Measuring Temporal Variation of Concentration of ALD Precursor TEMAZ in Process Chamber Using UV Absorption Method

Abstract

Atomic layer deposition (ALD) is becoming increasingly important to realize high-accurate atomic-scale deposition processes in semiconductor industry, where device dimensions approach atomic scale with the advances of the miniaturization. Basically, ALD is a process of forming a film using self-limited cyclic reactions, where precursor gas, purging gas, reaction gas, and purging gas are introduced sequentially [1]. To realize high-quality ALD with high-throughput, the reaction cycle should be shortened. However, this may cause incomplete gas replacement due to residual gas, resulting in the degradation in process quality. To develop high-quality ALD, precise control of temporal and spatial distribution of these gases is required. Although the total gas pressure in the chamber was usually measured as process monitor, partial pressure, or concentration of precursor as well as reaction gas was not be measured so far. This makes it impossible to evaluate gas replacement characteristics in the chamber, despite its importance for developing ALD technology.Our group developed a high-sensitivity, real-time, compact gas concentration sensor using UV light absorption method for the supply system of organometallic gases having UV absorption [2]. The gas concentration can be estimated from the UV absorption using Lambert-Beer law. Recently, to develop a new process monitoring tool in ALD, we apply this technique to in-situ measurement of precursor gas in the chamber. In this study, temporal variation of concentration of Tetrakis(ethylmethylamino)zirconium (TEMAZ) in cyclic introduction of TEMAZ and purging Ar gas was evaluated. Here, TEMAZ is widely used as a Zr precursor in ALD for zirconia films deposition, mainly in DRAM fabrication [3].Fig. 1 shows a schematic diagram of the experimental system used in this study. TEMAZ was supplied by a bubbling method. At the downstream of the TEMAZ tank, upstream pressure controller (UPCUS) was equipped to control the pressure in the tank to be constant. An inner diameter of a test chamber was 200 mm, and TEMAZ and purging Ar were supplied using a shower plate installed at the upside of the chamber. A densitometer using UV absorption method consisted of an UV light emitting part and an UV photometer, and these were placed at opposite side of the chamber. UV light with a wavelength of 250 nm, which corresponds to the absorption of TEMAZ, was used. Two densitometers (hereafter denoted as UVH and UVL) were set at 10mm and 60mm below the shower plate, respectively. A pressure gauge was also set at 10mm below the shower plate to measure total pressure in the chamber. In the measurement, the valve operation and data acquisition were conducted by a computer with a sampling period of 10 ms.Fig. 2(a) shows time evolution of chamber pressure and absorbance of UVH (which approximately proportional to TEMAZ partial pressure) in the case that TEMAZ and purging Ar were introduced alternatively every second for 50 times. TEMAZ was supplied to the chamber with diluted Ar (100 sccm) with a TEMAZ concentration of 2000 ppm. Purging Ar flow rate was 1500 sccm. In Fig. 2(b), extended graph of Fig. 2(a) for time range from 90 to 96 s were shown. It is clearly found that behavior of UVH signal and the chamber pressure was different, where UVH absorbance signal increased at TEMAZ introduction, and rapidly decreased when Ar purge started whereas the chamber pressure increased due to the Ar introduction. Such behavior cannot be obtained only from the pressure measurement.Fig. 3 compares UVH and UVL absorbance signals for 1 cycle of TEMAZ and purging Ar introductions. It can be seen that UVH absorbance began to increase at approximately 80.2 s (0.2 s after the TEMAZ introduction), and UHL absorbance began to increase with a delay at approximately 80.6 s (0.6 s after the TEMAZ introduction). This suggested that down flow of TEMAZ could be observed.The results demonstrated validity of the UV absorption method to evaluate temporal and spatial distribution of precursor gas, and this technology will greatly contribute to developing future ALD technology.[1] R. W. Johnson et al., Materials today, 17 (2014) 236.[2] H. Ishii et al., Jpn. J. Appl. Phys., 58 (2019) SBBL04.[3] S. K. Kim and C. S. Hwang, Electrochem. Solid-State Lett., 11 (2008) G9. Figure 1