Crystallinity Evaluation of Low Temperature Polycrystalline Silicon Thin Film Using UV/Visible Raman Spectroscopy

Abstract

1. Background and purpose Low temperature polycrystalline silicon (LTPS) has been developed as a material for thin film transistors (TFTs), because it has higher electron mobility than hydrogenated amorphous silicon (a-Si:H). To achieve high performance and reliability, superior crystallinity such as large grain size and low defect density are desired for poly-Si thin film. Thus, the nondestructive evaluation technique is indispensable to optimize film formation and crystallization conditions. However, conventional evaluation techniques such as scanning electron microscope (SEM) and X-ray diffraction (XRD) and so on, have poor sensitivity for the poly-Si crystallinity [1]. In this study, we evaluated crystallinity of LTPS fabricated by various processes using Raman spectroscopy. 2. Experimental method The 100 nm thick SiO­2 film was grown on Si as substrates. Table I shows a-Si deposition conditions. The a-Si layers were deposited at 510, 530 and 550oC by low pressure chemical vapor deposition (LPCVD) using the mono-silane (SiH4) gases under the pressure of 0.4 or 1.5 Torr. The a-Si layers on SiO2/Si substrate were then annealed at 625oC for 20, 40, 60, 90 and 180 minutes. A quasi-line excitation (with length of approximately 100 μm) was used for Raman spectroscopy. Excitation sources were visible (λ = 532 nm) and UV (λ = 355 nm) laser. The focal length of the Raman spectrometer was 2,000 mm, which provides wavenumber resolution of approximately 0.1 cm-1. One dimensional measurements with a quasi-line excitation source were performed at the same time [2]. Thus, 512 points Raman spectra for a sample were obtained simultaneously. For the crystallized samples, we evaluated the crystallinity in conjunction with the crystallization processes by Raman spectroscopy. 3. Results and Discussion Figure 1(a)-(d) show Raman images obtained by CCD with UV excitation. Here, the penetration depth is approximately 5 nm. The Raman peak for the single crystalline-Si (c-Si) should be at 520 cm-1. The images display 2-dimensional intensity profiles with the x-axis corresponding to the Raman shift wavenumber and the y-axis corresponding to the 100-mm-long 1-dimensional space mapping on the sample. We confirmed that there are different crystallinities depending on the fabrication processes. From Fig. 1(a)-(c), it is clearly recognized the crystallization process and its variation. As increasing the annealing duration and deposition temperature of a-Si, the clear bright line appeared at 520 cm-1 implying good crystallinity. It is also apparent that the a-Si deposition at 1.5 Torr resulted in the slower crystallization than that deposited at 0.4 Torr. Figure 2 statistically summarizes the median and variation (2σ) of the full-width-at-half-maximum (FWHM) extracted from the Raman spectra obtained in Fig. 1. It is well known that the FWHM of Raman spectra represents the crystallinity very well [3], the FWHM of Cz grown single Si crystal measured by the present system is approximately 3.0 cm-1. Therefore, we can conclude the crystallinity, i.e. grain size and the defects in the grains, was improved by higher temperature or lower pressure for the a-Si deposition. In conclusion, we have evaluated crystallization process and crystallinity in LTPS by Raman spectroscopy. We believe that Raman spectroscopy is useful evaluation methods for in-line process by adopting Raman imaging technique. Acknowledgements The authors thank to Dr. N. Sawamoto for her support in TEM observation.