應用掃描式光電流顯微術於半導體奈米線之載子傳輸特性研究

Abstract

具空間解析能力之電流量測技術如掃瞄式光電流顯微術(scanning photocurrent microscopy)已被應用於在製作成完整元件之前的半導體結構，如奈米線中，探索本質的載子傳輸特性。此光電流量測技術為許多以奈米線為基礎元件，包含了光偵測器、太陽能元件、雷射、感測器及電晶體等，提供了寶貴的載子傳輸資訊。在本論文中，我們將利用掃瞄式光電流顯微術來探討半導體奈米線中載子的傳輸特性。 首先，我們利用數值模擬的方式在PN接面型(p-n junction-based)及歐姆接觸型奈米線元件中解互相耦合的帕松方程式(Poisson’s equation)與兩條連續方程式。過去在文獻中，往往都使用一個簡單的指數遞減函數來擷取載子擴散長度或是載子衰減長度。我們的數值模擬結果驗證了對於有著巨大內建電場的PN接面型(p-n junction-based)元件，確實能基於掃瞄式光電流輪廓僅受到載子擴散機制影響的假設，利用一個簡單的指數遞減函數由掃瞄式光電流輪廓中擷取出載子擴散長度。然而我們發現，在無巨大內建電場的歐姆接觸型元件中，該傳統擷取技術將不再有效。有別於文獻中過於簡化的模型，我們保留了激發光感應電場(ΔE)的影響，由帕松方程式及連續方程式出發，成功的推導了掃瞄式光電流輪廓的新解析公式，我們發現掃瞄式光電流輪廓與少數載子衰減長度間的關係可以用一個特殊的解析公式來表示。我們先以數值模擬驗證了在歐姆接觸型元件中所推導出解析公式於砷化銦及矽奈米線中的有效性，並且在許多不同材料參數與操作條件下探討解析公式的適用範圍。我們也實際以N型砷化銦奈米線來演示載子衰減長度的擷取技術，於不同激發強度及不同外加電場的實驗結果也進一步確認了解析公式的有效性。 我們也以數值模擬的方式探討了歐姆接觸型元件於強光激發下的掃瞄式光電流輪廓，指出了文獻中對於歐姆接觸型元件中掃瞄式光電流輪廓的錯誤理解，我們發現了在強光激發下的掃瞄式光電流輪廓雖然會呈現一個簡單的指數遞減趨勢，但由掃瞄式光電流輪廓擷取出的光電流衰減長度卻與載子實際的衰減長度相差很多，故我們認為歐姆接觸型元件並不適合在強光激發條件下擷取載子衰減長度。

Spatially resolved current measurements such as scanning photocurrent microscopy (SPCM) serve as important tools to explore the intrinsic carrier transport properties directly in semiconductor structures, such as nanowires, before they are put into device fabrications. These techniques provide valuable information for a wide range of nanowire-based devices including photodetectors, photovoltaics, lasers, sensors, and transistors. However, there is not always a rigorous theoretical model established for every important device category, for example, ohmic-contact devices. In this dissertation, the carrier transport in semiconductor nanowires is studied both theoretically and experimentally using scanning photocurrent microscopy. First, scanning photocurrent profiles in both p-n junction and ohmic-contact nanowire devices are theoretically studied by solving the coupled Poisson’s and two continuity equations numerically. In the literature, a simple-exponential-decay formula is always used to extract the carrier diffusion length or carrier decay length from scanning photocurrent profiles in nanowire devices. This conventional fitting formula based on the assumption of carrier diffusion dominance in the scanning photocurrent profiles was verified numerically in p-n junction devices where a large built-in electric field exists. However, the scanning photocurrent profiles in ohmic-contact devices cannot be appropriately fitted by the conventional fitting formula. Furthermore, the scanning photocurrent profile and the carrier spatial distribution strikingly do not share the same functional form in such devices. In order to extract the transport parameters from the scanning photocurrent profiles, an analytic formula is derived by solving the coupled Poisson’s and continuity equations. Unlike the over-simplified model in the literature, the influence of photo-carrier-induced electric field ΔE is included in our analytical model. A surprising new analytic relation between the scanning photocurrent profile and the minority carrier decay length was established. The effectiveness of the formula in nanowires with different mobility values under different pumping and bias conditions was thoroughly discussed. The analytic formula was applied in n-InAs ohmic-contact nanowire devices, and the experimental photocurrent profiles also confirmed the adequacy of the derived analytic formula. Electric-field dependence of the carrier decay length was directly observed. Then the mobility-lifetime product and the carrier diffusion length can be obtained. With an accompanying pump-probe carrier lifetime measurement, the minority carrier mobility was obtained. The photocurrent profiles of ohmic-contact nanowire devices in strong excitation regime were also investigated using numerical simulation, and the misinterpretation of the scanning photocurrent profiles in literature is indicated. For the analysis of experimental results in SPCM and EBIC techniques, a simplified model based on the assumption of zero-ΔE is always used throughout the literature. Based on the over-simplified model in the literature, simple exponential decay function is applied to extract the carrier transport parameters in nanowire devices. According to our numerical simulation, the scanning photocurrent profiles in strong excitation regime did resemble simple exponential decay functions which are similar to those reported experimentally in the SPCM literature. However, the fitted photocurrent decay length was found to be remarkably different from the fitted carrier decay length. Carrier decay length extraction from such photocurrent decay profiles under strong excitation can suffer large inaccuracy in ohmic-contact nanowire devices. Therefore, SPCM should be avoided to operate under strong excitation.