Confocal Microscopy Scanned by Digital Micromirror Devise with. Stray Light Filters

Transcription

1 Confocal Microscopy Scanned by Digital Micromirror Devise with Stray Light Filters Chuan-Cheng Hung 1 Chang-Ching Lin 2 Koung-Ming Yeh 3 Yi-Chin Fang 4 Jia-Hua Wu 5 Hung-Chi Sun 6 Wei-Chi Lai 7 Yi-Liang Chen 8 1 Department of Electronic Engineering, Kao Yuan University 2 3 Metal Industries Research and Development Centre Institute Electro-Optical Engineering, Kaohsiung First University of Science and Technology ABSTRACT This research proposes a newly developed stray light filter, which might significantly eliminate the stray light and ghost image effect for the non-contact con-focal microscopy system handled by (Digital Micromirror Device, DMD) devise. DMD devise, which was introduced by Taxes instrument under advanced technology of Micro-Electro-Mechanical Systems, MEMS, could be the replacement of traditional scanning system. The employment of DMD system takes advantages of fast scanning rate, better resolution, simplifies optical system. However, traditional confocal microscopy system with pinhole will lead to light starving and potentially higher signal to noise ratio.without pinhole, the stray light and ghost image effect might complicate the signal measurement. A newly developed stray light filter will be presented in this research in order to eliminate the potential stray light and ghost image without sacrifice of luminance. It indicated that not only optical system could be much simplified but also resolution could be one step higher because neither pinhole nor CCD camera lens will be employed in this system. Experimental results will be shown in this research demonstrating an increase in contrast up to 60%. Keywords: confocal microscopy, surface profilometry, Digital Micromirror Devices (DMD), surface inspection 1. INTRODUCTION Automatic inspection technique could be divided into two categories: contact modeinspection and non--contact Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS VII, edited by Allyson L. Hartzell, Rajeshuni Ramesham, Proc. of SPIE Vol. 6884, 68840T, (2008) X/08/$18 doi: / SPIE Digital Library -- Subscriber Archive Copy Proc. of SPIE Vol T-1

2 mode inspection. Contact mode inspectionis widely used in the industrial field. Using the mode s point-by-point method, a comprehensive record of data on the sample s configuration could be obtained. However, it could be a time consuming process due to the fact that damage could be done to the sample s surface through the probe s contact. Furthermore, the probe s dimension prohibits it from yielding highly accurate results, thus it is not suitable for the inspection of micro-structures.non--contact mode inspectionconsists of different measuring methods such as sound-wave, magnetic-wave and light-wave. Within the different methods, light-wave has the advantage of being anti--interference, therefore most extensively applied. Optical measurement system selects unit device for high-resolution image comprising different types of light sources. It is both fast and accurate. Furthermore, because it does not come into contact with the sample, under circumstances where the micro-structures are not permitted to be touched, optical measurement method serves as an ideal choice. Confocal microscopy is an invention that has been successfully applied for many years. Its theory was created by Marvin Minsky in his 1957 patented design [1]. Due to a lack of capability in computer processed data, the development of this theory was forced to come to a halt. It was not till 1969 that the theory became validated. P. Avidovits and M.D. Egger actualized Minsky s theory and fabricated the early conventional optical microscope and fluorescence microscope [2]. A disadvantage of conventional microscope is that it used wide-view plane imaging. Confocal microscope employs the method of replacing the wide-view with space filter. This helps to obtain a more precise relationship between the focus of sample s surface and the focus of CCD camera s image. Light s intensity was traded-off for a better transverse and longitudinal resolution. Confocal microscopy was first applied in biologic measurements. Therefore, its application in the biology field is quite extensive. The traditional fluorescent stained samples are observed using fluorescence microscope. But due to its deep view, fluorescence microscopes cannot be adopted to observe a single particular boundary: the rendered images are often extremely vague, prohibiting the identification of the staining location. As a result, the technique of thin sectioning was born. However, thin section does not allow the observation of actual living samples, making it difficult to render the sample s original state. This difficulty was not solved until the creation of confocal microscopy system; theoretically speaking, this system provides the imaging mechanism to overcome the limit of the living samples [3]. 2. CONFOCAL MICROSCOPY S THEORY AND TECHNIQUE Traditional optical microscope utilizes wide-view method to observe samples, as shown in Figure 1. Sample s focal plane s reflecting light, excluding rays accompanied by non-focal plane s rays, causes the reduction of resolution power. Traditional optical microscope is also constrained by diffraction, therefore unable to provide high resolution power. Confocal microscopy s optical path and imaging theory resemble those of the conventional optical microscope, as shown in Figure 2. The main difference lies in the pinhole. Light source view s optical focus and sample s focal plane are conjugated through the pinhole in front of CCD. Compared to Figure 1, although it sacrifices view s intensity, optical resolution and optical section are improved. It also reduces background light noise and keeps the depth of view in control. Due to a lack of suitable light source and computer at the time, the development of confocal microscopy came to a temporary stop. Proc. of SPIE Vol T-2

3 Microscope has been widely used in biology and medical fields for the measurement of skeleton dimensions of micro samples. It has also been an important topic in recent years. Many optical outline measuring methods are created, including projection moiré, laser focusing technique, heterodyne interference, structure light, laser scan, laser scan confocal microscope, etc. All of these could be employed to measure 3D geometrical profile. During earlier periods, high-resolution confocal imaging system was built through turning Nipkow disc [5-8] at a very high speed to gather sample s information. Elements are formed using micromirror array and pinhole array to perform horizontal multi-points plane scan. Its disadvantage includes low-efficiency lamination and rigid rotating mechanism. 3. CONFOCAL MULTIBEAM SCAN MICROSCOPY SYSTEM The measurement of sample s geometrical space (x, y, z) in confocal scan could be divided into two mechanisms. One is targeted at fixed focal plane. Moving sample s depth direction (usually referred to as z ), performs scan on plane direction (x, y). The depth resolution is determined by the moving device s optical section capability on the scan position. The other is targeted at unfixed plane, rendering measurement for fixed sample s geometrical profile. Our system employs the former, projecting onto the traverse (x-y direction) structure, as shown in Figure 3. This system s application is quite similar to that of Nipkow disk. The system employs DMD technique, developed by Larry J. Hornbeck at TI in It improves the mechanically rotating Nipkow disk using the switch on DMD s micromirror. The speed of the switch is 15µs. The concept of confocal theory is to adopt the pinhole s filtration function, the larger the pinhole, the less favorable the effect, the more intense the light. Similarly, the smaller the pinhole, the more favorable the effect, and the less intense the light becomes. The low intensity in light can be compensated by a more powerful light source. This system s confocal multi-beam microscope adopts the light structure of the projector to measure different samples geometrical profile. Substituting pinhole with DMD product s multi-beam could be justified according to Lukosz s theory: reducing view to increase resolution. DMD can change projected pattern without any constraint. It can also divide projected space into to control the view size and make appropriate modifications in sample s resolution space when necessary. The system can be divided into three units. Below are the characteristic descriptions of each unit. 3.1 Light source projection The structure of projected light path is shown in Figure 4. Using UV-IR to filter the ultra violet and infra-red in white light (approximately nm)serves the purpose of avoiding possible damage in macromolecular module caused by ultra violet and heat produced by infra-red. First, the beam is focused on the green color filter. The green color filter then guides the beam into integrator rod. Inside the integrator rod, the beam is being reflected several times and transformed into square illuminants, each possessing the same intensity of light. Beam projected by integrator rod possessed of a wide view. This leads to the construction of relay system utilizing both non-spherical and spherical surface lens. Next, Relay Optics magnifies the beam through TIR (Total Internal Reflection) onto Proc. of SPIE Vol T-3

4 DMD s CMOS chip. After computer modifies the reflection of the array, appropriate patterns are projected. 3.2 Accurate displacements In this experiment, we utilized nano high-precision moving plate, as shown in Figure 5. Through the electric field, neighboring electric charges with different properties will move differently, rendering relative displacements and directly driv power. When employing nano high-precision displacement plate, moving stability can be established which permits the shifting of objects weighing 60kg. The system is programmed to shift 1µm at the time to perform depth measurement on samples. Capacitance sensor serves to determine whether shifted objects have the tendency to return to their original position, shown in Figure 6, is the Piezoelectric Transducer (PZT) block diagram. 3.3 Image selection During the analysis process, lens light intensity of depth position rendered through PZT is required. CCD camera is a high-velocity and high-resolution digital camera widely used in this field. It possesses 1.4 million pixels and its sensory CMOS chip s diagonal length is 11.1 mm. Its area possesses 1392 (H)*1040(V) pixels, wherein each pixel s size is as follows: 6.45 µm 6.45µm. Figure 7 shows the image of selected light point projecting onto the sample. 4. LOW PASS FILTER 4.1 Stray light s cause of formation and impact The existence of stray light is possible in many different optical systems. By analyzing the most probable cause of formation, the production of stray light in the system could be categorized into optical component surface s reflection, scatter formed on the component s surface, and stray light at the system s edge. Stray light affects the quality of imaging mainly because undesired light are formed after light rays are processed through multi-reflection inside the system. Two types of distribution can be found on the focus plane, divergent and convergent. Divergent distribution decreases the image s contrast while convergent distribution yields light spots on the image and affecting the image s quality. 4.2 Analysis of lowpass filter The purpose of the analysis of stray light is to prohibit any undesired light from entering the sensor, or at least lower its rate. To achieve these, light rays are processed through scattering surface to lower their power. Lowpass filter s main purpose is to let low-frequency light to pass through to avoid any impact made by stray light. Stray light s constraint results are then evaluated by PST (point source transmittance): PST ( θ ) i φ ( θi) ( ) c = (1) φ θ i φ ( θi ) represents the amount of light gathered at entrance pupil from angleθ i, c( i) φ θ represents the same Proc. of SPIE Vol T-4

5 light gathered by the sensor. This will be able to test the stray light s angular dependence. PST s purpose is not to identify any light source that causes stray light. Instead, it gathers data consisting of routes via which stray lights are produced and mechanisms that produce them to facilitate comparison of stray light constraint between different systems. Emulation mode uses optical software in different circumstances, which serves as a filter for stray lights. Using drawing software to build structure on the inner wall of the cylinder according to the standard structure processing specifications can enhance pathway for stray light. Plating method could increase absorption of stray light from different angels. Table 1 shows micro-arc oxidation characteristics. In optical software emulation, results could be rendered through the analysis of lights spots on the image s sensory surface, using PST. The amount of stray light could be determined by the distribution of power on the image s sensory surface and light rays projecting angle, using filters based upon two different kinds of structure. Figure 8 shows the basic analysis of stray light underneath simulation cylinder s inner wall. Experiment on practical measuring system s image contrast shall include systems with and without lowpass filters. Analyses on image contrast are shown in Figure 9 and Figure 10. These images show information rendered through the CCD camera. Results show that contrast of the system without lowpass filter is 0.11, and the contrast of the system with lowpass filter is Contrast differs by 60%. 5. CONFOCAL MULTIBEAM LIGHT PATH SYSTEM Light path system could be divided into three important units: light ray projection, precision plate, and image capture. Figure 3 shows the light path structure of confocal microscopy, images are rendered using light source projection device and DMD. DMD divided the projection space into to control the view. Band pass filter is used to determine light-wave s length needed for the light path. And stray light filter is used to constrain stray light and other unnecessary lights from the environment. Furthermore, light could be divided when passing through PBS due to a state of polarization. Afterwards, the light passes through λ/4 plate and turns into right-handed polarized light. Because the sample is not transparent in nature, the light signal is reflected by the focus surface. During the reflection, a potential difference of π/2 is rendered. This is the left-handed polarized light. After the light is reflected by λ/4 plate, P polarized light turns into S polarized light and reflected onto CCD camera. It will then control the vertical depth Z of a precision moving plate, and measure the image of the sample. At last, CCD camera receives the signal of light s intensity and measures the sample using confocal microscopy. 6. MEASUREMENTS Sample used in this proposed research is Ball Grid Array (BGA). The measurement for flip chip area is 32mm 32mm and BGA s height is around tens of micro meters. On confocal microscopy ray path, we used single light spot measuring method to project light pot pattern onto tin ball and substrate and recorded PZT s position and Proc. of SPIE Vol T-5

6 light spot intensity to render actual tin ball s height. Figure 11 shows the flow chart for single spot measurement. Substrate s highest intensity rendered is 16µm, tin ball s highest intensity rendered is 28µm, their difference is 12µm. According to the characteristics of confocal microscopy, project a single light pattern onto BGA and substrate, control precision displacement plate, shift 0.1µm at a time, and an image will be rendered. After rendering an image of tin ball s height according to the above steps, use calculation to render each image s largest gray value. Then using the method of interpolating, two Gauss curves distribution could be rendered. The comparison between the two curves light intensity s greatest value and displacement will then permit the calculation of tin ball s height, as shown in Figure CONCLUSION In designing light path, Digital Light Processor (DLP)is used to lower production cost. We modified DLP s original light path, results show that the contrast of system with lowpass filter is 0.68, image contrast increased by 60%. But because the measuring device was too large and employed white light at the source, coherence and intensity are not favorable, therefore unable to produce very good S/N ratio, shorter focus depth or increase resolution. However, there exist many measuring methods, calculations, and analyses which could be applied in confocal microscopy, therefore, still many topics to be researched upon. 8. REFERENCES [1] Minsky, M. Microscopy apparatus, US patent , [2] B. Herman, Fluorescence Microscopy 2nd, Bios Scientific Publishers, New York, pp. 15~38, [3] Yi-min Wang, Liang-an Wei, Fu-ren Gao, Confocal Microscopy Mischellanea, Physics bi-montly journal, 23 (2), pp , April, 2001 [4] C. J. R. Sheppard and D. M. Shotton, Confocal Laser Scanning Microscopy, Oxford, United Kingdom: BIOS Scientific Publishers, [5] G. Q. Xiao and C. Sheppard, Theory and Pratice of Scanning Optical Microscopy, Academic London, [6] S. Yin and G. Lu, J. Zhang, F. T. S. Yu, Joseph N. Mait, Kinoform-based Nipkow disk for a confocal microscope, Appl. Opt., Vol. 34, No. 25, [7] Jose-Angel Conchello and Jeff W.Lichtman, Theoretical analysis of a rotating-disk partially confocal scanning microscope, Appl. Opt., Vol. 33, No. 4, [8] Larry J. Hornbeck, Digital Light Processing and MEMS: Timely Convergence for a Bright Future, Texas Instrument Inc. [9] R. P. Brault, Control of Stray light, Handbook of optics, 2 nd ed., 5McGraw Hill, New York, Vol. 1, ch., [10] Ann. St. Clair Dinger, Stray light comparison of SIRTF designs with different aperture stop locations, Proc. SPIE, Vol. 1331, pp , Proc. of SPIE Vol T-6

7 [11] Marvin Minsky, Confocal Patent focal Scanning Microscope,US patent, serial number 3,013,467, TABLES AND FIGURES Eyepiece Illumination source Lens Beam Splitter Objective lens Specimen Figure 1: Traditional optical microscope imaging source aperture PBS lens CCD Figure 2: Theory of confocal microscopy [11] Proc. of SPIE Vol T-7

8 Project light source Computer Stray light filter Band pass filter Polarizer Stray light filter CCD camera PBS?/4 wave plate Objective lens?z Piezo actuator Figure 3: Confocal multi-beam microscopy system Color filter Relay optics Fold Mirror Integrator Rod Lamp UV-IR cut DMD TIR Prism Figure 4: Light source projection Figure 5: PZT controlled z-direction alignment plate Proc. of SPIE Vol T-8

9 Computer Serial Transmission Control Unit piezoelectric driver capacitance sensor PZT Specimen Sensor Y Z X Figure 6: PZT block diagram Figure 7: Image of projected light spots on sample (a) HH (b) Figure 8: Low pass filter simulation analysis (a) Unknown micro-structure s stray light distribution (b) Processed micro-structure s projected light distribution Proc. of SPIE Vol T-9

10 Figure 9: System image without lowpass filter Figure 10: System image with lowpass filter start DLP projects pattern onto sample s surface User designs desired projection PZT displacement, pass new image to CCD Set up PZT unit shift and total distance CCD selects unit s displacement image, light intensity Render surface height impact curve Highest surface point? NO YES Finish Figure 11: Flow chart of BGA height measurement curves Proc. of SPIE Vol T-10

11 Normalized Intensity ti sub Position (µm) Normalized Intensity Pmax= 16 Pmax= Position (µm) Figure 12: BGA height curve distribution (a) Record of each shift s light intensity (b) Using interpolating method to smooth the curves Table 1: MICRO-ARC OXIDATION CHARACTERISTICS Optical Coating of Property Material Type Absorption Coefficient α Degree of Reflection ρ Colour difference Brightness rating Emissivity ε KEPLA-COAT black for Al(8µm) KEPLA-COAT black for Ti(8µm) MAGOXID-COAT black(35µm) >95% 5% 29% 30% 65% >95% 5% 25% 15% 80% >95% 5% 24% 7% 81% Proc. of SPIE Vol T-11