A transparent optical fingerprint capture system based on grating couplers was proposed and demonstrated. Metallic gratings were used as high efficiency input/output couplers to control the light propagation in the planar waveguide. Via the comprehensive design of grating pitch and waveguide thickness, not only was light coupling achieved, but also the propagation angle of the light beam in the waveguide was well engineered to avoid the overlap of images; thus an undistorted full area of finger sensing is fulfilled in a slim and transparent structure. The proposed system was experimentally demonstrated with characteristics of considerable contrast as a potential candidate of a fingerprint sensor.
© 2016 Optical Society of America
Optical fingerprint recognition (OFR), widely used in access control system, is featured with low-cost and high image resolution. However such structure is hard to be integrated into mobile devices due to the bulky prism . Thus, current smart Phones or Notebooks use either capacitive or ultrasound fingerprint sensors . However, if higher resolution is required, the cost of those sensors will increase rapidly because of the finer sensing units. Also, the opaque of the sensing area brings about more difficulties for capacitive type to be embedded on a display screen. For example, Apple Touch ID  of a IPhone is placed on the non-transparent ‘home’ button. Although ultrasonic types  can capture 3D finger print image with high image quality, but their cost is even higher due to the need of piezoelectric materials. In view of these concerns, optical sensors are more attractive, providing device structure is appropriately designed. To reduce the size of the optical fingerprint sensors, optical fibers were proposed to replace the traditional prism to guide the light . However, this technique is costly because abundant fine fibers are needed to achieve high resolution image. M. Calmel reported a thin device using two-dimensional photo-electric imaging layer separated by strip-shaped gaps, but it is hard to manufacture due to the complicated structure . Using volume hologram gratings as couplers and planar optical waveguides as sensor areas is a promising approach [7–10], which has both advantages of high-resolution and slimness. However, current study on hologram grating based OFR mainly focused on diffraction efficiency, while how to obtain a full area fingerprint image via appropriate grating and waveguide parameters is seldom discussed, which is another key issue for the accuracy of OFR.
In this work, two large area of dielectric gratings at subwavelength pitch on a glass substrate were fabricated by laser interference lithography as input/output light couplers and the light propagation angle controller in the glass waveguide via diffraction effect. The dielectric gratings are coated with aluminum to enhance coupling efficiency. By choosing suitable grating pitch and waveguide thickness, light beams incident on the sensing surface are well controlled to avoid overlapping, thus a transparent optical fingerprint capture system (TOFCS) which can capture clear and full area fingerprint image is obtained. With the unique features of large transparent sensing area, the proposed TOFCS can be compactly embedded into a display screen without the need of extra space, especially for mobile phones.
2. Theory and simulation
A prototype of the proposed TOFCS that can be integrated into a display is presented in Fig. 1(a). The lights from a backlight source are split into two parts, the first one passing through a micro-prism film [11, 12] generates a collimated monochromatic light beam as detecting light, and the other is for displaying the information via the transparent sensing region to viewers. The details of the light coupling, sensing and propagating in the waveguide are presented in Fig. 0.1(b). The collimated light is coupled into the waveguide by the in-coupled metallic grating with a propagation angle larger than the total internal reflection angle (TIR) between air and waveguide at first. Then, in the detection region, TIR is not satisfied between finger and waveguide, thus the light is absorbed by the ridges of the finger pressed on the waveguide surface. In this way, the information of the fingerprint is carried by the reflected light where the dark zones is corresponding to the ridges. At last, the light is coupled out by the out-coupled metallic grating and is captured by a camera. The grating pitch, width, and thickness should be designed comprehensively to diffract the incident light to propagate in the waveguide for scanning the whole fingerprint.
Considering that the light should propagate in TIR mode and the finger should break TIR by touching the waveguide, the Eq. (1) should be satisfied13], thus we get 44.7°< θw <78.3°. θw is the diffracted angle by the in-couple grating which obeys Eq. (2) with negative first order diffraction where k0 is the wave vector of the light; θi is the incident angle of the collimated light ; T is the grating pitch and G = 2π/T is the unit of grating vector.Eq. (3).
To be simple, the center position of detection region with width of Lfin is designed at the center between the two coupling gratings with width of Lg. The widths Lg and Lfin should satisfy Eq. (4) for coupling the incident light to fully shine the detection region.Fig. 2), the finger should be scanned by the light beam only once. It means that the longitudinal distance of the light beam between the adjacent top and bottom of waveguide, i.e., should be larger than the width of the detection region and gratings:Eq. (5), it is clearly shown that the larger thediffraction angle is, i.e. the smaller pitch of the gratings, the thinner the device is. Considering the width of a fingerprint is about 15 mm, so the width of detection region and grating Lfin = Lg = 15 mm. From Eq. (1), the maximum of diffraction angle in the waveguide is 78.3°, thus the thickness of waveguide can be reduced to 2 mm, and we got 75.06°< θw <78.3°.The process of the light propagating in the waveguide and how it scans the finger surface is simulated by the commercial software LightTools7.2 which traces the light beams based on the Monte Carlo Method. The fingerprint is set as a three dimensional cylinder with diameter of 15 mm distributing with ridges on the surface, which is regarded as a Lambertian body with reflectance of 40% and absorption of 60% at wavelength of 530nm [14, 15]. The height of the ridge is 0.1 mm, and the distance between adjacent ridges is 0.3 mm. The thickness of the glass waveguide h = 2 mm. A light beam with diameter of 15 mm is incident to the waveguide with angle of 70°, 78°, and 79°, respectively, which corresponds to the diffracted angle θw generated by the in-coupling grating in Fig. 1.The simulated results are shown in Fig. 2. In the case of θw = 70° shown in Fig. 2(a1), serious image aliasing and background mottling appear. The light tracing image in the waveguide represented with red solid lines in Fig. 2(a2) clearly shows that the finger was repeatedly scanned by the light beam, thus the image becomes blurred due to the multiple superposition. Figure 2(b1) represents the full area fingerprint image with contrast of 10:1 in the case of θw = 78°. The light tracing image in the waveguide in Fig. 2(b2) clearly shows the full fingerprint capture, for that the detection region is shot by the light beam only once. For the case of θw = 79°, which does not satisfy Eq. (1), meaning the finger cannot break TIR, thus the light is unable to get the fingerprint information, as in Fig. 2(c1) and 2(c2).
3. Fabrication and measurement
According to the design principle and light trace simulation, we perform a verification experiment, where two areas (Lg = 15 mm) of metallic gratings with pitch T = 320 nm are usedas input and output couplers, the glass substrate as waveguide is 2mm in thickness. The space between the two gratings, i.e. the detection region is 15 mm. An un-polarized green laser diode (LD) with center wavelength of 530 nm is used as light source to incident on the input grating with incident angle of 15°.
3.1. Fabrication of the metallic grating
Figure 3(a) demonstrates the fabrication process of the device. (1) A layer of photoresist (PR) (ARP3500-6, all resist Co) was spin-coated on a 2 mm thick BK7 glass substrate. (2) Then, a two-beam Lloyd mirrored laser interference lithography system illustrated in Fig. 3(b) was built to expose the photoresist for making the two gratings with exposure time of 6 minutes. (3) Next, 0.35% NaOH solution in deionized water was used as an etchant with development time of 10 seconds. After rinsing with DI water, the PR relief grating was obtained. An aluminum
(Al) film with thickness of 50 nm was deposited on the PR grating by e-beam evaporation to form the bi-layer metallic grating. The scanning electron microscope (SEM) images of the topand side views of the fabricated grating are presented in Fig. 3(c), where the grating with height 150nm and Al film 50nm is clearly viewed. Figure 3(d) shows the fabricated device with a Shanghai Jiao Tong university's logo under the detection region which illustrates the transparency of the device.
3.2. Characterization of grating diffraction
The −1st order diffractive spectra of the bi-layer Al grating are shown in Fig. 4(a) for simulations and 4(b) for experiments. The simulated results are obtained via Rigorous Coupled Wave Analysis (Rsoft, Diffraction MOD). According to the experiment data shown in Fig. 3, the structure parameters in simulation are set in Table 1 as follows:
For the case with incident angle 15°represented with black and red solid lines in 4(a), the diffraction efficiencies reach 50% for TE polarization and nearly 27% for TM polarization, respectively for wavelengths from 450 nm to 570 nm, which cover the spectrum of the green laser light source, as in Fig. 4(c). The simulated results for the dielectric grating without metal are also depicted with red and black dotted lines in Fig. 4(a), which are obviously lower than that in the case of bi-layer metallic grating. The metallic grating is preferable for the higher diffraction efficiency. Since in the real application, light source is un-polarized, the simulated diffractive spectrum composed of 50% TE polarization and 50% TM polarization with incident angle 15° are shown in Fig. 4(b) with red solid line. The photo of the grating irradiated by a collimated un-polarized white light beam is presented in the inset in Fig. 4(b), which clearly shows the out-coupled green light from the side of the glass waveguide. The measured out-coupled spectrum of the grating is also presented in Fig. 4(b). Compared with the simulated results, the experimental efficiency is lower, which may be attributed to the un-perfect fabricated gratings and the absorption of the aluminum. Despite of the small discrepancy in the values, a good agreement is obtained between the experimental curve and the simulation result. To further investigate the effect of the parameters on the diffraction efficiency, we simulated −1th order reflective diffraction spectra for duty cycle 0.1-0.9 and grating height 50-500nm. As shown in Figs. 5(a1) and (a2) of diffraction efficiencies for the TE and TM wave versus the duty cycle, the diffraction efficiencies are optimized with the duty cycle of 0.5 for both polarizations at wavelength of 530nm. As in Fig. 5(b1) and (b2) of the diffraction efficiencies under different grating depth, due to the Fabry-Perot interference effect of the air slits in the gratings, the −1th order reflective diffraction efficiencies vary periodically with the grating depth. The efficiencies for TE and TM polarizations at wavelength 530nm are optimized when grating depth is 150nm. Further simulation comparisons of efficiencies versus refractive index of the PR layer shown in Fig. 5(c1) and Fig. 5(c2). Which have similar results due to the same mechanism of Fabry-Perot interference. It can be concluded that with appropriate grating parameters, the grating couplers can be designed to achieve optimized efficiency.
3.3 Acquisition experiment of the fingerprint
To demonstrate the feasibility of a transparent optical fingerprint sensor, a snapshot of the device irradiated by a collimated green laser diode with an incident angle of 15° is shown in Fig. 6. It is clear that before touching the incident light from the right side grating is coupled out from the left one. In comparison with Fig. 6(a), Fig. 6(b) shows that when the finger touches the waveguide surface, a legible and full area fingerprint image appears on the out-coupling zone,and then shines on a filter paper to view the image directly. The inset image in Fig. 6(b) shows the process. Although the total out-coupling efficiency of the device is about 4% due to the twice grating coupling, the output picture still well displays the micro profiles of the fingerprint image as shown in Fig. 6(c).
It is essential to process the original image reliably in order to make further feature extraction [16–20]. Although many thresholding techniques, such as global and local thresholding algorithms, multi thresholding methods and adaptive thresholding techniques have been developed in this region . We chose adaptive threshold algorithm to deal with the fingerprint image due to the feasibility with inhomogeneous background in this paper. During this process, the colorful image is transformed to a grey scale image and then the average value of the related 25*25 pixel region is computed by the algorithm. Subsequently, with the average value set to be the adaptive threshold in every block, the fingerprint image is decoupled from the background. As a result, the processed fingerprint image is a binary image which brings out a sharp contrast between ridges and valleys in the fingerprint as represented in Fig. 0.6(d). Compared with the previous works which used 15mm  or 8mm thick glass plate  and uncomfortable small sensing region, our device is thinner of 2 mm in thickness and is larger in sensing area for pleasant operation. In short, we have envisioned that the proposed sensor is applicable for touch screen based device such as smart-phones or tablets, where not only touch sensing but also fingerprint capture can be fulfilled. In addition, when quantum dot backlight  or laser light source is applied, the quality of the captured fingerprint image can be guaranteed [23, 24].
We have designed and fabricated an OTFCS based on metallic grating couplers with just about 2mm of total thickness. Through diffraction theory and light tracing simulation, the significance of avoiding image aliasing by controlling grating pitch and incident angle were investigated, and a transparent and slim structure which can capture full area fingerprint image was obtained. The experimental results further confirm the feasibility of the proposed structure. Our work discloses a new approach of fingerprint capture device compatible to mobile consumer electronics and featured with transparent and slim structure.
National Natural Science Foundation of China (NSFC) (61370047, 11374212, 51235007, and 11421064).
References and links
1. X. Xia and L. O’Gorman, “Innovations in fingerprint capture devices,” Pattern Recognit. 36(2), 361–369 (2003). [CrossRef]
2. B. A. Ganji and M. S. Nateri, “A high sensitive MEMS capacitive fingerprint sensor using slotted membrane,” Microsyst. Technol. 19(1), 121–129 (2013). [CrossRef]
3. B. Bhagavatula, Ur, K. Iacovino, S. M. Kywe, L. F. Cranor and M. Savvides. “Biometric authentication on iphone and android: Usability, perceptions, and influences on adoption,” in Proc. USEC. (Internet Society, 2015), pp. 1–2.
4. R. G. Maev, E. Y. Bakulin, E. Y. Maeva, and F. M. Severin, “High resolution ultrasonic method for 3D fingerprint representation in biometrics,” in Acoustical Imaging, Iwaki Akiyama, ed. (Springer Netherlands, 2008), pp. 279–285.
5. S. Memon, M. Sepasian, and W. Balachandran, “Review of fingerprint sensing technologies,” in Proceedings of IEEE Conference on Multitopic Conference (IEEE, 2008), 226–231.
6. C. Holz and P. Baudisch, “Fiberio: a touchscreen that senses fingerprints,” in Proceedings of the 26th Annual ACM Symposium on User interface Software and Technology, Sharam Izadi, Aaron Quigley, eds. (ACM 2013), pp. 41–50.
9. Y. Zheng, W. Y. Sun, and Q. Lan, “Experimental study on waveguide hologram fingerprint sensors,” Journal of Tsing Hua University 41(11), 68–70 (2001).
12. M. H. Lee, W. T. Moon, M. J. Kim, S. K. Lee, J. S. Yoon, Y. H. Jung, and B. C. Ahn, “New light collimation film (LCF) for direct‐lit BLU system,” in SID Symposium Digest of Technical Papers (SID, 2006) pp. 503–505. [CrossRef]
13. A. Knüttel and M. Boehlau-Godau, “Spatially confined and temporally resolved refractive index and scattering evaluation in human skin performed with optical coherence tomography,” J. Biomed. Opt. 5(1), 83–92 (2000). [CrossRef] [PubMed]
14. I. Lux and L. Koblinger, Monte Carlo Particle Transport Methods: Neutron and Photon Calculations (CRC, 1991).
15. M. Kerker, The Scattering of Light and Other Electromagnetic Radiation: Physical Chemistry: A Series of Monographs (Academic, 2013).
16. I. Emiroglu and M. B. Akhan, “Pre-processing of fingerprint images,” in Proceedings of IEEE Conference on Security and Detection (IEEE, 1997), pp. 147–151.
17. H. Lin, Y. F. Wan, and J. Anil, “Fingerprint image enhancement: algorithm and performance evaluation,” IEEE T. Pattern. Anal. 20(8), 777–789 (1998). [CrossRef]
18. L. O’Gorman and J. V. Nickerson, “An approach to fingerprint filter design,” Pattern Recognit. 22(1), 29–38 (1989). [CrossRef]
19. S. Jung, R. Thewes, T. Scheiter, K. F. Goser, and W. Weber, “A low-power and high-performance CMOS fingerprint sensing and encoding architecture,” J. Solid-State Circuits 34(7), 978–984 (1999). [CrossRef]
20. T. Tamir and S. Zhang, “Modal transmission-line theory of multilayered grating structures,” J. Lightwave Technol. 14(5), 914–927 (1996). [CrossRef]
21. Y. Yang and H. Yan, “An adaptive logical method for binarization of degraded document images,” Pattern Recognit. 33(5), 787–807 (2000). [CrossRef]
22. E. Jang, S. Jun, H. Jang, J. Lim, B. Kim, and Y. Kim, “White-light-emitting diodes with quantum dot color converters for display backlights,” Adv. Mater. 22(28), 3076–3080 (2010). [CrossRef] [PubMed]
23. T. Kurashima, K. Sakuma, T. Arai, A. Taqaya, and Y. Kolke, “A polarized laser backlight using a zero-zero-birefringence polymer for liquid crystal displays,” Opt. Rev. 19(6), 415–418 (2012). [CrossRef]
24. I. Fujieda, T. Kosugi, and Y. Inaba, “Speckle noise evaluation and reduction of an edge-lit backlight system utilizing laser diodes and an optical fiber,” J. Disp. Technol. 5(11), 414–417 (2009). [CrossRef]