| 摘要: | 物理氣相傳輸法(Physical vapor transport, PVT)為生長高品質大尺寸碳化矽單晶的主流製程，但該製程的長晶過程通常需要數十天，且長晶腔處於高溫低壓的密閉狀態難以透過實驗直接測量長晶腔內溫度、速度與濃度分佈，所以經由準確地數值模擬來了解其輸送現象是非常重要的，因此本研究發展合適模擬方法，透過數值模擬分析 PVT 長晶過程坩堝內溫度場、流場和反應氣體濃度的分佈，並對晶體進行長速(Growth rate)以及生長形狀的預測。本研究 透 過 改變感應線圈的位置與結構以及上溫度觀測點 (upper temperature)的溫度，來分析晶體生長初階段晶種(Seed)上的形貌以及長速，使系統達到最佳的長晶條件(Optimization)；接著將 PVT 法單晶成長過程變成多個穩態生長步階(Steady-state growth steps)，利用準穩態(Quasi-steady state)模擬碳化矽晶體在不同晶體厚度下坩堝內溫度場、流場和濃度分佈變化，並進一步對晶體形貌與長速進行預測，藉此導入長晶過程中可以降低晶體缺陷提高長速的優化製程技術。
模擬結果顯示晶體厚度增加時，晶體中心與邊緣皆有長速下降的趨勢，其
原因為腔體軸向溫度梯度與表面平均擴散係數降低，以及粉末表面物種分壓降低造成。而 PVT 製程上的改善依據為(1)調整線圈或改變坩堝構造來改善軸向溫度梯度或提高製程溫度以增加晶體的長速。(2)利用雙感應線圈降低晶體表面的徑向溫度梯度，使晶體中心與邊緣的長晶速率差距不要過大，避免晶體生長時在晶體內部產生熱應力。;Physical vapor transport (PVT) is the mainstream process for growing high quality large-size silicon carbide single crystals. However, the crystal growth process of this process usually takes several tens of days, Moreover, the crystal growth chamber is in a closed state of high temperature and low pressure, so it is difficult to directly measure the temperature, velocity and concentration distribution in the crystal growth chamber through experiments. Therefore, it is very important to understand the transport phenomenon through accurate numerical simulation. In this study, a suitable simulation method was developed, through numerical simulation to analyze the distribution of temperature field, flow field and reaction gas concentration in the crucible during PVT crystallization, and to predict the growth rate and the growth
shape of the crystal.
In this study, by changing the position and structure of the induction coil and the temperature of the upper temperature observation point, the morphology and growth rate of the seed crystal in the initial stage of crystal growth were analyzed, so that the system could achieve the best crystal growth conditions. Then, the PVT single crystal growth process is changed into multiple steady-state growth steps, and the quasi steady state is used to simulate the temperature field, flow field and concentration distribution changes in the crucible at different crystal thicknesses. Then, the crystal morphology and growth rate are predicted, so as to introduce an optimized process technology that can reduce crystal defects and increase the growth rate during the crystal growth process.
The simulation results show that when the thickness of the crystal increases, both center and edge of the crystal tend to reduce the growth rate. The reason is that the axial temperature gradient of the cavity and the average diffusion coefficient of the surface are reduced, and the partial pressure of species on the powder surface is reduced. The improvement in PVT process is based on (1) adjusting the coil or changing the crucible configuration to improve the axial temperature gradient or increase the process temperature to increase the growth rate of the crystal. (2) Use double induction coils to reduce the radial temperature gradient on the crystal surface,so that the difference in the growth rate between center and edge of the crystal is reduced to avoid thermal stress inside the crystal during crystal growth. |
