대면적 미세표면형상의 고정밀 광학식 측정

Abstract

In this study, surface data deformation of an optical measurement instrument by pixel size is quantified and precise measuring condition to choose the magnification of the objective lens is presented. For large area measurement with high precision, a 6-degree of freedom (d.o.f.) stitching method is proposed and a large areal data is possibly acquired within short time. Also, a synchronization method to match two different data sets with different resolutions which have an overlap area is proposed. The developed measuring condition, stitching method, and synchronization method are demonstrated by a multi-probe measurement system to verify their feasibilities. To quantify the distortion effect by pixel size, the 3D data distortion is newly modeled. A simulation test with the 3D distortion model was implemented for various soft gauges which had been acquired by an AMF and spectral analysis for the distortion results was done. Based on the analysis results, the effects of pixel size, the roughness of the surface on the distortion are specified so that magnification number of an objective lens can be chosen efficiently. To stitch adjacent data sets, a new stitching method using global least square method with correlation function to minimize the 6-d.o.f. position error is proposed. The stitching method is evaluated to have about 9 nm maximum error and ± 3.8 nm repeatability. A synchronization algorithm to minimize the 6-degree of freedom relative position error between two different data sets acquired by different resolutions was developed. Synchronizing the two different data sets acquired by a chromatic confocal which has low resolution and high measuring speed and a CLSM which has high resolution enables the efficiency of measurement and evaluation. The developed measuring condition, 6-d.o.f. stitching method, and synchronization method were applied to a multi-probe optical measurement system. In the roughness evaluation Rq of a milled surface, the error was 33.8 % by evaluation area suggested in ISO but the new method presented in this paper showed only 8.9 % error. The 6-d.o.f. stitching method was demonstrated with a LCD panel, a PDP panel, a bump wafer, a micro lens array, a Fresnel lens, an aspheric lens, and a model ship. In summary, a combination of the developed measurement condition, the 6-d.o.f. stitching method and synchronization method is an effective method to measure a large surface with high resolution and its feasibility for various measurement instruments was verified.|본 논문에서는, 광학식 측정기의 이미지 센서 내 화소의 크기로 인해 표면형상 측정데이터가 왜곡되는 현상을 정량화하고, 이를 기반으로 광학식 측정기의 정밀 측정조건을 제안하였다. 그리고 대면적 표면형상 측정을 위해 측정데이터의 6자유도 정합 방법을 제안하여, 빠른 속도로 누적오차 없이 대면적의 측정데이터를 획득할 수 있었다. 또한, 동일위치에 대해서 서로 다른 측정기 또는 분해능으로 측정된 데이터 사이의 기하학적 오차를 최소화할 수 있는 공간동기화 방법을 개발하여 효율적인 대면적 정밀 측정을 가능하게 하였다. 개발된 측정조건, 데이터 정합 및 공간동기화 방법을 멀티프로브 측정기에 적용하여 다양한 시편에 대해 대면적 정밀 측정이 가능함을 확인하였다. 광학식 측정기의 화소에 의한 데이터 왜곡 현상을 정량화하기 위해, 광학식 측정기의 3차원 측정모델을 설정하고, 이 측정모델에 적용될 피측정면 정보를 AFM(atomic force microscopy)을 통해 획득하였다. 획득된 피측정면과 측정모델을 기반으로 측정 모사실험을 수행하여 다양한 측정조건에서의 측정데이터를 수집하여, 이들을 스펙트럼 분석을 수행하였다. 분석된 결과를 기반으로 화소크기, 표면거칠기와 측정데이터의 왜곡정도 사이의 관계를 분명히하여 광학식 측정기의 대물렌즈 측정조건을 제안하였다. 또한, 3차원 형상 측정데이터에 대한 6자유도 정합 알고리즘을 개발하여, 고분해능 대면적 측정데이터를 용이하게 획득할 수 있도록 하였다. 누적오차를 최소화할 수 있는 글로벌 최소 제곱법과 상호상관함수를 이용하여 6자유도의 기하학적 오차를 최소화할 수 있는 정합 알고리즘을 제안하였다. 측정데이터와 모사실험을 기반으로 제안된 정합 알고리즘을 평가한 결과, 최대 오차는 9 nm, 반복정밀도는 ± 3.8 nm 정도로 광학식 측정기의 종방향 분해능보다 작은 오차가 발생하는 것을 확인하였다. 서로 다른 분해능을 가진 측정기를 이용하여 측정된 데이터들 사이의 기하학적 오차를 최소화할 수 있는 공간동기화 알고리즘을 개발하였다. 분해능은 낮지만 고속으로 대면적 측정이 용이한 색수차 현미경과 고분해능의 공초점 현미경에 의해 측정된 데이터를 공간적으로 동기화함으로써, 측정 및 검사의 효율성을 증대하였다. 개발된 측정조건, 6자유도 정합방법 및 공간동기화 방법을 멀티프로브 기반의 광학식 측정기에 적용하였다. 표면거칠기 Rq의 경우, ISO에서 제안하는 평가영역(evaluation area)를 적용하여, 밀링면을 측정/평가할 경우 오차가 33.8%였으나, 본 연구에서 제안한 측정방법을 적용하면 8.6 %의 오차를 보였다. LCD, PDP, Bump wafer, 마이크로 렌즈 어레이, 프레넬 렌즈, 비구면 렌즈, 모형선 등에 6자유도 정합 방법을 적용하였다. 제안된 측정조건, 6자유도 데이터 정합, 공간동기화 방법은 광학식 측정기를 이용한 대면적 고정밀 측정에 효과적인 방법이며, 다양한 측정기에 적용 가능한 방법임이 제시되었다.; In this study, surface data deformation of an optical measurement instrument by pixel size is quantified and precise measuring condition to choose the magnification of the objective lens is presented. For large area measurement with high precision, a 6-degree of freedom (d.o.f.) stitching method is proposed and a large areal data is possibly acquired within short time. Also, a synchronization method to match two different data sets with different resolutions which have an overlap area is proposed. The developed measuring condition, stitching method, and synchronization method are demonstrated by a multi-probe measurement system to verify their feasibilities. To quantify the distortion effect by pixel size, the 3D data distortion is newly modeled. A simulation test with the 3D distortion model was implemented for various soft gauges which had been acquired by an AMF and spectral analysis for the distortion results was done. Based on the analysis results, the effects of pixel size, the roughness of the surface on the distortion are specified so that magnification number of an objective lens can be chosen efficiently. To stitch adjacent data sets, a new stitching method using global least square method with correlation function to minimize the 6-d.o.f. position error is proposed. The stitching method is evaluated to have about 9 nm maximum error and ± 3.8 nm repeatability. A synchronization algorithm to minimize the 6-degree of freedom relative position error between two different data sets acquired by different resolutions was developed. Synchronizing the two different data sets acquired by a chromatic confocal which has low resolution and high measuring speed and a CLSM which has high resolution enables the efficiency of measurement and evaluation. The developed measuring condition, 6-d.o.f. stitching method, and synchronization method were applied to a multi-probe optical measurement system. In the roughness evaluation Rq of a milled surface, the error was 33.8 % by evaluation area suggested in ISO but the new method presented in this paper showed only 8.9 % error. The 6-d.o.f. stitching method was demonstrated with a LCD panel, a PDP panel, a bump wafer, a micro lens array, a Fresnel lens, an aspheric lens, and a model ship. In summary, a combination of the developed measurement condition, the 6-d.o.f. stitching method and synchronization method is an effective method to measure a large surface with high resolution and its feasibility for various measurement instruments was verified.