Key Engineering Materials Online: 005-10-15 ISSN: 166-9795, Vols. 95-96, pp 501-506 doi:10.408/www.scientific.net/kem.95-96.501 005 Trans Tech Publications, Switzerland A 3D Profile Parallel Detecting System Based on Differential Confocal Microscopy Y.H. Wang, X.F. Yu and Y.T. Fei School of Instrumentation, Hefei University of Technology, Hefei, 30009, China Keywords: Differential, Confocal Microscopy, Parallel Detecting, 3D Profile, Non-Scanning Abstract: Based on the differential confocal microscopy principle, a non-scanning 3D profile detecting system is proposed. A D light source array produced by micro-optic components is used to form a parallel micro-confocal system. The whole field detecting of the measuring plane is realized. The reflected light beam is divided into two paths. Two CCD cameras work together to detect the facula array in a differential arrangement. A differential algorithm of facula intensity is used. The noise and the shift of light source can be avoided effectively. Using the linearity characteristic of the differential confocal system, the high axial resolution is achieved at a larger sampling interval. The measuring efficiency and accuracy can be increased. The construction and working principle of the 3D parallel detecting system and the 3D profile reconstruction method are investigated. Experiment results agree with the ones of theory analysis. It indicates that the differential method is applicable for non-scanning 3D profile detection. Introduction Due to its special advantages of high resolution, high accuracy and its easiness to realize the digitization of 3D image, the confocal measurement has been applied in many fields, such as biomedicine, semiconductor and 3D surface topography detecting [1,]. However, traditional confocal microscopy methods cannot meet the demand of high-speed on line measuring due to their complex construction, low speed and vibration of scanning system. Therefore, a non-scanning 3D profile detecting probe is developed in recent years [3,4]. In this method, multiple-beam light parallel detecting is used instead of the single-point collimates and scans. It can simultaneously realize the whole-field detecting of measured surface. As an alternative to a pinhole diaphragm, the pixels on the CCD sensor directly capture and register the light intensities of the confocal point image. In the traditional parallel confocal detecting system, when the object shifts along the axial direction, the distances between the corresponding point and the focus plane are different because of difference of the heights of each measuring point. This distance can figure out the height of the profile. Although this method is convenient, the noise and shift of the light source will make the references of the micro-light-path different, thus will affect the measuring accuracy. To increase the measuring accuracy, a non-scanning 3D profile parallel detecting system based on differential confocal microscopy method is proposed. Working Principle of the Differential Confocal Microscopy In conventional confocal system light source, measuring point and pinhole detector are conjugated. When the object is located at the focal plane, the reflected light is accurately focused on the detector unit. The light intensity captured by the pinhole detector is the highest. When the object offsets from the focal plane, the reflected light is focused on the front or back position of detector unit. Then, the light intensity captured by detector is reduced because of the pinhole obstructing. Therefore, the axial displacement information can be obtained through the change of the light intensity. The axial light intensity responding to confocal system can be written as [5] All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of Trans Tech Publications, www.ttp.net. (ID: 130.03.136.75, Pennsylvania State University, University Park, USA-1/05/16,3:04:19)
50 Measurement Technology and Intelligent Instruments VI [ kz( 1 cosα )] ( 1 cosα ) sin I ( z) =, (1) kz where k is the wave number (k=π/λ, λ represents the wavelength of light wave), z is the offset from the focal plane, and sinα indicates the numerical aperture (NA) of the objective lens. However, the shift of the light source will cause the reference error of the light intensity. If used in the whole-field parallel measurement, the noise of light source will cause the light intensity variation of the different detecting point, thus bring the plane error. Focus plane 1 Detector 1 Detector Light source Beam-splitter Focus plane Object plane Beam-splitter 1 Objective lens Fig. 1 Schematic diagram of differential confocal microscopy The schematic diagram of differential confocal microscopy is shown in Fig. 1. Through the beam-splitter 1, the beam from the point light source is focused on the measured object by the object lens. The reflected beam from the object is divided into two ways by Beam-splitter, and detected by two detectors in a differential arrangement. When the object is located at the focal plane, the facula shape and light intensity captured by two detectors are the same, and the difference output between the signals of two detectors is zero. If the object offsets from the focal plane, the facula shape and light intensity will be different, and the differential signal of two detectors will not be zero. Then the axial shift information can be got through the change of the differential signal. When the offset of detector from the focal plane is ± Z, the axial response of the detector can be expressed as [6] Z sin k Z ± ( 1 cos α ) I ( Z ) ± = I. () 0 Z k Z ± ( 1 cos α ) When the detector offsets, the axial response curve has no change but an axial offset that produces a phase shift (shown in Fig. ). Obviously, the axial offset of the detector does not have much influence on axial response, which has been adopted in the differential focal method. The differential output of detector I D can be expressed as I = I( + Δ Z) I( Δ Z) D ΔZ ΔZ sin k Z + ( 1 cosα) sin k Z ( 1 cosα) = I ΔZ ΔZ k Z + ( 1 cosα) k Z ( 1 cosα) 0. (3)
Key Engineering Materials Vols. 95-96 503 Normalized light intensity 1 0-1 I(ΔZ) I D I(-ΔZ) ΔZ 0 -ΔZ Axial offset Fig. Normalized output curve of the detector The output curve consists of several linear ranges (shown in Fig. ). It also shows that normalized light intensity is zero in focal plane. The linearity range and sensitivity of output curve are related to the parameters of the optical system. Since the differential algorithm of facula intensity is used, the noise and the shift of light source act on two detectors simultaneously, so the differential output can effectively reduce the influence of light source. Another advantage of differential confocal method is to improve the detecting speed. In the conventional confocal measurement, the axial sampling interval must be decreased to improve the axial resolution. It will certainly slow the measuring speed. However, since there is satisfactory linear range in the output characteristic curve the measuring speed can be increased by the differential confocal method. As long as the sampling interval is within the linear range, the sample point height can be got from the detector differential output curve. Thus, we can get higher axial resolution than sampling interval. Non-Scanning 3D Profile Parallel Detecting System Light source Focus plane 1 CCD1 Beam-splitter Micro-lens-array Focus plane CCD Object plane HP559A Collimating lens D light source array plane Beam-splitter 1 Objective lens Axial stage Fig. 3 Structure of non-scanning 3D profile detecting system Figure 3 shows the structure of the non-scanning 3D profile detecting system. The D array of point light source is produced by the micro-lens-array. The micro-lens-array is used to separates and then focus the light beam, in order to increases the illumination efficiency. Through the beam-splitter 1, the
504 Measurement Technology and Intelligent Instruments VI beam from the point light source is focused on the measured object by the objective lens. Beam-splitter divides the reflected beam from the object into two ways. Two monochrome CCD detectors are arranged in the differential position to detect the light intensity of the differential defocusing facula array. The pixels of CCD detecting surface are divided into many detecting units according to the area of each facula. Each detecting unit corresponds to a facula. Here, each CCD detecting unit is equivalent to a pinhole detector. The CCD image is collected into the computer by the image-sampling card, and used to reconstruct the 3D profile. The displacement device is made up of the displacement table and the displacement driving circuit, which drives the workpiece to make axial movement. 3D Profile Reconstruction While the object moves along the axial direction, the distance between the corresponding point and the focal plane is different because the height of each object point is different. The above distance can figure out the profile height. Therefore, the obtained 3D profile can be transformed to get the sampling position where the output T is zero passage. During the measurement, point-light-array forms the parallel section n times. Suppose the distances between each section are all equal to ΔZ, which is determined by the Z-axial step of micro-displacement driver. The sampling matrix of sampling point (i, j) is expressed as [ Z,I ( i, j) ], [ Z,I ( i, j) ],..., [ Z,I ( i, j) ] D1 D n Dn 1, (4) where I D represents the output of CCD unit and Z n is Z-axial scanning height. The comparison between the outputs T of each section can be made. It shows that, if I D (i,j) is equal to 0, the object point height is calculated by Z k = kδz, (5) where k is the section where I D (i, j) = 0. If none of the output is zero passage, the height can be alternatively obtained through the linearity range of differential output curve. The linearity range and sensitivity of differential output curve are related to the parameters of optical system. We can choose the appropriate parameters of optical system based on the conclusions and the demand of the axial resolution. Experimental Results and Discussion HP559A CCD1 Beam-splitter CCD Light source Objective lens Beam-splitter 1 Micro-lens array Collimating lens Fig. 4 Hardware setup of the 3D profile detecting experiment system
Key Engineering Materials Vols. 95-96 505 In order to validate the feasibility of differential confocal method in the parallel detecting, we have done some primary works for the experiment system (shown in Fig. 4). The surface of a standard gauge was chosen as the object to detect the axial response of the differential facula array. Micro-lens array is 4 by 4 with 100µm dimension. An achromatic doublet objective lens is used. The area array monochrome CCD is used to capture the reflected facula image. The image acquisition card adopts METEORII card made in Canada Matrox Electronic Systems Ltd. A two-frequency laser interferometer (HP559A) is used to calibrate axial shift. The images of facula array, captured by the monochrome CCD in different location of the focal plane, are shown in Fig. 5. These CCD images indicate that the luminous intensity reflected by the detecting system is in good accordance with the characteristics of confocal measurement. In other words, when the object is located at the focal plane, the reflected light is focused on the CCD detecting unit accurately, and the light intensity captured by CCD is the highest. While being out of focus, the reflected light is focused on the front or the back of CCD detecting unit, causing the vague facula and the decreased light intensity captured by CCD. Thus, we can get the axial displacement according to the change of the light intensity. a) negative defocused b) focused c)positive defocused Fig. 5 CCD images of facula array 1.0 0.8 C D B Normalized light intensity 0.6 0.4 0. 0.0 0 5 50 75 100 15 150 175 00 5 50-0. Z/ Ì Z (µm) m -0.4-0.6-0.8 Fig. 6 Axial response curve of the facula The experiment result of the facula is shown in Fig. 6. Curve C and curve D are the axial response curves of the facula captured by two CCD respectively, and curve B is differential output between C and D. From the curves, we know that the experiment results are in agreement with theory analysis with better linearity in limited range. It demonstrates that effectiveness and applicability of differential confocal method for parallel detecting, and the linearity range and sensitivity of curve are
506 Measurement Technology and Intelligent Instruments VI related to the parameters of optical system and CCD offset. However, in the experiment, there is still some non-linearity error between differential output curve and theoretical curve because of optical aberration, CCD noise and CCD gain inequality. The uncertainty and resolution of the experiment system is micro order. If appropriate parameters of optical system, high-quality optical components and CCD are chosen, the higher accuracy and resolution will be obtained. Moreover, the differences of focal length between each micro lens result in different axial resolution of different transverse points, and cause certain error. However, the error is system error, which could be eliminated by calibration and the processing of subsequent software. Conclusions Compared with the conventional confocal system, the differential algorithm of facula intensity can effectively restrain the noise and the shift of light source. By using the linear characteristic of differential confocal system, the high axial resolution is achieved at a larger sampling interval. Furthermore, the method is beneficial to increase the measuring efficiency and accuracy. The linearity range and sensitivity of differential output curve are related to the parameters of optical system. We can choose the appropriate parameters of optical system based on the conclusion and the demand of the axial resolution to meet the requirements of some specified measurement. In high-accuracy measurement, optical aberration and CCD noise have great impact on the accuracy of detecting system. Therefore, further study is need such as how to reduce the effects of optical aberration and noise of CCD in practical application. Acknowledgments This research is sponsored by the National Natural Science Foundation of China (No. 5017504), the Important Research Foundation of National Education Ministry of China (No. 01103), and the Research Foundation of Hefei University of Technology (No. 030101F). References [1] T.M. Wilson: Confocal Microscopy (Academic Press, New York 1990). [] C.J.R. Sheppard and M. Gu: Journal of Microscopy, Vol. 165 (199), p. 377. [3] H.J. Tiziani and H.M. Uhde: Appl. Opt., Vol. 33 (1994), p. 567. [4] M. Ishihara and H. Sasaki: Opt. Eng., Vol. 38 (1994), p. 1035. [5] T.R. Corle and C.H. Chou: Optics Letters, Vol. 11 (1986), p. 770. [6] T.M. Wilson and A.R.Carlini: Appl. Opt., Vol. 7 (1988), p. 3791.
Measurement Technology and Intelligent Instruments VI 10.408/www.scientific.net/KEM.95-96 A 3D Profile Parallel Detecting System Based on Differential Confocal Microscopy 10.408/www.scientific.net/KEM.95-96.501 DOI References [5] T.R. Corle and C.H. Chou: Optics Letters, Vol. 11 (1986), p. 770. 10.1364/OL.11.000770