreted proteins quantified 7 [11]. As shown in Figure 2(c), single cells suspended 8 in media are deposited into a large array of microwells, 9 which are then sealed by a glass slide coated with a 10 specific capture reagent. After certain periods of 11 incubation, the glass slide is removed and further 12 interrogated with laser-based fluorescence scanners. 13 Although microengraving enables absolute 14 quantification of proteins secreted by individual cells, 15 compared to flow cytometry, it suffers from 16 complicated processes and limited throughput 17 18 19 In the fourth approach, large-array micro chambers 20 (single-cell barcoding microchips) were proposed to 21 assay both cytosolic and membrane proteins of single 22 cells [12]. As shown in Figure 2(d), the single-cell 23 barcoding microchips consist of thousands of 24 individually addressed micro chambers for single-cell 25 trapping, followed by cell lysis, capture of targeted 26 proteins by pre-printed antibodies on the surface of the 27 chambers, which are further measured using 28 immunosandwich assays. This approach is the most 29 Figure 2 – Key developments of optofluidic systems 30 powerful optofluidic system in the field of single-cell enabling high-throughput single-cell protein 31 protein analysis since it enables the absolute characterization, which includes (a) optofluidic 32 quantification of both surface and cytosolic proteins of fluorescent flow cytometry; (b) droplet based optofluidic 33 single cells. However, since this approach relies on the flow cytometry; (c) large-array micro wells 34 confinement of single cells within chambers, it cannot (microengraving) and (d) large-array micro chambers 35 be easily scaled up to further increase the throughput. (barcode micro chips). 36 Concluding Remarks and Perspectives 37 Advances in Science and Technology to Meet 38 Challenges 39 In this study, we summarize key developments of 40 optofluidic platforms enabling the quantification of 41 Absolute quantification is a key requirement for single- single-cell proteins. Although significant 42 cell protein assays since, without the capabilities of improvements have been made within the last decade, 43 absolute quantification, the protein levels measured by an ideal tool of single-cell protein quantification 44 different approaches cannot be effectively compared. featured with absolute quantification and high 45 From this perspective, new calibration approaches have throughput is still not available. 46 to be developed in fluorescent micro flow cytometry, 47 enabling the translation of raw fluorescent signals to Acknowledgments –This work was supported by the 48 protein copy numbers. 49 National Basic Research Program of China (973 50 Program, Grant No. 2014CB744600), National Natural 51 Throughput is also a key factor in single-cell protein Science Foundation of China (Grant No. 61431019, 52 quantification. Although optofluidic large-array 61671430), Key Projects of Chinese Academy of 53 devices are featured with parallel analysis of single Sciences (QYZDB-SSW-JSC011), Natural Science 54 cells, which are capable of processing hundreds of Foundation of Beijing (4152056), Instrument 55 single cells within one experiment, they still cannot be Development Program of the Chinese Academy of 56 used to process cell samples with heterogeneity (e.g., Sciences, and Beijing NOVA Program of Science and 57 tumor samples with at least 1 million cells) due to their Technology. 58 limited throughputs. Thus, tremendous efforts have to 59 be devoted to this field with the purpose of further References 60 scaling up these assays. As to micro flow cytometry, further technical improvements may focus on the 1. Kim, M.S., et al., A draft map of the human proteome. extension of this serial approach to function in parallel. Nature, 2014. 509(7502): p. 575-81. 2. Savage, N., Proteomics: High-protein research. Nature, AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 44 of 59

1 2 3 2015. 527(7576): p. S6-7. 4 3. Phizicky, E., et al., Protein analysis on a proteomic 5 scale. Nature, 2003. 422(6928): p. 208-15. 6 4. Hanash, S., Disease proteomics. Nature, 2003. 7 8 422(6928): p. 226-32. 9 5. Janes, M.R. and C. Rommel, Next-generation flow 10 cytometry. Nature Biotechnology, 2011. 29(7): p. 602-4. 11 6. Zenger, V.E., et al., Quantitative flow cytometry: inter- 12 laboratory variation. Cytometry, 1998. 33(2): p. 138-45. 13 7. Psaltis, D., S.R. Quake, and C. Yang, Developing 14 optofluidic technology through the fusion of 15 microfluidics and optics. Nature, 2006. 442(7101): p. 16 381-6. 17 8. Fan, B., et al., Development of Microfluidic Systems 18 Enabling High-Throughput Single-Cell Protein 19 Characterization. Sensors, 2016. 16(2): p. 232. 20 9. Chan, S.D., et al., Cytometric analysis of protein 21 expression and apoptosis in human primary cells with a 22 novel microfluidic chip-based system. Cytometry A, 23 2003. 55(2): p. 119-25. 24 10. Huebner, A., et al., Quantitative detection of protein 25 expression in single cells using droplet microfluidics. 26 Chemical Communications, 2007(12): p. 1218-1220. 27 11. Han, Q., et al., Polyfunctional responses by human T 28 cells result from sequential release of cytokines. 29 Proceedings of the National Academy of Sciences of the 30 United States of America, 2012. 109(5): p. 1607-12. 31 12. Ma, C., et al., A clinical microchip for evaluation of 32 single immune cells reveals high functional heterogeneity 33 in phenotypically similar T cells. Nature Medicine, 2011. 34 17(6): p. 738-43. 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 45 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 a-chip to identify genomic deletions or insertions 4 15. Optofluidics for DNA analysis – Markus associated with genetic illnesses, such as breast cancer 5 Pollnau 6 KTH−Royal Institute of Technology or anemia, critically depends on the detection of single 7 base-pair insertions or deletions from DNA fragments 8 Relevance of the topic in the diagnostically relevant range of 150−1000 base- 9 pairs. The sizing accuracy is related to DNA plug 10 Capillary electrophoresis (CE) is a powerful method formation and depends on many factors, including the 11 for biomolecule separation and analysis. The sorting range of DNA fragment sizes to be separated, the 12 and sizing of DNA molecules within the human choice of―and the changes in―the sieving gel matrix 13 genome project [1] has been enabled largely by CE inside the microfluidic channel and its inner-wall 14 separation and analysis. The human genome project coating, temperature and actuation voltage, as well as 15 has also lead to the genetic mapping of various human the choice of microfluidic channel cross section, dye 16 illnesses [2]. The traditional techniques, e.g. using a label and, finally, the calibration procedure. A number 17 bulk glass capillary filled with agarose gel [3], provide 18 of empirical studies have been performed to shed light a high separation resolution, but typically require on this complex situation, but results are sometimes 19 rather long analysis times and bulky instrumentation. 20 not reproducible and existing models are typically 21 These issues are addressed by the development of inaccurate and unable to predict device performance. 22 microchip CE. Integration of a lab on a chip, thereby Advances in Science and Technology to Meet 23 ensuring compactness and introducing the potential for Challenges 24 mass fabrication, paves the way for device portability 25 and in-field or point-of-care applications. Furthermore, With the general advancement in micro-fabrication, 26 DNA sequencing by microchip CE allows for cheap, significant technical progress has been made also in 27 high-speed analysis of low reagent volumes. In recent merging microfluidics and micro-optics in an 28 years on-chip integration of DNA sequencing [4] and optofluidic chip. For example, post-processing of 29 commercial microfluidic chips by femtosecond laser 30 genetic diagnostics [5] have become feasible. One of 31 its potential applications is the identification of writing of optical waveguides [7, 8] that can direct 32 genomic deletions or insertions associated with genetic laser light to the interaction point and/or collect 33 illnesses. fluorescence from the excited dye labels and deliver it 34 to the detection unit is a suitable method for optofluidic Current Status and Challenges 35 integration. This technique has the additional capability 36 Laser-induced fluorescence exploiting fluorescent dye of writing three-dimensional optical structures into the 37 labels is the most popular microchip CE monitoring microfluidic chip [9]. The light source and the 38 technique. Conventional instruments use confocal detection unit are preferably placed in a small portable 39 setups to focus the excitation light and to collect the 40 resulting fluorescence. Such schemes can provide high 41 sensitivity down to a detection limit of ~200 fM [6], 42 43 corresponding to merely eight molecules in the 44 excitation volume, making plausible the vision of (a) 45 going down to interrogating at the single-molecule 46 level. However, these schemes require accurate 47 mechanical alignment of the optics to the microfluidic 48 channels and are sensitive to mechanical vibrations and (b) 49 drifts. The need to use microchip CE in combination 50 with a massive benchtop instrument, such as an optical 51 microscope, frustrates many of the microchip CE 52 advantages, in particular it strongly limits device 53 portability and prevents in-field or point-of-care 54 applications. In the future optical components, 55 including light generation, transportation, interaction (c) (d) 56 Figure 1 – Multi-color fluorescence DNA analysis in an 57 with the reagents, collection and detection, need to be integrated, partly directly on the optofluidic chip. optofluidic chip: (a) Schematic of the optofluidic chip. 58 (b) Transient fluorescence from a molecule plug passing 59 Sizing accuracy poses an inherent challenge to CE and by the excitation waveguide (with background 60 particularly microchip CE and achieving sufficient and illumination outside the detection window) [13]. reliable sizing accuracy for the envisaged DNA Individual signals separated by Fourier analysis of (c) 12 analysis is probably the biggest hurdle to be taken. DNA molecules from a breast cancer gene and (d) 23 Application of CE-based DNA sequencing in a lab-on- DNA molecules from an anemia gene [11]. AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 46 of 59

1 2 3 instrument that also includes the steering unit for the 250 250 4 240 exp. 1, blue 240 exp. 1, red actuation voltages and to which the optofluidic chip 230 exp. 2, blue 230 exp. 2, red linear fit exp. 1 5 220 220 linear fit exp. 1 quadratic fit exp. 1 can be precisely attached. 210 210 quadratic fit exp. 1 6 200 200 Time (s) 190 Time (s) 190 7 180 The technique of multiplex ligation-dependent probe 180 8 170 170 amplification [10] allows for the simultaneous 160 160 150 150 9 4,8 5,0 5,2 5,4 5,6 5,8 6,0 6,2 4,8 5,0 5,2 5,4 5,6 5,8 6,0 6,2 extraction and individual end-labeling of DNA ln(base-pair size) ln(base-pair size) 10 fragments from independent human genomic segments. (a) (b) 11 4,0 4,0 In order to exploit this technique, parallel optical blue ref., same exp. blue ref., same exp. blue ref., diff. exp. blue ref., diff. exp. 12 red ref., same exp. red ref., same exp. ) ) 3,0 3 3,0 3 - red ref., diff. exp. - red ref., diff. exp.

processing of the prepared sample is required. A (10 (10 2 13 2 σ σ 14 demonstrated parallel optical processing method is 2,0 2,0 Variance Variance 15 modulation-frequency-encoded dual-wavelength laser 1,0 Variance 1,0 excitation, fluorescence detection with a single 0,0 0,0 16 blue sample, blue sample, red sample, red sample, blue sample, blue sample, red sample, red sample, exp. 1 exp. 2 exp. 1 exp. 2 exp. 1 exp. 2 exp. 1 exp. 2 17 ultrasensitive, albeit color-blind photomultiplier, and 18 Fourier analysis decoding [11, 12], see Fig. 1. To make (c) (c) 19 the laser excitation of dye labels selective, each label Figure 2 – Migration time (linear scale) versus base-pair 20 must absorb light in a different wavelength range. size (logarithmic scale) of (a) 12 blue-labeled and (b) 23 21 Typical dye labels exhibit rather large absorption red-labeled DNA molecules simultaneously migrated 22 bandwidths, thereby strongly limiting the number of and separated in experiment 1 (circles) and experiment 2 (squares) [14]. Linear and quadratic fits (solid lines) 23 exclusive labels that can be employed simultaneously. 24 shown only for experiment 1. Variance σ2 measured in 25 Consequently, the extension of dual-wavelength experiment 1 or 2 from (c) the linear fit and (d) the 26 analysis to a larger number of wavelengths faces the quadratic fit, obtained from the measured reference data 27 challenge of development of novel, narrow-excitation- from the same or the different experiment [14]. 28 band fluorescent labels. Alternatively, fluorescent 29 labels with spectrally overlapping absorption bands can reproduces the measured data significantly better than 30 be excited with a single laser, thereby reducing the a linear fit, indicating that the assumption of a simple 31 necessary infrastructure on the excitation side, and logarithmic dependence of migration time on base-pair 32 their individual fluorescence lines can be either size underlying the production of DNA ladders with a 33 detected by a color-sensitive detector array or logarithmic increase in base-pair size is not justified by 34 spectrally dispersed and then detected by a detector the experiment [14]. Nevertheless, the sizing accuracy 35 array. However, any complication on the detection side crucially depends on the applied law to fit the reference 36 that wastes fluorescence intensity or provides less data. At present, in the absense of a proper theoretical 37 understanding of DNA plug formation and flow 38 sensitivity than a photomultiplier impairs the detection limit. mechanism the choice of fit function is merely a best 39 guess and may even change with the underlying system 40 The dependence of sizing accuracy on the important parameters. 41 system parameters must be investigated more 42 systematically and its theoretical background needs to Choice of a suitable dye label, combined with 43 be improved. For example, recently it has been reference calibration and sample investigation by 44 fluorescent detection in consecutive experiments and 45 identified that the choice of fluorescent dye label can significantly influence the sizing accuracy and that a application of a best-guess fit function, has resulted in 46 capillary electrophoretic separation of fluorescent- 47 proper choice can largely improve the accuracy [14], 48 see Fig. 2, but the reason for this effect is unknown and labeled DNA molecules in the 150−1000 base-pair 49 systematic investigations are lacking. In the same work range with single-base-pair resolution [14], see Fig. 2, 50 it has been demonstrated that the calibration strategy thereby paving the way towards the detection of single 51 influences the sizing accuracy [14]. On the one hand, it base-pair insertion or deletion in a lab-on-a-chip with 52 may not matter much whether the optofluidic device is low reagent volumes in a few-minute experiment. 53 calibrated by flow of a known reference sample in the Concluding Remarks and Perspectives 54 same or in a separate experiment and whether the DNA 55 fragments of the reference sample are almost identical Through the efforts of many scientists in this research 56 or slightly different from those of the unknown sample area, in-field and point-of-care applications have 57 to be investigated. On the other hand, applying the become feasible, but a considerable amount of work 58 remains to be performed to better understand the 59 same fluorescent dye label to the two samples appears dependence of sizing accuracy on the various system 60 to be crucial. parameters by establishing a better theoretical Most importantly, the dependence of migration time on framework and to improve the measurement base-pair size needs to be understood theoretically. The repeatability and reliability. results displayed in Fig. 2 prove that a quadratic fit Page 47 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 Acknowledgments – The author gratefully [12] Alaverdian L, Alaverdian S, Bilenko O, 4 5 acknowledges collaboration with his co-workers C. Bogdanov I, Filippova E, Gavrilov D, Gorbovitski 6 Dongre and H. J. W. M. Hoekstra, the Politecnico di B, Gouzman M, Gudkov G, Domratchev S, 7 Milano, LioniX B.V. and Zebra Bioscience B.V. Kosobokova O, Lifshitz N, Luryi S, 8 Ruskovoloshin V, Stepoukhovitch A, 9 References Tcherevishnick M, Tyshko G and Gorfinkel V 10 [1] International Human Genome Sequencing 2002 A family of novel DNA sequencing 11 Consortium 2001 Initial sequencing and analysis instruments based on single-photon detection 12 of the human genome Nature 409 860-921 Electrophoresis 23 2804-2817 13 [2] Altshuler D, Daly MJ and Lander ES 2008 [13] Dongre C, van Weerd J, Besselink GAJ, van 14 Genetic mapping of human disease Science 322 Weeghel R, Martínez Vázquez R, Osellame R, 15 881-888 Cerullo G, Cretich M, Chiari M, Hoekstra HJWM 16 [3] Brody JR and Kern SE 2004 History and and Pollnau M 2010 High-resolution electrophoretic 17 principles of conductive media for standard DNA separation and integrated-waveguide excitation of 18 electrophoresis Anal. Biochem. 333 1-13. fluorescent DNA molecules in a lab on a chip 19 [4] Fredlake CP, Hert DG, Kan CW, Chiesl TN, Root Electrophoresis 31 2584-2588 20 [14] Pollnau M, Hammer M, Dongre C and Hoekstra 21 BE, Forster RE and Barron AE 2008 Ultrafast HJWM 2016 Combined microfluidic-optical DNA 22 DNA sequencing on a microchip by a hybrid analysis with single-base-pair sizing capability 23 separation mechanism that gives 600 bases in 6.5 Biomed. Opt. Express 7 5201-5207 24 minutes Proc. Natl. Acad. Sci. USA 105 476-481 25 [5] Easley CJ, Karlinsey JM, Bienvenue JM, 26 Legendre LG, Roper MG, Feldman SH, Hughes 27 MA, Hewlett EL, Merkel TJ, Ferrance JP and 28 Landers JP 2006 A fully integrated microfluidic 29 genetic analysis system with sample-in-answer- 30 out capability Proc. Natl. Acad. Sci. USA 103 31 32 19272-19277 33 [6] Dongre C, Pollnau M and Hoekstra HJWM 2011 34 All-numerical noise filtering of fluorescence 35 signals for achieving ultra-low limit of detection 36 in biomedical applications Analyst 136 1248-1251 37 [7] Martínez Vázquez R, Osellame R, Nolli D, Dongre 38 C, van den Vlekkert HH, Ramponi R, Pollnau M 39 and Cerullo G 2009 Integration of femtosecond laser 40 written optical waveguides in a lab-on-chip Lab 41 Chip 9 91-96 42 [8] Osellame R, Hoekstra HJWM, Cerullo G and 43 Pollnau M 2011 Femtosecond laser 44 microstructuring: an enabling tool for optofluidic 45 lab-on-chips Laser Photonics Rev. 5 442-463 46 [9] Crespi A, Gu Y, Ngamson B, Hoekstra HJWM, 47 Dongre C, Pollnau M, Ramponi R, van den Vlekkert 48 HH, Watts P, Cerullo G and Osellame R 2010 49 Three-dimensional Mach-Zehnder interferometer in 50 a microfluidic chip for spatially-resolved label-free 51 detection Lab Chip 10 1167-1173 52 [10] Schouten JP, McElgunn CJ, Waaijer R, 53 54 Zwiinenburg D, Diepvens F and Pals G 2002 55 Relative quantification of 40 nucleic acid 56 sequences by multiplex ligation-dependent probe 57 amplification Nucleic Acids Res. 30 57-69 58 [11] Dongre C, van Weerd J, Besselink GAJ, Martínez 59 Vázquez R, Osellame R, Cerullo G, van Weeghel R, 60 van den Vlekkert HH, Hoekstra HJWM and Pollnau M 2011 Modulation-frequency encoded multi-color fluorescent DNA analysis in an optofluidic chip Lab Chip 11 679-683 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 48 of 59

1 2 3 4 16. Nanoscale optofluidics – Sha Xiong and Ai- 5 Qun Liu control layer 6 Nanyang Technological University 7 nanofluidic flow layer

8 Relevance of the topic photonic layer 9 10 Over the past decades, optofluidic devices have been waveguide 11 extensively exploited for novel functions and

12 promising applications in chemical and biological H2O CaCl2 nanochannel 13 analysis, imaging and energy by combining the merits photonic crystal 14 of both fluids and optics in the microscale platforms [1, (a) 15 2]. Recently, researchers have shown great interest in

16 force scaling down the channel size and studying both the force 17 outlet fundamental and applied science on the nanoscale. 18 When the channel dimension is comparable with the 19 size of molecules, the fluid transport and molecular 20 biomolecules behaviors are different [3]. Up to now, the most 21 promising applications and phenomena related to 22 23 nanofluidic devices are based on the electrokinetic inlet nanochannels 24 effects, such as water desalination, fluidic diodes and nanochannel 25 energy harvesting. In contrast, the big potential of 26 nanofluidics has not been exploited in the realm of Figure 1 – Illustration of nanochannel-based optofluidic 27 optics. The functions of nanochannels in the devices. (a) Nanochannel bonded onto a planar photonic- 28 optofluidic devices are mainly in two aspects as shown crystal for optical tuning. Adapted with permission from Ref. [4] Copyright (2006) OSA. (b) Nanochannels for 29 in Fig. 1. First, the nanochannels provide a solution for biomolecule confinement. Adapted by permission from 30 local tunability over photonic circuits. Their sensitive 31 Macmillan Publishers Ltd: Nature Materials Ref. [7] response to the change of the refractive index can also copyright 2007. 32 be used for biosensing. Second, the nanochannels offer 33 effective confinement of biomolecules to facilitate the optical properties of the device by controlling fluid 34 biological analysis, especially for DNA related flows with various refractive indices. The nanochannels 35 can infiltrate the liquid into well-defined defects of a 36 genomic applications. Despite the great expectation of photonic crystal [4], or form a tunable grating by 37 expanding optofluidics on the nanoscale, it leaves 38 ample room not only for basic discovery but also for filling liquid into the nanochannel array [5]. 39 technical improvement in current directions. Introducing the nanofluidic modulation into the photonic circuits provides a powerful approach for 40 Current Status and Challenges 41 realizing compact and reconfigurable optical 42 Here, we discuss the state of art and highlight the functionalities with precise localized control. The 43 technical barriers impeding the development of major challenge for the tuning of photonic functions is 44 nanoscale optofluidics. the nanofluidic handling. The speed and volumetric 45 Although enormous progress has been made in the flow rate of the liquids transported through the 46 development of nanofabrication over recent years, nanochannel is extremely low. Conventional transport 47 fabricating well-defined nanochannels with low cost methods such as capillary filling, pressure drop or 48 electroosmosis are not effective. 49 and high throughput is still a challenge. The typical Another hot research topic on nanoscale 50 nanofabrication is completed in a well-equipped clean 51 room using a lithography process, such as electron- optofluidics is the analysis and detection of a single 52 beam lithography or focused-ion beam lithography. It molecule. Nanochannels are used to confine the sample 53 is usually expensive and laborious with a long volume to match the dimensions of the detection 54 processing time. Inspired by the soft lithography with volume, increase its concentration for optical detection, 55 PDMS molding, nano-imprint lithography is employed or do precise manipulation of individual molecules, 56 for large-scale fabrication of nanochannels. Although which enhance the sensitivity and resolution of the 57 the replication method improves the throughput, it has detection. Researchers have achieved length 58 limitations in fabricating the nano-molds. Researchers measurement, sizing and optical mapping of DNA by 59 are looking for alternative materials and techniques for stretching or separating the molecules in nanochannels 60 fabricating nanodevices, in order to make them [6]. In most cases, the detection depends on the commercially viable. fluorescent labeling, which is costly and time- Dynamic reconfiguration of photonic functions is consuming. Alternatively, scattering signals can also one of the representative topics of optofluidics. It tunes be used for detecting molecules in the nanochannels, Page 49 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 which, however, has stringent requirements on the 4 thin liquid film 5 optical detection system. Improving the detection limit interface 6 remains a bottleneck. Although new phenomena keep being discovered, 7 bubble 8 it is difficult to precisely probe the transport properties 9 on the nanoscale and get good control over the jet 10 nanoflow. The existing techniques are mostly based on 11 an electrical measurement. Optical detection may offer 12 alternative strategies, but have to overcome several

13 limitations in nanochannels, such as the influence of gas 14 Brownian motion and wall effects. pulsed laser 15 16 Advances in Science and Technology to Meet gas liquid droplet 17 Challenges 18 jet 10 μm 2 μs bubble 20 μs 200 μs 19 A revolution on the nanofabrication is expected, just 20 as the sprout out of microfluidics following the Figure 2 – Jetting caused by the laser at a gas/liquid 21 development of PDMS based soft lithography interface in the nanochannel. Adapted from Ref. [8] with 22 techniques. Recently, mechanical deformation of permission from the Royal Society of Chemistry.

23 elastomeric materials (e.g. wrinkling, cracking and 24 and offer sophisticated manipulation on individual 25 microchannel collapsing) was taken advantage of in molecules. 26 fabricating nanochannels. These techniques are cost- A better understanding of mass transport at the 27 effective and can adjust the channel size by controlling nanoscale is required to reach the full potential of 28 the stress applied on the materials [7]. Meanwhile, nanodevices. The next generation of cameras is 29 wide attention has been drawn towards bottom-up expected to have higher frame rate and resolution, 30 methods, in which the nano-structures are directly which is possible to improve the flow characterization 31 assembled by atoms and molecules. Novel materials in the nanochannels. Combining other imaging 32 like carbon nanotubes and graphene have shown their techniques such as plasmonics and nanophotonics hold 33 unique properties in nanofluidic transport. One could 34 great promise to further improve detection. Moreover, exploit the new emerging 3D printing technology to optical induced perturbation on the nanoscale can be an 35 fabricate nanochannels in the optofluidic system. 36 alternative strategy to gain new insight in the flow Optical actuation is a burgeoning method for fluid 37 dynamics. 38 control, which includes optical driven flow through the The integration of nanofluidics and optical 39 photo-thermal effect of nanoparticles, direct optical components in a fully automated system is a key point 40 manipulation of particles/structure in the fluid, and for transferring lab-based technologies into the real 41 laser-induced cavitation, etc. The optical driven market. Advanced techniques are expected to integrate 42 nanofluidic transport is insensitive to surface or the fluidic modulators, light source and optical 43 solution conditions. Even though all the dynamic detection system on the same miniaturized chip, as 44 processes slow down due to the dominating surface well as facilitate large-scale production with parallel 45 forces, it is still possible to generate a high-speed jet by nanochannels. 46 using a pulsed laser to drive flow as shown in Fig. 2 47 [8]. Concluding Remarks and Perspectives 48 Besides developing a new optical detection concept 49 Nanoscale optofluidics is an exciting frontier field, and optimizing the optical system, sorting and pre- 50 which carries the hope that new opportunities will 51 concentration of the targeted samples before detection emerge with the reducing scales. In the point of 52 is a feasible approach to improve the detection limit. application, genomic studies are still a hot direction. Its 53 Conventional approaches cannot precisely separate the potential in the detection and analysis of exosomes and 54 samples. A more robust solution is offered by viruses opens new windows in the diagnostics and 55 continuous flow separation of structures such as an environment monitoring. The advanced techniques for 56 entropic trap [9] and nanoscale Deterministic Lateral optical detection and manipulation should be taken 57 Displacement (DLD) pillar arrays [10], which advantage of to go deeper on the fundamental research 58 successfully demonstrated the sorting of DNA and of nanofluidics such as flow characterization and 59 exosomes. It can enhance the specificity of 60 precise control of flow on the nanoscale. Nanoscale biomolecules, and therefore increase the sensitivity and optofluidics has the potential impact to bring specificity of detection. Advanced techniques are breakthrough discoveries, new devices and functions expected to increase the throughput of the separation by studying the molecule level interaction.

AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 50 of 59

1 2 3 Acknowledgments – This work was supported by the 4 5 Singapore National Research Foundation under its 6 Environmental & Water Technologies Strategic 7 Research Programme (1102-IRIS-05-01), which is 8 administered by PUB. 9 10 References 11 [1] Monat C, Domachuk P and Eggleton B J 2007 12 Integrated Optofluidics: A new river of light Nature 13 Photonics 1 106-114. 14 [2] Yang Y, Liu A Q, Chin L K, Zhang X M, Tsai D P, 15 Lin C L, Lu C, Wang G P and Zheludev N I 2012 16 Optofluidic waveguide as a transformation optics 17 device for lightwave bending and manipulation 18 Nature Communications 3 651. 19 [3] Sparreboom W, van den Berg A and Eijkel J C T 20 2008 Principles and applications of nanofluidic 21 transport Nature Nanotechnology 4 713-720. 22 [4] Erickson D, Rockwood T, Emery T, Scherer A and 23 Psaltis D 2006 Nanofluidic tuning of photonic 24 crystal circuits Optics Letters 31 59-61. 25 [5] Yasui T, Ogawa K, Kaji N, Nilsson M, Ajiri T, 26 Tokeshi M, Hriike Y and Baba Y 2016 Label-free 27 detection of real-time DNA amplification using a 28 29 nanofluidic diffraction grating Scientific Reports 6 30 31642. 31 [6] Friedrich S M, Zec H C and Wang T H 2016 32 Analysis of single nucleic acid molecules in micro- 33 and nano-fluidics Lab on a Chip 16 790-811. 34 [7] Huh D, Mills K L, Zhu X, Burns M A, Thouless M 35 D and Takayama S 2007 Tuneable elastomeric 36 nanochannels for nanofluidic manipulation Nature 37 Materials 6 424-428. 38 [8] Xiong S, Chin L K, Ando K, Tandiono T, Liu A Q 39 and Ohl C D 2015 Droplet generation via a single 40 bubble transformation in a nanofluidic channel Lab 41 on a Chip 15 1451-1457. 42 [9] Fu J P, Schoch R B, Stevens A L, Tannenbaum S R 43 and Han J Y A 2007 A patterned anisotropic 44 nanofluidic sieving structure for continuous-flow 45 separation of DNA and proteins Nature 46 Nanotechnology 2 121-128. 47 [10] Wunsch B H, Smith J T, Gifford S M, Wang C, 48 Brink M, Bruce R L, Austin R H, Stolovitzky G and 49 Astier Y 2016 Nanoscale lateral displacement arrays 50 for the speration of exosmes and colloids down to 51 20 nm Nature Nanotechnology 11 936-940. 52 53 54 55 56 57 58 59 60 Page 51 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 17. Optofluidic immunoassays – Chia-Chann (a) Competitive 5 Shiue and Shih-Kang Fan 6 National Taiwan University 7 8 Relevance of the topic 9

10 Integrating and combining optics and microfluidics,

11 optofluidics [1] has been investigated to achieve

12 precise and adaptive photonic components that (b) Sandwich 13 generate, shape, route, switch and discriminate light for

14 integrated optical systems. With spatial and temporal

15 control of a fluid on a microscale, the optical properties

16 can be altered and monitored; this capability has been 17 developed further for imaging, light routing, bio- 18 sensors, energy and other purposes. For applications in 19 biological and chemical analysis, various mechanisms (c) Indirect 20 involving photonic detection have been studied in 21 integrated optofluidic systems [2], including metallic 22 nanohole array plasmonics, photonic crystals, photonic 23 24 crystal fibers, ring resonators, Mach-Zehnder interferometers, and Fabry-Perot cavities to sense the 25 refractive index that are employed as label-free 26 27 immunosensors. Beyond improving the performance of 28 photonic sensing, the precise and automated Figure 1 – Heterogeneous immunoassay configurations. 29 manipulations of fluids and suspended particles with coupled device (CCD) or CMOS image sensor. 30 microfluidic techniques [3] facilitate assay protocols Without labeled molecules, various optical 31 and enhance biomedical analyses. Among diverse immunoassays were demonstrated, such as surface 32 analyses, an optofluidic immunoassay [4] is an plasmon resonance (SPR), interferometers, ring 33 essential and common practice in biomedical study and 34 resonators, and photonic crystals [5]. Other non-optical in vitro diagnosis (IVD) to analyze a target analyte, for 35 means were achieved with various techniques, for 36 example proteins and small molecules, by exploiting example electrochemical, piezoelectric (quartz-crystal 37 the sensitivity, specificity and affinity of antibody- microbalance (QCM) and microcantilever), 38 antigen interactions. calorimetric methods and with field-effect transistors 39 Current Status and Challenges (FET). 40 41 Immunoassays are generally classified as Concurrently with advances of the antigen/antibody, 42 homogeneous or heterogeneous. In homogeneous labeled molecules of immunoassays and optical 43 immunoassays, antigens and antibodies are dispersed methods of detection, microfluidic devices drive fluids, 44 and interact in solutions, whereas in heterogeneous liquids or gas, typically along microchannels fabricated 45 immunoassays the target antigens or antibodies on silicon, glass or poly(dimethylsiloxane) (PDMS), to 46 suspended in the solution are detected with specific offer a miniaturized, automatic, rapid and cheap 47 antigens or antibodies immobilized on a solid support analysis with decreased consumption of reagents. 48 surface. Figure 1 shows various heterogeneous Hatch et al. investigated a homogeneous immunoassay 49 conducted along T-shaped microchannels by flowing 50 immunoassays that occur on a solid surface with appropriately immobilized and labeled antigen or two fluid streams containing antibody and labeled and 51 target antigen [6]; the diffusion and interaction 52 antibody molecules that have been developed. In a between the two streams under a laminar flow regime 53 competitive configuration, the sensing signal is 54 inversely proportional to the amount of target antibody with a small Reynolds number was monitored with a 55 present in the sample solution. In the non-competitive CCD to analyze the concentration of the target antigen. 56 manners of both sandwich and indirect configurations, Although the detection was performed within 1 min, 57 a detectable signal from the labeled antibody or the laminar flow maintained by the continuous flow 58 secondary antibody is proportional to the amount of kept consuming the samples and reagents during the 59 target antigen. With optical labels, the signal can be measurement. Depending on the tubing between the 60 detected with colorimetry, fluorescence, pump and chip, the dead volume might waste more photoluminescence (PL), chemiluminescence (CL) and sample and reagents. electrochemiluminescence (ECL) methods simply with Heterogeneous immunoassays with antigens or a photodiode, photomultiplier tube (PMT), charge- antibodies absorbed physically or immobilized AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 52 of 59

1 2 3 covalently on a microchannel surface have also been 4 5 studied. To simplify the fabrication of a microchannel, 6 and modification and immobilization of a surface, 7 conjugated microbeads, especially magnetic particles, 8 have been widely adopted for the development of 9 optofluidic heterogeneous immunoassays that have 10 been applied successfully to clinical samples; to 11 complete the immunoassay, repeated microfluidic

12 procedures, including mixing, incubation, and washing 13 of samples and reagents with precisely metered 14 volumes, are necessary but challenging for 15 microfluidic devices with fixed microchannels that are 16 not reconfigurable during operation. 17

18 Advances in Science and Technology to Meet 19 Challenges 20

21 An alternative means to perform optofluidic 22 immunoassay is based on digital microfluidics (DMF) Figure 2 – Parallel plate device configuration of a 23 that has demonstrated promising DNA analysis, drug digital microfluidic (DMF) platform performing 24 discovery and real-time biomolecular detection with magnetic bead-based heterogeneous optofluidic 25 precise, reprogrammable and individual control of immunoassay with electrowetting driven droplets and 26 multiple droplets. Among techniques of droplet precisely metered volumes. 27 actuation, electrowetting is the most efficient, and has 28 With the sandwich heterogeneous immunoassay been widely adopted in DMF devices composed of configuration using magnetic beads, the fluorescence 29 parallel plates free from issues of complicated 30 signal from the fluorophore labeled with secondary microchannel fabrication and dead volume. Figure 2 31 antibody was enhanced with appropriately collected 32 shows how droplets are generated, transported, mixed and gathered magnetic beads (diameter 6 μm) in a 33 and split between two plates of glass, silicon or printed shallow fluidic space (height 10 μm) between plates 34 circuit board (PCB) on applying electric signals to the [8]. The volume of the sample was 2.5 nL; the limit of 35 electrodes that are covered with a dielectric or detection (LOD) was 15 pg/mL with coefficient of 36 hydrophobic layer. For instance, Sista et al. reported a variability (CV) 3 %. Moreover, without the top plate a 37 heterogeneous immunoassay to detect insulin and sessile droplet typically performing as a liquid lens was 38 cytokine interleukin-6 (IL-6) on a DMF platform applied to focus the fluorescence on the detector and to 39 within 7 min [7]; a droplet of sample containing insulin enhance the detection sensitivity [9]. The detection was 40 or IL-6 was first mixed with the reagent droplet 41 performed in a compact and shielded light-proof box (7 containing magnetic beads coated with a primary × 7 × 7 cm3) with all components, achieving a 42 antibody, a secondary antibody labeled with alkaline 43 sensitive, cost-effective and portable optofluidic phosphate (ALP) and blocking proteins. The antibody- 44 bioassay system. 45 magnetic bead-antigen sandwich complex was then Concluding Remarks and Perspectives 46 collected and fixed on transporting the droplet near a 47 permanent magnet placed either above or beneath the To date, various microchannel-based and droplet-based 48 two plates for the following washing steps. The digital microfluidic techniques have been demonstrated 49 supernatant liquid was removed with droplet splitting; for homogeneous and heterogeneous optofluidic 50 one part of the split droplet containing unbound immunoassays. Label-free photonic sensors generally 51 components was discarded. The magnetic beads in the require advanced fabrication techniques and are less 52 remaining droplet were resuspended on mixing with an suitable for the application of disposable IVD medical 53 additional fresh washing buffer droplet. The washing 54 devices. DMF with modified magnetic beads to was repeated until all unbound components were perform bioassays with individually driven droplets 55 removed. The CL signal was eventually measured with 56 simplifies the fabrication and the antigen or antibody a PMT on mixing the droplet containing the antibody- 57 immobilization to realize optofluidic immunoassays for 58 magnetic bead-antigen complex with a droplet of CL clinical diagnostics, enabling the performance of robust 59 substrate, Lumigen APS-5. Such a DMF platform fluidic procedures with reliable droplet and magnetic- 60 capable of automated fluid actuations to conduct bead manipulation without volume variation, bead loss multiple assay protocols has been applied to multiplex or biofouling. The integration of detectors in the DMF newborn screening. platform [10] to eliminate optical alignment is essential to the development of disposable cartridges for IVD. Page 53 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 References 4 [1] Psaltis D, Quake S R and Yang C 2006 Developing 5 optofluidic technology through the fusion of 6 microfluidics and optics Nature 442 381-386 7 8 [2] Fan X and White I M 2011 Optofluidic 9 microsystems for chemical and biological analysis 10 Nat Photonics 5 591-597 11 [3] Sackmann E K, Fulton A L and Beebe D J 2014 The 12 present and future role of microfluidics in 13 biomedical research Nature 507 181-189 14 [4] He J-L, Wang D-S and Fan S-K 2016 Opto- 15 microfluidic immunosensors: from colorimetric to 16 plasmonic Micromachines 7 doi: 10.3390/ 17 mi7020029 18 [5] Fan X, White I M, Shopova S I, Zhu H, Suter J D 19 and Sun Y 2008 Sensitive optical biosensors for 20 unlabeled targets: A review Anal Chim Acta 620 8- 21 26 22 [6] Hatch A, Kamholz A E, Hawkins K R, Munson M 23 S, Schilling E A, Weigl B H and Yager P 2001 A 24 rapid diffusion immunoassay in a T-sensor Nat 25 Biotechnol 19 461-465 26 [7] Sista R S, Eckhardt A E, Srinivasan V, Pollack M 27 G, Palankib S and PamulaV K 2008 Heterogeneous 28 immunoassays using magnetic beads on a digital 29 microfluidic platform Lab Chip 8 2188-2196 30 [8] Huang C-Y, Shih P-H, Tsai P-Y, Lee I-C, Hsu H-Y, 31 Huang H-Y, Fan S-K and Hsu W2015 AMPFLUID: 32 aggregation magnified post-assay fluorescence for 33 ultrasensitive immunodetection on digital 34 microfluidics Proc IEEE 103 225-235 35 [9] Zeng X, Zhang K, Pan J, Chen G, Liu A-Q, Fan S-K 36 and Zhou J 2013 Chemiluminescence detector based 37 38 on a single planar transparent digital microfluidic 39 device Lab Chip 13 2714-2720 40 [10] Royal M W, Jokerst N M and Fair R B 2013 41 Droplet-based sensing: Optical microresonator 42 sensors embedded in digital electrowetting 43 microfluidics systems IEEE Sens J 13 4733-4742 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 54 of 59

1 2 3 least some form of liquid based sample processing 4 18. Optofluidics and point-of-need involved. Additionally, optical methods, like 5 diagnostics for precision medicine and florescence, have been used in lab-on-chip devices for 6 David Erickson global health – many years prior to the advent of the term 7 Cornell University Optofluidics. As such one could view the goal of the 8 9 field in this area over the last dozen years as examining Relevance of the topic 10 ways in which one might be able to: (1) use light and 11 Point-of-Care and Point-of-Need Diagnostics: Point- fluidics synergistically or (2) take advantage of 12 of-Care diagnostics refers, formally, to the ability to advanced photonic/optical methods in ways that lead to 13 identify the nature of an illness or health concern at the better: cost, reliability, deployment, accuracy or 14 location where care is provided. From the Engineer’s outcome advantages than would otherwise be possible. 15 or Technologist’s point of view this typically concerns 16 the development of devices or systems which can 17 Current Status and Challenges 18 provide actionable information to a healthcare provider 19 within a timeframe compatible with the likely patient- There are numerous excellent examples of Optofluidic 20 provider interaction. A variation on this, which we diagnostic technology including recent work on: 21 will focus on here, is point-of-need diagnostics. This diagnostics for concussion recovery [2], rapid single 22 differs from point-of-care in an important way, in that virus detection [3], and minimally invasive malaria 23 the best place for administering the diagnostic may not diagnostics [4]. Limitations on the length of this 24 be where patient care is provided. There are numerous article prevent providing a comprehensive review, so 25 examples of this including: mobile breast cancer rather I will provide two case studies in involving 26 screening programs, precision medicine where time precision medicine and global health. In the final 27 tracking of health markers requires both sample subsection I will point to some needed technological 28 collection and analysis to be done in a personalized 29 advancements that could have significant impact on the unit, and infectious diseases monitoring where cases 30 field. 31 can originate far away from a proper healthcare site 32 and preliminary screening can help ensure referral is Case Study I - Optofluidics for Precision Medicine 33 done properly. The technologies that enable these 34 diagnostics can exploit chemical, biological or physical The US National Institutes of Health defines Precision 35 measurements of health, be single point measurement Medicine as “an emerging approach for disease 36 or involve monitoring over time, or address acute or 37 chronic health issues. (A) NutriPhone 38 39 Optics and Optofluidics for Point-of-Care/Need 40 Diagnostics: Since the early days of what we now call 41 microfluidics, roughly the mid-1990s, point-of- 42 care/need diagnostics was seen as one of the key 43 applications. The concept was, and is, simple; in the 44 same way that modern integrated electronics enabled 45 mobile, portable computers, microfluidics was going to 46 put the power of a traditional analytical lab into a 47 physician’s office through the “Lab-on-a-Chip”. 48 49 When Optofluidics first emerged as a formal term 50 roughly 10 years later, in the mid-2000s, diagnostics 51 were still viewed as one of the major drivers. Indeed (B) KS-Detect 52 the seminal paper in the field by Psaltis et al. [1] lists 53 an anticipated market demand for the development of 54 “portable devices for environmental monitoring, 55 medical diagnostics and chemical-weapon detection” 56 57 as one of the key applications Optofluidics is trying to 58 realize. 59 60 The need for the “fluidics” part of Optofluidics in diagnostics is fairly obvious in that if one is taking, processing, and analyzing a biological sample (e.g. Figure 1 (A) NutriPhone – Personalized Nutrition Tracking and (b) KS-Detect – SolarThermal PCR. blood, sweat, saliva, biopsy, etc.) there is likely to be at Page 55 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 treatment and prevention that takes into account be able to be manufactured, serviced, distributed, and 4 operated in a way that is compatible with the existing 5 individual variability in environment, lifestyle and 6 genes for each person.” There are a number of ways in healthcare system. Since these requirements (including 7 which Optofluidics can play a number of roles in but not limited to price-point) vary from location to 8 precision medicine with the most relevant to this article location, it is key to involve local expertise as early on 9 being to enable the point-of-need diagnostics that can in the process as possible. enable the better, easier, and more accurate tracking of 10 An example of how we have used Optofluidics to 11 biomarkers at the individual level. Being able to fuse address some of these concerns is our KS-Detect 12 biomarker data with changes in behavior or system [9, 10] (see Figure 1B). The technology 13 environment is key to understanding how individuals focuses solar or other light through a shadow mask 14 react. 15 onto an optically absorbing microfluidic chip. The 16 Given the current worldwide near-ubiquity of mobile system is designed such that the projection of this light 17 technology, it is reasonable to pair these types of onto the chip creates a radial thermal pattern in steps 18 diagnostics with mobile technologies, such as mobile whose temperatures match those required to conduct a 19 phones and tablets [5]. As one example, our group and PCR reaction. The channels on the microfluidic chip 20 the Mehta group in Nutritional Sciences at Cornell are designed so as to move the sample through the 21 have been working on coupling Optofluidics based static temperature zones with the proper residence 22 micronutrient level tracking diagnostics with mobile time. This technique, coupled with the use of a mobile 23 phones. Our “NutriPhone”, see Figure 1(A), uses a phone to read and interpret the results, allows the 24 25 specially developed paper microfluidics card coupled device to be operated under conditions where power 26 with an Optofluidic reader system to enable individuals may not be available or reliable. This enables one to 27 to easily track how changes in diet and lifestyle can have more flexibility in performing the diagnosis at the 28 affect circulating levels of different micronutrients. true point-of-need rather than just the point-of-care. 29 Monitoring nutritional outcomes is a particularly good As can be seen in the above mentioned papers, we have 30 candidate for personalized medicine because responses demonstrated the technology for the diagnosis of 31 to changes in diet and lifestyle can be very Kaposi’s Sarcoma. 32 individualized and often difficult to quantify the Advances in Science and Technology to Meet 33 magnitude of changes a priori. To date we have 34 Challenges developed tests for vitamin D (25-OH-D3) [6], iron 35 (ferritin), vitamin B12 [7], and others. There are numerous opportunities for Optofluidic 36 technology in point-of-need diagnostics. The key 37 Case Study II - Optofluidics for Global Health technological and scientific advancements that will 38 39 Formally “Global Health” refers to the health of likely be required include: further and more seamless 40 everyone on the planet, independent of country of integration of Optofluidic systems with mobile 41 origin. When used in the context of technology technology, reduction or elimination of the need for 42 development, however, it often refers to systems for peripheral equipment needed to operate technology 43 use in so-called “limited resource settings” or places (e.g. tunable lasers), reduced material and fabrication 44 where access to reliable infrastructure ideally required costs for consumables to hit consumer-level or 45 to operate the technology may not be available. In internationally acceptable price points, and an 46 2012 our group published a paper broadly describing increased emphasis on the skill set of the end user 47 the Optofluidic opportunities within Global Health [8], resulting in lower complexity easier to use devices and 48 including point-of-care diagnostics. systems. 49 50 Numerous agencies and foundations have published Concluding Remarks and Perspectives 51 criteria for what is required for a successful point-of- A transformative change would be the ability to use 52 need diagnostic in limited resource settings. These Optofluidic or optical/spectroscopic methods to reduce 53 requirements vary slightly depending on the source but the invasiveness of diagnostics. This has proven 54 most include requirements for devices to be: operable difficult in the past due to challenges with accuracy 55 in locations with limited or no medical infrastructure, and high per-unit cost of the resulting instrumentation 56 simple to operate and interpret, and able to provide 57 and thus technologies that can address these two actionable information with fidelity similar to what one 58 challenges would be highly desirable. could expect with the state-of-the-art test. Coupled 59 with these technical requirements, there is an equally 60 Acknowledgments – DE acknowledges support for important need that the technology be coupled with a his efforts in the areas discussed above from the US sustainable business/deployment model. This means National Institutes of Health through grants R01-EB- that the system and any required consumables need to 021331 (FeverPhone) and UH2-CA-202723 (KS- AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 56 of 59

1 2 3 Detect) and the US National Science Foundation 4 5 through grants CBET-1343058 (Mobile Health) and 6 IIP-1430092 (NutriPhone). 7 8 The author declares the following competing financial 9 interests(s): DE has equity interest in VitaMe 10 Technologies Inc. which is commercializing 11 micronutrient diagnostic technology similar to that 12 described herein. 13 14 References 15 1. Psaltis, D., S.R. Quake, and C.H. Yang, Developing 16 optofluidic technology through the fusion of 17 microfluidics and optics. Nature, 2006. 442(7101): 18 p. 381-386. 19 2. Ko, J., et al., Smartphone-enabled optofluidic 20 exosome diagnostic for concussion recovery. 21 Scientific Reports, 2016. 6. 22 3. Ozcelik, D., et al., Optofluidic wavelength division 23 multiplexing for single-virus detection. Proceedings 24 of the National Academy of Sciences of the United 25 States of America, 2015. 112(42): p. 12933-12937. 26 4. Lukianova-Hleb, E.Y., et al., Hemozoin-generated 27 vapor nanobubbles for transdermal reagent- and 28 needle-free detection of malaria. Proceedings of the 29 National Academy of Sciences, 2014. 111(3): p. 30 900-905. 31 5. Erickson, D., et al., Smartphone technology can be 32 transformative to the deployment of lab-on-chip 33 diagnostics. Lab on a Chip, 2014. 14(17): p. 3159- 34 3164. 35 6. Lee, S., et al., A smartphone platform for the 36 quantification of vitamin D levels. Lab on a Chip, 37 2014. 38 7. Lee, S., et al., NutriPhone: a mobile platform for 39 low-cost point-of-care quantification of vitamin 40 B12 concentrations. Under review, 2016. 41 8. Chen, Y.-F., et al., Optofluidic opportunities in 42 global health, food, water and energy. Nanoscale, 43 2012. 4(16): p. 4839-4857. 44 9. Snodgrass, R., et al., KS-Detect – Validation of 45 Solar Thermal PCR for the Diagnosis of Kaposi’s 46 Sarcoma Using Pseudo-Biopsy Samples. Plos One, 47 2016. 11(1): p. e0147636. 48 10. Jiang, L., et al., Solar thermal polymerase chain 49 reaction for smartphone-assisted molecular 50 diagnostics. Sci. Rep., 2014. 4. 51 52 53 54 55 56 57 58 59 60 Page 57 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 4 19. Optofluidics in energy – David Sinton 5 University of Toronto 6 7 Relevance of the topic 8 With broad economic and environmental impacts, the 9 energy sector is at the forefront in the minds of policy 10 makers, technology developers and the general public. 11 The dominant demand for global energy usage is in fluids 12 – oil and gas – and the ultimate source of global energy is 13 Figure 1 – Breathable waveguides for combined light light – solar radiation. The tight integration of light and 14 and fluid delivery to photosynthetic microorganisms. fluids via optofluidics thus presents various opportunities 15 Reproduced from Pierobon et al. [10]. 16 for current and future energy technologies. The small- 17 scale of traditional optofluidics technologies, however, is combination of horizontal drilling and hydraulic 18 in stark contrast with the scale required for impact in the fracturing in shale and tight oil reservoirs. As a result, 19 energy sector. Thus, as with microfluidics for energy biofuel producers and technology developers are focusing 20 applications, the challenge here is to develop optofluidic on niche high-value chemical outputs, such as 21 technologies that can leverage the integration of light and neutraceuticals, in which optofluidics photobioreactors 22 fluids fundamental to the area, while enabling scaling of can better compete. Optofluidics has also enabled very 23 the technology. The two established routes to scale are: controlled multiplexed experiments to determine optimal 24 leveraging already-scaled materials such as fiber optics conditions for photosynthetic energy conversion in 25 (energy or energy conversion is the product); and operations of all scales [4], [5]. 26 employing optofluidics to inform processes at larger 27 scales (information is the product). Converting light energy into chemical energy using 28 photocatalysts is an alternative to the biological approach. 29 An early vision for the field of optofluidics for energy Integrating photocatalysts and fluid in optofluidic systems 30 was mapped in our perspective paper five years ago [1]. presents several attractive features, most notably 31 My focus here is the recent developments in these areas, enhanced transport of reactants to photocatalysts and 32 newly emerging energy applications and the roadmap products away, as well as leveraging the excellent control 33 ahead to broader application of optofluidics in energy. associated with optofluidics systems to ensure conditions 34 Specifically, I outline first recent progress and a roadmap throughout [6]. Analogous to the photosynthetic energy 35 for optofluidics in photosynthetic and photocatalytic conversion (above), however, there are challenges of both 36 energy conversion – examples of tight integration of cost and efficiency for optofluidics photocatalysis. 37 fluids and light – followed by emerging and hybrid Specifically, there is a mismatch between the goals of 38 approaches that leverage elements of the optofluidics light collection (favoring large areas) and conversion 39 toolbox for application in energy. In this short overview, 40 (favoring dense reactors). This mismatch can be I unfortunately cannot do justice to other exciting 41 technologically addressed by separately collecting and 42 developments in this arena such as liquid lensing for solar concentrating solar energy and feeding light to, and 43 energy collection, photothermochemical conversion, and distributing within, a dense reactor system using 44 solar steam generation, to name a few. waveguides. There are associated challenges of 45 minimizing losses and costs in this waveguiding approach, but there are exciting developments in this 46 Current Status and Challenges 47 area. 48 Photosynthesis is nature’s method of converting light 49 energy into chemical energy, and is well-suited to Separating light collection and reactor functions by using 50 optofluidics [1-2]. Most of the challenges in this approach conventional solar photovoltaics to collect and convert 51 stem from the fundamental low efficiency of the solar energy into electrical energy is another approach 52 photosynthetic process – effectively the overhead [7]. The result is not the tight integration of light and 53 associated with biological machinery in series with losses fluids germane to optofluidics, but the advantages are 54 of light energy at both the reactor scale and the organism worth considering. Most notably, this approach leverages 55 scale. Although these challenges can be overcome to the massive scaling and cost reductions already achieved 56 varying degrees with engineering, implementation in solar photovoltaics, as well as the ease with which 57 increases cost in an already-strained sector [3]. Fuel electrical output can be controlled and distributed. The 58 production represents the largest volume potential market fluids-half of the challenge is then in designing effective 59 for photobioreactor products, but it is also provides the electrochemical reactors for which there is a long history 60 lowest margin. The low price of conventional from fuel cells and electrolyzers. Electrocatalytic hydrocarbons made biofuels a challenging market, and conversion of CO2 into (or back into) useful chemical the economics became bleaker with the advent of feedstocks using solar-generated renewable electrons is unconventional oil and gas made possible by the an attractive approach that is gaining traction – AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195 Page 58 of 59

1 2 3 particularly with advancements in both nanostructure 4 electrocatalysts [8] and gas diffusion electrode cell 5 designs [9]. 6 7 8 Advances in Science and Technology to Meet 9 Challenges 10 11 In the area of optofluidics photobioreactors, a promising 12 approach is unifying light and fluid delivery within single 13 materials, such as light-guiding gas-permeable 14 membranes illustrated in Figure 1 [10]. This approach 15 removes the necessity for separate light/fluid 16 infrastructure and associated costs. The next logical step 17 in this progression of simplification might be to integrate 18 immobilized photosynthetic organisms within such 19 materials. Challenges of this approach are many, Figure 2 – Demonstration of wavelength-dependant 20 particularly as the one material must serve many separate light penetration in dense cultures. The productivity of 21 functions, but there is an opportunity for cost reductions green light was ~ 4x that of red due to better light 22 through the integration. A less exotic and more distribution of the weaker absorbed wavelength. 23 accessible approach would be to improve the optical Reproduced from Ooms et al. [ref] 24 function of existing medium- or large-scale facilities, by 25 for instance, providing excitation light that is off- photosynthesis accomplishes). Although these long-chain 26 absorption-peak. Providing light that matches the liquid hydrocarbons are the preferred energy currency 27 absorption peak of the photosynthetic organism is a worldwide, producing them from CO2 is a veritable, 28 common practice, which is grounded in the idea of multistep photo/electrochemical challenge. Particularly 29 feeding the organism the light it is tuned to absorb best. with the increase in activity in this area of late, there is 30 The result, however, is rapid absorption of light by the great potential for rapid advances in here short order. 31 culture and very poor light distribution resulting in poor 32 light use efficiencies overall. The penetration of various The role of optofluidic systems to study and inform on 33 photosynthetic and photocatatlyic processes is noted 34 wavelengths into identical cultures is shown in Figure 2. Illuminating cultures with wavelengths that are off-peak- above. We see additional, broader opportunities for this 35 approach here. Specifically, these same methods are very 36 absorbance, can allow much deeper penetration of the exciting light while maintaining photosynthetic well suited to quantify the environmental impacts of 37 anthropogenic CO emissions on organisms and 38 productivity and boosting overall performance [11]. 2 microcosms, in combination with light variables and/or a 39 variety of local stressors. One can imagine independently 40 Photocatalytic reactors have made impressive advances controlled optofluidic reactors as multiplexed micro- 41 over the last few years – both in the context of wastewater 42 treatment and light-to-chemical energy conversion. The greenhouses, or micro-ecosystems, that can provide the 43 photocatalysis community has the challenge of massive amount of data necessitated by fundamentally 44 engineering catalysts that efficiently manage both light complex, multivariable coupled interactions of the natural 45 utilization and chemical conversion while minimizing environment. 46 electron-hole recombination, bandgap, activity loss, Concluding Remarks and Perspectives 47 toxicity and material cost. Despite these fundamental 48 challenges there is much progress and reason for A major barrier to wider application of renewable energy 49 optimism in this area. Analogous to photosynthesis, there is the lack of storage. The optofluidic approaches 50 are also roles for optofluidics systems in informing both outlined here aim to overcome this barrier by converting 51 the synthesis of photocatalysts and how they might be – directly or in hybrid – our most plentiful energy source, 52 used in larger systems, such as reactors with disperse solar energy, into our most demanded energy form, liquid 53 photocatalytic nanoparticles [12]. hydrocarbons. Converting these concepts and 54 demonstrations into useful, scalable approaches will take 55 Hybrid methods that separate solar-PV and a sustained effort from the science and engineering 56 electrocatalytic conversion have challenges to address communities, as well as funding from policy makers and 57 particularly in the electrochemical reactions and reactors. venture capital. This short overview covered some recent 58 Electrocatalytically upgrading CO2 is energy intensive developments and opportunities in photosynthetic, 59 and advances are needed in catalysts and cells to increase photocatalytic and hybrid solar-PV-electrocatalysis 60 Faradaic efficiency for intended products, increase approaches. All three of these approaches have shared current densities, and reduce the cost of catalysts. A challenges of cost and present opportunities for both challenge shared with photocatalysis is the production of fundamental science and engineering advancement, the long-chain hydrocarbons from CO2 (something Page 59 of 59 AUTHOR SUBMITTED MANUSCRIPT - JOPT-104195

1 2 3 most promising of which will keep cost minimization as 2107–2115, 2015. 4 the guiding and motivating principle. 5 6 Acknowledgments – We gratefully acknowledge 7 support from the NSERC E.W.R. Steacie Memorial 8 Fellowship, Discovery program, MEET CREATE 9 program, and infrastructure funding from CFI. 10 11 References (separate from the two page limit) 12 [1] D. Erickson, D. Sinton, and D. Psaltis, “Optofluidics 13 for energy applications,” Nat. Photonics, vol. 5, no. 14 10, pp. 583–590, Sep. 2011. 15 [2] S. S. Ahsan, A. Gumus, A. Jain, L. T. Angenent, and 16 D. Erickson, “Integrated hollow fiber membranes for 17 gas delivery into optical waveguide based 18 photobioreactors,” Bioresour. Technol., vol. 192, pp. 19 845–849, 2015. 20 [3] M. D. Ooms, C. T. Dinh, E. H. Sargent, and D. 21 Sinton, “Photon management for augmented 22 photosynthesis,” Nat. Commun., vol. 7, p. 12699, 23 Sep. 2016. 24 [4] P. J. Graham, J. Riordon, and D. Sinton, 25 “Microalgae on display: A microfluidic pixel-based 26 irradiance assay for photosynthetic growth,” Lab 27 Chip, vol. 15, no. 15, pp. 3116–3124, 2015. 28 [5] B. Nguyen, P. J. Graham, and D. Sinton, “Dual 29 gradients of light intensity and nutrient 30 concentration for full-factorial mapping of 31 photosynthetic productivity,” Lab Chip, vol. 16, pp. 32 2785–2790, 2016. 33 [6] S. S. Ahsan, A. Gumus, and D. Erickson, “Redox 34 mediated photocatalytic water-splitting in 35 optofluidic microreactors,” Lab Chip, vol. 13, no. 3, 36 pp. 409–414, 2013. 37 [7] C. A. Rodriguez, M. A. Modestino, D. Psaltis, and 38 C. Moser, “Design and cost considerations for 39 practical solar- hydrogen generators,” Energy 40 Environ. Sci., vol. 7, pp. 3828–3835, 2014. 41 [8] M. Liu, Y. Pang, B. Zhang, P. De Luna, O. Voznyy, 42 J. Xu, X. Zheng, C. T. Dinh, F. Fan, C. Cao, F. P. G. 43 De Arquer, T. S. Safaei, A. Mepham, A. Klinkova, 44 E. Kumacheva, T. Filleter, D. Sinton, S. O. Kelley, 45 and E. H. Sargent, “Enhanced electrocatalytic CO2 46 reduction via field-induced reagent concentration,” 47 Nature, vol. 537, no. 7620, pp. 382–386, 2016. 48 [9] H. M. Jhong, S. Ma, and P. J. Kenis, 49 “Electrochemical conversion of CO2 to useful 50 chemicals: current status, remaining challenges, and 51 future opportunities,” Curr. Opin. Chem. Eng., vol. 52 2, no. 2, pp. 191–199, May 2013. 53 [10] S. C. Pierobon, J. Riordon, B. Nguyen, and D. 54 Sinton, “Breathable waveguides for combined light 55 and CO2 delivery to microalgae,” Bioresour. 56 Technol., vol. 209, pp. 391–396, Jun. 2016. 57 [11] M. D. Ooms, P. J. Graham, B. Nguyen, E. H. 58 Sargent, and D. Sinton, “Light dilution via 59 wavelength management for efficient high-density 60 photobioreactors,” Biotechnol. Bioeng., vol. xx, p. (under review), 2017. [12] T. Burdyny, J. Riordon, C.-T. Dinh, E. H. Sargent, and D. Sinton, “Self-assembled nanoparticle- stabilized photocatalytic reactors.,” Nanoscale, pp.

View publication stats