Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Foldscope: Origami-Based Paper Microscope

Foldscope: Origami-Based Paper Microscope Here we describe an ultra-low-cost origami-based approach for large-scale manufacturing of microscopes, specifically demonstrating brightfield, darkfield, and fluorescence microscopes. Merging principles of optical design with origami enables high-volume fabrication of microscopes from 2D media. Flexure mechanisms created via folding enable a flat compact design. Structural loops in folded paper provide kinematic constraints as a means for passive self-alignment. This light, rugged instrument can survive harsh field conditions while providing a diversity of imaging capabilities, thus serving wide-ranging applications for cost-effective, portable microscopes in science and education. Citation: Cybulski JS, Clements J, Prakash M (2014) Foldscope: Origami-Based Paper Microscope. PLoS ONE 9(6): e98781. doi:10.1371/journal.pone.0098781 Editor: Lennart Martens, UGent/VIB, Belgium Received January 25, 2014; Accepted May 7, 2014; Published June 18, 2014 Copyright:  2014 Cybulski et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: Manu Prakash acknowledges support from Terman Fellowship, The Baxter Foundation, Coulter Foundation, Spectrum Foundation (NIH CTSA UL1 TR000093), C-Idea (National Institutes of Health grant RC4 TW008781-01), Bill and Melinda Gates Foundation, Pew Foundation, and Gordon and Betty Moore foundation for financial support. Jim Cybulski is supported by NIH Fogarty Institute Global Health Equity Scholars (GHES) Fellowship. James Clements was supported by NSF Graduate fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: This work is covered under patent application PCT/US2013/025612 (Title: Optical Device). Further details, including the 10,000 microscope project and updates on the latest status of this work, are available on www.foldscope.com. This does not alter the authors’ adherence to PLOS ONE policies on sharing data and materials. * Email: [email protected] By combining principles of optical design with origami [9–11], Introduction here we present a novel platform for the fabrication of flat Microscopes are ubiquitous tools in science, providing an microscopes cheaply in bulk (figure 1). The Foldscope is an essential, visual connection between the familiar macro-world and origami-based optical microscope that can be assembled from a the remarkable underlying micro-world. Since the invention of the flat sheet of paper in under 10 minutes (see video S1, figure S8). microscope, the field has evolved to provide numerous imaging Although it costs less than a dollar in parts (see Bill of Materials in modalities with resolution approaching 250 nm and smaller [1]. table 1), it can provide over 2,0006magnification with submicron However, some applications demand non-conventional solutions resolution, weighs less than two nickels (8.8 g), is small enough to due to contextual challenges and tradeoffs between cost and fit in a pocket (7062062mm ), requires no external power, and performance. For example, in situ examination of specimens in the can survive being dropped from a 3-story building or stepped on field provides important opportunities for ecological studies, by a person (see figure 1G and video S2). Its minimalistic, scalable biological research, and medical screening. Further, ultra-low cost design is inherently application-specific instead of general-purpose, DIY microscopes provide means for hands-on science education in providing less functionality at dramatically reduced cost. Using this schools and universities. Finally, this platform could empower a platform, we present our innovations for various imaging worldwide community of amateur microscopists to capture and modalities (brightfield, darkfield, fluorescence, lens-array) and share images of a broad range of specimens. scalable manufacturing strategies (capillary encapsulation lens Cost-effective and scalable manufacturing is an integral part of mounting, carrier tape lens mounting, self-alignment of micro- ‘‘frugal science and engineering’’ [2]. For example, manufacturing optics by folding, paper microscope slide). via folding has emerged as a powerful and general-purpose design The Foldscope is operated by inserting a sample mounted on a strategy with applications from nanoscale self-assembly [3] to microscope slide (figure 1B), turning on the LED (figure 1C), and large-aperture space telescopes [4]. More recently, possibilities of viewing the sample while panning and focusing with one’s thumbs. folding completely functional robots have been explored [5–7], The sample is viewed by holding the Foldscope with both hands with actuators, sensors and flexures integrated in a seamless and placing one’s eye close enough to the micro-lens so one’s fashion. Modern micro-lens fabrication technology is another eyebrow is touching the paper (figure 1F). Panning is achieved by prime example of scalable manufacturing. Although the use of placing one’s thumbs on opposite ends of the top stage (colored high-curvature miniature lenses traces back to Antony van yellow in figure 1A–C) and moving them in unison, thus Leeuwenhoek’s seminal discovery of microbial life forms [8], translating both optics and illumination stages while keeping the manufacturing micro-lenses in bulk was not possible until recently. stages aligned (figure 1B). Focusing is achieved using the same Modern techniques such as micro-scale plastic molding and positioning of one’s thumbs, except the thumbs are pulled apart (or centerless ball-grinding have grown to serve numerous applica- pushed together). This causes tension (or compression) along the tions, including telecommunication fiber couplers, cell phone optics stage, resulting in 2Z (or +Z) deflection of the micro-lens cameras, and medical endoscopes. due to flexure of the supporting structure of the sample-mounting PLOS ONE | www.plosone.org 1 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope Figure 1. Foldscope design, components and usage. (A) CAD layout of Foldscope paper components on an A4 sheet. (B) Schematic of an assembled Foldscope illustrating panning, and (C) cross-sectional view illustrating flexure-based focusing. (D) Foldscope components and tools used in the assembly, including Foldscope paper components, ball lens, button-cell battery, surface-mounted LED, switch, copper tape and polymeric filters. (E) Different modalities assembled from colored paper stock. (F) Novice users demonstrating the technique for using the Foldscope. (G) Demonstration of the field-rugged design, such as stomping under foot. doi:10.1371/journal.pone.0098781.g001 stage (figure 1C). Unlike traditional microscopes, the Foldscope micro-lenses), lens-holder apertures, an LED with diffuser or anchors the sample at a fixed location while the optics and condenser lens, a battery, and an electrical switch (see figure 1D). illumination stages are moved in sync. The three stages are weaved together to form an assembled Foldscope (figure 1B,C,E) with the following features: fully- constrained X–Y panning over a 20620 mm region (figure 1B), Results flexure-based focusing via Z-travel of the optics stage relative to Design Platform the sample-mounting stage (figure 1C), and a vernier scale for Construction from flat media. The Foldscope is comprised measuring travel distances across the sample slide with 0.5 mm of three stages cut from paper _ illumination, sample-mounting, resolution. The total optical path length from the light source to and optics _ and assembled via folding (figure 1A–C, video S1). the last lens surface is about 2.7 mm (figure S1), only 1% that of a Other primary components include a spherical ball lens (or other conventional microscope. Flat polymeric sheets and filters can also PLOS ONE | www.plosone.org 2 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope Table 1. Bill of Materials. Component Unit cost (10,000 pcs) Paper (400 cm ) $0.06 Ball Lens (low mag/high mag) $0.17/$0.56 3V button battery (CR2016) $0.06 LED $0.21 Switch $0.05 Copper Tape (5 cm ) $0.03 Foldscope $0.58/$0.97 Summary of unit costs for Foldscope components in volumes of 10,000 units, not including assembly costs. This assumes a Foldscope in brightfield constructed from the following: polypropylene sheets (Press Sense 10mil Durapro); a 140X low-mag lens (Winsted Precision Ball 2.4 mm borosilicate ball, P/N 3200940F1ZZ00A0, from www.mcmaster.com, P/N 8996K21) or a 2,180X high-mag lens (Swiss Jewel Co. 0.2 mm sapphire ball lens); a 3V CR2016 button cell (Sanyo CR2016-TT1B #8565 from Batteriesandbutter.com); a white LED (Avago ASMT CW40 from Mouser.com); an electrical slider switch (‘‘Off/On MINI SMD Switch’’ from AliExpress.com); and copper tape (Sparkfun P/N 76555A648). doi:10.1371/journal.pone.0098781.t001 be inserted into the optical path, including diffusion filters for focal length. Equally important for image quality, the illumination improving illumination uniformity, Fresnel lenses as condensers source (LED plus diffuser and/or condenser lens) should provide for concentrating illumination intensity, color filters for fluores- even illumination over the field, ample intensity, narrow intensity cence imaging, and linear polarizers for polarization imaging. profile, and high CRI (color rendering index). The LED used in Alignment by Folding. Folding provides a passive alignment the Foldscope consumes only 6 mW of electrical power and can mechanism that is used here to align the micro-lens with the light operate over 50 hours on a CR2032 button cell battery (figure source. A sharp crease in a thin sheet of inextensible material, such S3A). Precise control over the illumination profile is required for as paper, of thickness h introduces elastic energy of bending of the high-quality microscopy [17], so integration of a condenser lens is order ,h [12]. Thus, a fold introduces buckling at the inner edge, crucial for optimal imaging (figure S3C). For low-magnification giving variation in the exact location of the hinge and resulting in imaging applications not requiring optimal imaging, the illumina- random alignment error of the order ,h. To minimize this error, tion source can be removed and the Foldscope can be operated we introduce folding features that form a closed structural loop while facing an external light source. between the optics stage and the illumination stage. This improves alignment repeatability through elastic averaging within kinematic Design Innovations constraints (figure 1A; [13]). We characterized alignment accuracy High-Resolution Brightfield Microscopy. For some appli- and repeatability by constructing twenty independent Foldscopes cations, extending the resolution limit of the Foldscope to out of 350 mm thick black cardstock and manually folding and submicron length scales is a practical necessity. For this reason, unfolding them twenty times each (see Materials and Methods the resolution of the single-ball-lens Foldscope was further section), while measuring absolute X–Y alignment (figure S2). optimized and empirically characterized. The analytical optimi- Assembly repeatability was assessed as the mean value of twice the zation was carried out for a single field point at the optical axis to standard deviation for each Foldscope (65 mm in X and 25 mmin assess the best achievable resolution (see Modeling and Charac- Y), while assembly accuracy was assessed as the mean value of all terization and table 2). A 1,450X Brightfield Foldscope with the trials (59 mm in X and 67 mm in Y). A higher skew in X-axis configuration depicted in figure 2E was used to capture the image repeatability results from structurally distinct constraints imple- in figure 2A, empirically confirming submicron resolution. As mented for the X- and Y-axes. The small assembly accuracy errors shown in figure S4A, spherical ball lenses have significant (less than 20% of the paper thickness) in both directions are wavefront error at the edge of the field defined by the aperture consequences of the design which can be compensated by feature (aperture shown in figure S1). As a result, not all regions can be shifts in future designs. simultaneously in focus within this field. The center portion of the Micro-Optics and Illumination. The Foldscope design field, with wavefront error less than 1/5 wavenumber and low accommodates different optical configurations, including spherical curvature and distortion, is denoted the ‘‘optimal field of view’’ ball lenses, spherical micro-lens doublets (such as a Wollaston (figure S4A–C). Thus, the best achievable resolution is attained at doublet), and more complex assemblies of aspheric micro-lenses. the expense of a reduced field of view. When a digital sensor is While more optical elements generally provide reduced aberration used in place of the naked eye, the lens fixture effectively reduces and improved field of view, spherical ball lenses have distinct the field of view to roughly the optimal field of view. advantages for high-volume manufacturing, including reduced Fluorescence. Conventional fluorescence microscopy typi- part count and simplified assembly due to rotational symmetry cally requires an expensive illumination source for high-intensity [14–16]. Since magnification varies inversely with ball-lens broad-spectrum illumination and multiple optical elements with diameter, commonly available ball lenses provide an ample range precisely defined spectral profiles. The simplified configuration of of magnifications (under 100X to over 2,000X, as seen in table 2). the Fluorescence Foldscope uses a high-intensity colored LED of The back focal length of these lenses varies drastically, thus narrow spectral width and polymeric sheets inserted in the optical motivating alternative lens-mounting schemes (above the optics path for a shortpass excitation filter and a longpass emission filter stage, as in figure 1C, or below) and requiring samples with no (figure 2B,F). A blue LED light source and commonly available gel coverslip for lenses with less than approximately 140 mm back filters (with spectral transmissivities plotted in figure S3B) were PLOS ONE | www.plosone.org 3 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope PLOS ONE | www.plosone.org 4 June 2014 | Volume 9 | Issue 6 | e98781 Table 2. Foldscope Analytical Model Parameter Summary Table. Parameter Values at Select Combinations of r,n Parameter Functional Form in Optimized System r = 1200 r = 500 r = 400 r = 150 r = 150 r = 100 n = 1.517 n = 1.517 n = 1.517 n = 1.517 n = 1.77 n = 1.77 2:5e5mm n{1 MAG 140 340 430 1140 1450 2180 MAG~ ~ðÞ 5e5mm EFL n r BFL 1 rðÞ 2{n 561 234 187 70 22 15 BFL~EFL{r~ 2 n{1 1=4 RES 1.90 1.52 1.44 1.13 0.86 0.77 3: : : : 1=4 l r n½ nzðÞ 2{nðÞ 2n{1 : : 3: : RES~k f l jj s ~k 2 2 128ðÞ n{1 1=4 1=4 nOAR : 3 0.294 0.366 0.387 0.495 0.510 0.565 a k l 8l n : : nOAR~ ~ ~k : : : r r jj s rðÞ n{1½ nzðÞ 2{nðÞ 2n{1 1=4 OAR 1=4 : 3: 3 353 183 155 74 77 56 l 8l n r : : : OAR~nOAR r~k ~k 1 1 : : s ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 jj EFL 1 n r 1761 734 587 220 172 115 EFL~ 2 ðÞ n{1 1=4 NA 3 0.200 0.249 0.264 0.337 0.444 0.491 : : 2aðÞ n{1 128 lðÞ n{1 NA~ ~k : : : : r n r n½ nzðÞ 2{nðÞ 2n{1 1=4 FOV 518 268 227 109 88 65 : : 3: 7 n a l r n FOV~ ~k : : 2ðÞ n{1 2ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DOF : : : : 13.7 8.8 7.9 4.8 2.8 2.3 l 1 l r n½ nzðÞ 2{nðÞ 2n{1 DOF~ ~ 2 2 3 NA k 128ðÞ n{1 {1=8 SR 0.8825 0.8825 0.8825 0.8825 0.8825 0.8825 SR~e Functional form and select numerical values for the following dependent parameters: Magnification (MAG), Back Focal Length (BFL), Resolution (RES), nOAR (Normalized Optimal Aperture Radius), OAR (Optimal Aperture Radius), Effective Focal Length (EFL), Numerical Aperture (NA), Field of View (FOV), Depth of Field (DOF), Strehl Ratio (SR). These are calculated for infinite ‘‘object’’ distance per analytical model RM2, with aperture radius 1=4 3 a~OAR~kðÞ l=s , k %0:9321, k %0:7415, and with aberration coefficient s~{ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 ðÞ 2rn . All calculations assume an incident wavelength of l = 0.55 mm, and all specified distances are in units of mm. 1 1 2 The indices of refraction n = 1.517 and n = 1.77 correspond to borosilicate glass and sapphire, respectively. doi:10.1371/journal.pone.0098781.t002 Origami-Based Paper Microscope Figure 2. Foldscope imaging modalities. (A) Brightfield Foldscope image of a monolayer of 1 mm polystyrene microspheres (Polysciences 07310- 15) using a 1,450X lens. (B) Fluorescent Foldscope image of 2 mm polyfluorescent microspheres (Polysciences 19508-2) using a 1,140X lens with Roscolux gel filters #19 and #80. (C) 2X2 lens-array Brightfield Foldscope image of Giemsa-stained thin blood smear using 1,450X lenses. (D) 140X Darkfield Foldscope images of 6 mm polystyrene microspheres (Polysciences 15714-5), using a 140X lens for the darkfield condenser. Darkfield condenser aperture shown in inset has 1.5 mm inner diameter and 4.0 mm outer diameter. (E–H) Schematic cross-sections of Brightfield, Fluorescence, Lens-Array, and Darkfield Foldscope configurations, showing the respective arrangements of ball lenses, filters, and LEDs. See table 2 for ball lenses used for specific magnifications. doi:10.1371/journal.pone.0098781.g002 used to image 2 mm diameter red poly-fluorescent polystyrene configured in a single Foldscope (figure 2C,G). Such a lens array beads as shown in figure 2B. For fluorescent imaging requiring may be comprised of identical lenses or of different lenses with higher contrast, small pieces (3 mm square or smaller) of different magnifications and/or back focal lengths. This provides interference filters can be used in place of the polymeric sheets for the design of a lens-array Foldscope with several key features. at reasonable cost due to the small size. For non-contiguous samples such as blood smears, a larger field of Lens-array and Multi-modality. Since a micro-lens has a view can be obtained by overlapping a number of small fields of very small footprint, multiple optical paths can be independently view. Alternatively, an optimal array pitch will give tangential non- Figure 3. Manufacturing innovations for lens- and specimen- mounting. (A) Fabrication, mounting, and characterization of capillary- encapsulation process for lens-mounted apertures. X and Y error bars for all measurements are 2.5 mm. (B) Reel of polystyrene carrier tape with custom pockets and punched holes for mounting over 2,000 ball lenses with optimal apertures. The first ten pockets include mounted ball lenses. Inset shows sectioned view from CAD model of carrier tape mounted lenses. Note the aperture is the punched hole shown on the bottom side of the ball lens. This tape is 16 mm wide and is designed for 2.4 mm ball lenses (aperture diameter is 0.7 mm). (C) Top: Paper microscope slide shown next to standard glass slide with coverslip, both with wet mount algae specimens. Bottom: Schematic of paper microscope slide, showing specimen containment cavity formed between upper tape and lower tape in middle of slide. doi:10.1371/journal.pone.0098781.g003 PLOS ONE | www.plosone.org 5 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope PLOS ONE | www.plosone.org 6 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope Figure 4. Analytical, numerical, and empirical characterization of Foldscope. (A,B) Analytical ‘‘design curves’’ for normalized optimal aperture radius (nOAR) and optimal resolution (RES) versus magnification (MAG) over index of refraction (range 1.33–1.91) and ball lens radius (range 40–1200 mm). (C) Comparison of analytical (3D surface) and numerical (plotted as points) results for RES versus index of refraction and ball lens radius. (D) Modulus of the Optical Transfer Function (MTF) over the optimal field of view for a 300 mm sapphire lens with optimal aperture. (E,F) Image of USAF 1951 resolution target taken with 430X ball lens, including an enlarged caption of Group 9, and an intensity profile plot along path denoted by green line in image caption. This demonstrates resolvability for Group 9, Element 4 corresponding to 724 Line Pairs/mm or 1.38 mm resolution. (G,H) Image of USAF 1951 resolution target taken with 140X ball lens, including an enlarged caption of Group 8, and an intensity profile plot along path denoted by green line in image caption. This demonstrates resolvability for Group 8, Element 6 corresponding to 456 Line Pairs/mm or 2.19 mm resolution. The data was taken using GUPPY Pro 503C scientific camera, with 259261944 pixels and pixel size 2.262.2 mm . doi:10.1371/journal.pone.0098781.g004 overlapping fields of view (as seen in figure 2C), thus reducing the removed from the synthetic paper and replaced so that the slide time required to scan a slide for a feature such as a parasite. Since can be reused for many specimens. Note that the specimen individual lenses have independent optical paths, the novel depicted in figure 2D was mounted on a paper microscope slide. capability of building multi-modality lens-array microscopes arises. One such combination is a two-by-one array of brightfield and Modeling and Characterization fluorescence modalities, which could be used to scan a sample for Theory and analysis. For a brightfield Foldscope, basic the presence of fluorescence markers and then identify the non- measures of optical performance can be described in terms of the fluorescent surrounding structures. ball radius (r), index of refraction (n), aperture radius (a), and Darkfield. The Darkfield Foldscope configuration, shown in incident wavelength (l; see text S1). Assuming the paraxial figure 2H, requires a diffuser, a darkfield condenser aperture (inset approximation, these include effective focal length (EFL), back in figure 2D), and a condenser lens. The diffuser helps to evenly focal length (BFL), and magnification (MAG). For a 300 mm distribute light from the small LED over the aperture area, while sapphire ball lens: EFL = 172 mm, BFL = 22 mm, and the condenser focuses a hollow cone of light onto the specimen. MAG = 1,450X. Thus substantial magnification can be obtained, Since the specimen must be placed at the focal point of the but the sample must be separated from the lens by only a fraction condenser, the slide thickness has to match the back focal length of of the thickness of a human hair. Three additional optical the condenser plus the spacing from the condenser to the slide. A performance metrics include field of view radius (FOV), numerical 140X Darkfield Foldscope was used to image 6 mm polystyrene aperture (NA), and depth of field (DOF). These depend on microspheres as shown in figure 2D. aperture radius (a), the optimization of which is discussed below. Capillary Encapsulation Lens Mounting. The process of For the previous example, the normalized optimal aperture radius precisely mounting micro-optics to an aperture crucially governs is nOAR = a/r = 0.51, giving: FOV = 88 mm, NA = 0.44, lens performance. Therefore, a capillary encapsulation process was DOF = 2.8 mm (see table 2). developed to automatically mount a ball lens while forming a The aperture radius controls the balance between diffraction circular aperture of precisely tunable diameter (see figure 3A). By effects from the edges of the aperture with spherical aberration partially engulfing ball lenses in an opaque polymer held between effects from the lens. Therefore, a complete analytical model was two glass substrates coated with flat nonstick PDMS (see top left of created to predict the normalized optimal aperture radius (nOAR) figure 3A), a precise aperture is self-assembled around the ball-lens and optimal resolution (RES), as well as the aberration coefficient (see Materials and Methods section for details). Pressure applied (s) for a ball lens (see Supplementary Materials and table 2), between the substrates is used to precisely tune aperture size, with yielding: greater pressure providing a larger aperture. The epoxy encap- sulated lens is then adhesively mounted to a paper aperture and pffiffiffiffiffi 1=4 inserted in the Foldscope (see bottom right of figure 3A). 1=4 nOAR~ðÞ k =rðÞ l=jj s , k ~ 3 10 4p %0:9321 1 1 Carrier Tape Lens Mounting. Black polystyrene carrier tape is commonly used for low-cost reel-to-reel packaging of electronic components. This pre-existing infrastructure was 1=4 1=4 1=8 leveraged to create low-cost mounting structures for ball lenses RES~k fðÞ l=jj s , k ~1:22ðÞ p=12 ðÞ e=10 %0:7415 2 2 with optimized apertures. As depicted in figure 3B, the custom thermoformed pocket holds the lens in place with a press fit, and a punched hole in the bottom of the pocket precisely defines the s~{ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 ðÞ 2rn aperture. A single lens is cut from the carrier tape and adhesively attached to a paper aperture which is then inserted into the optics The expressions for nOAR and RES are depicted as 2D design stage of the Foldscope, analogous to that shown for the epoxy plots as a function of desired MAG in figure 4A,B and figure 4C encapsulated lens in figure 3A. shows a 3D plot of RES over n and r. For the example discussed Paper microscope slide. A low-cost microscope slide earlier, the values for normalized aperture radius and resolution (similar to that in [18]) was constructed out of 18mil polystyrene are found by locating the intersection of the lines for r = 150 mm synthetic paper and transparent tape as shown in figure 3C. If tape and n = 1.77 in the design plots. This gives nOAR = 0.51 and is placed on only one side (either upper or lower), specimens are RES = 0.86 mm, and corresponds to MAG = 1,450X. Note that conveniently mounted on the exposed sticky surface of the tape. the regions enclosed by the curves in the 2D design plots represent Using both upper and lower pieces of transparent tape creates a the available design space for nOAR, RES, and MAG as defined cavity for mounting wet specimens such as live algae suspended in by the range of possible values for n and r. The design curves thus water. Since the paper microscope slide is less than half the make it a simple exercise to pick optimal design parameters within thickness of a standard glass slide, a spacer is required to elevate the space of interest. Also, one can see from figure 4B that the the sample closer to the lens. This is achieved by inserting the specimen slide together with two blank paper slides beneath it. lower limit for the best achievable resolution in ball lenses appears Once the specimen has been viewed, the transparent tape can be PLOS ONE | www.plosone.org 7 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope to be near 0.5 mm, based on the range of parameters identified for For the 430X lens, note that the empirical value is 4.2% smaller this figure. than the theoretical value. A less conservative theoretical model Numerical Modeling. A ray-tracing numerical model was (see text S1) predicts values 11.5% smaller than those in table 2, developed for the Foldscope using Zemax software to confirm the indicating the empirical value is reasonable and near the limits results of the analytical model and to evaluate points across the predicted by theory. For the 140X lens, note that the empirical field of view (see Materials and Methods section for details). The value is 15.3% larger than the theoretical value. This difference is results for nOAR and RES show very good agreement, with attributed to the fact that this particular lens was selected for its 2 2 correlation coefficients of R = 0.985 for nOAR and R = 0.998 low cost at the expense of larger tolerances on diameter and for RES (see figure 4C). The numerical modeling results across the sphericity (grade 48) compared to the 430X lens (grade 10). For field of view are shown in figure S4A, where the optimal field of example, diameter tolerance per ball sphericity is 0.25 mm for view used for calculating the MTF (Modulus of the Optical grade 10 and scales linearly with grade number. Transfer Function) is defined. The MTF for this system is plotted Since the USAF 1951 resolution target has a glass covering with in figure 4D for a field point on the optical axis and another at the thickness comparable to a standard coverslip (140 mm), ball lenses edge of the optimal field of view. The field point at the optical axis with magnification higher than 430X could not be assessed with shows near-diffraction-limited response, and the tangential and this resolution target due to short back focal length. Instead, a sagittal curves for the edge of the field drop to half of their low- monolayer of 1 mm polystyrene beads was imaged using a 1450X frequency value at a spatial frequency of about 300 cycles/mm. ball lens (figure 2A) to demonstrate sub-micron resolution for this Empirical Characterization of Resolution. The resolution lens (theoretical value is 0.86 mm). of the 430X and 140X lenses were empirically characterized using a standard USAF 1951 resolution target and a GUPPY Pro 503C Discussion and Conclusions scientific camera (with bare sensor) as shown in figure S7. Images taken with these lenses are shown in figure 4E,G, and the relevant By removing cost barriers, Foldscope provides new opportuni- ties for a vast user base in both science education and field work intensity profiles are shown in figure 4F,H. The formula for the resolution (in line pairs/mm) of a given group and element of the for science and medicine. Many children around the world have Group+(Element21)/6 never used a microscope, even in developed countries like the USAF 1951 target is given by RES =2 , and LP the center-to-center distance between the lines in microns is given United States. A universal program providing ‘‘a microscope for by RES = 1000/RES . From table 2, we see the theoretical every child’’ could foster deep interest in science at an early age. CC LP While people have known for decades that hands-on examination values for the resolution of the lenses are 1.44 mm and 1.90 mm, respectively. Based on the critically resolved intensity profiles in and inquiry is crucial in STEM (Science, Technology, Engineer- figure 4F,H, the empirical values for RES are found to be ing, and Mathematics) education [19–20], the challenge posed by CC 1.38 mm and 2.19 mm. J. M. Bower to engage ‘‘all teachers and all children’’ [21] requires Figure 5. Mosaic of Foldscope Images. Bright field images of (A) Giardia lamblia (2,180X), (B) Leishmania donovani (1,450X), (C) Trypanosoma cruzi (1,450X), (D) gram-negative Escherichia coli (1,450X), (E) gram-positive Bacillus cereus (1,450X), (F) Schistosoma haematobium (140X), and (G) Dirofilaria immitis (140X). Unstained (H) leg muscles and (I) tarsi of an unidentified ladybug (genus Coccinella). (J) Unstained leg muscles (fixed in formaldehyde) of an unidentified red ant (genus Solenopsis). An LED diffuser (Roscolux #111) was added for (A) and an LED condenser (2.4 mm borosilicate ball lens) was used for (C). Images (H–J) were taken by novice user using a self-made Foldscope (140X). See table 2 for ball lenses used for specific magnifications. White scale bar: 5 mm; black scale bar: 100 mm. doi:10.1371/journal.pone.0098781.g005 PLOS ONE | www.plosone.org 8 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope large-scale adoption of practices and broad availability of tools that 2D Media and Filters were previously cost-prohibitive [22]. Moreover, the opportunity The 2D media used in constructing Foldscopes included black to make microscopes both approachable and accessible can inspire 105 lb card stock (ColorMates Smooth & Silky Black Ice Dust children to examine the rich bio-diversity on our planet as Card Stock, purchased from thePapermillstore.com), polypropyl- amateur microscopists and to make discoveries of their own, as ene (PressSense Durapro CC 10mil), and others. Foldscope parts already seen in the field of amateur astronomy ([23]; see images were cut from 2D media using a CO laser (Epilog Elite, Mini24). taken by novice user with self-made Foldscope in figure 5H–J). Copper tape was used for providing connectivity (by soldering) Disease-specific Foldscope designs are an important vision for between the LED, battery, and switch. The filters used in future development [24–25]. Figure 5 depicts early bench-test constructing Foldscopes included Roscolux colored gel filters data, including high-magnification brightfield images of Giardia (including Primary Blue #80 and Fire Red #19, which lamblia, Leishmania donovani, Trypanosoma cruzi (Chagas parasite), approximate an Acridine Orange filter set), Roscolux diffuser Escherichia coli, and Bacillus cereus (figure 5A–E), and low-magnifi- filters (Tough Rolex #111), and polymeric linear polarizers cation brightfield images of Schistosoma haematobium and Dirofilaria (Edmund Optics P/N 86181). Each type of filter is assembled to immitis (figure 5F–G). Note that these include magnifications the Foldscope by cutting out a 3–5 mm square piece and ranging from 140X to 2,180X, none of which require immersion adhesively attaching it to the appropriate stage with single-sided oil. In the future, darkfield and fluorescence Foldscopes will also be or double-stick Scotch tape. Paper microscope slides were adapted for diagnostics, and sensitivity and specificity will be constructed from polypropylene sheets (PressSense Durapro CC measured for various disease-specific Foldscopes in the field as 18mil) and transparent scotch tape. clinical validations against existing diagnostic standards. Constructing instruments from 2D media provides other unique LEDs, Switches and Power Sources advantages and opportunities. Embedding flat rare-earth magnets The LEDs used in constructing Foldscopes included the Avago in paper provides means for magnetic self-alignment, allowing the HSMW-CL25 (now replaced by P/N Avago ASMT CW40) white Foldscope to be reversibly coupled to a conventional smartphone LED for brightfield Foldscopes, the Kingbright APTD1608QBC/ for image capture, for smartphone-based diagnostics, or for D blue LED for fluorescence Foldscopes. The electrical slider telemedicine [26–29]. By printing text and images on the paper, switch was purchased from AliExpress.com (‘‘Off/On MINI SMD this platform provides an efficient information-delivery method for Switch’’, Product ID: 665019103). The power sources included specific staining protocols, pathogen identification guides, or Duracell 3V CR2032 button cells, Sanyo 3V CR2016 button cells language-free folding instructions (figure S6). Some applications (Sanyo CR2016-TT1B #8565 from Batteriesandbutter.com), and in highly infectious diseases may benefit from a disposable a GW Instek DC power supply (Model GPD-3303D). Button cells microscope _ or ‘‘use-and-throw’’ microscopy _ where the entire were used with no resistors for Foldscopes. microscope can be incinerated. Also, in place of a glass slide, the 2D media also allows direct addition of the sample to a paper- Aperture Manufacturing based micro-fluidic assay [30] for automated staining and/or This method produces inexpensive apertures through polymer pathogen-concentration, thus yielding an independent fully- encapsulation of ball lenses while preserving the optical quality of functional diagnostic system. the lens and allowing multiple lenses to be encapsulated at once. Future work will build upon the key features of this platform. The experimental setup shown in the top left of figure 3A was used Roll-to-roll processing of flat components and automated ‘‘print- to encapsulate 300 mm sapphire ball lenses with aperture and-fold’’ assembly make yearly outputs of a billion units diameters ranging from 100 mm to 214 mm. The lens was attainable. Ongoing work with advanced micro-optics and sandwiched between parallel substrates (glass or silicon) coated illumination design _ including spherical GRIN lenses [31–32], with planar films of PDMS with thickness greater than 1 mm aspheric multi-lens optics, and condenser lens provisions for (formed from Dow Corning Sylgard 184 PDMS). A micrometer Ko¨hler illumination _ is expected to improve both resolution and stage was used to precisely apply pressure between the substrates field of view at low cost. International field-work in both to adjust the diameter of the resulting elastic deformation of the diagnostics and education will provide vital inputs for further PDMS film. This diameter was measured in situ using phase improvements. Our long-term vision is to universalize frugal contrast microscopy to set the target value for the aperture. A fast- science, using this platform to bring microscopy to the masses. curing opaque polymer (Smooth-On Smooth-Cast Onyx Fast Polyurethane) was then injected into the cavity and allowed to Materials and Methods cure. Reflected light microscopy was used to measure the dimensions of the final aperture formed. Once removed from Ball Lenses the non-stick PDMS films, the encapsulated lens was attached to The ball lenses used in constructing Foldscopes included the underside of the optics stage of a Foldscope material types borosilicate, BK7 borosilicate, sapphire, ruby, and S-LAH79. The vendors included Swiss Jewel Co, Edmund Optics, Characterization of Self-Alignment by Folding and Winsted Precision Ball. Part numbers for some select lenses Twenty independent microscopes were cut out of black 105 lb include: 300 mm sapphire lens from Swiss Jewel Co. (Model cardstock, each marked with a cross-hair in both the optics and B0.30S), 200 mm sapphire lenses from Swiss Jewel Co. (Model illumination stages (see figure S2C). After folding, alignment was B0.20S), 2.4 mm borosilicate lenses from Winsted Precision Ball measured using a dissection microscope (Olympus upright, 306 (P/N 3200940F1ZZ00A0), 300 mm BK7 borosilicate lenses from magnification) via digitizing the cross-hair images, drawing lines Swiss Jewel Co. (Model BK7-0.30S), and 1.0 mm BK7 borosilicate through the center of each cross-hair (X and Y cross-hairs on both lenses from Swiss Jewel Co. (Model BK7-1.00). Note that half-ball stages), and digitally measuring the X and Y displacements to lenses from both Edmund Optics and Swiss Jewel Co. were also characterize the alignment. Every Foldscope was iteratively folded, tested for use as condenser lenses for the LEDs. imaged to record X–Y alignment, and unfolded twenty times. The PLOS ONE | www.plosone.org 9 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope data was then used to assess accuracy and repeatability (see figure length from the LED to the lens is almost an order of magnitude S2A,B). smaller than the size of the human eye. (TIF) Sample Preparation Figure S2 Characterization of Self-Alignment by Fold- Thin-blood smears of Plasmodium falciparum (ring stage), Trypano- ing. Twenty independent Foldscopes were constructed out of soma cruzi, Giardia lamblia, Leishmania donovani, and Dirofilaria immitis 350 mm thick black cardstock and manually folded and unfolded were freshly prepared from cultures provided by Center for twenty times each, with alignment measured after each assembly. Discovery and Innovation in Parasitic Diseases (CDIPD) at UCSF. The data was used to produce plots of (A) assembly repeatability The samples were fixed in methanol and stained in freshly (distribution of all 400 values, adjusted to give zero mean for each prepared Giemsa solution (Sigma Aldrich, #48900-500ML-F) Foldscope) and (B) assembly accuracy (distribution of 20 mean using standard protocols before imaging. Once fixed, the slides values calculated per Foldscope) using (C) cross-hair alignment could be used for several weeks. Bacterial samples of Bacillus cereus features on the optics and illumination stages. Note that the span and Escherichia coli were provided by KC Huang Lab at Stanford of the data in both plots is less than the thickness of the paper used University. The samples were heat fixed onto glass slides using to construct the Foldscopes. Based on the data shown in the plots, standard procedures and gram stained using standard protocols assembly repeatability was assessed as the mean value of twice the (Fisher Scientific Gram Stain Set, Catalog No. 23-255-959). standard deviation for each Foldscope (65 mm in X and 25 mmin Plasmodium-infected red blood cells were taken from cultures Y), while assembly accuracy was assessed as the mean value of all provided by the Center for Discovery and Innovation in Parasitic trials (59 mm in X and 67 mm in Y). A higher skew in X-axis Diseases (CDIPD). Schistosoma haematobium were provided by the repeatability results from structurally distinct constraints imple- Michael Hsieh Lab at Stanford University. Insects used for mented for the X- and Y-axes, while the assembly accuracy errors imaging were caught on Stanford campus and imaged after fixing in both directions are consequences of the design which can be in formaldehyde without any stain. No human samples were compensated by feature shifts in future designs. Note that the X utilized in the current work. and Y error bars for all measurements are 8.4 mm. (TIF) Image-Capture Protocol Figure S3 Component Characterization. (A) LED voltage Brightfield images were taken using a Canon EOS 5D Mark II and intensity versus time for a white LED (Avago HSMW-CL25) with the Foldscope placed 3 cm away from the 100 mm focal powered by a Duracell CR2032 battery with no resistor. (B) Filter length lens and using the following settings: F/3.2, 1/30 sec. transmission spectra of three Roscolux filters _ Tough Rolex exposure, ISO-2000. An initial image was first captured using diffuser (#111), Fire Red (#19), and Primary Blue (#80) _ automatic white-balance and then used as a reference white measured with Ocean Optics Photo spectrometer USB4000. (C) balance image during data collection. Fluorescence images were Intensity profile of a white LED (Avago HSMW-CL25) as taken in a similar fashion to the brightfield images with typical visualized in water with dissolved fluorescein. The left image is camera settings: F/2.8, 15 sec. exposure, ISO-1000. Although not taken with the bare LED while the right image is taken with a presented, images were also obtained by coupling the Foldscope to condenser lens (2.4 mm borosilicate ball lens) placed adjacent to cell-phones including an iPhone using a custom magnetic coupler. the LED in the optical path, demonstrating that a ball lens can be USAF 1951 resolution target data was taken using GUPPY Pro used to effectively collimate the light emitted by this LED. 503C scientific camera, with 259261944 pixels and pixel size (TIF) 2.262.2 mm , using the setup shown in Figure S7. Figure S4 Numerical Modeling Characterization of Optimal Field of View. (A) Plot of Wavefront Error over the Numerical Model full field of view defined by the aperture for a 300 mm Sapphire Zemax software was used to model the Foldscope optics to ball lens with a 147 mm aperture. With increasing field coordinate, assess optimal aperture radius and resolution. The basic model of the Wavefront Error becomes very large and the image will be out the system consists of a ball lens, an aperture, an object at infinity, of focus. An ‘‘optimal field of view’’ is defined at a field coordinate and an image plane (see figure S5A). This model requires two of 21 mm, where the Wavefront Error is approximately 1/5 wave parameters to be independently optimized _ lens-image distance number. (B,C) Plots of Field Curvature and Distortion over the and aperture radius. The analysis is carried out in four steps: 1) optimal field of view. (D) Plot of RMS spot size over the optimal optimize lens-image distance in model by minimizing focusing field of view depicting four cases: optimized solution treated as metric (figure S5C); 2) determine search space for aperture radius reference with zero defocus (red line), defocus of 3 mm (green line), as defined by empirically chosen limits on Strehl Ratio, 0.75–0.98; defocus of 3 mm (green line), diffraction limit (dashed black line). 3) optimize aperture radius using resolution metric (figure S5D); The reference solution provides the best achievable resolution at and 4) use Matlab surface-fitting tool to fit data for optical the center of the field of view (approximately equal to the performance parameters as functions F(n,r, l) and compare with diffraction limit for this choice of aperture), while other plots show analytical model. that increasing defocus moves the region of best resolution radially out from the center in an annular ring. (E) Plot of RMS spot size Supporting Information over the optimal field of view depicting optimal aperture predicted by analytical model (red lines) and adjusted aperture giving Figure S1 Foldscope Schematics. (A) Real image formation uniform RMS spot size over the field of view (purple lines). via projection. (B) Virtual image formation via direct observation (TIF) with the eye. Note the drawings are not to scale. The indicated lengths are example values that show the versatility of this design Figure S5 Diagrams and Plots for Numerical and as well as its extreme space efficiency. For example, the same Analytical Models. (A) Schematic of time-reversed Zemax system can be used for projecting or imaging simply by changing model showing collimated light coming from an object at infinity, the object-lens distance by about 20 mm. Also, notice the total path passing through aperture, and focused by the ball lens onto a focal PLOS ONE | www.plosone.org 10 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope point in the image plane. (B) Schematic of time-reversed model Video S1 Foldscope Assembly. A short video of 140X showing key parameters used in some derivations for the analytical Brightfield Foldscope assembly process. model. (C) Plot of Focusing Metric versus Lens-Image Distance for (M4V) l = 0.55 mm, r = 150 mm, n = 1.517. This illustrates how focusing Video S2 Foldscope Drop Test and ruggedness. A short metrics FM1, FM2, and FM3 select different values for the optimal video of a three story drop test and ruggedness of Foldscope lens-image distance. (D) Plot of Resolution Metric versus Aperture demonstrated by stomping under feet. Radius for l = 0.55 mm, r = 150 mm, n = 1.517. This illustrates (AVI) how resolution metrics RM1 and RM2 select nearly the same aperture radius but yield different values for resolution. Text S1 Supplementary information. Analytical expres- (TIF) sions and derivations related to optical performance metrics. (DOCX) Figure S6 Artistic Layout of Foldscope Paper Compo- nents. Artistic version of Foldscope layout with integrated Acknowledgments universal folding instructions based on color coding, where like colors are matched during the folding process to leave a single We thank all members of the Prakash lab for valuable suggestions. We solid color in the final folded configuration. acknowledge Ioana Urama for assistance with supplementary figure S6, (TIF) Anika Radiya for assistance with magnetic couplers and Marisa Borja for taking images in figure 5H–J, and Center for Discovery and Innovation in Figure S7 Foldscope Image Capture Setup for Resolu- Parasitic Diseases (CDIPD) at UCSF and the Michael Hsieh Lab at tion Metric. Picture of the experimental setup used to caputure Stanford University for supplying samples. This work is covered under images of the USAF 1951 resolution target viewed with the patent application PCT/US2013/025612 (Title: Optical Device). Further Foldscope. The data was taken using GUPPY Pro 503C scientific details, including the 10,000 microscope project and updates on the latest camera, with 259261944 pixels and pixel size 2.262.2 mm . status of this work, are available on ?www.foldscope.com. (TIF) Author Contributions Figure S8 Quick Reference Guide for Foldscope Assem- bly. One-page handout to facilitate users in guided assembly of a Conceived and designed the experiments: JSC MP. Performed the Foldscope. experiments: JSC JC MP. Analyzed the data: JSC JC MP. Wrote the (PDF) paper: JSC MP. References 1. Keller E, Goldman R (2006) Light Microscopy. Woodbury, NY: Cold Spring inservice elementary school teachers. Journal of Elementary Science Education Harbor Laboratory Press. p. 8. 1(2): 14–34. 2. Whitesides GM (2011) The Frugal Way. The Economist - The World in 2012. 20. Bredderman T (1983) The effects of activity-based science on student outcomes: A quantitative synthesis. Review of Education Research 53: 499–518. 3. Rothemund PW (2006) Folding DNA to create nanoscale shapes and patterns. 21. Bower JM (1996) Scientists and science education reform: myths, methods, and Nature 440: 297–302. madness. National Academies of Science Resources for Involving Scientists in 4. Hyde RA (1999) Eyeglass, a large-aperture space telescope. Appl Opt 38(19): Education. Available: http://www.nas.edu/rise/backg2a.htm. Accessed: 1 May 4198–4212. 5. Hoover AM, Fearing RS (2008) Fast scale prototyping for folded millirobots. 22. Kremer M, Brannen C, Glennerster R (2013) The Challenge of Education and IEEE International Conference on Robotics and Automation 5: 886–892. Learning in the Developing World. Science 340(6130): 297–300. 6. Whitney JP, Sreetharan PS, Ma KY, Wood RJ (2011) Pop-up book MEMS. 23. O’Neill I (2011) Supernova Discovered by 10-Year-Old. Discovery News. J Micromech Microeng 21(115021): 1–7. Available: http://news.discovery.com/space/astronomy/supernova-discovered- 7. Sreetharan PS, Whitney JP, Strauss MD, Wood RJ (2012) Monolithic by-10-year-old.htm. Accessed: 2014 May 1. fabrication of millimeter-scale machines. J Micromech Microeng 22(055027): 24. Garcia LS (2007) Diagnostic medical parasitology, 5th ed. Washington, DC: 1–6. ASM Press, 142–180. 8. Porter JR (1976) Antony van Leeuwenhoek: tercentenary of his discovery of 25. Baron EJ, Miller JM, Weinstein MP, Richter SS, Gilligan PH, et al. (2013) A bacteria. Bacteriol 40: 260–269. Guide to Utilization of the Microbiology Laboratory for Diagnosis of Infectious 9. McArthur M, Lang R (2012) Folding paper: The infinite possibilities of Origami. Diseases: 2013 Recommendations by the Infectious Diseases Society of America Washington, DC: International Arts & Artists. 96 p. (IDSA) and the American Society for Microbiology (ASM). Clinical Infectious 10. Demaine ED, O’Rourke J (2007) Geometric Folding Algorithms: Linkages, Diseases, 57(4): e22–e121. doi:10.1093/cid/cit278. Origami, Polyhedra. Cambridge, MA: Cambridge Univ Press. 472 p. 26. Smith ZJ, Chu K, Espenson AR, Rahimzadeh M, Gryshuk A, et al. (2011) Cell- 11. Myer JH (1969) Optigami - A Tool for Optical System Design. Appl Opt 8: Phone-Based Platform for Biomedical Device Development and Education 260–260. Applications. PLOS ONE 6(3): e17150. doi:10.1371/journal.pone.0017150. 12. Cerda E, Mahadevan L (2005) Confined developable elastic surfaces: cylinders, 27. Breslauer DN, Maamari RN, Switz NA, Lam WA, Fletcher DA (2009) Mobile cones and the Elastica. Proc R Soc A 461: 671–700. phone based clinical microscopy for global health applications. PLOS ONE 4(7): 13. Slocum AH (1992) Precision Machine Design. Englewood Cliffs, NJ: Prentice e6320. doi:10.1371/journal.pone.0006320. Hall. 750 p. 28. Zhu H, Isikman SO, Mudanyali O, Greenbaum A, Ozcan A (2013) Optical 14. Doushkina V, Fleming E (2009) Optical and mechanical design advantages using imaging techniques for point-of-care diagnostics. Lab Chip 13: 51–67. polymer optics. Advances in Optomechanics 7424–31. 29. Bogoch II, Andrews JR, Speich B, Utzinger J, Ame SM, et al (2013) Mobile 15. Lee C-C, Hsiao S-Y, Fang W (2009) Formation and integration of a ball lens Phone Microscopy for the Diagnosis of Soil-Transmitted Helminth Infections: A utilizing two-phase liquid technology. IEEE 22nd International Conference on Proof-of-Concept Study. Am J Trop Med Hyg 88(4): 626–629. Micro Electro Mechanical Systems: pp. 172–175. 30. Martinez AW, Phillips ST, Whitesides GM (2010) Diagnostics for the 16. Jeong K-H, Liu GL, Chronis N, Lee LP (2004) Tunable microdoublet lens array Developing World: Microfluidic Paper-Based Analytical Devices. Anal Chem integrated with microfluidic network. Optics Express 12(11): 2494–2500. 82(1): 3–10. 17. Goldberg O (1980) Ko¨hler illumination. The Microscope 28: 15–22. 31. Kikuchi K, Morikawa T, Shimada J, Sakurai K (1981) Cladded radially 18. Brady S, Willard C (1998) Microscopic Explorations. UC Berkeley: Lawrence inhomogeneous sphere lenses. Appl Opt 20(3): 388–394. Hall of Science. pp. 20–21. 32. Koike Y, Kanemitsu A, Shioda Y, Nihei E, Ohtsuka Y (1994) Spherical 19. Cox CA, Carpenter JR (1989) Improving attitudes toward teaching science and gradient-index polymer lens with low spherical aberration. Appl Opt 33(16): reducing science anxiety through increasing confidence in science ability in 3394–3400. PLOS ONE | www.plosone.org 11 June 2014 | Volume 9 | Issue 6 | e98781 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png PLoS ONE Unpaywall

Foldscope: Origami-Based Paper Microscope

Cybulski, James S.; Clements, James; Prakash, Manu
PLoS ONE – Jun 18, 2014
Download PDF
11 pages

Loading next page...
 
/lp/unpaywall/foldscope-origami-based-paper-microscope-uHhNs4aJet

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
1932-6203
DOI
10.1371/journal.pone.0098781
Publisher site
See Article on Publisher Site

Abstract

Here we describe an ultra-low-cost origami-based approach for large-scale manufacturing of microscopes, specifically demonstrating brightfield, darkfield, and fluorescence microscopes. Merging principles of optical design with origami enables high-volume fabrication of microscopes from 2D media. Flexure mechanisms created via folding enable a flat compact design. Structural loops in folded paper provide kinematic constraints as a means for passive self-alignment. This light, rugged instrument can survive harsh field conditions while providing a diversity of imaging capabilities, thus serving wide-ranging applications for cost-effective, portable microscopes in science and education. Citation: Cybulski JS, Clements J, Prakash M (2014) Foldscope: Origami-Based Paper Microscope. PLoS ONE 9(6): e98781. doi:10.1371/journal.pone.0098781 Editor: Lennart Martens, UGent/VIB, Belgium Received January 25, 2014; Accepted May 7, 2014; Published June 18, 2014 Copyright:  2014 Cybulski et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: Manu Prakash acknowledges support from Terman Fellowship, The Baxter Foundation, Coulter Foundation, Spectrum Foundation (NIH CTSA UL1 TR000093), C-Idea (National Institutes of Health grant RC4 TW008781-01), Bill and Melinda Gates Foundation, Pew Foundation, and Gordon and Betty Moore foundation for financial support. Jim Cybulski is supported by NIH Fogarty Institute Global Health Equity Scholars (GHES) Fellowship. James Clements was supported by NSF Graduate fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: This work is covered under patent application PCT/US2013/025612 (Title: Optical Device). Further details, including the 10,000 microscope project and updates on the latest status of this work, are available on www.foldscope.com. This does not alter the authors’ adherence to PLOS ONE policies on sharing data and materials. * Email: [email protected] By combining principles of optical design with origami [9–11], Introduction here we present a novel platform for the fabrication of flat Microscopes are ubiquitous tools in science, providing an microscopes cheaply in bulk (figure 1). The Foldscope is an essential, visual connection between the familiar macro-world and origami-based optical microscope that can be assembled from a the remarkable underlying micro-world. Since the invention of the flat sheet of paper in under 10 minutes (see video S1, figure S8). microscope, the field has evolved to provide numerous imaging Although it costs less than a dollar in parts (see Bill of Materials in modalities with resolution approaching 250 nm and smaller [1]. table 1), it can provide over 2,0006magnification with submicron However, some applications demand non-conventional solutions resolution, weighs less than two nickels (8.8 g), is small enough to due to contextual challenges and tradeoffs between cost and fit in a pocket (7062062mm ), requires no external power, and performance. For example, in situ examination of specimens in the can survive being dropped from a 3-story building or stepped on field provides important opportunities for ecological studies, by a person (see figure 1G and video S2). Its minimalistic, scalable biological research, and medical screening. Further, ultra-low cost design is inherently application-specific instead of general-purpose, DIY microscopes provide means for hands-on science education in providing less functionality at dramatically reduced cost. Using this schools and universities. Finally, this platform could empower a platform, we present our innovations for various imaging worldwide community of amateur microscopists to capture and modalities (brightfield, darkfield, fluorescence, lens-array) and share images of a broad range of specimens. scalable manufacturing strategies (capillary encapsulation lens Cost-effective and scalable manufacturing is an integral part of mounting, carrier tape lens mounting, self-alignment of micro- ‘‘frugal science and engineering’’ [2]. For example, manufacturing optics by folding, paper microscope slide). via folding has emerged as a powerful and general-purpose design The Foldscope is operated by inserting a sample mounted on a strategy with applications from nanoscale self-assembly [3] to microscope slide (figure 1B), turning on the LED (figure 1C), and large-aperture space telescopes [4]. More recently, possibilities of viewing the sample while panning and focusing with one’s thumbs. folding completely functional robots have been explored [5–7], The sample is viewed by holding the Foldscope with both hands with actuators, sensors and flexures integrated in a seamless and placing one’s eye close enough to the micro-lens so one’s fashion. Modern micro-lens fabrication technology is another eyebrow is touching the paper (figure 1F). Panning is achieved by prime example of scalable manufacturing. Although the use of placing one’s thumbs on opposite ends of the top stage (colored high-curvature miniature lenses traces back to Antony van yellow in figure 1A–C) and moving them in unison, thus Leeuwenhoek’s seminal discovery of microbial life forms [8], translating both optics and illumination stages while keeping the manufacturing micro-lenses in bulk was not possible until recently. stages aligned (figure 1B). Focusing is achieved using the same Modern techniques such as micro-scale plastic molding and positioning of one’s thumbs, except the thumbs are pulled apart (or centerless ball-grinding have grown to serve numerous applica- pushed together). This causes tension (or compression) along the tions, including telecommunication fiber couplers, cell phone optics stage, resulting in 2Z (or +Z) deflection of the micro-lens cameras, and medical endoscopes. due to flexure of the supporting structure of the sample-mounting PLOS ONE | www.plosone.org 1 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope Figure 1. Foldscope design, components and usage. (A) CAD layout of Foldscope paper components on an A4 sheet. (B) Schematic of an assembled Foldscope illustrating panning, and (C) cross-sectional view illustrating flexure-based focusing. (D) Foldscope components and tools used in the assembly, including Foldscope paper components, ball lens, button-cell battery, surface-mounted LED, switch, copper tape and polymeric filters. (E) Different modalities assembled from colored paper stock. (F) Novice users demonstrating the technique for using the Foldscope. (G) Demonstration of the field-rugged design, such as stomping under foot. doi:10.1371/journal.pone.0098781.g001 stage (figure 1C). Unlike traditional microscopes, the Foldscope micro-lenses), lens-holder apertures, an LED with diffuser or anchors the sample at a fixed location while the optics and condenser lens, a battery, and an electrical switch (see figure 1D). illumination stages are moved in sync. The three stages are weaved together to form an assembled Foldscope (figure 1B,C,E) with the following features: fully- constrained X–Y panning over a 20620 mm region (figure 1B), Results flexure-based focusing via Z-travel of the optics stage relative to Design Platform the sample-mounting stage (figure 1C), and a vernier scale for Construction from flat media. The Foldscope is comprised measuring travel distances across the sample slide with 0.5 mm of three stages cut from paper _ illumination, sample-mounting, resolution. The total optical path length from the light source to and optics _ and assembled via folding (figure 1A–C, video S1). the last lens surface is about 2.7 mm (figure S1), only 1% that of a Other primary components include a spherical ball lens (or other conventional microscope. Flat polymeric sheets and filters can also PLOS ONE | www.plosone.org 2 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope Table 1. Bill of Materials. Component Unit cost (10,000 pcs) Paper (400 cm ) $0.06 Ball Lens (low mag/high mag) $0.17/$0.56 3V button battery (CR2016) $0.06 LED $0.21 Switch $0.05 Copper Tape (5 cm ) $0.03 Foldscope $0.58/$0.97 Summary of unit costs for Foldscope components in volumes of 10,000 units, not including assembly costs. This assumes a Foldscope in brightfield constructed from the following: polypropylene sheets (Press Sense 10mil Durapro); a 140X low-mag lens (Winsted Precision Ball 2.4 mm borosilicate ball, P/N 3200940F1ZZ00A0, from www.mcmaster.com, P/N 8996K21) or a 2,180X high-mag lens (Swiss Jewel Co. 0.2 mm sapphire ball lens); a 3V CR2016 button cell (Sanyo CR2016-TT1B #8565 from Batteriesandbutter.com); a white LED (Avago ASMT CW40 from Mouser.com); an electrical slider switch (‘‘Off/On MINI SMD Switch’’ from AliExpress.com); and copper tape (Sparkfun P/N 76555A648). doi:10.1371/journal.pone.0098781.t001 be inserted into the optical path, including diffusion filters for focal length. Equally important for image quality, the illumination improving illumination uniformity, Fresnel lenses as condensers source (LED plus diffuser and/or condenser lens) should provide for concentrating illumination intensity, color filters for fluores- even illumination over the field, ample intensity, narrow intensity cence imaging, and linear polarizers for polarization imaging. profile, and high CRI (color rendering index). The LED used in Alignment by Folding. Folding provides a passive alignment the Foldscope consumes only 6 mW of electrical power and can mechanism that is used here to align the micro-lens with the light operate over 50 hours on a CR2032 button cell battery (figure source. A sharp crease in a thin sheet of inextensible material, such S3A). Precise control over the illumination profile is required for as paper, of thickness h introduces elastic energy of bending of the high-quality microscopy [17], so integration of a condenser lens is order ,h [12]. Thus, a fold introduces buckling at the inner edge, crucial for optimal imaging (figure S3C). For low-magnification giving variation in the exact location of the hinge and resulting in imaging applications not requiring optimal imaging, the illumina- random alignment error of the order ,h. To minimize this error, tion source can be removed and the Foldscope can be operated we introduce folding features that form a closed structural loop while facing an external light source. between the optics stage and the illumination stage. This improves alignment repeatability through elastic averaging within kinematic Design Innovations constraints (figure 1A; [13]). We characterized alignment accuracy High-Resolution Brightfield Microscopy. For some appli- and repeatability by constructing twenty independent Foldscopes cations, extending the resolution limit of the Foldscope to out of 350 mm thick black cardstock and manually folding and submicron length scales is a practical necessity. For this reason, unfolding them twenty times each (see Materials and Methods the resolution of the single-ball-lens Foldscope was further section), while measuring absolute X–Y alignment (figure S2). optimized and empirically characterized. The analytical optimi- Assembly repeatability was assessed as the mean value of twice the zation was carried out for a single field point at the optical axis to standard deviation for each Foldscope (65 mm in X and 25 mmin assess the best achievable resolution (see Modeling and Charac- Y), while assembly accuracy was assessed as the mean value of all terization and table 2). A 1,450X Brightfield Foldscope with the trials (59 mm in X and 67 mm in Y). A higher skew in X-axis configuration depicted in figure 2E was used to capture the image repeatability results from structurally distinct constraints imple- in figure 2A, empirically confirming submicron resolution. As mented for the X- and Y-axes. The small assembly accuracy errors shown in figure S4A, spherical ball lenses have significant (less than 20% of the paper thickness) in both directions are wavefront error at the edge of the field defined by the aperture consequences of the design which can be compensated by feature (aperture shown in figure S1). As a result, not all regions can be shifts in future designs. simultaneously in focus within this field. The center portion of the Micro-Optics and Illumination. The Foldscope design field, with wavefront error less than 1/5 wavenumber and low accommodates different optical configurations, including spherical curvature and distortion, is denoted the ‘‘optimal field of view’’ ball lenses, spherical micro-lens doublets (such as a Wollaston (figure S4A–C). Thus, the best achievable resolution is attained at doublet), and more complex assemblies of aspheric micro-lenses. the expense of a reduced field of view. When a digital sensor is While more optical elements generally provide reduced aberration used in place of the naked eye, the lens fixture effectively reduces and improved field of view, spherical ball lenses have distinct the field of view to roughly the optimal field of view. advantages for high-volume manufacturing, including reduced Fluorescence. Conventional fluorescence microscopy typi- part count and simplified assembly due to rotational symmetry cally requires an expensive illumination source for high-intensity [14–16]. Since magnification varies inversely with ball-lens broad-spectrum illumination and multiple optical elements with diameter, commonly available ball lenses provide an ample range precisely defined spectral profiles. The simplified configuration of of magnifications (under 100X to over 2,000X, as seen in table 2). the Fluorescence Foldscope uses a high-intensity colored LED of The back focal length of these lenses varies drastically, thus narrow spectral width and polymeric sheets inserted in the optical motivating alternative lens-mounting schemes (above the optics path for a shortpass excitation filter and a longpass emission filter stage, as in figure 1C, or below) and requiring samples with no (figure 2B,F). A blue LED light source and commonly available gel coverslip for lenses with less than approximately 140 mm back filters (with spectral transmissivities plotted in figure S3B) were PLOS ONE | www.plosone.org 3 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope PLOS ONE | www.plosone.org 4 June 2014 | Volume 9 | Issue 6 | e98781 Table 2. Foldscope Analytical Model Parameter Summary Table. Parameter Values at Select Combinations of r,n Parameter Functional Form in Optimized System r = 1200 r = 500 r = 400 r = 150 r = 150 r = 100 n = 1.517 n = 1.517 n = 1.517 n = 1.517 n = 1.77 n = 1.77 2:5e5mm n{1 MAG 140 340 430 1140 1450 2180 MAG~ ~ðÞ 5e5mm EFL n r BFL 1 rðÞ 2{n 561 234 187 70 22 15 BFL~EFL{r~ 2 n{1 1=4 RES 1.90 1.52 1.44 1.13 0.86 0.77 3: : : : 1=4 l r n½ nzðÞ 2{nðÞ 2n{1 : : 3: : RES~k f l jj s ~k 2 2 128ðÞ n{1 1=4 1=4 nOAR : 3 0.294 0.366 0.387 0.495 0.510 0.565 a k l 8l n : : nOAR~ ~ ~k : : : r r jj s rðÞ n{1½ nzðÞ 2{nðÞ 2n{1 1=4 OAR 1=4 : 3: 3 353 183 155 74 77 56 l 8l n r : : : OAR~nOAR r~k ~k 1 1 : : s ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 jj EFL 1 n r 1761 734 587 220 172 115 EFL~ 2 ðÞ n{1 1=4 NA 3 0.200 0.249 0.264 0.337 0.444 0.491 : : 2aðÞ n{1 128 lðÞ n{1 NA~ ~k : : : : r n r n½ nzðÞ 2{nðÞ 2n{1 1=4 FOV 518 268 227 109 88 65 : : 3: 7 n a l r n FOV~ ~k : : 2ðÞ n{1 2ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DOF : : : : 13.7 8.8 7.9 4.8 2.8 2.3 l 1 l r n½ nzðÞ 2{nðÞ 2n{1 DOF~ ~ 2 2 3 NA k 128ðÞ n{1 {1=8 SR 0.8825 0.8825 0.8825 0.8825 0.8825 0.8825 SR~e Functional form and select numerical values for the following dependent parameters: Magnification (MAG), Back Focal Length (BFL), Resolution (RES), nOAR (Normalized Optimal Aperture Radius), OAR (Optimal Aperture Radius), Effective Focal Length (EFL), Numerical Aperture (NA), Field of View (FOV), Depth of Field (DOF), Strehl Ratio (SR). These are calculated for infinite ‘‘object’’ distance per analytical model RM2, with aperture radius 1=4 3 a~OAR~kðÞ l=s , k %0:9321, k %0:7415, and with aberration coefficient s~{ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 ðÞ 2rn . All calculations assume an incident wavelength of l = 0.55 mm, and all specified distances are in units of mm. 1 1 2 The indices of refraction n = 1.517 and n = 1.77 correspond to borosilicate glass and sapphire, respectively. doi:10.1371/journal.pone.0098781.t002 Origami-Based Paper Microscope Figure 2. Foldscope imaging modalities. (A) Brightfield Foldscope image of a monolayer of 1 mm polystyrene microspheres (Polysciences 07310- 15) using a 1,450X lens. (B) Fluorescent Foldscope image of 2 mm polyfluorescent microspheres (Polysciences 19508-2) using a 1,140X lens with Roscolux gel filters #19 and #80. (C) 2X2 lens-array Brightfield Foldscope image of Giemsa-stained thin blood smear using 1,450X lenses. (D) 140X Darkfield Foldscope images of 6 mm polystyrene microspheres (Polysciences 15714-5), using a 140X lens for the darkfield condenser. Darkfield condenser aperture shown in inset has 1.5 mm inner diameter and 4.0 mm outer diameter. (E–H) Schematic cross-sections of Brightfield, Fluorescence, Lens-Array, and Darkfield Foldscope configurations, showing the respective arrangements of ball lenses, filters, and LEDs. See table 2 for ball lenses used for specific magnifications. doi:10.1371/journal.pone.0098781.g002 used to image 2 mm diameter red poly-fluorescent polystyrene configured in a single Foldscope (figure 2C,G). Such a lens array beads as shown in figure 2B. For fluorescent imaging requiring may be comprised of identical lenses or of different lenses with higher contrast, small pieces (3 mm square or smaller) of different magnifications and/or back focal lengths. This provides interference filters can be used in place of the polymeric sheets for the design of a lens-array Foldscope with several key features. at reasonable cost due to the small size. For non-contiguous samples such as blood smears, a larger field of Lens-array and Multi-modality. Since a micro-lens has a view can be obtained by overlapping a number of small fields of very small footprint, multiple optical paths can be independently view. Alternatively, an optimal array pitch will give tangential non- Figure 3. Manufacturing innovations for lens- and specimen- mounting. (A) Fabrication, mounting, and characterization of capillary- encapsulation process for lens-mounted apertures. X and Y error bars for all measurements are 2.5 mm. (B) Reel of polystyrene carrier tape with custom pockets and punched holes for mounting over 2,000 ball lenses with optimal apertures. The first ten pockets include mounted ball lenses. Inset shows sectioned view from CAD model of carrier tape mounted lenses. Note the aperture is the punched hole shown on the bottom side of the ball lens. This tape is 16 mm wide and is designed for 2.4 mm ball lenses (aperture diameter is 0.7 mm). (C) Top: Paper microscope slide shown next to standard glass slide with coverslip, both with wet mount algae specimens. Bottom: Schematic of paper microscope slide, showing specimen containment cavity formed between upper tape and lower tape in middle of slide. doi:10.1371/journal.pone.0098781.g003 PLOS ONE | www.plosone.org 5 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope PLOS ONE | www.plosone.org 6 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope Figure 4. Analytical, numerical, and empirical characterization of Foldscope. (A,B) Analytical ‘‘design curves’’ for normalized optimal aperture radius (nOAR) and optimal resolution (RES) versus magnification (MAG) over index of refraction (range 1.33–1.91) and ball lens radius (range 40–1200 mm). (C) Comparison of analytical (3D surface) and numerical (plotted as points) results for RES versus index of refraction and ball lens radius. (D) Modulus of the Optical Transfer Function (MTF) over the optimal field of view for a 300 mm sapphire lens with optimal aperture. (E,F) Image of USAF 1951 resolution target taken with 430X ball lens, including an enlarged caption of Group 9, and an intensity profile plot along path denoted by green line in image caption. This demonstrates resolvability for Group 9, Element 4 corresponding to 724 Line Pairs/mm or 1.38 mm resolution. (G,H) Image of USAF 1951 resolution target taken with 140X ball lens, including an enlarged caption of Group 8, and an intensity profile plot along path denoted by green line in image caption. This demonstrates resolvability for Group 8, Element 6 corresponding to 456 Line Pairs/mm or 2.19 mm resolution. The data was taken using GUPPY Pro 503C scientific camera, with 259261944 pixels and pixel size 2.262.2 mm . doi:10.1371/journal.pone.0098781.g004 overlapping fields of view (as seen in figure 2C), thus reducing the removed from the synthetic paper and replaced so that the slide time required to scan a slide for a feature such as a parasite. Since can be reused for many specimens. Note that the specimen individual lenses have independent optical paths, the novel depicted in figure 2D was mounted on a paper microscope slide. capability of building multi-modality lens-array microscopes arises. One such combination is a two-by-one array of brightfield and Modeling and Characterization fluorescence modalities, which could be used to scan a sample for Theory and analysis. For a brightfield Foldscope, basic the presence of fluorescence markers and then identify the non- measures of optical performance can be described in terms of the fluorescent surrounding structures. ball radius (r), index of refraction (n), aperture radius (a), and Darkfield. The Darkfield Foldscope configuration, shown in incident wavelength (l; see text S1). Assuming the paraxial figure 2H, requires a diffuser, a darkfield condenser aperture (inset approximation, these include effective focal length (EFL), back in figure 2D), and a condenser lens. The diffuser helps to evenly focal length (BFL), and magnification (MAG). For a 300 mm distribute light from the small LED over the aperture area, while sapphire ball lens: EFL = 172 mm, BFL = 22 mm, and the condenser focuses a hollow cone of light onto the specimen. MAG = 1,450X. Thus substantial magnification can be obtained, Since the specimen must be placed at the focal point of the but the sample must be separated from the lens by only a fraction condenser, the slide thickness has to match the back focal length of of the thickness of a human hair. Three additional optical the condenser plus the spacing from the condenser to the slide. A performance metrics include field of view radius (FOV), numerical 140X Darkfield Foldscope was used to image 6 mm polystyrene aperture (NA), and depth of field (DOF). These depend on microspheres as shown in figure 2D. aperture radius (a), the optimization of which is discussed below. Capillary Encapsulation Lens Mounting. The process of For the previous example, the normalized optimal aperture radius precisely mounting micro-optics to an aperture crucially governs is nOAR = a/r = 0.51, giving: FOV = 88 mm, NA = 0.44, lens performance. Therefore, a capillary encapsulation process was DOF = 2.8 mm (see table 2). developed to automatically mount a ball lens while forming a The aperture radius controls the balance between diffraction circular aperture of precisely tunable diameter (see figure 3A). By effects from the edges of the aperture with spherical aberration partially engulfing ball lenses in an opaque polymer held between effects from the lens. Therefore, a complete analytical model was two glass substrates coated with flat nonstick PDMS (see top left of created to predict the normalized optimal aperture radius (nOAR) figure 3A), a precise aperture is self-assembled around the ball-lens and optimal resolution (RES), as well as the aberration coefficient (see Materials and Methods section for details). Pressure applied (s) for a ball lens (see Supplementary Materials and table 2), between the substrates is used to precisely tune aperture size, with yielding: greater pressure providing a larger aperture. The epoxy encap- sulated lens is then adhesively mounted to a paper aperture and pffiffiffiffiffi 1=4 inserted in the Foldscope (see bottom right of figure 3A). 1=4 nOAR~ðÞ k =rðÞ l=jj s , k ~ 3 10 4p %0:9321 1 1 Carrier Tape Lens Mounting. Black polystyrene carrier tape is commonly used for low-cost reel-to-reel packaging of electronic components. This pre-existing infrastructure was 1=4 1=4 1=8 leveraged to create low-cost mounting structures for ball lenses RES~k fðÞ l=jj s , k ~1:22ðÞ p=12 ðÞ e=10 %0:7415 2 2 with optimized apertures. As depicted in figure 3B, the custom thermoformed pocket holds the lens in place with a press fit, and a punched hole in the bottom of the pocket precisely defines the s~{ðÞ n{1½ nzðÞ 2{nðÞ 2n{1 ðÞ 2rn aperture. A single lens is cut from the carrier tape and adhesively attached to a paper aperture which is then inserted into the optics The expressions for nOAR and RES are depicted as 2D design stage of the Foldscope, analogous to that shown for the epoxy plots as a function of desired MAG in figure 4A,B and figure 4C encapsulated lens in figure 3A. shows a 3D plot of RES over n and r. For the example discussed Paper microscope slide. A low-cost microscope slide earlier, the values for normalized aperture radius and resolution (similar to that in [18]) was constructed out of 18mil polystyrene are found by locating the intersection of the lines for r = 150 mm synthetic paper and transparent tape as shown in figure 3C. If tape and n = 1.77 in the design plots. This gives nOAR = 0.51 and is placed on only one side (either upper or lower), specimens are RES = 0.86 mm, and corresponds to MAG = 1,450X. Note that conveniently mounted on the exposed sticky surface of the tape. the regions enclosed by the curves in the 2D design plots represent Using both upper and lower pieces of transparent tape creates a the available design space for nOAR, RES, and MAG as defined cavity for mounting wet specimens such as live algae suspended in by the range of possible values for n and r. The design curves thus water. Since the paper microscope slide is less than half the make it a simple exercise to pick optimal design parameters within thickness of a standard glass slide, a spacer is required to elevate the space of interest. Also, one can see from figure 4B that the the sample closer to the lens. This is achieved by inserting the specimen slide together with two blank paper slides beneath it. lower limit for the best achievable resolution in ball lenses appears Once the specimen has been viewed, the transparent tape can be PLOS ONE | www.plosone.org 7 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope to be near 0.5 mm, based on the range of parameters identified for For the 430X lens, note that the empirical value is 4.2% smaller this figure. than the theoretical value. A less conservative theoretical model Numerical Modeling. A ray-tracing numerical model was (see text S1) predicts values 11.5% smaller than those in table 2, developed for the Foldscope using Zemax software to confirm the indicating the empirical value is reasonable and near the limits results of the analytical model and to evaluate points across the predicted by theory. For the 140X lens, note that the empirical field of view (see Materials and Methods section for details). The value is 15.3% larger than the theoretical value. This difference is results for nOAR and RES show very good agreement, with attributed to the fact that this particular lens was selected for its 2 2 correlation coefficients of R = 0.985 for nOAR and R = 0.998 low cost at the expense of larger tolerances on diameter and for RES (see figure 4C). The numerical modeling results across the sphericity (grade 48) compared to the 430X lens (grade 10). For field of view are shown in figure S4A, where the optimal field of example, diameter tolerance per ball sphericity is 0.25 mm for view used for calculating the MTF (Modulus of the Optical grade 10 and scales linearly with grade number. Transfer Function) is defined. The MTF for this system is plotted Since the USAF 1951 resolution target has a glass covering with in figure 4D for a field point on the optical axis and another at the thickness comparable to a standard coverslip (140 mm), ball lenses edge of the optimal field of view. The field point at the optical axis with magnification higher than 430X could not be assessed with shows near-diffraction-limited response, and the tangential and this resolution target due to short back focal length. Instead, a sagittal curves for the edge of the field drop to half of their low- monolayer of 1 mm polystyrene beads was imaged using a 1450X frequency value at a spatial frequency of about 300 cycles/mm. ball lens (figure 2A) to demonstrate sub-micron resolution for this Empirical Characterization of Resolution. The resolution lens (theoretical value is 0.86 mm). of the 430X and 140X lenses were empirically characterized using a standard USAF 1951 resolution target and a GUPPY Pro 503C Discussion and Conclusions scientific camera (with bare sensor) as shown in figure S7. Images taken with these lenses are shown in figure 4E,G, and the relevant By removing cost barriers, Foldscope provides new opportuni- ties for a vast user base in both science education and field work intensity profiles are shown in figure 4F,H. The formula for the resolution (in line pairs/mm) of a given group and element of the for science and medicine. Many children around the world have Group+(Element21)/6 never used a microscope, even in developed countries like the USAF 1951 target is given by RES =2 , and LP the center-to-center distance between the lines in microns is given United States. A universal program providing ‘‘a microscope for by RES = 1000/RES . From table 2, we see the theoretical every child’’ could foster deep interest in science at an early age. CC LP While people have known for decades that hands-on examination values for the resolution of the lenses are 1.44 mm and 1.90 mm, respectively. Based on the critically resolved intensity profiles in and inquiry is crucial in STEM (Science, Technology, Engineer- figure 4F,H, the empirical values for RES are found to be ing, and Mathematics) education [19–20], the challenge posed by CC 1.38 mm and 2.19 mm. J. M. Bower to engage ‘‘all teachers and all children’’ [21] requires Figure 5. Mosaic of Foldscope Images. Bright field images of (A) Giardia lamblia (2,180X), (B) Leishmania donovani (1,450X), (C) Trypanosoma cruzi (1,450X), (D) gram-negative Escherichia coli (1,450X), (E) gram-positive Bacillus cereus (1,450X), (F) Schistosoma haematobium (140X), and (G) Dirofilaria immitis (140X). Unstained (H) leg muscles and (I) tarsi of an unidentified ladybug (genus Coccinella). (J) Unstained leg muscles (fixed in formaldehyde) of an unidentified red ant (genus Solenopsis). An LED diffuser (Roscolux #111) was added for (A) and an LED condenser (2.4 mm borosilicate ball lens) was used for (C). Images (H–J) were taken by novice user using a self-made Foldscope (140X). See table 2 for ball lenses used for specific magnifications. White scale bar: 5 mm; black scale bar: 100 mm. doi:10.1371/journal.pone.0098781.g005 PLOS ONE | www.plosone.org 8 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope large-scale adoption of practices and broad availability of tools that 2D Media and Filters were previously cost-prohibitive [22]. Moreover, the opportunity The 2D media used in constructing Foldscopes included black to make microscopes both approachable and accessible can inspire 105 lb card stock (ColorMates Smooth & Silky Black Ice Dust children to examine the rich bio-diversity on our planet as Card Stock, purchased from thePapermillstore.com), polypropyl- amateur microscopists and to make discoveries of their own, as ene (PressSense Durapro CC 10mil), and others. Foldscope parts already seen in the field of amateur astronomy ([23]; see images were cut from 2D media using a CO laser (Epilog Elite, Mini24). taken by novice user with self-made Foldscope in figure 5H–J). Copper tape was used for providing connectivity (by soldering) Disease-specific Foldscope designs are an important vision for between the LED, battery, and switch. The filters used in future development [24–25]. Figure 5 depicts early bench-test constructing Foldscopes included Roscolux colored gel filters data, including high-magnification brightfield images of Giardia (including Primary Blue #80 and Fire Red #19, which lamblia, Leishmania donovani, Trypanosoma cruzi (Chagas parasite), approximate an Acridine Orange filter set), Roscolux diffuser Escherichia coli, and Bacillus cereus (figure 5A–E), and low-magnifi- filters (Tough Rolex #111), and polymeric linear polarizers cation brightfield images of Schistosoma haematobium and Dirofilaria (Edmund Optics P/N 86181). Each type of filter is assembled to immitis (figure 5F–G). Note that these include magnifications the Foldscope by cutting out a 3–5 mm square piece and ranging from 140X to 2,180X, none of which require immersion adhesively attaching it to the appropriate stage with single-sided oil. In the future, darkfield and fluorescence Foldscopes will also be or double-stick Scotch tape. Paper microscope slides were adapted for diagnostics, and sensitivity and specificity will be constructed from polypropylene sheets (PressSense Durapro CC measured for various disease-specific Foldscopes in the field as 18mil) and transparent scotch tape. clinical validations against existing diagnostic standards. Constructing instruments from 2D media provides other unique LEDs, Switches and Power Sources advantages and opportunities. Embedding flat rare-earth magnets The LEDs used in constructing Foldscopes included the Avago in paper provides means for magnetic self-alignment, allowing the HSMW-CL25 (now replaced by P/N Avago ASMT CW40) white Foldscope to be reversibly coupled to a conventional smartphone LED for brightfield Foldscopes, the Kingbright APTD1608QBC/ for image capture, for smartphone-based diagnostics, or for D blue LED for fluorescence Foldscopes. The electrical slider telemedicine [26–29]. By printing text and images on the paper, switch was purchased from AliExpress.com (‘‘Off/On MINI SMD this platform provides an efficient information-delivery method for Switch’’, Product ID: 665019103). The power sources included specific staining protocols, pathogen identification guides, or Duracell 3V CR2032 button cells, Sanyo 3V CR2016 button cells language-free folding instructions (figure S6). Some applications (Sanyo CR2016-TT1B #8565 from Batteriesandbutter.com), and in highly infectious diseases may benefit from a disposable a GW Instek DC power supply (Model GPD-3303D). Button cells microscope _ or ‘‘use-and-throw’’ microscopy _ where the entire were used with no resistors for Foldscopes. microscope can be incinerated. Also, in place of a glass slide, the 2D media also allows direct addition of the sample to a paper- Aperture Manufacturing based micro-fluidic assay [30] for automated staining and/or This method produces inexpensive apertures through polymer pathogen-concentration, thus yielding an independent fully- encapsulation of ball lenses while preserving the optical quality of functional diagnostic system. the lens and allowing multiple lenses to be encapsulated at once. Future work will build upon the key features of this platform. The experimental setup shown in the top left of figure 3A was used Roll-to-roll processing of flat components and automated ‘‘print- to encapsulate 300 mm sapphire ball lenses with aperture and-fold’’ assembly make yearly outputs of a billion units diameters ranging from 100 mm to 214 mm. The lens was attainable. Ongoing work with advanced micro-optics and sandwiched between parallel substrates (glass or silicon) coated illumination design _ including spherical GRIN lenses [31–32], with planar films of PDMS with thickness greater than 1 mm aspheric multi-lens optics, and condenser lens provisions for (formed from Dow Corning Sylgard 184 PDMS). A micrometer Ko¨hler illumination _ is expected to improve both resolution and stage was used to precisely apply pressure between the substrates field of view at low cost. International field-work in both to adjust the diameter of the resulting elastic deformation of the diagnostics and education will provide vital inputs for further PDMS film. This diameter was measured in situ using phase improvements. Our long-term vision is to universalize frugal contrast microscopy to set the target value for the aperture. A fast- science, using this platform to bring microscopy to the masses. curing opaque polymer (Smooth-On Smooth-Cast Onyx Fast Polyurethane) was then injected into the cavity and allowed to Materials and Methods cure. Reflected light microscopy was used to measure the dimensions of the final aperture formed. Once removed from Ball Lenses the non-stick PDMS films, the encapsulated lens was attached to The ball lenses used in constructing Foldscopes included the underside of the optics stage of a Foldscope material types borosilicate, BK7 borosilicate, sapphire, ruby, and S-LAH79. The vendors included Swiss Jewel Co, Edmund Optics, Characterization of Self-Alignment by Folding and Winsted Precision Ball. Part numbers for some select lenses Twenty independent microscopes were cut out of black 105 lb include: 300 mm sapphire lens from Swiss Jewel Co. (Model cardstock, each marked with a cross-hair in both the optics and B0.30S), 200 mm sapphire lenses from Swiss Jewel Co. (Model illumination stages (see figure S2C). After folding, alignment was B0.20S), 2.4 mm borosilicate lenses from Winsted Precision Ball measured using a dissection microscope (Olympus upright, 306 (P/N 3200940F1ZZ00A0), 300 mm BK7 borosilicate lenses from magnification) via digitizing the cross-hair images, drawing lines Swiss Jewel Co. (Model BK7-0.30S), and 1.0 mm BK7 borosilicate through the center of each cross-hair (X and Y cross-hairs on both lenses from Swiss Jewel Co. (Model BK7-1.00). Note that half-ball stages), and digitally measuring the X and Y displacements to lenses from both Edmund Optics and Swiss Jewel Co. were also characterize the alignment. Every Foldscope was iteratively folded, tested for use as condenser lenses for the LEDs. imaged to record X–Y alignment, and unfolded twenty times. The PLOS ONE | www.plosone.org 9 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope data was then used to assess accuracy and repeatability (see figure length from the LED to the lens is almost an order of magnitude S2A,B). smaller than the size of the human eye. (TIF) Sample Preparation Figure S2 Characterization of Self-Alignment by Fold- Thin-blood smears of Plasmodium falciparum (ring stage), Trypano- ing. Twenty independent Foldscopes were constructed out of soma cruzi, Giardia lamblia, Leishmania donovani, and Dirofilaria immitis 350 mm thick black cardstock and manually folded and unfolded were freshly prepared from cultures provided by Center for twenty times each, with alignment measured after each assembly. Discovery and Innovation in Parasitic Diseases (CDIPD) at UCSF. The data was used to produce plots of (A) assembly repeatability The samples were fixed in methanol and stained in freshly (distribution of all 400 values, adjusted to give zero mean for each prepared Giemsa solution (Sigma Aldrich, #48900-500ML-F) Foldscope) and (B) assembly accuracy (distribution of 20 mean using standard protocols before imaging. Once fixed, the slides values calculated per Foldscope) using (C) cross-hair alignment could be used for several weeks. Bacterial samples of Bacillus cereus features on the optics and illumination stages. Note that the span and Escherichia coli were provided by KC Huang Lab at Stanford of the data in both plots is less than the thickness of the paper used University. The samples were heat fixed onto glass slides using to construct the Foldscopes. Based on the data shown in the plots, standard procedures and gram stained using standard protocols assembly repeatability was assessed as the mean value of twice the (Fisher Scientific Gram Stain Set, Catalog No. 23-255-959). standard deviation for each Foldscope (65 mm in X and 25 mmin Plasmodium-infected red blood cells were taken from cultures Y), while assembly accuracy was assessed as the mean value of all provided by the Center for Discovery and Innovation in Parasitic trials (59 mm in X and 67 mm in Y). A higher skew in X-axis Diseases (CDIPD). Schistosoma haematobium were provided by the repeatability results from structurally distinct constraints imple- Michael Hsieh Lab at Stanford University. Insects used for mented for the X- and Y-axes, while the assembly accuracy errors imaging were caught on Stanford campus and imaged after fixing in both directions are consequences of the design which can be in formaldehyde without any stain. No human samples were compensated by feature shifts in future designs. Note that the X utilized in the current work. and Y error bars for all measurements are 8.4 mm. (TIF) Image-Capture Protocol Figure S3 Component Characterization. (A) LED voltage Brightfield images were taken using a Canon EOS 5D Mark II and intensity versus time for a white LED (Avago HSMW-CL25) with the Foldscope placed 3 cm away from the 100 mm focal powered by a Duracell CR2032 battery with no resistor. (B) Filter length lens and using the following settings: F/3.2, 1/30 sec. transmission spectra of three Roscolux filters _ Tough Rolex exposure, ISO-2000. An initial image was first captured using diffuser (#111), Fire Red (#19), and Primary Blue (#80) _ automatic white-balance and then used as a reference white measured with Ocean Optics Photo spectrometer USB4000. (C) balance image during data collection. Fluorescence images were Intensity profile of a white LED (Avago HSMW-CL25) as taken in a similar fashion to the brightfield images with typical visualized in water with dissolved fluorescein. The left image is camera settings: F/2.8, 15 sec. exposure, ISO-1000. Although not taken with the bare LED while the right image is taken with a presented, images were also obtained by coupling the Foldscope to condenser lens (2.4 mm borosilicate ball lens) placed adjacent to cell-phones including an iPhone using a custom magnetic coupler. the LED in the optical path, demonstrating that a ball lens can be USAF 1951 resolution target data was taken using GUPPY Pro used to effectively collimate the light emitted by this LED. 503C scientific camera, with 259261944 pixels and pixel size (TIF) 2.262.2 mm , using the setup shown in Figure S7. Figure S4 Numerical Modeling Characterization of Optimal Field of View. (A) Plot of Wavefront Error over the Numerical Model full field of view defined by the aperture for a 300 mm Sapphire Zemax software was used to model the Foldscope optics to ball lens with a 147 mm aperture. With increasing field coordinate, assess optimal aperture radius and resolution. The basic model of the Wavefront Error becomes very large and the image will be out the system consists of a ball lens, an aperture, an object at infinity, of focus. An ‘‘optimal field of view’’ is defined at a field coordinate and an image plane (see figure S5A). This model requires two of 21 mm, where the Wavefront Error is approximately 1/5 wave parameters to be independently optimized _ lens-image distance number. (B,C) Plots of Field Curvature and Distortion over the and aperture radius. The analysis is carried out in four steps: 1) optimal field of view. (D) Plot of RMS spot size over the optimal optimize lens-image distance in model by minimizing focusing field of view depicting four cases: optimized solution treated as metric (figure S5C); 2) determine search space for aperture radius reference with zero defocus (red line), defocus of 3 mm (green line), as defined by empirically chosen limits on Strehl Ratio, 0.75–0.98; defocus of 3 mm (green line), diffraction limit (dashed black line). 3) optimize aperture radius using resolution metric (figure S5D); The reference solution provides the best achievable resolution at and 4) use Matlab surface-fitting tool to fit data for optical the center of the field of view (approximately equal to the performance parameters as functions F(n,r, l) and compare with diffraction limit for this choice of aperture), while other plots show analytical model. that increasing defocus moves the region of best resolution radially out from the center in an annular ring. (E) Plot of RMS spot size Supporting Information over the optimal field of view depicting optimal aperture predicted by analytical model (red lines) and adjusted aperture giving Figure S1 Foldscope Schematics. (A) Real image formation uniform RMS spot size over the field of view (purple lines). via projection. (B) Virtual image formation via direct observation (TIF) with the eye. Note the drawings are not to scale. The indicated lengths are example values that show the versatility of this design Figure S5 Diagrams and Plots for Numerical and as well as its extreme space efficiency. For example, the same Analytical Models. (A) Schematic of time-reversed Zemax system can be used for projecting or imaging simply by changing model showing collimated light coming from an object at infinity, the object-lens distance by about 20 mm. Also, notice the total path passing through aperture, and focused by the ball lens onto a focal PLOS ONE | www.plosone.org 10 June 2014 | Volume 9 | Issue 6 | e98781 Origami-Based Paper Microscope point in the image plane. (B) Schematic of time-reversed model Video S1 Foldscope Assembly. A short video of 140X showing key parameters used in some derivations for the analytical Brightfield Foldscope assembly process. model. (C) Plot of Focusing Metric versus Lens-Image Distance for (M4V) l = 0.55 mm, r = 150 mm, n = 1.517. This illustrates how focusing Video S2 Foldscope Drop Test and ruggedness. A short metrics FM1, FM2, and FM3 select different values for the optimal video of a three story drop test and ruggedness of Foldscope lens-image distance. (D) Plot of Resolution Metric versus Aperture demonstrated by stomping under feet. Radius for l = 0.55 mm, r = 150 mm, n = 1.517. This illustrates (AVI) how resolution metrics RM1 and RM2 select nearly the same aperture radius but yield different values for resolution. Text S1 Supplementary information. Analytical expres- (TIF) sions and derivations related to optical performance metrics. (DOCX) Figure S6 Artistic Layout of Foldscope Paper Compo- nents. Artistic version of Foldscope layout with integrated Acknowledgments universal folding instructions based on color coding, where like colors are matched during the folding process to leave a single We thank all members of the Prakash lab for valuable suggestions. We solid color in the final folded configuration. acknowledge Ioana Urama for assistance with supplementary figure S6, (TIF) Anika Radiya for assistance with magnetic couplers and Marisa Borja for taking images in figure 5H–J, and Center for Discovery and Innovation in Figure S7 Foldscope Image Capture Setup for Resolu- Parasitic Diseases (CDIPD) at UCSF and the Michael Hsieh Lab at tion Metric. Picture of the experimental setup used to caputure Stanford University for supplying samples. This work is covered under images of the USAF 1951 resolution target viewed with the patent application PCT/US2013/025612 (Title: Optical Device). Further Foldscope. The data was taken using GUPPY Pro 503C scientific details, including the 10,000 microscope project and updates on the latest camera, with 259261944 pixels and pixel size 2.262.2 mm . status of this work, are available on ?www.foldscope.com. (TIF) Author Contributions Figure S8 Quick Reference Guide for Foldscope Assem- bly. One-page handout to facilitate users in guided assembly of a Conceived and designed the experiments: JSC MP. Performed the Foldscope. experiments: JSC JC MP. Analyzed the data: JSC JC MP. Wrote the (PDF) paper: JSC MP. References 1. Keller E, Goldman R (2006) Light Microscopy. Woodbury, NY: Cold Spring inservice elementary school teachers. Journal of Elementary Science Education Harbor Laboratory Press. p. 8. 1(2): 14–34. 2. Whitesides GM (2011) The Frugal Way. The Economist - The World in 2012. 20. Bredderman T (1983) The effects of activity-based science on student outcomes: A quantitative synthesis. Review of Education Research 53: 499–518. 3. Rothemund PW (2006) Folding DNA to create nanoscale shapes and patterns. 21. Bower JM (1996) Scientists and science education reform: myths, methods, and Nature 440: 297–302. madness. National Academies of Science Resources for Involving Scientists in 4. Hyde RA (1999) Eyeglass, a large-aperture space telescope. Appl Opt 38(19): Education. Available: http://www.nas.edu/rise/backg2a.htm. Accessed: 1 May 4198–4212. 5. Hoover AM, Fearing RS (2008) Fast scale prototyping for folded millirobots. 22. Kremer M, Brannen C, Glennerster R (2013) The Challenge of Education and IEEE International Conference on Robotics and Automation 5: 886–892. Learning in the Developing World. Science 340(6130): 297–300. 6. Whitney JP, Sreetharan PS, Ma KY, Wood RJ (2011) Pop-up book MEMS. 23. O’Neill I (2011) Supernova Discovered by 10-Year-Old. Discovery News. J Micromech Microeng 21(115021): 1–7. Available: http://news.discovery.com/space/astronomy/supernova-discovered- 7. Sreetharan PS, Whitney JP, Strauss MD, Wood RJ (2012) Monolithic by-10-year-old.htm. Accessed: 2014 May 1. fabrication of millimeter-scale machines. J Micromech Microeng 22(055027): 24. Garcia LS (2007) Diagnostic medical parasitology, 5th ed. Washington, DC: 1–6. ASM Press, 142–180. 8. Porter JR (1976) Antony van Leeuwenhoek: tercentenary of his discovery of 25. Baron EJ, Miller JM, Weinstein MP, Richter SS, Gilligan PH, et al. (2013) A bacteria. Bacteriol 40: 260–269. Guide to Utilization of the Microbiology Laboratory for Diagnosis of Infectious 9. McArthur M, Lang R (2012) Folding paper: The infinite possibilities of Origami. Diseases: 2013 Recommendations by the Infectious Diseases Society of America Washington, DC: International Arts & Artists. 96 p. (IDSA) and the American Society for Microbiology (ASM). Clinical Infectious 10. Demaine ED, O’Rourke J (2007) Geometric Folding Algorithms: Linkages, Diseases, 57(4): e22–e121. doi:10.1093/cid/cit278. Origami, Polyhedra. Cambridge, MA: Cambridge Univ Press. 472 p. 26. Smith ZJ, Chu K, Espenson AR, Rahimzadeh M, Gryshuk A, et al. (2011) Cell- 11. Myer JH (1969) Optigami - A Tool for Optical System Design. Appl Opt 8: Phone-Based Platform for Biomedical Device Development and Education 260–260. Applications. PLOS ONE 6(3): e17150. doi:10.1371/journal.pone.0017150. 12. Cerda E, Mahadevan L (2005) Confined developable elastic surfaces: cylinders, 27. Breslauer DN, Maamari RN, Switz NA, Lam WA, Fletcher DA (2009) Mobile cones and the Elastica. Proc R Soc A 461: 671–700. phone based clinical microscopy for global health applications. PLOS ONE 4(7): 13. Slocum AH (1992) Precision Machine Design. Englewood Cliffs, NJ: Prentice e6320. doi:10.1371/journal.pone.0006320. Hall. 750 p. 28. Zhu H, Isikman SO, Mudanyali O, Greenbaum A, Ozcan A (2013) Optical 14. Doushkina V, Fleming E (2009) Optical and mechanical design advantages using imaging techniques for point-of-care diagnostics. Lab Chip 13: 51–67. polymer optics. Advances in Optomechanics 7424–31. 29. Bogoch II, Andrews JR, Speich B, Utzinger J, Ame SM, et al (2013) Mobile 15. Lee C-C, Hsiao S-Y, Fang W (2009) Formation and integration of a ball lens Phone Microscopy for the Diagnosis of Soil-Transmitted Helminth Infections: A utilizing two-phase liquid technology. IEEE 22nd International Conference on Proof-of-Concept Study. Am J Trop Med Hyg 88(4): 626–629. Micro Electro Mechanical Systems: pp. 172–175. 30. Martinez AW, Phillips ST, Whitesides GM (2010) Diagnostics for the 16. Jeong K-H, Liu GL, Chronis N, Lee LP (2004) Tunable microdoublet lens array Developing World: Microfluidic Paper-Based Analytical Devices. Anal Chem integrated with microfluidic network. Optics Express 12(11): 2494–2500. 82(1): 3–10. 17. Goldberg O (1980) Ko¨hler illumination. The Microscope 28: 15–22. 31. Kikuchi K, Morikawa T, Shimada J, Sakurai K (1981) Cladded radially 18. Brady S, Willard C (1998) Microscopic Explorations. UC Berkeley: Lawrence inhomogeneous sphere lenses. Appl Opt 20(3): 388–394. Hall of Science. pp. 20–21. 32. Koike Y, Kanemitsu A, Shioda Y, Nihei E, Ohtsuka Y (1994) Spherical 19. Cox CA, Carpenter JR (1989) Improving attitudes toward teaching science and gradient-index polymer lens with low spherical aberration. Appl Opt 33(16): reducing science anxiety through increasing confidence in science ability in 3394–3400. PLOS ONE | www.plosone.org 11 June 2014 | Volume 9 | Issue 6 | e98781

Journal

PLoS ONE – Unpaywall

Published: Jun 18, 2014

There are no references for this article.