Note: Real time three-dimensional topography measurement of microfluidic devices with pillar structures using confocal microscope Kar Tien Ang, Zhong Ping Fang, and Arthur Tay Citation: Review of Scientific Instruments 85, 026108 (2014); doi: 10.1063/1.4865112 View online: http://dx.doi.org/10.1063/1.4865112 View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/85/2?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Three-dimensional surface profile measurement using a beam scanning chromatic confocal microscope Rev. Sci. Instrum. 80, 073706 (2009); 10.1063/1.3184023 Three-dimensional observations of polar domain structures using a confocal second-harmonic generation interference microscope J. Appl. Phys. 104, 054112 (2008); 10.1063/1.2975218 Multifunctional fluorescence correlation microscope for intracellular and microfluidic measurements Rev. Sci. Instrum. 78, 053711 (2007); 10.1063/1.2740053 Real-time three-dimensional surface measurement by color encoded light projection Appl. Phys. Lett. 89, 111108 (2006); 10.1063/1.2352729 Error budget of step height and pitch measurement using a scanning tunneling microscope with a threedimensional interferometer J. Vac. Sci. Technol. B 15, 1494 (1997); 10.1116/1.589481
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
REVIEW OF SCIENTIFIC INSTRUMENTS 85, 026108 (2014)
Note: Real time three-dimensional topography measurement of microfluidic devices with pillar structures using confocal microscope Kar Tien Ang,1 Zhong Ping Fang,2 and Arthur Tay1,a) 1 Electrical and Computer Engineering Department, National University of Singapore, 4 Engineering Drive 3, Singapore 117576 2 Singapore Institute of Manufacturing Technology, 71 Nanyang Drive, Singapore 638075
(Received 16 October 2013; accepted 25 January 2014; published online 10 February 2014) Miniature pillars are three-dimensional (3D) features commonly found in microfluidic device. These features are usually employed as filters. Non-confocal profilometers have difficulties in measuring 3D topography of pillar structures in transparent microfluidic devices. Confocal sensors can be used to measure the 3D topography of pillar structures but they are usually time consuming due to the scanning process. We have developed a technique to measure 3D topography using a modified confocal microscope with a spinning Nipkow disk and chromatic confocal technique. Experimental results on a microfluidic device with pillar structures demonstrate the feasibility of the proposed technique. Our technique is suitable for in situ, real time measurement of microfluidic device at production speed since it requires only one confocal image to complete a measurement. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4865112] Microfluidic devices have found widespread applications in biology, chemistry, and photonics. As the dimensions of microfluidic devices continue to decrease and the geometric complexity of the devices increases, profilometers that were initially used to measure surface profiles have now become important tools for three-dimensional (3D) topography measurement in these microfluidic devices. Miniature pillars are a kind of 3D feature of microfluidic device, widely used as filters. Fig. 1 shows an example of miniature pillars on a microfluidic device chip. The diameter, pitch, and height of the pillars are about 5 μm, 15 μm, and 50 μm, respectively. In mass production, all of the dimensions are required to be monitored in situ and real-time at production speed. Most commercial profilometers are not suitable to measure these pillar structures. For example, profilometers which are based on phase-shifting interferometer1 principle are not able to measure surface with pillar structure because of the phase unwrapping problem due to the large step height (i.e., pillar). Profilometers based on focus variation2 are not able to measure the surface due to the smooth surfaces of microfluidic chips. Profilometers based on point-autofocus3 techniques are not suitable because light path will be blocked by the pillars during the scanning process. Similarly, profilometers that use triangulation, structure light, or pattern projection4 methods cannot measure transparent surface as parts of light path are blocked by the pillars. The confocal technique5 is able to measure the surface profile of the pillars, but its measurement speed is too slow because it is a point by point measurement technique in which lateral and axial scanning is required to scan over the whole 3D surface. In this note, we demonstrate a technique for in situ inspection of the microfluidic devices by modifying a commercial confocal microscope which takes less than a
minute to complete measurement for an area of 0.359 mm × 0.266 mm. Carl Zeiss’s Axiotron 2 Visible light-Ultra Violet (VISUV) Confocal Scanning Module (CSM) is a commercial confocal microscope that uses the spinning Nipkow disk and chromatic confocal technique.6 The optical system of the microscope is shown in Fig. 2. The halogen light is collimated and focused on the Nipkow spinning disk after passing through a motorized aperture diaphragm and a gray filter. Each pinhole on the Nipkow disk is regarded as a point light source. The output light from the pinhole is then focused by a tube lens and an objective lens onto the sample. An optical element called “Chromat C” is inserted between the tube lens and the objective lens to produce the chromatic aberration. Chromatic aberration causes the light of different wavelengths focus at different focal planes. Combination of “Chromat C” and the objective lens causes the light of shorter wavelength to have shorter focal length while the light with longer wavelength to have longer focal length. The difference between the focal length of red light (wavelength, λ ≈ 700 nm) and violet light (wavelength, λ ≈ 400 nm) depends on the magnification of the objective lens used. For example, the red and violet light focal length difference for magnification of 20× microscope objective is about 55 μm, while it is 8.8 μm for 50× and 2.2 μm for 100×.
a) Author to whom correspondence should be addressed. Electronic mail:
FIG. 1. The pillar structures of a microfluidic device (filter), objective magnification 5× (left) and 20× (right).
[email protected]. 0034-6748/2014/85(2)/026108/3/$30.00
85, 026108-1
© 2014 AIP Publishing LLC
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
026108-2
Ang, Fang, and Tay
Rev. Sci. Instrum. 85, 026108 (2014)
FIG. 4. Calibrated RGB curves.
FIG. 2. Schematic of Carl Zeiss’s Axiotron 2 CSM VIS-UV confocal microscope.
The wavelengths which are focused on the sample surface will be reflected and pass through the pinholes on the Nipkow disk, while the wavelengths which are out-of-focus will be filtered out by the pinholes on the Nipkow disk. The reflected light beams that passed through the pinholes will be captured by color camera. To convert the confocal microscope to become a profilometer, the confocal microscope must first go through a calibration process to find out the relationship between Red, Green, Blue (RGB) values of a pixel and the surface height level. The experimental setup for the calibration process is shown in Fig. 3. During calibration, a piezoelectric transducer driven linear translation stage (PZT stage) is placed on the stage of the confocal microscope. A piece of microfluidic device material with flat surface is used as the sample. The sample is placed on the PZT stage. The magnification of the objective lens used is 20×. The arbitrary initial depth position is labeled as (D = 0). The confocal microscopic image of the sample is captured. Next, the sample is manually moved down by 1 μm using the PZT controller. The new depth position is labeled as D = 1. The confocal microscopic image of the sample is
FIG. 3. Experiment setup for calibration process.
captured. The procedure was repeated until the depth position of D = 100 is reached; a total of 101 confocal microscopic images at different depth positions are collected. From these images, the relationship between the depth positions and RGB values was plotted as shown in Fig. 4. From the calibrated RGB curves, it is clear that the vertical measurement range is about 55 μm (depth position D = 20 to D = 75) because the “Chromat-C” can only introduce chromatic aberration in the range of 55 μm for 20× microscope objective for visible light. Therefore, the calibration points whose position is higher than D = 20 or lower than D = 75 are out of the focal range and appear dark. Only 56 calibration points, i.e., from D = 20 to D = 75, would be used for extracting surface height information. When these 56 calibration points are plotted in hue, saturation, value (HSV) color space, we get a calibrated curve as shown in Fig. 5. The normalized hue, saturation, and value can only vary from 0 to 1.0. To extract surface height information of a pixel, first we convert RGB of the pixel to HSV. Then the Euclidean distances of the pixel to the 56 calibration points in HSV color space are computed. Finally, the depth
FIG. 5. Fifty-six calibration points in HSV color space.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
026108-3
Ang, Fang, and Tay
FIG. 6. Confocal microscopic image (20×) of the microfluidic device sample.
position of the pixel is given by the calibration point which has the minimum Euclidean distances from the pixel. After calibration, a confocal microscopic image of the microfluidic device sample is captured as shown in Fig. 6. The 3D topography of the sample was successfully retrieved as shown in Fig. 7 and it is verified to be correct based on the information given by the manufacturer. From the experimental results, it is clear that our technique is suitable to measure 3D topography of a transparent microfluidic device that has micro-pillar structures. After proper calibration, the technique requires only one confocal image to retrieve 3D topography information of specimen and the process only takes a few seconds. The technique
Rev. Sci. Instrum. 85, 026108 (2014)
is thus very fast compared to other commercial profilometers that normally take a few minutes to complete a 3D topography measurement of an area with the similar size (0.359 mm × 0.266 mm) which contains a total of 3 207 360 measurement points (2080 pixels × 1543 pixels). In addition, the modified confocal microscope is compared with a commercial profilometer, Alicona InfiniteFocus.7 The two systems are used to measure the height of the letter “J” on a Singapore 10¢. The average heights measured by the modified confocal profilometer and the commercial profilometer are 45.0 μm and 45.42 μm, respectively. Currently, the modified confocal microscope is calibrated with the step size of 1 μm and its vertical resolution is 1 μm. The images of the current system are 8 bits per channel images. It can differentiate about 16.8 × 106 colors. The vertical resolution can be further improved by employing a higher bit per channel camera. For example, about 4.4 × 1012 colors can be differentiated by using 14 bits per channel. The limitations in measuring sidewalls and sharp edges are common issues for all optical microscopes.8 For the modified confocal microscope, it is currently not possible to measure the sidewalls, sharp edges, and surface with the slope greater than 15◦ . The pixels at the sidewalls, shape edges, and steep slope surface will appear dark and similar to those pixels at out-of-focus area. We acknowledge the financial support by the Singapore Institute of Manufacturing Technology (SIMTech) and the National University of Singapore. Furthermore, we would like to thank Microfluidic Manufacturing and Precision Measurement groups at SIMTech for providing the microfluidic device sample and confocal microscope. 1 J.
FIG. 7. The 3D topography of the microfluidic device retrieved using the calibrated confocal microscope.
Lim and S. Rah, “Absolute measurement of the reference surface profile of a phase shifting interferometer,” Rev. Sci. Instrum. 77, 086107 (2006). 2 R. Danzl, F. Helmli, P. Rubert, and M. Prantl, “Optical roughness measurements on specially designed roughness standards,” Proc. SPIE 7102, 71020M (2008). 3 H. Fukatsu and K. Yanagi, “Development of an optical stylus displacement sensor for surface profiling instruments,” Microsyst. Technol. 11, 582–589 (2005). 4 M. Yamamoto and T. Yoshizawa, “Surface profile measurement by grating projection method with dual-projection optics,” Proc. SPIE 6000, 60000I (2005). 5 B. S. Chun, K. Kim, and D. Gweon, “Three-dimensional surface profile measurement using a beam scanning chromatic confocal microscope,” Rev. Sci. Instrum. 80, 073706 (2009). 6 M. Hildebrandt, Zeiss Axiotron Inspection Microscope Operating Manual preliminary version, version 0.1, HSEB Dresden GmbH, 2005. 7 See http://www.alicona.com/home/products/infinitefocus-standard.html for InfiniteFocus for form and roughness measurement, Alicona Imaging GmbH, Alicona (Cited on 20 November 2013). 8 S. Li, Z. Xu, I. Reading, S. F. Yoon, Z. P. Fang, and J. Zhao, “Three dimensional sidewall measurements by laser fluorescent confocal microscopy,” Opt. Express 16(6), 4001 (2008).
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
REVIEW OF SCIENTIFIC INSTRUMENTS 85, 026108 (2014)
Note: Real time three-dimensional topography measurement of microfluidic devices with pillar structures using confocal microscope Kar Tien Ang,1 Zhong Ping Fang,2 and Arthur Tay1,a) 1 Electrical and Computer Engineering Department, National University of Singapore, 4 Engineering Drive 3, Singapore 117576 2 Singapore Institute of Manufacturing Technology, 71 Nanyang Drive, Singapore 638075
(Received 16 October 2013; accepted 25 January 2014; published online 10 February 2014) Miniature pillars are three-dimensional (3D) features commonly found in microfluidic device. These features are usually employed as filters. Non-confocal profilometers have difficulties in measuring 3D topography of pillar structures in transparent microfluidic devices. Confocal sensors can be used to measure the 3D topography of pillar structures but they are usually time consuming due to the scanning process. We have developed a technique to measure 3D topography using a modified confocal microscope with a spinning Nipkow disk and chromatic confocal technique. Experimental results on a microfluidic device with pillar structures demonstrate the feasibility of the proposed technique. Our technique is suitable for in situ, real time measurement of microfluidic device at production speed since it requires only one confocal image to complete a measurement. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4865112] Microfluidic devices have found widespread applications in biology, chemistry, and photonics. As the dimensions of microfluidic devices continue to decrease and the geometric complexity of the devices increases, profilometers that were initially used to measure surface profiles have now become important tools for three-dimensional (3D) topography measurement in these microfluidic devices. Miniature pillars are a kind of 3D feature of microfluidic device, widely used as filters. Fig. 1 shows an example of miniature pillars on a microfluidic device chip. The diameter, pitch, and height of the pillars are about 5 μm, 15 μm, and 50 μm, respectively. In mass production, all of the dimensions are required to be monitored in situ and real-time at production speed. Most commercial profilometers are not suitable to measure these pillar structures. For example, profilometers which are based on phase-shifting interferometer1 principle are not able to measure surface with pillar structure because of the phase unwrapping problem due to the large step height (i.e., pillar). Profilometers based on focus variation2 are not able to measure the surface due to the smooth surfaces of microfluidic chips. Profilometers based on point-autofocus3 techniques are not suitable because light path will be blocked by the pillars during the scanning process. Similarly, profilometers that use triangulation, structure light, or pattern projection4 methods cannot measure transparent surface as parts of light path are blocked by the pillars. The confocal technique5 is able to measure the surface profile of the pillars, but its measurement speed is too slow because it is a point by point measurement technique in which lateral and axial scanning is required to scan over the whole 3D surface. In this note, we demonstrate a technique for in situ inspection of the microfluidic devices by modifying a commercial confocal microscope which takes less than a
minute to complete measurement for an area of 0.359 mm × 0.266 mm. Carl Zeiss’s Axiotron 2 Visible light-Ultra Violet (VISUV) Confocal Scanning Module (CSM) is a commercial confocal microscope that uses the spinning Nipkow disk and chromatic confocal technique.6 The optical system of the microscope is shown in Fig. 2. The halogen light is collimated and focused on the Nipkow spinning disk after passing through a motorized aperture diaphragm and a gray filter. Each pinhole on the Nipkow disk is regarded as a point light source. The output light from the pinhole is then focused by a tube lens and an objective lens onto the sample. An optical element called “Chromat C” is inserted between the tube lens and the objective lens to produce the chromatic aberration. Chromatic aberration causes the light of different wavelengths focus at different focal planes. Combination of “Chromat C” and the objective lens causes the light of shorter wavelength to have shorter focal length while the light with longer wavelength to have longer focal length. The difference between the focal length of red light (wavelength, λ ≈ 700 nm) and violet light (wavelength, λ ≈ 400 nm) depends on the magnification of the objective lens used. For example, the red and violet light focal length difference for magnification of 20× microscope objective is about 55 μm, while it is 8.8 μm for 50× and 2.2 μm for 100×.
a) Author to whom correspondence should be addressed. Electronic mail:
FIG. 1. The pillar structures of a microfluidic device (filter), objective magnification 5× (left) and 20× (right).
[email protected]. 0034-6748/2014/85(2)/026108/3/$30.00
85, 026108-1
© 2014 AIP Publishing LLC
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
026108-2
Ang, Fang, and Tay
Rev. Sci. Instrum. 85, 026108 (2014)
FIG. 4. Calibrated RGB curves.
FIG. 2. Schematic of Carl Zeiss’s Axiotron 2 CSM VIS-UV confocal microscope.
The wavelengths which are focused on the sample surface will be reflected and pass through the pinholes on the Nipkow disk, while the wavelengths which are out-of-focus will be filtered out by the pinholes on the Nipkow disk. The reflected light beams that passed through the pinholes will be captured by color camera. To convert the confocal microscope to become a profilometer, the confocal microscope must first go through a calibration process to find out the relationship between Red, Green, Blue (RGB) values of a pixel and the surface height level. The experimental setup for the calibration process is shown in Fig. 3. During calibration, a piezoelectric transducer driven linear translation stage (PZT stage) is placed on the stage of the confocal microscope. A piece of microfluidic device material with flat surface is used as the sample. The sample is placed on the PZT stage. The magnification of the objective lens used is 20×. The arbitrary initial depth position is labeled as (D = 0). The confocal microscopic image of the sample is captured. Next, the sample is manually moved down by 1 μm using the PZT controller. The new depth position is labeled as D = 1. The confocal microscopic image of the sample is
FIG. 3. Experiment setup for calibration process.
captured. The procedure was repeated until the depth position of D = 100 is reached; a total of 101 confocal microscopic images at different depth positions are collected. From these images, the relationship between the depth positions and RGB values was plotted as shown in Fig. 4. From the calibrated RGB curves, it is clear that the vertical measurement range is about 55 μm (depth position D = 20 to D = 75) because the “Chromat-C” can only introduce chromatic aberration in the range of 55 μm for 20× microscope objective for visible light. Therefore, the calibration points whose position is higher than D = 20 or lower than D = 75 are out of the focal range and appear dark. Only 56 calibration points, i.e., from D = 20 to D = 75, would be used for extracting surface height information. When these 56 calibration points are plotted in hue, saturation, value (HSV) color space, we get a calibrated curve as shown in Fig. 5. The normalized hue, saturation, and value can only vary from 0 to 1.0. To extract surface height information of a pixel, first we convert RGB of the pixel to HSV. Then the Euclidean distances of the pixel to the 56 calibration points in HSV color space are computed. Finally, the depth
FIG. 5. Fifty-six calibration points in HSV color space.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
026108-3
Ang, Fang, and Tay
FIG. 6. Confocal microscopic image (20×) of the microfluidic device sample.
position of the pixel is given by the calibration point which has the minimum Euclidean distances from the pixel. After calibration, a confocal microscopic image of the microfluidic device sample is captured as shown in Fig. 6. The 3D topography of the sample was successfully retrieved as shown in Fig. 7 and it is verified to be correct based on the information given by the manufacturer. From the experimental results, it is clear that our technique is suitable to measure 3D topography of a transparent microfluidic device that has micro-pillar structures. After proper calibration, the technique requires only one confocal image to retrieve 3D topography information of specimen and the process only takes a few seconds. The technique
Rev. Sci. Instrum. 85, 026108 (2014)
is thus very fast compared to other commercial profilometers that normally take a few minutes to complete a 3D topography measurement of an area with the similar size (0.359 mm × 0.266 mm) which contains a total of 3 207 360 measurement points (2080 pixels × 1543 pixels). In addition, the modified confocal microscope is compared with a commercial profilometer, Alicona InfiniteFocus.7 The two systems are used to measure the height of the letter “J” on a Singapore 10¢. The average heights measured by the modified confocal profilometer and the commercial profilometer are 45.0 μm and 45.42 μm, respectively. Currently, the modified confocal microscope is calibrated with the step size of 1 μm and its vertical resolution is 1 μm. The images of the current system are 8 bits per channel images. It can differentiate about 16.8 × 106 colors. The vertical resolution can be further improved by employing a higher bit per channel camera. For example, about 4.4 × 1012 colors can be differentiated by using 14 bits per channel. The limitations in measuring sidewalls and sharp edges are common issues for all optical microscopes.8 For the modified confocal microscope, it is currently not possible to measure the sidewalls, sharp edges, and surface with the slope greater than 15◦ . The pixels at the sidewalls, shape edges, and steep slope surface will appear dark and similar to those pixels at out-of-focus area. We acknowledge the financial support by the Singapore Institute of Manufacturing Technology (SIMTech) and the National University of Singapore. Furthermore, we would like to thank Microfluidic Manufacturing and Precision Measurement groups at SIMTech for providing the microfluidic device sample and confocal microscope. 1 J.
FIG. 7. The 3D topography of the microfluidic device retrieved using the calibrated confocal microscope.
Lim and S. Rah, “Absolute measurement of the reference surface profile of a phase shifting interferometer,” Rev. Sci. Instrum. 77, 086107 (2006). 2 R. Danzl, F. Helmli, P. Rubert, and M. Prantl, “Optical roughness measurements on specially designed roughness standards,” Proc. SPIE 7102, 71020M (2008). 3 H. Fukatsu and K. Yanagi, “Development of an optical stylus displacement sensor for surface profiling instruments,” Microsyst. Technol. 11, 582–589 (2005). 4 M. Yamamoto and T. Yoshizawa, “Surface profile measurement by grating projection method with dual-projection optics,” Proc. SPIE 6000, 60000I (2005). 5 B. S. Chun, K. Kim, and D. Gweon, “Three-dimensional surface profile measurement using a beam scanning chromatic confocal microscope,” Rev. Sci. Instrum. 80, 073706 (2009). 6 M. Hildebrandt, Zeiss Axiotron Inspection Microscope Operating Manual preliminary version, version 0.1, HSEB Dresden GmbH, 2005. 7 See http://www.alicona.com/home/products/infinitefocus-standard.html for InfiniteFocus for form and roughness measurement, Alicona Imaging GmbH, Alicona (Cited on 20 November 2013). 8 S. Li, Z. Xu, I. Reading, S. F. Yoon, Z. P. Fang, and J. Zhao, “Three dimensional sidewall measurements by laser fluorescent confocal microscopy,” Opt. Express 16(6), 4001 (2008).
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 129.120.242.61 On: Wed, 03 Dec 2014 18:21:50
