3468
OPTICS LETTERS / Vol. 39, No. 12 / June 15, 2014
Quantitative phase imaging using a deep UV LED source Alok Kumar Singh,* Ahmad Faridian, Peng Gao, Giancarlo Pedrini, and Wolfgang Osten Institut für Technische Optik, Universität Stuttgart, Pfaffenwaldring 9, D-70569 Stuttgart, Germany *Corresponding author: [email protected]‑stuttgart.de Received February 7, 2014; revised May 7, 2014; accepted May 8, 2014; posted May 9, 2014 (Doc. ID 206085); published June 5, 2014 We propose a method for high resolution phase imaging of biological and non-biological samples using an incoherent deep ultraviolet (DUV) LED source. The diffraction pattern of the object wave is recorded at different axial planes and the phase is retrieved by propagation of the angular spectrum. To maintain enough light intensity, we avoided using a pinhole or spectral filter for increasing the coherence of the DUV LED source. This makes the setup very simple and cost effective. The short wavelength (285 nm) of the DUV light, tuned to the absorption peak of the biological samples, allows simultaneously high resolution and high contrast images. The experimental results are presented to verify this principle. © 2014 Optical Society of America OCIS codes: (090.1995) Digital holography; (030.1640) Coherence; (110.4980) Partial coherence in imaging; (100.5070) Phase retrieval. http://dx.doi.org/10.1364/OL.39.003468
Imaging with deep-ultraviolet (DUV) light sources has many practical advantages. According to the Abbe’s criteria, the diffraction-limited lateral resolution is given by 0.61λ∕NA, where λ is the wavelength of the light source and NA is the numerical aperture of the imaging system. Thus, the lateral resolution can be increased by reducing the wavelength. In addition, since the change of phase Δφ 2πdn∕λ of a beam propagating through a medium having length d and refraction index n is inversely proportional to the wavelength, the use of ultraviolet light increases the sensitivity to the optical thickness distribution or height variation of the sample under investigation. Such short wavelength sources are being used, for instance, in scatterometry for quantitative characterization of periodic structures [1–4]. Despite its toxicity toward biological samples, DUV light has been successfully used for high contrast biological imaging and extracting the protein mass map within a cell [5]. DUV holography was successfully applied for high resolution imaging using a short coherence laser source with wavelength 193 nm [6,7]. Although, the system performance proved promising for high resolution imaging, the presence of a separate reference beam made the setup expensive and sensitive. A single beam in-line holographic system (Gabor configuration) in the same wavelength regime was also presented [8] but can be used only when most of the incident beam is not modulated by the object and acts as reference. Other single beam phase retrieval techniques based on iterative methods using a 193 nm excimer laser are also reported [9]. In all later methods, short coherent lasers were utilized, although the inherent coherent noise could not be thoroughly removed. Other drawbacks of these methods are the high cost of the laser sources (over €20 K) and the complex technical constraints in this spectral range. Although the optical components used in shortwavelength regime are still more expensive than the visible ones, utilizing an incoherent DUV LED (costs only about 150 €) instead of the laser would help to overcome the existing problems and reduce the system cost significantly. Recently, some digital holographic systems utilizing LEDs in the visible range were reported [10–14]; however, compensating for the optical path difference between the object and the reference wave in such systems requires many optical elements, which increases 0146-9592/14/123468-04$15.00/0
the complexity of the system and makes it very sensitive to vibration. In this Letter, we report a single beam phase imaging system that avoids the shortcomings of the existing techniques using an incoherent DUV LED source. Thanks to recent technology developments, LEDs are now available in this spectral range and are cost effective. For the phase retrieval, we apply a technique using intensity diffraction patterns recorded at different planes and an iterative algorithm [15–19]. The proposed system can be used for imaging technical structures as well as biological samples. Figure 1 shows the experimental setup for the phase retrieval. The light source is a 0.8 mW UV LED (LED285W–UVTOP UV LED) with central wavelength λ 285 nm and the full width at half-maximum (FWHM) is Δλ 12 nm. The resulting temporal coherence is λ2 ∕Δλ 6.7 μm. The emitting area of the source is less than 0.3 mm × 0.3 mm. A lens of diameter 10 mm and focal length 15 mm was inserted between the LED and the sample and the light was loosely focused onto the sample. This increases the spatial coherence at the sample plane. We did not use any other component, e.g., pinhole or spectral filter, for increasing the coherence of the source. The sample is magnified and imaged onto the CCD using a microscope objective (MO) having a focal length of 1 mm, NA 0.7, and corrected for infinity at the wavelength 193 nm. All the optical components are made of fused silica, to be transparent to the UV light. The CCD camera used in the experiment is a PCO Sensicam em680; quantum efficiency: 30% for 285 nm, pixels 1002 × 1024, pixel size 8 μm × 8 μm, and 10 bit. The separation between the CCD and the MO was kept relatively large (typically 80 cm) so that a very small area of the
Fig. 1. Schematic of the experimental setup. I1 ; I2 ; I3 …In are the intensity samplings at S1 ; S2 ; S3 …Sn planes, respectively; MO is microscope objective. © 2014 Optical Society of America
June 15, 2014 / Vol. 39, No. 12 / OPTICS LETTERS
3469
sample is imaged onto the CCD. The magnification was more than 200 times and the field-of-view approximately 35 μm. To detect the changes of the phase due to the variation of the refraction index or thickness, it is advantageous to illuminate the sample with a beam having constant phase. For the configuration used in our experiment the spatial coherence at the sample plane was 40 μm (measured by a shearing interferometer), which is larger than the field-of-view. Thus, the phase of the illumination beam inside the field-of-view may be considered as constant. A sequence of intensity diffraction patterns I1 , I2 , I3 …In are recorded by moving the CCD at several axial planes S1 , S2 ; S3 …Sn , spaced by Δz. The phase information is retrieved from the intensities by using an iterative method. We assume a random phase at the first imaging plane S1 and multiply it with the amplitude at this plane (square root of the measured intensity I1 ) and then propagate it to the next image plane S2 using the angular spectrum method: Ux; y; z Δz ZZ ~ x ; f y ; zPF expi2πf x x f y ydf x df y ; Uf
(1)
where f x and f y are the spatial frequencies, Ux; y; z Δz is the complex amplitude at a distance Δz from ~ x ; f y ; f z is the angular specthe sampling plane, and Uf trum of the complex amplitude at the given sampling q
plane. PF expikΔz 1 − λf x 2 − λf y 2 is the free space propagation function. Here, the angular spectrum method is preferred over Fresnel propagation, as the long distance propagation is not required. The resulting phase in the plane S2 is then combined with the measured amplitude at this plane and propagated to the next sampling plane S3 . This process continues until we reach the sampling plane Sn . To reduce the number of sampling planes, this algorithm is iterated on a fixed number of intensity images. After reaching the last sampling plane, the wavefront is propagated to the first sampling plane by the same, but inverted process. This process continues until the difference between two neighboring reconstructions is smaller than a threshold value. It should be noted that the phase was guessed only once at the first sampling plane. To check the convergence, the intensity images obtained by the propagation are compared with the sampled intensity images. This method is capable of giving fast convergence and accurate solutions; thus, there is no need to assume that the phase satisfies periodic boundary conditions, as in the case when solving the transport of intensity equation. Once the complex amplitude is obtained, it can be used to focus at different planes. Figure 2 shows the experimental results obtained using the iterative phase retrieval method and digital holography, where the test sample is a standard target with different circular and elliptical holes with sizes ranging from 1 to 8 μm (QUANTIFOIL Multi A). The camera was mounted on a movable stage, 80 cm away from the object plane and can be moved with a precision of 0.5 mm. Five images were acquired at different axial
Fig. 2. High resolution imaging of QUANTIFOIL mesh: (a) reconstructed amplitude image; (b) phase image at 285 nm by using the iterative method; (c) phase profile of the dashed line segment shown in (b); and (d) the 3D phase profile of the specimen. (e) and (f) are, respectively, the reconstructed amplitude and phase images from the digital hologram recorded using a 266 nm wavelength laser source. The contrast of the reconstructed amplitudes in (a) and (e) looks inverted because of the angle of illumination while recording the hologram.
planes at an interval of 10 mm. The initial guess of the phase was random with values ranging from 0 to π. The retrieved amplitude and phase images are shown in Figs. 2(a) and 2(b), respectively. From the images, it is obvious that using this method, clear amplitude and phase images can be obtained. Figure 2(c) shows the plot of the dashed red line section of Fig. 2(b), where the holes are separated by 1 μm, and can be easily resolved. The length of the dashed line section is shown on the x axis. The phase values and corresponding thickness are shown on the y axis, on the left and right hand sides of the figure, respectively. The thickness of the specimen, as specified by the manufacturer, is 12 nm (20%) and the measured thickness is 11.3 nm (9%). Figure 2(d) shows the 3D phase profile of the specimen. For comparison of the coherent noise reduction between the laser illumination and the LED illumination, imaging of the same sample was also performed using holography with a 266 nm wavelength laser source. Figures 2(e) and 2(f) show the reconstructed amplitude and phase images using digital holography, respectively. The quality of the images is obviously degraded by the presence of coherent noise. The reconstructed amplitude image looks grainy and the shapes of the structures are not clear. To compare the sensitivity of the two methods, the signal-to-noise ratio (SNR) is calculated. The SNR for the reconstructed amplitudes with the LED and the laser illumination are 40 and 30 dB, respectively. The higher
3470
OPTICS LETTERS / Vol. 39, No. 12 / June 15, 2014
value of SNR verifies our claim that reconstruction with the LED illumination reduces the noise in the images. We also performed the experiments on a nanostructured template made of a 35 nm gold layer, coated by ion-beam sputtering on a fused silica substrate. It included square and line structures, ranging from 100 to 500 nm in width. The SEM image of the sample is shown in Fig. 3(a). Because of the ultra-small size of the sample and low power illumination, a very small amount of the LED light could transmit through the sample. Therefore, the integration time of the camera was kept at 4 s for the imaging purpose. Here also, only five on-axis diffraction patterns were acquired at a separation of 10 mm for retrieving the phase, and the initial guess of the phase varies from 0 to π. The amplitude and phase obtained by the phase retrieval method using a 285 nm LED are shown in Figs. 3(b) and 3(c), respectively. The phase image provides finer detail. Figure 3(d) shows an image of the sample under white light illumination in the same configuration, only the LED is replaced by a white light source (FLEXILUX 2000). The image is noisy because of less transmitted light, and the resolution is also poor. By virtue of the shorter wavelength, resolution of the DUV image is better. For a similar comparison, a reconstructed amplitude image of the same sample, using off-axis holography at 193 nm, is shown in Fig. 3(e). As can be seen from the images, the resolution of the reconstructed image at 193 nm is better; however, the image quality is poor because of the inherent coherent noise. The phase profile of the dashed line segment shown in Fig. 3(c) is plotted in Fig. 3(f). The width of the lines is 500 nm and they are well-resolved. The phase information can be used to represent the 3D surface profile, as shown in Fig. 3(g). To test the system for biological samples, we performed imaging and phase retrieval on a 3 μm thick slice of mouse brain, which is kept on a fused silica substrate. The sample was mounted onto the holder and imaging was performed similar to that described above. The obtained amplitude and phase images are shown in Figs. 4(a) and 4(b), respectively. Since the proteins and nucleic acids have their absorption peaks near 285 nm,
Fig. 4. Biological sample: high contrast images of the 3 μm thick slice of mouse brain; (a) reconstructed amplitude image in the image plane and (b) reconstructed phase map. An enlarged view of the marked regions is shown in each figure. Different structures of the brain tissue are identified and shown in the magnified part of the amplitude image. The scale bar is 6 μm.
the contrast of the images is high and the structures on the sample can be distinguished properly. For example, in the magnified part of the amplitude image, the neurons, nucleus, and nucleolus are identified. The phase information can be used to measure and visualize dynamic changes in the structure, which could be useful for many medical applications. It took a total of four intensity images recorded at an interval of 10 mm each, 50 iterations and only 37 s to reconstruct the complex amplitude. For reconstruction, we used an Intel core i5-3570 processor with 3.40 GHz clock speed. The quality of the images can be further improved using an objective that is corrected for a 285 nm wavelength. In summary, we have successfully presented a singlebeam phase imaging method using an incoherent deep UV LED source for retrieving the phase of a complex specimen, including biological cells. This not only reduces the coherent noise and the complexity of the experimental setup but has all the advantages of digital holography as well as the capability of numerical focusing and wavefront reconstruction. Most importantly, the setup is less sensitive to vibrations. Since the low irradiance reduces the risk of damaging the biological specimen, it can be used for minimally invasive testing. The authors would like to acknowledge the support of the DFG-Deutsche Forschungsgemeinschaft (German Research Foundation) under grant no. OS111/19-3. The authors also gratefully thank Prof. Dr. Hermann J. Schluesener and Dr. Azadeh Ebrahimi from the Department of Neuropathology at University of Tübingen, for kindly providing the mouse brain samples.
Fig. 3. (a) SEM image of the “ITO” logo; (b) reconstructed amplitude image using phase retrieval method at 285 nm in the image plane; (c) phase image; (d) white light image of the sample in the same configuration; (e) reconstructed amplitude image using an off-axis hologram at 193 nm; (f) height variation of the dashed line segment shown in (c); and (g) phase profile of the sample in 3D. The scale bar is 3 μm.
References 1. M. Wurm, F. Pilarski, and B. Bodermann, Rev. Sci. Instrum. 81, 023701 (2010). 2. X. Niu, N. H. Jakatdar, J. Bao, C. J. Spanos, and S. K. Yedur, Proc. SPIE 3677, 159 (1999). 3. W. Osten, V. F. Paz, K. Frenner, T. Schuster, and H. Bloess, Proc. AIP 1173, 371 (2009). 4. B. Bodermann, E. Buhr, G. Ehret, F. Scholze, and M. Wurm, Proc. SPIE 7155, 71550V (2008). 5. B. J. Zeskind, C. D. Jordan, W. Timp, L. Trapani, G. Waller, V. Horodincu, D. J. Ehrlich, and P. Matsudaira, Nat. Methods 4, 567 (2007).
June 15, 2014 / Vol. 39, No. 12 / OPTICS LETTERS 6. A. Faridian, D. Hopp, G. Pedrini, U. Eigenthaler, M. Hirscher, and W. Osten, Opt. Express 18, 14159 (2010). 7. A. Faridian, D. Hopp, G. Pedrini, and W. Osten, Proc. SPIE 7718, 771803 (2010). 8. U. Gopinathan, G. Pedrini, B. Javidi, and W. Osten, J. Display Technol. 6, 479 (2010). 9. A. Anand, A. Faridian, V. Chhaniwal, G. Pedrini, W. Osten, and B. Javidi, Opt. Lett. 36, 4362 (2011). 10. R. Guo, B. Yao, P. Gao, J. Min, M. Zhou, J. Han, X. Yu, X. Yu, M. Lie, S. Yan, Y. Yang, D. Dan, and T. Ye, Appl. Opt. 52, 8233 (2013). 11. S. Stürwald, B. Kemper, C. Remmersmann, P. Langehanenberg, and G. von Bally, Proc. SPIE 6995, 699507 (2008).
3471
12. F. Dubois and C. Yourassowsky, Opt. Lett. 37, 2190 (2012). 13. F. Dubois, L. Joannes, and J. C. Legros, Appl. Opt. 38, 7085 (1999). 14. B. Kemper, S. Stürwald, C. Remmersmann, P. Langehanenberg, and G. von Bally, Opt. Lasers Eng. 46, 499 (2008). 15. G. Pedrini, W. Osten, and Y. Zhang, Opt. Lett. 30, 833 (2005). 16. P. Almoro, G. Pedrini, and W. Osten, Appl. Opt. 45, 8596 (2006). 17. F. Zhang, G. Pedrini, and W. Osten, Phys. Rev. A 75, 043805 (2007). 18. A. Mazine and K. Heggarty, Appl. Opt. 50, 2679 (2011). 19. L. J. Allen and M. P. Oxley, Opt. Commun. 199, 65 (2001).
OPTICS LETTERS / Vol. 39, No. 12 / June 15, 2014
Quantitative phase imaging using a deep UV LED source Alok Kumar Singh,* Ahmad Faridian, Peng Gao, Giancarlo Pedrini, and Wolfgang Osten Institut für Technische Optik, Universität Stuttgart, Pfaffenwaldring 9, D-70569 Stuttgart, Germany *Corresponding author: [email protected]‑stuttgart.de Received February 7, 2014; revised May 7, 2014; accepted May 8, 2014; posted May 9, 2014 (Doc. ID 206085); published June 5, 2014 We propose a method for high resolution phase imaging of biological and non-biological samples using an incoherent deep ultraviolet (DUV) LED source. The diffraction pattern of the object wave is recorded at different axial planes and the phase is retrieved by propagation of the angular spectrum. To maintain enough light intensity, we avoided using a pinhole or spectral filter for increasing the coherence of the DUV LED source. This makes the setup very simple and cost effective. The short wavelength (285 nm) of the DUV light, tuned to the absorption peak of the biological samples, allows simultaneously high resolution and high contrast images. The experimental results are presented to verify this principle. © 2014 Optical Society of America OCIS codes: (090.1995) Digital holography; (030.1640) Coherence; (110.4980) Partial coherence in imaging; (100.5070) Phase retrieval. http://dx.doi.org/10.1364/OL.39.003468
Imaging with deep-ultraviolet (DUV) light sources has many practical advantages. According to the Abbe’s criteria, the diffraction-limited lateral resolution is given by 0.61λ∕NA, where λ is the wavelength of the light source and NA is the numerical aperture of the imaging system. Thus, the lateral resolution can be increased by reducing the wavelength. In addition, since the change of phase Δφ 2πdn∕λ of a beam propagating through a medium having length d and refraction index n is inversely proportional to the wavelength, the use of ultraviolet light increases the sensitivity to the optical thickness distribution or height variation of the sample under investigation. Such short wavelength sources are being used, for instance, in scatterometry for quantitative characterization of periodic structures [1–4]. Despite its toxicity toward biological samples, DUV light has been successfully used for high contrast biological imaging and extracting the protein mass map within a cell [5]. DUV holography was successfully applied for high resolution imaging using a short coherence laser source with wavelength 193 nm [6,7]. Although, the system performance proved promising for high resolution imaging, the presence of a separate reference beam made the setup expensive and sensitive. A single beam in-line holographic system (Gabor configuration) in the same wavelength regime was also presented [8] but can be used only when most of the incident beam is not modulated by the object and acts as reference. Other single beam phase retrieval techniques based on iterative methods using a 193 nm excimer laser are also reported [9]. In all later methods, short coherent lasers were utilized, although the inherent coherent noise could not be thoroughly removed. Other drawbacks of these methods are the high cost of the laser sources (over €20 K) and the complex technical constraints in this spectral range. Although the optical components used in shortwavelength regime are still more expensive than the visible ones, utilizing an incoherent DUV LED (costs only about 150 €) instead of the laser would help to overcome the existing problems and reduce the system cost significantly. Recently, some digital holographic systems utilizing LEDs in the visible range were reported [10–14]; however, compensating for the optical path difference between the object and the reference wave in such systems requires many optical elements, which increases 0146-9592/14/123468-04$15.00/0
the complexity of the system and makes it very sensitive to vibration. In this Letter, we report a single beam phase imaging system that avoids the shortcomings of the existing techniques using an incoherent DUV LED source. Thanks to recent technology developments, LEDs are now available in this spectral range and are cost effective. For the phase retrieval, we apply a technique using intensity diffraction patterns recorded at different planes and an iterative algorithm [15–19]. The proposed system can be used for imaging technical structures as well as biological samples. Figure 1 shows the experimental setup for the phase retrieval. The light source is a 0.8 mW UV LED (LED285W–UVTOP UV LED) with central wavelength λ 285 nm and the full width at half-maximum (FWHM) is Δλ 12 nm. The resulting temporal coherence is λ2 ∕Δλ 6.7 μm. The emitting area of the source is less than 0.3 mm × 0.3 mm. A lens of diameter 10 mm and focal length 15 mm was inserted between the LED and the sample and the light was loosely focused onto the sample. This increases the spatial coherence at the sample plane. We did not use any other component, e.g., pinhole or spectral filter, for increasing the coherence of the source. The sample is magnified and imaged onto the CCD using a microscope objective (MO) having a focal length of 1 mm, NA 0.7, and corrected for infinity at the wavelength 193 nm. All the optical components are made of fused silica, to be transparent to the UV light. The CCD camera used in the experiment is a PCO Sensicam em680; quantum efficiency: 30% for 285 nm, pixels 1002 × 1024, pixel size 8 μm × 8 μm, and 10 bit. The separation between the CCD and the MO was kept relatively large (typically 80 cm) so that a very small area of the
Fig. 1. Schematic of the experimental setup. I1 ; I2 ; I3 …In are the intensity samplings at S1 ; S2 ; S3 …Sn planes, respectively; MO is microscope objective. © 2014 Optical Society of America
June 15, 2014 / Vol. 39, No. 12 / OPTICS LETTERS
3469
sample is imaged onto the CCD. The magnification was more than 200 times and the field-of-view approximately 35 μm. To detect the changes of the phase due to the variation of the refraction index or thickness, it is advantageous to illuminate the sample with a beam having constant phase. For the configuration used in our experiment the spatial coherence at the sample plane was 40 μm (measured by a shearing interferometer), which is larger than the field-of-view. Thus, the phase of the illumination beam inside the field-of-view may be considered as constant. A sequence of intensity diffraction patterns I1 , I2 , I3 …In are recorded by moving the CCD at several axial planes S1 , S2 ; S3 …Sn , spaced by Δz. The phase information is retrieved from the intensities by using an iterative method. We assume a random phase at the first imaging plane S1 and multiply it with the amplitude at this plane (square root of the measured intensity I1 ) and then propagate it to the next image plane S2 using the angular spectrum method: Ux; y; z Δz ZZ ~ x ; f y ; zPF expi2πf x x f y ydf x df y ; Uf
(1)
where f x and f y are the spatial frequencies, Ux; y; z Δz is the complex amplitude at a distance Δz from ~ x ; f y ; f z is the angular specthe sampling plane, and Uf trum of the complex amplitude at the given sampling q
plane. PF expikΔz 1 − λf x 2 − λf y 2 is the free space propagation function. Here, the angular spectrum method is preferred over Fresnel propagation, as the long distance propagation is not required. The resulting phase in the plane S2 is then combined with the measured amplitude at this plane and propagated to the next sampling plane S3 . This process continues until we reach the sampling plane Sn . To reduce the number of sampling planes, this algorithm is iterated on a fixed number of intensity images. After reaching the last sampling plane, the wavefront is propagated to the first sampling plane by the same, but inverted process. This process continues until the difference between two neighboring reconstructions is smaller than a threshold value. It should be noted that the phase was guessed only once at the first sampling plane. To check the convergence, the intensity images obtained by the propagation are compared with the sampled intensity images. This method is capable of giving fast convergence and accurate solutions; thus, there is no need to assume that the phase satisfies periodic boundary conditions, as in the case when solving the transport of intensity equation. Once the complex amplitude is obtained, it can be used to focus at different planes. Figure 2 shows the experimental results obtained using the iterative phase retrieval method and digital holography, where the test sample is a standard target with different circular and elliptical holes with sizes ranging from 1 to 8 μm (QUANTIFOIL Multi A). The camera was mounted on a movable stage, 80 cm away from the object plane and can be moved with a precision of 0.5 mm. Five images were acquired at different axial
Fig. 2. High resolution imaging of QUANTIFOIL mesh: (a) reconstructed amplitude image; (b) phase image at 285 nm by using the iterative method; (c) phase profile of the dashed line segment shown in (b); and (d) the 3D phase profile of the specimen. (e) and (f) are, respectively, the reconstructed amplitude and phase images from the digital hologram recorded using a 266 nm wavelength laser source. The contrast of the reconstructed amplitudes in (a) and (e) looks inverted because of the angle of illumination while recording the hologram.
planes at an interval of 10 mm. The initial guess of the phase was random with values ranging from 0 to π. The retrieved amplitude and phase images are shown in Figs. 2(a) and 2(b), respectively. From the images, it is obvious that using this method, clear amplitude and phase images can be obtained. Figure 2(c) shows the plot of the dashed red line section of Fig. 2(b), where the holes are separated by 1 μm, and can be easily resolved. The length of the dashed line section is shown on the x axis. The phase values and corresponding thickness are shown on the y axis, on the left and right hand sides of the figure, respectively. The thickness of the specimen, as specified by the manufacturer, is 12 nm (20%) and the measured thickness is 11.3 nm (9%). Figure 2(d) shows the 3D phase profile of the specimen. For comparison of the coherent noise reduction between the laser illumination and the LED illumination, imaging of the same sample was also performed using holography with a 266 nm wavelength laser source. Figures 2(e) and 2(f) show the reconstructed amplitude and phase images using digital holography, respectively. The quality of the images is obviously degraded by the presence of coherent noise. The reconstructed amplitude image looks grainy and the shapes of the structures are not clear. To compare the sensitivity of the two methods, the signal-to-noise ratio (SNR) is calculated. The SNR for the reconstructed amplitudes with the LED and the laser illumination are 40 and 30 dB, respectively. The higher
3470
OPTICS LETTERS / Vol. 39, No. 12 / June 15, 2014
value of SNR verifies our claim that reconstruction with the LED illumination reduces the noise in the images. We also performed the experiments on a nanostructured template made of a 35 nm gold layer, coated by ion-beam sputtering on a fused silica substrate. It included square and line structures, ranging from 100 to 500 nm in width. The SEM image of the sample is shown in Fig. 3(a). Because of the ultra-small size of the sample and low power illumination, a very small amount of the LED light could transmit through the sample. Therefore, the integration time of the camera was kept at 4 s for the imaging purpose. Here also, only five on-axis diffraction patterns were acquired at a separation of 10 mm for retrieving the phase, and the initial guess of the phase varies from 0 to π. The amplitude and phase obtained by the phase retrieval method using a 285 nm LED are shown in Figs. 3(b) and 3(c), respectively. The phase image provides finer detail. Figure 3(d) shows an image of the sample under white light illumination in the same configuration, only the LED is replaced by a white light source (FLEXILUX 2000). The image is noisy because of less transmitted light, and the resolution is also poor. By virtue of the shorter wavelength, resolution of the DUV image is better. For a similar comparison, a reconstructed amplitude image of the same sample, using off-axis holography at 193 nm, is shown in Fig. 3(e). As can be seen from the images, the resolution of the reconstructed image at 193 nm is better; however, the image quality is poor because of the inherent coherent noise. The phase profile of the dashed line segment shown in Fig. 3(c) is plotted in Fig. 3(f). The width of the lines is 500 nm and they are well-resolved. The phase information can be used to represent the 3D surface profile, as shown in Fig. 3(g). To test the system for biological samples, we performed imaging and phase retrieval on a 3 μm thick slice of mouse brain, which is kept on a fused silica substrate. The sample was mounted onto the holder and imaging was performed similar to that described above. The obtained amplitude and phase images are shown in Figs. 4(a) and 4(b), respectively. Since the proteins and nucleic acids have their absorption peaks near 285 nm,
Fig. 4. Biological sample: high contrast images of the 3 μm thick slice of mouse brain; (a) reconstructed amplitude image in the image plane and (b) reconstructed phase map. An enlarged view of the marked regions is shown in each figure. Different structures of the brain tissue are identified and shown in the magnified part of the amplitude image. The scale bar is 6 μm.
the contrast of the images is high and the structures on the sample can be distinguished properly. For example, in the magnified part of the amplitude image, the neurons, nucleus, and nucleolus are identified. The phase information can be used to measure and visualize dynamic changes in the structure, which could be useful for many medical applications. It took a total of four intensity images recorded at an interval of 10 mm each, 50 iterations and only 37 s to reconstruct the complex amplitude. For reconstruction, we used an Intel core i5-3570 processor with 3.40 GHz clock speed. The quality of the images can be further improved using an objective that is corrected for a 285 nm wavelength. In summary, we have successfully presented a singlebeam phase imaging method using an incoherent deep UV LED source for retrieving the phase of a complex specimen, including biological cells. This not only reduces the coherent noise and the complexity of the experimental setup but has all the advantages of digital holography as well as the capability of numerical focusing and wavefront reconstruction. Most importantly, the setup is less sensitive to vibrations. Since the low irradiance reduces the risk of damaging the biological specimen, it can be used for minimally invasive testing. The authors would like to acknowledge the support of the DFG-Deutsche Forschungsgemeinschaft (German Research Foundation) under grant no. OS111/19-3. The authors also gratefully thank Prof. Dr. Hermann J. Schluesener and Dr. Azadeh Ebrahimi from the Department of Neuropathology at University of Tübingen, for kindly providing the mouse brain samples.
Fig. 3. (a) SEM image of the “ITO” logo; (b) reconstructed amplitude image using phase retrieval method at 285 nm in the image plane; (c) phase image; (d) white light image of the sample in the same configuration; (e) reconstructed amplitude image using an off-axis hologram at 193 nm; (f) height variation of the dashed line segment shown in (c); and (g) phase profile of the sample in 3D. The scale bar is 3 μm.
References 1. M. Wurm, F. Pilarski, and B. Bodermann, Rev. Sci. Instrum. 81, 023701 (2010). 2. X. Niu, N. H. Jakatdar, J. Bao, C. J. Spanos, and S. K. Yedur, Proc. SPIE 3677, 159 (1999). 3. W. Osten, V. F. Paz, K. Frenner, T. Schuster, and H. Bloess, Proc. AIP 1173, 371 (2009). 4. B. Bodermann, E. Buhr, G. Ehret, F. Scholze, and M. Wurm, Proc. SPIE 7155, 71550V (2008). 5. B. J. Zeskind, C. D. Jordan, W. Timp, L. Trapani, G. Waller, V. Horodincu, D. J. Ehrlich, and P. Matsudaira, Nat. Methods 4, 567 (2007).
June 15, 2014 / Vol. 39, No. 12 / OPTICS LETTERS 6. A. Faridian, D. Hopp, G. Pedrini, U. Eigenthaler, M. Hirscher, and W. Osten, Opt. Express 18, 14159 (2010). 7. A. Faridian, D. Hopp, G. Pedrini, and W. Osten, Proc. SPIE 7718, 771803 (2010). 8. U. Gopinathan, G. Pedrini, B. Javidi, and W. Osten, J. Display Technol. 6, 479 (2010). 9. A. Anand, A. Faridian, V. Chhaniwal, G. Pedrini, W. Osten, and B. Javidi, Opt. Lett. 36, 4362 (2011). 10. R. Guo, B. Yao, P. Gao, J. Min, M. Zhou, J. Han, X. Yu, X. Yu, M. Lie, S. Yan, Y. Yang, D. Dan, and T. Ye, Appl. Opt. 52, 8233 (2013). 11. S. Stürwald, B. Kemper, C. Remmersmann, P. Langehanenberg, and G. von Bally, Proc. SPIE 6995, 699507 (2008).
3471
12. F. Dubois and C. Yourassowsky, Opt. Lett. 37, 2190 (2012). 13. F. Dubois, L. Joannes, and J. C. Legros, Appl. Opt. 38, 7085 (1999). 14. B. Kemper, S. Stürwald, C. Remmersmann, P. Langehanenberg, and G. von Bally, Opt. Lasers Eng. 46, 499 (2008). 15. G. Pedrini, W. Osten, and Y. Zhang, Opt. Lett. 30, 833 (2005). 16. P. Almoro, G. Pedrini, and W. Osten, Appl. Opt. 45, 8596 (2006). 17. F. Zhang, G. Pedrini, and W. Osten, Phys. Rev. A 75, 043805 (2007). 18. A. Mazine and K. Heggarty, Appl. Opt. 50, 2679 (2011). 19. L. J. Allen and M. P. Oxley, Opt. Commun. 199, 65 (2001).
