sensors

Article Pursuing the Limit with Nano-LED Scanning Transmission Optical Microscopy

Sergio Moreno 1,* , Joan Canals 1 , Victor Moro 1, Nil Franch 1, Anna Vilà 1,2 , Albert Romano-Rodriguez 1,2 , Joan Daniel Prades 1,2, Daria D. Bezshlyakh 3, Andreas Waag 3, Katarzyna Kluczyk-Korch 4,5 , Matthias Auf der Maur 4 , Aldo Di Carlo 4,6 , Sigurd Krieger 7, Silvana Geleff 7 and Angel Diéguez 1,2

1 Electronic and Biomedical Engineering Department, University of Barcelona, 08028 Barcelona, Spain; [email protected] (J.C.); [email protected] (V.M.); [email protected] (N.F.); [email protected] (A.V.); [email protected] (A.R.-R.); [email protected] (J.D.P.); [email protected] (A.D.) 2 Institute for Nanoscience and Nanotechnology-IN2UB, University of Barcelona, 08028 Barcelona, Spain 3 Institute of Semiconductor Technology, Technische Universität Braunschweig, 38106 Braunschweig, Germany; [email protected] (D.D.B.); [email protected] (A.W.) 4 Department of Electronic Engineering, University of Rome “Tor Vergara”, 00133 Roma, Italy; [email protected] (K.K.-K.); [email protected] (M.A.d.M.); [email protected] (A.D.C.) 5 Faculty of Physics, University of Warsaw, 00-662 Warsaw, Poland 6 CNR-ISM, 00128 Rome, Italy 7 Department of Pathology, Medical University of Vienna, 1210 Wien, Austria; [email protected] (S.K.); [email protected] (S.G.) * Correspondence: [email protected]   Abstract: Recent research into miniaturized illumination sources has prompted the development Citation: Moreno, S.; Canals, J.; of alternative microscopy techniques. Although they are still being explored, emerging nano-light- Moro, V.; Franch, N.; Vilà, A.; emitting-diode (nano-LED) technologies show promise in approaching the optical resolution limit in Romano-Rodriguez, A.; Prades, J.D.; a more feasible manner. This work presents the exploration of their capabilities with two different Bezshlyakh, D.D.; Waag, A.; prototypes. In the first version, a resolution of less than 1 µm was shown thanks to a prototype based Kluczyk-Korch, K.; et al. Pursuing the on an optically downscaled LED using an LED scanning transmission optical microscopy (STOM) Diffraction Limit with Nano-LED Scanning Transmission Optical technique. This research demonstrates how this technique can be used to improve STOM images by Microscopy. Sensors 2021, 21, 3305. oversampling the acquisition. The second STOM-based was fabricated with a 200 nm https://doi.org/10.3390/s21103305 GaN LED. This demonstrates the possibilities for the miniaturization of on-chip-based .

Academic Editor: Bruno Tiribilli Keywords: CMOS sensor; nano-LED; optical downscaling; nanopositioners; miniaturization

Received: 4 March 2021 Accepted: 5 May 2021 Published: 11 May 2021 1. Introduction Optical microscopy systems have been diversified over past centuries with a va- Publisher’s Note: MDPI stays neutral riety of techniques (phase contrast [1–3], dark field [4,5], confocal [6,7], fluorescence with regard to jurisdictional claims in microscopy [8–10], etc.) that have allowed for the easy observation of most processes published maps and institutional affil- and cellular structures down to a few tenths of a micrometer. For the exploration of objects iations. on a smaller scale, photonic systems have been hindered by the light diffraction limit, theorized by Ernst Abbe to be approximately 200 nm. More recently, some super-resolution (SR) techniques, i.e., techniques with a resolving power beyond the diffraction limit, such as stimulated emission depletion (STED) [11–13], Copyright: © 2021 by the authors. structured illumination microscopy (SIM) [14,15], stochastic optical reconstruction mi- Licensee MDPI, Basel, Switzerland. croscopy (STORM) [14,15], photo-activated localization microscopy (PALM) [16–18], or This article is an open access article scanning near-field microscopy (SNOM) [19], have demonstrated resolutions of as small as distributed under the terms and a few tenths of a nanometer. However, the overall complexity and cost of these imaging conditions of the Creative Commons systems have increased significantly, limiting the widespread use of some of the more Attribution (CC BY) license (https:// advanced optical imaging techniques beyond well-equipped laboratories. creativecommons.org/licenses/by/ 4.0/).

Sensors 2021, 21, 3305. https://doi.org/10.3390/s21103305 https://www.mdpi.com/journal/sensors Sensors 2021, 21, x FOR PEER REVIEW 2 of 15

Sensors 2021, 21, 3305 imaging systems have increased significantly, limiting the widespread use of some 2of of the 14 more advanced optical imaging techniques beyond well-equipped laboratories. Alternatively, cheaper and simpler techniques are being investigated. For instance, lenslessAlternatively, microscopy cheaper [20] takes and advantage simpler techniques of the large are field-of-view being investigated. (FOV) and For small instance, lenslesssize provided microscopy by large [20 commercial] takes advantage cameras of the(CMOS large or field-of-view CCD). Various (FOV) methods and small of lensless pixel sizeimaging provided can be by found large commercialtoday, whether cameras based (CMOS on shadowing or CCD). [21–23], Various fluorescence methods of lensless[24–26], imagingholography can [27–29], be found or today, 3D imaging whether [30,31]. based The on shadowingrefinement [of21 high-resolution–23], fluorescence (HR) [24 –tech-26], holographyniques and the [27 –improvement29], or 3D imaging of computational [30,31]. The resources refinement [32–34] of high-resolution have boosted (HR)the spatial tech- niquesresolution and below the improvement the pixel size of limitation. computational Thanks resources to techniques [32–34] such have as boosted pixel super-reso- the spatial resolutionlution (PSR), below based the on pixel the displacement size limitation. of the Thanks image to sensor techniques [35–37], such the as sample pixel super-or the resolutionlight source, (PSR), they based can reach on the resolutions displacement of lower of the than image 1 µm sensor [38]. [ 35However,–37], the the sample diffraction or the lightlimit source,remains they a barrier can reach for low-cost resolutions SR alternatives. of lower than 1 µm [38]. However, the diffraction limit remainsA novel amicroscopy barrier for low-costimaging SRtechnique alternatives. has emerged, which harnesses the key fea- turesA of novel lensless microscopy microscopy imaging and technique scanning hastechniques. emerged, As which in lensless harnesses microscopy, the key features LEDs ofare lensless the basic microscopy elements. andHowever, scanning instead techniques. of fixing As its in position, lensless microscopy,the LED is moved LEDs areto scan the basic elements. However, instead of fixing its position, the LED is moved to scan the the sample, as in scanning microscopes. This scan can be performed with one LED alone, sample, as in scanning microscopes. This scan can be performed with one LED alone, by moving it over the sample or by moving the sample over the LED; alternatively, its by moving it over the sample or by moving the sample over the LED; alternatively, its apparent position can be electronically controlled using the LEDs available in an array or apparent position can be electronically controlled using the LEDs available in an array or in an LED micro-display (see Figure 1). The other key component of the microscope is an in an LED micro-display (see Figure1). The other key component of the microscope is optical photodetector used to measure the light traversing the sample, and a camera can an optical photodetector used to measure the light traversing the sample, and a camera typically be used. The first prototype using this technique was already presented in [39], can typically be used. The first prototype using this technique was already presented where every LED in an 8 × 8 gallium nitride (GaN) array (5 µm in size and 10 µm in pitch) in [39], where every LED in an 8 × 8 gallium nitride (GaN) array (5 µm in size and was sequentially turned on and off, mapping the sample to prove the principle of opera- 10 µm in pitch) was sequentially turned on and off, mapping the sample to prove the tion of nano-illumination microscopy (NIM). The main features of this technique are that principle of operation of nano-illumination microscopy (NIM). The main features of this the resolution is given by the LED pitch and that the FOV is given by the LED array size technique are that the resolution is given by the LED pitch and that the FOV is given by thewhen LED the array sample size is when placed the in sample direct contact is placed with in directthe array. contact Therefore, with the the array. sensing Therefore, device theused sensing does not device play usedas important does not a role play as as in important conventional a role lensless as in microscopy. conventional NIM lensless only microscopy.needs two chips NIM and only benefits needs from two chips being and relati benefitsvely close from together. being relatively On the other close hand, together. LED Onminiaturization the other hand, is a LED very miniaturization active field of research, is a very activewhich field seems of research,to allow for which sizes seems smaller to allowthan Abbe’s for sizes diffraction smaller than limit Abbe’s [40–42]. diffraction In this context, limit [40 the–42 fabrication]. In this context, of new the compact fabrication low- ofcost new microscopes compact low-cost based on microscopes nano-LED basedscanning on nano-LEDtransmission scanning optical transmissionmicroscopy (STOM) optical microscopyis pursued, capable (STOM) of is sub-micron pursued, capable resolutions. of sub-micron With the decrease resolutions. in LED With size, the further decrease res- inolutions LED size, can furtherbe achieved resolutions [43]. Nevertheless can be achieved, high-resolution [43]. Nevertheless, and large-FOV high-resolution microscopes and large-FOVremain a still-distant microscopes prospect remain because a still-distant of the prospectresearch becauserequired, of which the research is limited required, to ad- whichvancements is limited in both to advancements directions. in both directions.

FigureFigure 1.1. SchematicSchematic diagramdiagram ofof electronicelectronic scanningscanning transmissiontransmission opticaloptical microscopy.microscopy. (a(a)) Perspective Perspective view. view. ( b(b)) Side Side view. view.

InIn thisthis work,work, thethe secondsecond andand thirdthird generationsgenerations ofof NIMNIM microscopesmicroscopes areare presented.presented. BothBoth areare basedbased onon [[39],39], wherewhere thethe proof-of-conceptproof-of-concept andand thethe characterizationcharacterization ofof thethe imagingimaging techniquetechnique were were described. described. The The first first microscope microscope presented presented was was assembled assembled with with a single a single 5 µm 5 LED,µm LED, which which was opticallywas optically downscaled downscaled with awith a tolens further to further investigate investigate the possibilities the possibili- of theties STOM of the STOM technique. technique. As this As approach this approach integrated integrated an optical an optical lens, thelens, resulting the resulting spot wasspot

Sensors 2021, 21, 3305 3 of 14

diffracted, creating a virtual LED with a size determined by the first minimum of the light intensity from the optical axis, i.e., the size of the diffracted spot is given by

1.22·λ D = , (1) spot NA where λ is the of the LED-emitted light and NA is the numerical of the lens. Nevertheless, as the distance between LEDs, or their position, is also downscaled, the light can be shifted onto the sample with steps that are smaller than the diffraction limit, allowing for an oversampling of the image. Contrary to [39], where the sample is scanned by the light sequentially emitted by the 8 × 8 nano-LED array, a single LED and a set of nanopositioners are used in order to carry it out. Therefore, the nanopositioners are integrated to move the sample, i.e., they emulate the light scan, with nanometric precision. While the demagnified version of the 5 µm LEDs is used for the study of the STOM microscope with optical downscaling, the limits of the technique by building a second STOM microscope are explored, based on the 200 nm LEDs below the diffraction limit for the wavelength emitted. The fabrication and characterization of the 200 nm GaN-based nano-LED array used were reported in [44]. To demonstrate the potential of the methods, a number of metal nanostructures with different sizes and shapes were fabricated by electron-beam lithography (EBL) and were observed with the two microscopes. Additional experiments were performed with the aim of potential application in the biology field. Therefore, the resolution limit has been pursued with different versions of STOM. In the following sections, the Materials and Methods are detailed, followed by the Results and Discussion sections. This includes the description of the STOM prototype, using an optically downscaling LED array, and the experimental results obtained with it, to delve into the role of the scanning step. In the second section, the STOM prototype is described using 200 nm nano-LEDs, together with the experimental results. Finally, the main conclusions are summarized.

2. Materials and Methods 2.1. CMOS Sensor The board camera used (TIS-DMM-22BUC03-ML) integrates an Aptina MT9V024 CMOS sensor (from OnSemiconductor®, Phoenix, AZ, USA) with a sensing area of 4.55 mm (H) × 2.97 mm (V). The sensor has a frame rate of 76 fps @ 744 H × 480 V resolution (higher frame rates at lower resolutions), a 6 µm × 6 µm global shutter pixel with a quantum efficiency of ~50% in the blue range (450–495 nm), and an 8-bit dynamic range. The board camera includes a selectable 10- to 8-bit ADC and a bit-serial LVDS or parallel data interface. In addition, it also supports an 8-bit monochrome pixel format. Therefore, this format transmits data using one byte for each pixel. The Python interface controls the sensor and the electromechanical setup to perform the image acquisition.

2.2. Nanopositioning System The fine nanopositioning system is formed by a compact XY piezo stage and a vertical Z piezo stage from Physik Instrumente (PI). P-621.2CD and P-621.2ZCD nanopositioners (from PhysikInstrumente GmbH & Co. KG, Karlsruhe, Germany) are used on the XY plane and in the Z vertical direction, respectively. In the closed-loop configuration, both actuators had 0.4 and 0.3 nm resolutions. As for the travel range, i.e., the maximum FOV, this reaches up to 100 µm. In addition, a linearity error of 0.02% and a repeatability error of 1 nm are taken into account when performing the scans. The actuator step unit defines the pitch of the emulated LED array. For every complete scan in the XY direction, a frame is captured and processed. The Z-axis actuator is used to place the sample in direct contact with the surface of the LED chip. The coarse system is composed of three N-470 PiezoMike linear actuators coupled to triaxial mechanical stages. Its resolution reaches 20 nm and its Sensors 2021, 21, 3305 4 of 14

travel range reaches 7 mm. However, this system lacks an encoder, so the accuracy of the coarse-motor steps is limited.

2.3. Image Reconstruction In STOM-based microscopes with LED scanning, the sample is placed between the optical detector and the nano-LED array. Then, the sample is mapped by the LED, moving the sample sequentially along the X and Y axes to emulate a nano-LED array. The CMOS detector senses the light transmitted (or blocked) by the object. The image is reconstructed following the same order as the sample movement sequence. The morphology of the sample affects the light received by the sensor. These variations, in terms of light intensity, allow for the reconstruction of the image of the sample in grayscale, according to the sample’s physical properties.

2.4. Simulations The simulations were performed by using the finite-difference time-domain method implemented in the commercial software, CST Studio. All of the calculations were made for equal to 450 nm. The optical properties of the materials were modelled with the experimental data available in the literature accounting for real and imaginary parts of the dielectric function. The values used in the calculations were as follows: GaN [45] (ε’ = 5.825, ε” = 0.407), Au [46](ε’ = −1.756, ε” = 5.299), SiO2 [47](ε’ = 2.165, ε” = 0.007). The emission multiple-quantum-well (MQW) layer was approximated with a dipole source placed in the middle of the MQW layer (i.e., 300 nm underneath the top surface of the LED) and below the center of the Ti/Au contact. The calculations were performed for the x and y polarizations of the dipole source and were then summed up to simulate the unpolarized characteristics of the light emission from the LED. To avoid reflection from the borders of the computational area, open boundary conditions were used in all directions.

2.5. Observed Samples The fabrication of the EBL sample included a device-quality starting material of 4” fused silica wafers, which were 0.525 mm thick. First, dehydration was carried out in an oven at 250 ◦C for 2 h. Then, a nominally 180 nm-thick CSAR-P6200.09 positive photoresist (from Allresist GmbH, Strausberg, Germany) was spun at a speed of 4000 rpm for 1 min and was cured at 180 ◦C for 3 min. Prior to resist exposure, a 20 nm-thick aluminum layer was thermally evaporated at 0.3 nm/s to reduce charge build-up in the wafers, while the exposure was performed at a beam current of 2 nA. The sacrificial aluminum layer was removed by a 6000 single puddle in a 2.38% tetramethyl-ammonium hydroxide solution, and development of the CSAR photoresist was achieved in the AR 600-546 developer (from Allresist GmbH, Strausberg, Germany) for 1 min. The sample was further covered by a 40 nm-thick electron-beam-evaporated chromium layer, deposited at 0.5 nm/s. The lift-off of the deposited metal was performed in Remover 1165 at 45 ◦C for 20 min, followed by rinsing in isopropyl alcohol. Human lung fibroblast LL47 (Mado, ECACC, #90102538, Salisbury, UK) were cultured under standard cell culture conditions following the manufacturer’s protocols. Samples for microscopy were prepared in short by tryptic release. After washing with minimum essen- tial medium (MEM, Gibco, 11095080), containing 10% fetal calf serum and 1% penicillin- streptomycin (Gibco, 10378016), cells were counted and 3000 cells were cultured overnight in droplets of 300 µL on 24 × 50 mm coverslips (VWR International, #631-0148) coated with 5 µg/mL of bovine collagen-I (Gibco, A1064401). On the next day, cells were fixed for 20 min at room temperature with MEM containing 2% paraformaldehyde, they were washed twice with phosphate-buffered saline, and they were stained with hematoxylin and eosin, following standard procedures. Finally, cells were embedded with Fluoromount-G (Invitrogen, #00-4958-02) using 22 × 22 mm coverslips (Corning, #CORN2850-22). Cell specimens were assessed using a microscope using a 488 nm LED laser at 2% transmission and an EC Plan Neofluar 10×/0.30 lens (Zeiss). Images with a size of 1612 × 1612 Sensors 2021, 21, x FOR PEER REVIEW 5 of 15

22). Cell specimens were assessed using a microscope using a 488 nm LED laser at 2% transmission and an EC Plan Neofluar 10×/0.30 lens (Zeiss). Images with a size of 1612 × 1612 pixels (1.28 × 1.28 mm) and a spatial resolution of 0.79 × 0.79 µm were produced using the transmission photo multiplier of the microscope.

3. Results and Discussion 3.1. Scanning Transmission with an Optically Downscaled LED In order to explore how scanning is affected by shifting the light source in steps smaller than the diffraction limit, the first prototype with a lens to provide demagnifica- tion was constructed. To do this, the sample was scanned with the downscaled image of Sensors 2021, 21, 3305 a 5 µm LED from the 8 × 8 GaN array described formerly in [39]. Optical downscaling5 ofwas 14 performed with a x60 objective with an NA of 0.85. Consequently, a reduced and dif- fracted image of the LED was obtained, ~750 nm in size, according to Equation (1). The (1.28Materials× 1.28 and mm) Methods and asection spatial describes resolution the of fine 0.79 positioning× 0.79 µm system were formed produced by usingtwo piezo the transmissionstages from Physik photo multiplierInstrumente of the(PI), microscope. with the coarse system used to extend the FOV up to several millimeters and the CMOS camera used for light sensing. The core of this setup 3.is Resultspresented and in DiscussionFigure 2a. 3.1. ScanningA key improvement Transmission of Optical this setup, Microscope following with an the Optically results Downscaledof [39], was LEDthe ability to lo- cate Inthe order sample to precisely explore howwhere scanning the spot is was affected of minimum by shifting size, the by lightplacing source it at the in stepsfocal smallerplane of than the lens the diffraction(see Figure limit, 2b1). the As firstthe spot prototype size, as with well a lensas the to pitch, provide is critical demagnification for image wasformation, constructed. this allows To do for this, improved the sample resoluti wasons. scanned Additionally, with the samples downscaled that imagepresent of an a 5inhomogeneousµm LED from thesurface 8 × 8can GaN be arrayexplored described by scanning formerly them in [39at ].different Optical height downscaling levels. wasTherefore, performed more withinformation a x60 objective is compiled with from an NAthe scanned of 0.85. sample Consequently, due to the a reduced in-focus andand diffractedout-of-focus image areas. of An the additional LED was obtained, strength ~750of this nm microscope in size, according was the toflexibility Equation in (1).finding The Materialsthe region-of-interest and Methods (ROI). section Thanks describes to the the large fine travel positioning range of system the coarse formed piezo-actuators, by two piezo stagesthe sample from could Physik be Instrumente brought closer (PI), to with the the optical coarse detector, system thus used obtaining to extend an the image FOV up with to severala larger millimetersFOV, only limited and the by CMOS the detector camera size, used as for in lightconventional sensing. lensless The core microscopy of this setup (see is presentedFigure 2b2). in This Figure allowed2a. for the easier scrolling through the sample.

FigureFigure 2.2. ((aa)) SetupSetup ofof thethe scanningscanning transmissiontransmission opticaloptical microscopemicroscope withwith opticaloptical downscaling.downscaling. ((bb)) SchematicSchematic diagramsdiagrams ofof thethe STOMSTOM withwith opticaloptical downscalingdownscaling inin differentdifferent operating operating modes. modes. ((b1b1)) NIMNIM mode. mode. ((b2b2)) LenslessLensless mode.mode.

AThe key effect improvement of the step size of this on the setup, image following reconstruction the results with ofthis [39 setup], was can the be abilityobserved to locatein Figures the sample 3 and 4. precisely Figure where3 presents the spotseveral was images of minimum of an size,800 nm by placingsquare pattern it at the array focal planetaken in of steps the lens of 750 (see nm Figure (Figure2b1). 3a), As 400 the nm spot (Figure size, 3b), as well 200 nm as the (Figure pitch, 3c), is and critical 100 fornm image(Figure formation, 3d) with the this objective allows for of improved x60 with an resolutions. NA of 0.85. Additionally, Figure 3a shows samples that, that when present the anstep inhomogeneous used was near surfacethe size canof the be exploreddiffracted by spot, scanning the structures them at could different be estimated height levels. but Therefore, more information is compiled from the scanned sample due to the in-focus and out-of-focus areas. An additional strength of this microscope was the flexibility in finding the region-of-interest (ROI). Thanks to the large travel range of the coarse piezo-actuators, the sample could be brought closer to the optical detector, thus obtaining an image with a larger FOV, only limited by the detector size, as in conventional lensless microscopy (see Figure2b2). This allowed for the easier scrolling through the sample. The effect of the step size on the image reconstruction with this setup can be observed in Figures3 and4. Figure3 presents several images of an 800 nm square pattern array taken in steps of 750 nm (Figure3a), 400 nm (Figure3b), 200 nm (Figure3c), and 100 nm (Figure3d) with the objective of x60 with an NA of 0.85. Figure3a shows that, when the step used was near the size of the diffracted spot, the structures could be estimated but not resolved. The starting position of the scan influenced the reconstruction of the image, as it determines whether light intensity is captured just above each square or in the separation Sensors 2021, 21, x FOR PEER REVIEW 6 of 15

Sensors 2021, 21, x FOR PEER REVIEW 6 of 15 not resolved. The starting position of the scan influenced the reconstruction of the image, as it determines whether light intensity is captured just above each square or in the sepa- ration between the two. In this case, the size of the LED was equal to the size of the pitch not resolved.used. The However, starting position for a pitch of the smaller scan influenced than or equa the reconstructionl to the size of of the the virtual image, LED, the images as it determines whether light intensity is captured just above each square or in the sepa- were resolved with more clarity (Figure 3b–d), i.e., the measurement was oversampled, ration between the two. In this case, the size of the LED was equal to the size of the pitch which allowed us to improve the acquisition of the STOM image. used. However, for a pitch smaller than or equal to the size of the virtual LED, the images were resolved Figurewith more 4 shows clarity the (Figure corresponding 3b–d), i.e., theedge measurement spread function was oversampled, (ESF) and line spread func- which allowedtion (LSF) us to [48].improve The the ESF acquisition is the contrast of the STOMprofil eimage. across an edge in the image, and the LSF is Figureits 4derivative. shows the correspondingIn the images edgeobtained, spread both function were (ESF) calculated and line at spread one edge func- of a square in the tion (LSF)pattern [48]. The in ESF Figure is the 3. contrast The resolution profile across can bean edgecalculated in the image,from both and theESF LSF and is LSF. This corre- Sensors 2021, 21, 3305 its derivative.sponds In the to theimages distance obtained, between both we 10%re calculated and 90% atof onethe edgeESF. ofOn a squarethe other in the hand, this6 of can 14 also pattern inbe Figure evaluated 3. The resolutionas the FWHM can be of ca lculatedthe LSF. from The both ESF ESF data and were LSF. normalizedThis corre- and averaged sponds to the distance between 10% and 90% of the ESF. On the other hand, this can also (dashedbetween lines) the two.to obtain In this a case,uniform the size LSF of (solid the LED lines). was As equal can to be the observed size of the from pitch the used. data in be evaluated as the FWHM of the LSF. The ESF data were normalized and averaged FigureHowever, 4, the for resolution a pitch smaller was thannot oraffected equal toby the the size step of the size, virtual although LED, the oversampling images were pro- (dashed lines) to obtain a uniform LSF (solid lines). As can be observed from the data in ducedresolved visually with improved more clarity images. (Figure Fo3b–d),r all i.e.,step the sizes, measurement the resoluti wason oversampled, of all of them which was main- Figure 4, the resolution was not affected by the step size, although oversampling pro- tainedallowed at around us to improve 800 nm, the i.e., acquisition around ofthe the spot STOM size. image. duced visually improved images. For all step sizes, the resolution of all of them was main- tained at around 800 nm, i.e., around the spot size.

Figure 3. STOM images of an 800 nm-wide squared pattern using a diffracted spot of 750 nm with an objective of x60, with Figure 3. STOMFigure images 3. STOM of an 800 images nm-wide of an squared 800 nm-wide pattern squared using patterna diffracted using spot a diffracted of 750 nm spot with of an 750 objective nm with of an x60, objective with of x60, with steps ofsteps (a) 750, stepsof ((ba)) of 400,750, (a )( c 750,()b 200,) 400, (b )and 400, (c )( d200, (c) )100 200, and nm. and ( d ()d 100) 100 nm. nm.

Figure 4. The measured ESF from a sharpened line pattern with steps of 400, 200, and 100 nm. The LSF has been fitted and Figure 4. The measured ESF from a sharpened line pattern with steps of 400, 200, and 100 nm. The illustrated withLSF a solid has been line andfitted has and been illustrated normalized with to a unity.solid line The and peaks has at been the topnormalized of the sample to unity. profile The correspond peaks at to the diffraction of the object. The raw curves correspond to the same sharp profile, but for easy visual understanding, they have the top of theFigure sample 4. The profile measured correspond ESF to from the di a ffractionsharpened of the line object. patte Thern withraw curves steps ofcorrespond 400, 200, and 100 nm. The been divided along the X-axis. The resolution calculated from the LSF is shown in the inset. LSF has been fitted and illustrated with a solid line and has been normalized to unity. The peaks at the top ofFigure the sample4 shows profile the corresponding correspond to edge the spreaddiffraction function of the (ESF) object. and The line raw spread curves function correspond (LSF) [48]. The ESF is the contrast profile across an edge in the image, and the LSF is its derivative. In the images obtained, both were calculated at one edge of a square in the pattern in Figure3. The resolution can be calculated from both ESF and LSF. This corresponds to the distance between 10% and 90% of the ESF. On the other hand, this can also be evaluated as the FWHM of the LSF. The ESF data were normalized and averaged (dashed lines) to obtain a uniform LSF (solid lines). As can be observed from the data in Figure4, the resolution was not affected by the step size, although oversampling produced Sensors 2021, 21, x FOR PEER REVIEW 7 of 15

Sensors 2021to, 21the, 3305 same sharp profile, but for easy visual understanding, they have been divided along the X- 7 of 14 axis. The resolution calculated from the LSF is shown in the inset. visually improved images. For all step sizes, the resolution of all of them was maintained The final experimentsat around 800 demonstrated nm, i.e., around the the capabi spot size.lities of the technique for biological microscopy. Figure 5aThe shows final experimentsthe raw view demonstrated of a scanned the capabilities fly wing ofin the the technique camera. for Figure biological 5b,c show STOM imagesmicroscopy. with Figure the5 aobjective shows the of raw x60 view with of a scannedan NA flyof wing0.85 inand the camera.a step size Figure of5b,c show STOM images with the objective of x60 with an NA of 0.85 and a step size of 300 nm. 300 nm. As shownAs in shownFigure in 5b, Figure regions5b, regions of the of thesample sample were were out out ofof focus focus while while others others were were perfectly perfectly focused. focused.Nevertheless, Nevertheless, as the as focus the focus could could be be controlled controlled byby placing placing the the sample sample at different at different heights, itheights, could it be could refocused, be refocused, as shown as shown in inFigure Figure 55c.c.

FigureFigure 5. (a )5. Raw (a) imageRaw image in lensless in lensless mode of themode fly wing of the sample. fly wi Theng refocusingsample. The images refocusing for the selected images region for the of ( b,c) correspondingselected region to the height of (b, differencec) corresponding of 2 µm at 17to andthe 19heightµm, respectively. difference of 2 µm at 17 and 19 µm, respec- tively. In addition, a second sample, composed of a human lung fibroblast, was explored. Figure6a shows the raw image, recorded with a Zeiss Confocal Laser Scanning Microscope In addition, a(CLSM) second with sample, a pixel resolutioncomposed of 0.79of aµ m,human which islung similar fibroblast, to that of ourwas optical explored. demagnifi- Figure 6a shows thecation raw STOM image, microscope. recorded A STOM with image,a Zeiss with Confocal the objective Laser of x60, Scanning an NA of Micro- 0.85, and a scope (CLSM) withstep a sizepixel of resolution 600 nm of the of same 0.79 set µm, of cells, which is presented is similar in Figure to that6b. of Both our the optical nuclei and the cell bodies, as well as the sample projections, were very similar in size and structure demagnification STOMin both microscope. images. Therefore, A STOM it was imag possiblee, with to analyze the objective the state of of the x60, cells an thanks NA of to the 0.85, and a step sizefact of that 600 the nm nuclei, of the i.e., same the center set ofof mass cells, of is the presented cell, could bein identifiedFigure 6b. despite Both the the lower nuclei and the cellcontrast bodies, of as the well STOM as image.the samp The twole projections, images do not were correspond very similar exactly to in the size same and region structure in both images.because, Therefore, once the sample it was is scanned, possible it is to dried. analyze Therefore, the state the images of the were cells obtained thanks with to the fact that the differentnuclei, setsi.e., ofthe cells center that were of mass part ofof the the same cell, cell could culture. be identified despite the lower contrast of the STOM image. The two images do not correspond exactly to the same region because, once the sample is scanned, it is dried. Therefore, the images were ob- tained with different sets of cells that were part of the same cell culture.

3.2. Scanning Transmission Optical Microscope with a 200 nm LED The setup of the microscope used in the second STOM prototype is presented in Fig- ure 7a. The nano-LED light source was obtained from a single GaN chip with an array of 32 × 2200 nm LEDs [44], specifically targeting this new approach to optical microscopy. One of these 32 × 2 LEDs was chosen for the experiments. The scan operation was per- formed by the displacement of the sample over the LED with a nano-positioning system

Sensors 2021, 21, x FOR PEER REVIEW 8 of 15

in raster scan mode, emulating a virtual LED array of arbitrary size with only one LED. Sensors 2021, 21, 3305 When the cover was closed, the camera and the LED chip had an observation chamber of 8 of 14 ~1.8 mm.

Figure 6. HumanFigure lung 6. fibroblasts Human lung observed fibroblasts with a observed (a) confocal with laser a ( scanninga) confocal microscope laser scanning (CLSM) microscope and (b) STOM (CLSM) microscope and (b) STOM microscope using a diffracted spot of 750 nm and a step size of 600 nm. using a diffracted spot of 750 nm and a step size of 600 nm.

Figure 3.2.7b presents Scanning the Transmission sample positioning Optical Microscope in closer with detail. a 200 The nm LEDsample holder was designed with two main openings. The small one was used to fit the sample. The large one was used asThe an opening setup of for the the microscope LED chip bonding. used in theThis second solution STOM allowed prototype for directly is presented in Figure7a. The nano-LED light source was obtained from a single GaN chip with an array placing the sample on the LED chip to perform the scan in order to obtain the maximum of 32 × 2200 nm LEDs [44], specifically targeting this new approach to optical microscopy. resolution. Two additional micrometer heads were integrated in the microscope base to One of these 32 × 2 LEDs was chosen for the experiments. The scan operation was make a coarse exploration of the sample. Finally, the LED current driver and the control performed by the displacement of the sample over the LED with a nano-positioning system electronics were placed under the LED chip. The current driver allowed for the control of in raster scan mode, emulating a virtual LED array of arbitrary size with only one LED. the current of the nano-LED to provide the optimal operation conditions. When the cover was closed, the camera and the LED chip had an observation chamber of Figure 8a shows the image obtained using an optical microscope (background) with ~1.8 mm. a x60 magnification from the objective, together with the image obtained by the lensless Figure7b presents the sample positioning in closer detail. The sample holder was STOM, scanning a region of 80 × 65 pixels using the 200 nm LED (front) of an electron- designed with two main openings. The small one was used to fit the sample. The large beam lithography (EBL) sample with 1.6 µm-width squared features. It can be observed one was used as an opening for the LED chip bonding. This solution allowed for directly that the squaresplacing were the roughly sample resolved on the LED in the chip scanning to perform microscope. the scan in order to obtain the maximum As in theresolution. previous Two section, additional the ESF micrometer and the LSF heads were were considered integrated to indetermine the microscope the base to spatial resolution.make a Therefore, coarse exploration the ESF and of the the sample. LSF of Finally,the region the marked LED current in Figure driver 8a and are the control presented onelectronics Figure 8b. were The placedbest obtained under theresolution LED chip. was The 1.53 current µm, close driver to the allowed actual forspac- the control of ing betweenthe squares current in ofFigure the nano-LED 8a, but this to wa provides noticeably the optimal larger operationthan the LEDs’ conditions. dimension and pitch.

Sensors 2021, 21, x FOR PEER REVIEW 9 of 15 Sensors 2021, 21, 3305 9 of 14

Sensors 2021, 21, x FOR PEER REVIEW 9 of 15

Figure 7. (a) PrototypeFigure 7. ( ofa) thePrototype mechanical-based of the mechanical-based microscope in microsco comparisonpe in with comparison a 20-cent with coin. a ( b20-cent) Placement coin. ( ofb) the sample Placement of the sample on the LED array for scanning with the mechanical-based STOM micro- on the LED array for scanning with the mechanical-based STOM microscope. The operating LED had a size of 200 nm and scope. The operating LED had a size of 200 nm and emitted at a wavelength of 465 nm. emitted at a wavelength of 465 nm.

Figure8a shows the image obtained using an optical microscope (background) with a x60 magnification from the objective, together with the image obtained by the lensless STOM, scanning a region of 80 × 65 pixels using the 200 nm LED (front) of an electron- Figure 7. (a) Prototype of the mechanical-based microscope in comparison with a 20-cent coin. (b) µ beamPlacement lithography of the sample (EBL) on sample the LED with array 1.6 for scanningm-width with squared the mechanical-based features. It can STOM be observed micro- thatscope. the The squares operating were LED roughly had a size resolved of 200 innm the and scanning emitted at microscope. a wavelength of 465 nm.

Figure 8. (a) STOM image obtained with an LED size of 200 nm biased at 800 nA and with a pitch of 200 nm, superimposed on the standard optical microscopy image at x60 of an EBL matrix of 1.6 µm squares. (b) ESF and LSF from a sharpened line with steps of 200 nm calculated in the region marked in (a). Both functions were fitted and normalized to unity, and they are illustrated with dashed and solid lines, respectively.

To evaluate this resolution discrepancy, we present below a theorical analysis of LED behavior. A modelFigureFigure based 8. 8.( a(a)) STOM onSTOM a reduced image image obtained obtained version with ofwith the an an 200 LED LED nm size size LED of of 200 200chip nm nm was biased biased created. at at 800 800 nAFigure nA and and with with a pitcha pitch of 9a presents the200of scheme 200 nm, nm, superimposed of superimposed this proposed on theon themodel, standard standard consisting optical optical microscopy ofmi twocroscopy lines image image of at nine x60 at x60LED of an of EBLpixels an EBL matrix matrix of 1.6 ofµ 1.6m (with a pixel widthsquares.µm squares. of (b 200) ESF ( bnm) and ESF and LSF and pitch from LSF of afrom sharpened 400 a nm)sharpened in line the with line line steps with direction. ofsteps 200 of nm In200 calculated Figure nm calculated 9b, in the region in the markedregion schematic cross-sectioninmarked (a). Both in functions(ofa). the Both LED functions were model fitted were is and shown. fitted normalized and The normalized system to unity, was andto unity, simulated they areand illustrated they using are theillustrated with dashed with and dashed and solid lines, respectively. finite differencesolid time lines, domain respectively. method, employing the commercial software, CST Studio (see the Materials and Method section). The model includes the Pd/Ti/Au contact on top AsTo inevaluate the previous this resolution section, discrepancy, the ESF and we the present LSF were below considered a theorical to analysis determine of LED the of each LED. The emitted optical power, predicted by the simulations, is presented in Fig- spatialbehavior. resolution. A model Therefore, based on a thereduced ESF and version the LSFof the of 200 the nm region LED marked chip was in created. Figure8 Figurea are ure 9c. The results show that there was a local minimum of light intensity above the Au presented9a presents on the Figure scheme8b. The of this best proposed obtained model, resolution consisting was 1.53 of µtwom, closelines toof thenine actual LED pixels spac- contact, which was due to parasitic absorption by the metal. The shape of the light spot ing(with between a pixel squares width of in Figure200 nm8 a,and but pitch this wasof 400 noticeably nm) in the larger line thandirection. the LEDs’ In Figure dimension 9b, the near the top surface of the LED was ring-like around the pad, and the light was emitted andschematic pitch. cross-section of the LED model is shown. The system was simulated using the mostly between contacts. The same simulation was represented on the XZ plane (Figure finiteTo difference evaluate thistime resolution domain method, discrepancy, employing we present the belowcommercial a theorical software, analysis CST of Studio LED 9d). The dipole was placed 500 nm below the LED chip surface. The propagation of light behavior.(see the Materials A model and based Method on a section). reduced The version model of includes the 200 nmthe Pd/Ti/Au LED chip contact was created. on top was calculated between the black lines across the Z axis. The width of this area defines the Figureof each9a LED.presents The emitted the scheme optical of this power, proposed predict model,ed by consistingthe simulations, of two is lines presented of nine in LED Fig- spot size, i.e., the FWHM. Thus, the measured spot size was significantly larger than that ure 9c. The results show that there was a local minimum of light intensity above the Au contact, which was due to parasitic absorption by the metal. The shape of the light spot near the top surface of the LED was ring-like around the pad, and the light was emitted mostly between contacts. The same simulation was represented on the XZ plane (Figure 9d). The dipole was placed 500 nm below the LED chip surface. The propagation of light was calculated between the black lines across the Z axis. The width of this area defines the spot size, i.e., the FWHM. Thus, the measured spot size was significantly larger than that

Sensors 2021, 21, 3305 10 of 14

pixels (with a pixel width of 200 nm and pitch of 400 nm) in the line direction. In Figure9b , the schematic cross-section of the LED model is shown. The system was simulated using the finite difference time domain method, employing the commercial software, CST Studio (see the Materials and Method section). The model includes the Pd/Ti/Au contact on top of each LED. The emitted optical power, predicted by the simulations, is presented in Figure9c . The results show that there was a local minimum of light intensity above the Au contact, which was due to parasitic absorption by the metal. The shape of the light spot near the top surface of the LED was ring-like around the pad, and the light was emitted mostly between contacts. The same simulation was represented on the XZ plane (Figure9d ). The dipole was placed 500 nm below the LED chip surface. The propagation of light was calculated between the black lines across the Z axis. The width of this area defines the spot size, i.e., the FWHM. Thus, the measured spot size was significantly larger Sensors 2021, 21, x FOR PEER REVIEW than that of the pixel, due to the influence of parasitic absorption of the11 metalof 15 contacts on

light propagation.

Figure 9. Scheme of the LED model. (a) Perspective. (b) Cross section along an LED line (scale is not respected). The dipole Figure 9. Scheme of the LED model. (a) Perspective. (b) Cross section along an LED line (scale is source was placed at h1 = 300 nm under the GaN surface. (c) Logarithm of absolute value of power intensity distribution on not respected). The dipole source was placed at h1 = 300 nm under the GaN surface. (c) Logarithm the XY plane.of The absolute cross sectionvalue of was power taken intensity at z = 975distribution nm (i.e., 100on the nm XY above plane. the The top cross surface section of the was LED). taken The at blackz lines indicate the field= 975 values nm (i.e., equal 100 to nm half above of the the maximum top surface of of the th signal.e LED). ( dThe) Logarithm black lines of indicate absolute the value field of values power intensity distribution inequal the XZ to plane.half of Thethe maximum cross section of the was signal. taken at(d) x Logarithm = 1800 nm. of absolute value of power intensity distribution in the XZ plane. The cross section was taken at x = 1800 nm. Finally, the previous increase in the FWHM in the X and Y axes as a function of Finally, distancethe previous in the increase Z-axis isin presentedthe FWHM more in the clearly X and in Y Figure axes as 10 a. function The LED of surface dis- (SiO2/air) tance in the Z-axiswas 400 is nmpresented above themore emission clearly region,in Figure giving 10. The a minimum LED surface light (SiO spot2/air) size was of ~800 nm with 400 nm abovethe the sample emission laying region, completely giving a flat mi onnimum the surface. light spot Therefore, size of ~800 according nm with to the simulations, sample layingthe completely sample was flat placed on the between surface. 400 Therefore, and 500 according nm with respect to the tosimulations, the LED surface. the Although sample was placed between 400 and 500 nm with respect to the LED surface. Although the maximum achievable resolution would theoretically be 800 nm, this was because the sample could not be moved closer to the emission zone. This limitation can be explained either because the distance of the metal pattern from the LED structure was roughly 500 nm or because the effective emitting area was larger than that assumed for the simulations as a result of lateral carrier diffusion.

Sensors 2021, 21, 3305 11 of 14

the maximum achievable resolution would theoretically be 800 nm, this was because the sample could not be moved closer to the emission zone. This limitation can be explained either because the distance of the metal pattern from the LED structure was roughly 500 nm Sensors 2021, 21, x FOR PEER REVIEW 12 of 15 or because the effective emitting area was larger than that assumed for the simulations as a result of lateral carrier diffusion.

FigureFigure 10.10. Full-widthFull-width at at half-maximum half-maximum as as a afunction function of ofthe the vertical vertical coordinate coordinate z. Dashed z. Dashed lines lines indicateindicate thethe materialmaterial interfaces inside inside the the LED LED structure. structure. The The p-GaN p-GaN layer’s layer’s thickness thickness is isequal equal to to 300300 nm.nm.

ComparedCompared toto conventionalconventional confocalconfocal andand lenslesslensless microscopy,microscopy, somesome similaritiessimilarities andand differencesdifferences areare evident.evident. InIn thethe casecase ofof lenslesslensless microscopy,microscopy,a a minimumminimumdistance distance to to thethe LEDLED isis neededneeded for for the the sample sample to to be be fully fully illuminated. illuminated. In In confocal confocal microscopes, microscopes, an opticalan optical system sys- istem required is required to focus to focus the sample. the sample. However, However, the STOM the STOM technique technique makes itmakes possible it possible to reduce to thereduce distance the distance between between the sensor the sensor and the and light the source. light source. This is This highly is highly advantageous, advantageous, as it allowsas it allows for the for manufacturing the manufacturing of smaller of smaller high-resolution high-resolution devices, devices, as the totalas the height total height of the microscopeof the microscope is equivalent is equivalent to the thicknessto the thickness needed needed to introduce to introduce the sample. the sample. Nevertheless, Never- thetheless, current the statecurrent of STOM state of technology STOM technology requires nanopositionersrequires nanopositioners that increase that itsincrease cost and its volume.cost and Onevolume. solution One issolution to replace is to the replace scanning the scanning system,based system, on based a single on LED,a single and LED, the mechanicaland the mechanical system with system LED with arrays. LED These arrays. systems These currentlysystems currently have pitches have similar pitches to similar image sensorto image pixels sensor but pixels still have but still the potentialhave the forpotential further for reduction further reduction as LEDs areas LEDs continuously are con- beingtinuously miniaturized. being miniaturized. Finally, and Finally, despite and having despite a prototype having a that prototype incorporates that incorporates LEDs with sizesLEDs belowwith sizes the below diffraction the diffraction limit, the limit, behavior the behavior of the LEDs of the was LEDs not was as not expected, as expected, and furtherand further development development needs needs to be realizedto be realized to compare to compare their resolution their resolution with super-resolution with super-res- microscopes.olution microscopes.

4.4. ConclusionsConclusions In this work, we investigated different aspects of scanning transmission optical micro- In this work, we investigated different aspects of scanning transmission optical mi- scopes based on nano-LED light sources. The method of using a single LED for scanning croscopes based on nano-LED light sources. The method of using a single LED for scan- the sample was validated, as with the prototype based on optical demagnification and the ning the sample was validated, as with the prototype based on optical demagnification 200 nm LED-based microscope. and the 200 nm LED-based microscope. The former presents some advantages compared with the previous lensless setup. The former presents some advantages compared with the previous lensless setup. The possibility of exploring a sample, both inorganic and organic, placed at the lens The possibility of exploring a sample, both inorganic and organic, placed at the lens focus, focus, is interesting because it allows for more than just flat samples to be investigated. is interesting because it allows for more than just flat samples to be investigated. Com- Compared with the CLSM system, the use of a less energetic LED source can avoid cell pared with the CLSM system, the use of a less energetic LED source can avoid cell damage damage in the study of living tissue [49]. Resolutions below 1 µm were achieved using the in the study of living tissue [49]. Resolutions below 1 µm were achieved using the appro- appropriate . Image processing was not required in any of the performed experiments. priate optics. Image processing was not required in any of the performed experiments. Scanning the sample with a movement step below the pixel size gave rise to oversampling andScanning improved the sample image with reconstruction. a movement However, step below the the use pixel of nanopositioners size gave rise to wasoversampling a serious and improved image reconstruction. However, the use of nanopositioners was a serious drawback, as it limited the prototype in terms of cost. As noted earlier, this can be solved by using LED arrays with large dimensions instead of a single LED. On the other hand, the second prototype was fabricated with 200 nm nanoLEDs, i.e., with a size below the diffraction limit. At current, our LED manufacturing procedure pro- vided us with an effective resolution of around 1.6 µm. The proper operation of the STOM microscopes without optical elements requires the sample to be in direct contact with the

Sensors 2021, 21, 3305 12 of 14

drawback, as it limited the prototype in terms of cost. As noted earlier, this can be solved by using LED arrays with large dimensions instead of a single LED. On the other hand, the second prototype was fabricated with 200 nm nanoLEDs, i.e., with a size below the diffraction limit. At current, our LED manufacturing procedure provided us with an effective resolution of around 1.6 µm. The proper operation of the STOM microscopes without optical elements requires the sample to be in direct contact with the surface of the point of emission of each LED; however, with the passivation and structures present in current technology, this is not possible, which negatively affects the resolving power. The resolution of the STOM prototype was worse than expected. Further experiments integrating arrays of nano-LEDs fabricated with a different structure, i.e., transparent contacts (TCO) [50], will prove promising in terms of spot shape and FWHM. It is expected that a more symmetric dot shape and a narrower FWHM can be obtained, thus improving the resolution. Moreover, in combination with the rapid development of new display technologies [51–53], i.e., large electronically controlled arrays of LEDs, we could create new prototypes with resolutions close to the diffraction limit with low computational effort and relatively low dimensions and cost.

Author Contributions: The conceptualization of the technique was produced by A.D., J.D.P. and A.W.; D.D.B. designed the LED chips; A.D. and N.F. designed the EBL samples that were manu- factured by A.R.-R.; V.M. designed the current drivers; S.M. and J.C. constructed the devices; S.K. and S.G. prepared the biological samples and validated the biological results; K.K.-K. and M.A.d.M. performed the simulations; S.M. implemented the software, performed the experiments and wrote the article; A.D. and A.V. reviewed the paper. A.D.C. has participated in the simulations with K.K.-K. and M.A.d.M. However, A.D. is the principal investigator who supervised the research. A.D. also participated in the final approval of the submitted manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by CHIPSCOPE. The APC was funded by the European Union’s Horizon 2020 research and innovation program under grant agreement No. 737089. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Data sharing not applicable. Acknowledgments: This work was partially supported by the European Union’s Horizon 2020 research and innovation program under grant agreement No. 737089—ChipScope. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Fogazzi, G.; Delanghe, J. Microscopic examination of urine sediment: Phase contrast versus bright field. Clin. Chim. Acta 2018, 487, 168–173. [CrossRef] 2. Mann, C.J.; Yu, L.; Lo, C.M.; Kim, M.K. High-resolution quantitative phase-contrast microscopy by digital . Opt. Express 2005, 13, 8693–8698. [CrossRef] 3. Ru, Q.; Lai, G.; Aoyama, K.; Endo, J.; Tonomura, A. Principle and application of phase-shifting electron holography. Ultrami- croscopy 1994, 55, 209–220. [CrossRef] 4. Liu, M.; Chao, J.; Deng, S.; Wang, K.; Li, K.; Fan, C. Dark-field microscopy in imaging of plasmon resonant nanoparticles. Colloids Surf. B Biointerfaces 2014, 124, 111–117. [CrossRef] 5. Harutyunyan, H.; Palomba, S.; Renger, J.; Quidant, R.; Novotny, L. Nonlinear Dark-Field Microscopy. Nano Lett. 2010, 10, 5076–5079. [CrossRef][PubMed] 6. Pawley, J.B. Biological ; Springer: Boston, MA, USA, 2006. [CrossRef] 7. Peterson, D. Confocal Microscopy. Encycl. Mov. Disord. 2010, 250–252. [CrossRef] 8. Lichtman, J.W.; Conchello, J.-A. Fluorescence microscopy. Nat. Methods 2005, 2, 910–919. [CrossRef][PubMed] 9. Dobrucki, J.W.; Kubitscheck, U. Fluorescence Microscopy. In Fluorescence Microscopy: From Principles to Biological Applications, 2nd ed.; Kubitscheck, U., Ed.; Wiley: Berlin, Germany, 2017; pp. 85–132. 10. Hill, M. Principles and Technique of Fluorescence Microscopy. J. Cell Sci. 1961, 3, 419–449. Sensors 2021, 21, 3305 13 of 14

11. Hell, S.W.; Wichmann, J. Breaking the diffraction resolution limit by stimulated emission: Stimulated-emission-depletion fluorescence microscopy. Opt. Lett. 1994, 19, 780–782. [CrossRef][PubMed] 12. Gao, M.; Maraspini, R.; Beutel, O.; Zehtabian, A.; Eickholt, B.; Honigmann, A.; Ewers, H. Expansion Stimulated Emission Depletion Microscopy (ExSTED). ACS Nano 2018, 12, 4178–4185. [CrossRef] 13. Blom, H.; Widengren, J. Stimulated Emission Depletion Microscopy. Chem. Rev. 2017, 117, 7377–7427. [CrossRef][PubMed] 14. Heintzmann, R.; Huser, T. Super-Resolution Structured Illumination Microscopy. Chem. Rev. 2017, 117, 13890–13908. [CrossRef] [PubMed] 15. Wu, Y.; Shroff, H. Faster, sharper, and deeper: Structured illumination microscopy for biological imaging. Nat. Methods 2018, 15, 1011–1019. [CrossRef] 16. Godin, A.G.; Lounis, B.; Cognet, L. Super-resolution Microscopy Approaches for Live Cell Imaging. Biophys. J. 2014, 107, 1777–1784. [CrossRef][PubMed] 17. Leung, B.O.; Chou, K.C. Review of Super-Resolution Fluorescence Microscopy for Biology. Appl. Spectrosc. 2011, 65, 967–980. [CrossRef] 18. Henriques, R.; Griffiths, C.; Rego, E.H.; Mhlanga, M.M. PALM and STORM: Unlocking live-cell super-resolution. Biopolymers 2011, 95, 322–331. [CrossRef] 19. Heinzelmann, H.; Pohl, D.W. Scanning near-field optical microscopy. Appl. Phys. A 1994, 59, 89–101. [CrossRef] 20. Ozcan, A.; McLeod, E. Lensless Imaging and Sensing. Annu. Rev. Biomed. Eng. 2016, 18, 77–102. [CrossRef] 21. Roy, M.; Seo, D.; Oh, S.; Yang, J.-W.; Seo, S. A review of recent progress in lens-free imaging and sensing. Biosens. Bioelectron. 2017, 88, 130–143. [CrossRef] 22. Yang, C.; Bu, X.; Ma, H.; Zhang, L.; Cao, X.; Yue, T.; Hua, X.; Yan, F. Resolution-Enhanced Lensless Color Shadow Imaging Microscopy Based on Large Field-of-View Submicron-Pixel Imaging Sensors. In Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Salt Lake City, UT, USA, 18–22 June 2018; pp. 2327–2334. [CrossRef] 23. Fang, Y.; Yu, N.; Wang, R.; Su, D. An on-chip instrument for white blood cells classification based on a lens-less shadow imaging technique. PLoS ONE 2017, 12, e0174580. [CrossRef] 24. Imai, K.; Nishigaki, M.; Onozuka, Y.; Akimoto, Y.; Nagai, M.; Matsumoto, S.; Kousai, S. A lens-free single-shot fluorescent imaging system using CMOS image sensors with dielectric multi-layer filter. In Proceedings of the 2017 19th International Conference on Solid-State Sensors, Actuators Microsystems (TRANSDUCERS), Kaohsiung, Taiwan, 18–22 June 2017; IEEE: Piscataway, NJ, USA, 2017; pp. 139–142. [CrossRef] 25. Sasagawa, K.; Kimura, A.; Haruta, M.; Noda, T.; Tokuda, T.; Ohta, J. Highly sensitive lens-free fluorescence imaging device enabled by a complementary combination of interference and absorption filters. Biomed. Opt. Express 2018, 9, 4329–4344. [CrossRef][PubMed] 26. Coskun, A.F.; Su, T.W.; Ozcan, A. Wide field-of-view lens-free fluorescent imaging on a chip. Lab Chip 2010, 10, 824–827. [CrossRef] 27. Pedrini, G.; Tiziani, H.J. Short-coherence digital microscopy by use of a lensless holographic imaging system. Appl. Opt. 2002, 41, 4489–4496. [CrossRef] 28. Restrepo, J.F.; Garcia-Sucerquia, J. Automatic three-dimensional tracking of particles with high-numerical-aperture digital lensless holographic microscopy. Opt. Lett. 2012, 37, 752–754. [CrossRef][PubMed] 29. Garcia-Sucerquia, J. Color lensless digital holographic microscopy with micrometer resolution. Opt. Lett. 2012, 37, 1724–1726. [CrossRef] 30. Adams, J.K.; Boominathan, V.; Avants, B.W.; Vercosa, D.G.; Ye, F.; Baraniuk, R.G.; Robinson, J.T.; Veeraraghavan, A. Single-frame 3D fluorescence microscopy with ultraminiature lensless FlatScope. Sci. Adv. 2017, 3, e1701548. [CrossRef][PubMed] 31. Berdeu, A.; Momey, F.; Dinten, J.; Gidrol, X.; Allier, C.; Berdeu, A.; Momey, F.; Dinten, J.; Picollet-d, N.; Gidrol, X. 3D Lensfree Microscopy for 3D Cell Culture. In Proceedings of the SPIE 10414, Advances in Microscopic Imaging, Munich, Germany, 28 July 2017. [CrossRef] 32. Zuo, C.; Sun, J.; Li, J.; Chen, Q. Computational microscopy with programmable illumination and coded aperture. In Proceedings of the International Conference on Optical and Photonics Engineering (icOPEN 2016), Chengdu, China, 10 February 2017; Volume 10250, p. 102501K. [CrossRef] 33. Rivenson, Y.; Wu, Y.; Ozcan, A. Deep learning in holography and coherent imaging. Light. Sci. Appl. 2019, 8, 1–8. [CrossRef] 34. Shen, H.; Gao, J. Portable deep learning singlet microscope. J. Biophotonics 2020, 13, e202000013. [CrossRef] 35. Luo, W.; Zhang, Y.; Feizi, A.; Göröcs, Z.; Ozcan, A. Pixel super-resolution using wavelength scanning. Light. Sci. Appl. 2016, 5, e16060. [CrossRef] 36. Zhang, J.; Sun, J.; Chen, Q.; Li, J.; Zuo, C. Adaptive pixel-super-resolved lensfree in-line digital holography for wide-field on-chip microscopy. Sci. Rep. 2017, 7, 11777. [CrossRef] 37. Zhang, H.; Wang, W.; Liu, C.; Liu, J.; Chenguang, L. Pixel super-resolved lens-free on-chip microscopy based on dual laterally shifting modulation. Appl. Opt. 2020, 59, 3411–3416. [CrossRef] 38. Wu, Y.; Ozcan, A. Lensless digital holographic microscopy and its applications in biomedicine and environmental monitoring. Methods 2018, 136, 4–16. [CrossRef] Sensors 2021, 21, 3305 14 of 14

39. Franch, N.; Canals, J.; Moro, V.; Vilà, A.; Romano-Rodríguez, A.; Prades, J.D.; Guelink, J.; Bezshlyakh, D.; Waag, A.; Kluczyk- Korch, K.; et al. Nano illumination microscopy: A technique based on scanning with an array of individually addressable nanoLEDs. Opt. Express 2020, 28, 19044. [CrossRef] 40. Wasisto, H.S.; Prades, J.D.; Gülink, J.; Waag, A. Beyond solid-state lighting: Miniaturization, hybrid integration, and applications of GaN nano- and micro-LEDs. Appl. Phys. Rev. 2019, 6, 041315. [CrossRef] 41. Wun, J.M.; Lin, C.W.; Chen, W.; Sheu, J.K.; Lin, C.L.; Li, Y.L.; Bowers, J.E.; Shi, J.W.; Vinogradov, J.; Kruglov, R.; et al. GaN-Based Miniaturized Cyan Light-Emitting Diodes on a Patterned Sapphire Substrate With Improved Fiber Coupling for Very High-Speed Plastic Optical Fiber Communication. IEEE Photon. J. 2012, 4, 1520–1529. [CrossRef] 42. Ding, K.; Avrutin, V.; Izyumskaya, N.; Özgür, Ü.; Morkoç, H. Micro-LEDs, a Manufacturability Perspective. Appl. Sci. 2019, 9, 1206. [CrossRef] 43. Kluczyk-Korch, K.; Palazzo, D.; Waag, A.; Dieguez, A.; Prades, J.D.; Di Carlo, A.; Der Maur, M.A. Optical design of InGaN/GaN nanoLED arrays on a chip: Toward: Highly resolved illumination. Nanotechnology 2021, 32, 105203. [CrossRef] 44. Bezshlyakh, D.D.; Spende, H.; Weimann, T.; Hinze, P.; Bornemann, S.; Gülink, J.; Canals, J.; Prades, J.D.; Dieguez, A.; Waag, A. Directly addressable GaN-based nano-LED arrays: Fabrication and electro-optical characterization. Microsyst. Nanoeng. 2020, 6, 1–10. [CrossRef] 45. Kawashima, T.; Yoshikawa, H.; Adachi, S.; Fuke, S.; Ohtsuka, K. Optical properties of hexagonal GaN. J. Appl. Phys. 1997, 82, 3528–3535. [CrossRef] 46. Johnson, P.B.; Christy, R.W. Optical Constant of the Nobel Metals. Phys. Rev. B 1972, 6, 4370–4379. [CrossRef] 47. Rodríguez-De Marcos, L.V.; Larruquert, J.I.; Méndez, J.A.; Aznárez, J.A. Self-consistent optical constants of SiO2 and Ta2O5 films. Opt. Mater. Express 2016, 6, 3622–3637. [CrossRef] 48. Function, E.S. Edge Spread Function. J. Opt. Soc. Am. 1965, 55, 1217–1221. 49. Wäldchen, S.; Lehmann, J.; Klein, T.; Van De Linde, S.; Sauer, M. Light-induced cell damage in live-cell super-resolution microscopy. Sci. Rep. 2015, 5, srep15348. [CrossRef] 50. Tsai, Y.L.; Huang, H.H.; Lin, C.-C.; Huang, Y.M.; Yang, S.M.; Kuo, W.H.; Fang, Y.H.; Hsu, S.C.; Huang, C.P.; Shih, H.Y.; et al. High Performance Ultraviolet Micro-LED Arrays for Fine-Pitch Micro Displays. In Proceedings of the 2019 IEEE Photonics Conference (IPC), San Antonio, TX, USA, 29 September–3 October 2019; Institute of Electrical and Electronics Engineers (IEEE): Piscataway, NJ, USA, 2019; Volume 2, pp. 1–2. 51. Olivier, F.; Tirano, S.; Dupré, L.; Aventurier, B.; Largeron, C.; Templier, F. Influence of size-reduction on the performances of GaN-based micro-LEDs for display application. J. Lumin. 2017, 191, 112–116. [CrossRef] 52. Herrnsdorf, J.; McKendry, J.J.D.; Zhang, S.; Xie, E.; Ferreira, R.; Massoubre, D.; Zuhdi, A.M.; Henderson, R.K.; Underwood, I.; Watson, S.; et al. Active-Matrix GaN Micro Light-Emitting Diode Display With Unprecedented Brightness. IEEE Trans. Electron Devices 2015, 62, 1918–1925. [CrossRef] 53. Liu, Z.; Chong, W.C.; Wong, K.M.; Lau, K.M. GaN-based LED micro-displays for wearable applications. Microelectron. Eng. 2015, 148, 98–103. [CrossRef]