銅化學氣相沉積及其應用在深次微米銅栓之研究

Abstract

本論文以自行組建之多腔體式低壓銅化學氣相沉積系統，使用Cu(hfac)TMVS + 2.4 wt% TMVS當作先驅物(precursor)，研究銅化學氣相沉積技術，內容包括銅化學氣相沉積的基本特性探討、銅膜在不同基板(TiN、Ta和TaN)之沉積特性探討、銅膜的熱應力行為、基板的電漿前處理效應、以及管洞填充(via filling，銅栓)技術；再者，本論文亦探討非晶態TaSixNy薄膜的擋銅阻障特性。 首先，本研究針對銅膜沉積所需之各種功能及需求考量，自行設計並組建一部多腔體、溫壁式、具基板in-situ電漿前處理及自動化控制功能的低壓銅化學氣相沉積系統。本設備含有四個腔體(置入室、傳輸室、前處理室及銅膜沉積室)，並預留擴充空間可加掛三個其他用途的腔體。本設備的銅源傳輸採用液態銅源直接注入式(direct liquid injection，簡稱DLI)，可提高銅膜的沉積速率及沉積可控性和穩定性。前處理室可供基板作表面處理後，不破真空in-situ的進行銅膜沉積。在監控系統方面，本設備採用具獨立性的插槽(slot)式模組化設計，在拆裝與系統擴充方面都具有方便性。另外，本設備亦可透過網路而具有遠端監控之功能。 在銅化學氣相沉積的基本特性探討方面，本設備可在基板溫度 200°C及液態銅源流量為0.4 ml/min的條件下，獲得100 nm/min以上的銅膜沉積率，符合產業生產之需求。在沉積壓力為150 mTorr及基板溫度為160°C (或高於160°C)的條件下，沉積在4吋直徑之Ta基板上的銅膜之膜厚均勻性在5%以內。銅膜之晶粒大小，依沉積時基板溫度的升高而變大。銅膜沉積在不同基板(TiN、Ta和TaN)的效應方面，吾人發現沉積在TiN基板的銅膜具有較佳的薄膜特性，包括最低的電阻率(1.90 μΩ-cm at 160°C)、最低的雜質含量及最高的Cu(111)結晶取向。此外，在TiN基板的銅膜沉積亦具有最短的成核潛伏期 (incubation time)及較佳的銅膜附著力。因此，就銅膜的物性而言，TiN為化學氣相沉積銅膜的最佳沉積基板。 沉積在Si和TaN基板上的銅膜顯示不同的熱應力(thermal stress)行為。銅膜分別沉積在Si和TaN/Si兩種基板後，作室溫至400°C的來回真空熱循環處理，吾人發現沉積在Si基板的銅膜具有較小的熱應力變化，但是，沉積在TaN基板的銅膜在室溫時具有較低的張應力(tensile stress)。此外，厚度較薄的銅膜具有較高的張應力及較大的熱應力變化。與濺鍍銅膜的應力行為比較，化學氣相沉積銅膜具有較小的熱應力變化，熱循環處理後的室溫殘留張應力也比較小。 本論文研究亦探討基板之電漿前處理效應，在銅膜沉積之前，以氬氣(Ar)、氫氣(H2)、或氬氣加氫氣(Ar+H2) (先以氬氣電漿處理再作氫氣電漿處理)分別對TiN和TaN 基板作電漿處理。在Ar電漿處理過的TiN和TaN基板上，銅之成核數大量增加，並具有較大的沉積率，而且所沉積之銅膜具有較大的Cu(111)/Cu(200) 晶向比、較小但形狀較有規則的晶粒、以及表面較平整等特性。另外，在H2及Ar+H2電漿處理過的TaN基板上所沉積的銅膜具有較低的雜質含量。然而，在電漿處理過的基板上所沉積的銅膜電阻率變大，推測係由於晶粒較小的緣故。在Ar、H2和Ar+H2三種電漿基板前處理中，就以沉積在Ar+H2電漿處理過的TaN基板上之銅膜具有最佳特性，包括最低的電阻率及最高的Cu(111)/Cu(200) 晶向比。銅膜沉積後，若再加以400°C/30分鐘的氮氣中熱處理，則可進一步提升銅膜的(111)結晶取向，並且降低銅膜的電阻率；例如，沉積在Ar+H2電漿處理過的TaN基板上的銅膜之Cu(111)/Cu(200) 晶向比從4.90提升為5.56，而電阻率則從2.35下降至2.06 μm-cm。 銅栓的製作涉及微小管洞(via)的銅金屬填充。欲對微小尺寸之高深寬比(aspect ratio)管洞達成無空隙(void free)之銅金屬填充，必需使Cu(hfac)TMVS在基板表面的黏著係數(sticking coefficient)減小，再發射率(re-emission frequency)增大。欲達到此一有利管洞填充的條件，銅膜沉積時的壓力和溫度都必需降低，但銅源(precursor)流量必需增加。對於深寬比AR = 9.1的0.11 μm管洞而言，在沉積壓力60 mTorr和銅源流量0.4 ml/min的條件下，只要基板溫度在160°C以下，都可達到無空隙的銅金屬管洞填充。此外，吾人發現使用氦氣(He)當作銅沉積的載氣(carrier gas)比使用氫氣(H2)當作載氣具有較好的填洞能力；這是因為以氫氣為載氣時，在銅沉積的化學反應中，因自身的氧化還原反應及氫氣還原反應同時進行而增加銅粒子的黏著係數，因而減低銅粒子的再發射率，導致無空隙填洞能力的降低。 最後本論文探討TaSixNy的抗銅阻障特性。TaSixNy薄膜係以反應濺鍍法，用TaSi2當作濺鍍靶，在Ar/N2混合氣中濺鍍所得之非晶態TaSixNy(x=1.4，y＝2.5) 薄膜(成份係由RBS鑑定所得)。吾人以各種材料分析及電性量測對TaSixNy阻障層的抗銅阻障溫度加以評估，其中電性量測是最靈敏、最嚴格的測試法；以Cu/TaSixNy/p+n接面二極體的電性量測結果顯示，5 nm厚度的TaSixNy阻障層可以有效阻擋銅金屬的擴散至400°C (以30分鐘的氮氣熱處理為準)而不致使二極體的電性發生劣化，而10、20、40 nm厚度的TaSixNy阻障層則可以分別有效阻擋銅金屬的擴散至500、550和650°C。再者，吾人分別以氮氣電漿處理(N2 plasma treatment)、氮氣熱處理(N2 thermal annealing)及氮氣快速回火處理(N2 RTA)等三種方式對TaSixNy阻障層作濺鍍後ex-situ處理，發現TaSixNy阻障層的抗銅阻障能力都有所提升。從最小熱預算(thermal budget)以及阻障溫度提升的幅度加以綜合考量，以快速回火處理的效果最佳。例如，以300°C/60秒之快速回火處理的10和20 nm厚度之 TaSixNy阻障層的有效抗銅阻障溫度分別提升至600和650°C。

This thesis study includes the design and construction of a multi-chamber copper chemical vapor deposition (Cu CVD) system, the basic characterization of Cu CVD and its variation on different underlayer substrates, the stress-temperature properties of CVD Cu film, the effects of substrate plasma treatment on Cu CVD, and the study on deep sub-micron via-filling by CVD Cu film (i.e. Cu-plug). Moreover, the thesis also includes the study on the thermal stability of reactively sputtered amorphous TaSixNy (x=1.4, y=2.5) films serving as a diffusion barrier between Cu and Si substrate. The first stage of this thesis study is to build the multi-chamber Cu CVD system used in this work because there was no commercial and inexpensive equipment available. This multi-chamber Cu CVD apparatus includes an in-situ substrate plasma pre-treatment chamber, a precursor direct liquid injection (DLI) system, and a fully automatic remote monitor and controller system. Copper chemical vapor deposition (Cu CVD) using a liquid metalorganic compound of Cu(hfac)TMVS with 2.4 wt% TMVS additive as the Cu precursor, was investigated with respect to its basic characteristics. It is found that the effective deposition rate of CVD Cu film about 100 nm/min can be obtained at a deposition temperature of 200°C when the precursor flow rate of 0.4 ml/min is used. Good uniformity (less than 5%) of film thickness can be obtained at a deposition pressure of 150 mTorr and a deposition temperature of 160°C or above. The grain size of the Cu film increases with the deposition temperature, being 0.32 μm at 140°C and 0.50 μm at 200°C. Cu CVD on different substrates, including TiN, Ta, and TaN, was studied with regard to the physical properties, nucleation, and adhesion of the deposited Cu films. The Cu films deposited on TiN substrate have a number of favorable properties over the films deposited on Ta and TaN substrates. These include lower electrical resistivity (~ 1.90 μΩ-cm), lower impurity contamination, higher (111)-preferred orientation and better film-adhesion. Moreover, CVD of Cu films on TiN substrate has a shorter incubation time. On the stress-temperature behavior of CVD-Cu film, it is found that the Cu film deposited on Si substrate has a smaller thermal stress variation than that deposited on TaN substrate for a thermal cycling process between room temperature and 400°C. On the other hand, the Cu film deposited on TaN substrate has a smaller residual tensile stress at room temperature. However, the CVD-Cu films, either on Si or on TaN substrate, exhibit smaller thermal stress variation and smaller residual tensile stress at room temperature than the corresponding PVD-Cu films. As for the effect of film thickness, it is found that the thinner Cu film (on Si substrate) has a higher tensile stress and a larger thermal stress variation than the thicker Cu film. The effects of plasma substrate pretreatment on the CVD-Cu film were investigated. The Cu films deposited on the Ar-, H2- or Ar+H2-plasma-treated TaN substrate have a number of favorable properties over the films deposited on the substrate without the plasma treatment. These include a smoother film surface, regular arrangement of Cu grains, enhanced (111) preferential orientation, and larger effective deposition rate. In addition, lower impurity contamination was obtained for the Cu films deposited on the H2- and Ar+H2-plasma-treated TaN substrates. However, the Cu films deposited on the plasma-treated substrate have a higher electrical resistivity, presumably due to smaller grain size and thus higher grain boundary density. Among the three plasma treating schemes, the Ar+H2 plasma treatment produced the most favorable effect. Post-deposition thermal annealing resulted in the decrease of film resistivity and the increase of Cu(111)/Cu(200) reflection peak ratio. A combined process including an Ar+H2 plasma substrate treatment prior to the Cu film deposition and a post-deposition thermal annealing at an appropriate temperature (e.g. 400°C) in N2 ambient, is believed to be favorable for the low-resistivity and high (111)-oriented Cu film deposition. The Cu CVD is a superior method for fabricating miniature size Cu-plug. A superior void-free via-filling of Cu film can be achieved at low temperature, low pressure and high concentration of precursor species in the gas phase of Cu CVD. However, these favorable conditions for void-free via-filling of Cu film also lead to a somewhat inferior film property, including higher film resistivity, higher content of impurities, degraded adhesion to substrate, and lower deposition rate. Thus, a trade-off is needed between the optimum deposition condition for via-filling and the optimum deposition condition for a superior film property. From the viewpoint of void-free via-filling of Cu film, helium (He) is a preferable carrier gas. In this thesis study, we have achieved the void-free filling of CVD Cu film on deep sub-micron 0.11-μm-diameter vias with an aspect ratio of 9.1 at and below the deposition temperature of 160°C, using a precursor flow rate of 0.4 ml/min and at a deposition pressure of 60 mTorr. Finally, the thermal stability of the reactively sputtered amorphous TaSixNy (x=1.4, y=2.5) films serving as a diffusion barrier between Cu and Si substrate was investigated. The Cu/TaSixNy/p+n junction diodes with a 5-nm-thick TaSixNy barrier are able to sustain a 30 min thermal annealing at temperatures up to 400□C without degrading the device’s electrical characteristics. With thicker barriers of 10-, 20- and 40-nm-thick TaSixNy layers, the thermal stability temperatures of the Cu/TaSixNy/p+n junction diodes are raised to 500, 550 and 650□C, respectively. With N2 plasma treatment on the TaSixNy barrier layer, the thermal stability temperatures of the Cu/TaSixNy/p+n junction diodes with 10- and 20-nm-thick TaSixNy barrier can be increased to 550 and 600□C, respectively. With N2 thermal annealing (at 300 to 500°C for 15 min) or N2 RTA annealing (300°C/60 sec) on the TaSixNy barrier layer, the thermal stability temperatures of the Cu/TaSixNy/p+n junction diodes with 10- and 20-nm-thick TaSixNy barrier are further raised to 600 and 650□C, respectively. From the viewpoint of minimal thermal budge and less impurity contamination, the 300°C/60sec N2 RTA treatment is believed to be a superior scheme for the improvement of the barrier capability of the amorphous TaSixNy barrier layer.