Comparison of EUV Photomask Metrology Between CD-AFM and TEM

Accurate metrology of extreme ultraviolet (EUV) photomask is a crucial task. In this paper, two different methods for reference EUV photomask metrology are compared. One is the critical dimension atomic force microscopy (CD-AFM). In the measurements, the contribution of its AFM tip geometry is usually the dominant error source, as measured AFM images are the dilated results of measured structures by the AFM tip geometry. To solve this problem, a bottom-up approach has been applied in calibrating the (effective) AFM tip geometry where the result is traceably calibrated to the lattice constant of silicon crystals. The other is transmission electron microscopy (TEM). For achieving measurement traceability, structure features are measured in pairs in TEM images;thus the distance between the structure pair calibrated by a metrological AFM in prior can be applied to determine the magnification of the TEM image. In this study, selected photomask structures are calibrated by the CD-AFM, and then sample prepared and measured by high-resolution TEM nearly at the same location. The results are then compared. Of six feature groups compared, the results agree well within the measurement uncertainty, indicating excellent performance of the developed methodology. This research supports the development of a photomask standard, which is applied as a “reference ruler” with improved low measurement uncertainty in photomask fabs.

Correction of Interferometric High-Order Nonlinearity Error in Metrological Atomic Force Microscopy

Metrological atomic force microscopes(Met.AFMs)with built-in interferometers are one of the main workhorses for versatile dimensional nanometrology.The interferometric nonlinearity error,particularly the high-order(i.e.,3rd-and 4th-order)nonlinearity errors,is a dominant error source for further improving their metrology performance,which cannot be corrected using the conventional Heydemann correction method.To solve this problem,two new methods were developed.One uses a capacitive sensor embedded in the Met.AFM,and the other applies an external physical artifact with a flat surface.Both methods can be applied very conveniently and can effectively reduce the nonlinearity error.In this paper,the propagation of the(residual)nonlinearity error in step height calibrations is examined.Finally,the performance of the improved tool is verified in the calibration of a highly demanding industrial sample.For the measurements performed at 25 different positions and repeated six times,the standard deviation of the total 150 measured values is 0.08 nm,which includes the contributions from the reproducibility of the metrology tool and sample inhomogeneity.This research has significantly improved our dimensional nanometrology service.For instance,the extended measurement uncertainty(k=2)is reduced from 1.0 to 0.3 nm for the step height or etching depth calibrations.