利用奈米壓痕機量測BOAC(Bonding Pad Over Active Circuits)結構的機械強度和接合性

Abstract

本論文著重於建立一套利用奈米壓痕機量測BOAC銲墊下結構機械抵抗力的方法，以準確預測BOAC接合性及改善開發時效。本研究利用基材效應來判別不同BOAC結構的機械強度，並利用King’s 方程式回歸實驗數據以量化所有BOAC結構的機械強度。從實驗的結果，我們成功的利用基材效應計算出4種不同影響因素對BOAC機械強度的影響，分別為(1) 銲墊結構: 具高金屬密度及trench/via堆壘的傳統銲墊比BOAC強。(2) 銅導線密度: 最上金屬層的密度對整體BOAC強度的影響不大，表示最上層的鋁和oxide吸收了大部分的衝擊力量而保護了下面的結構免受壓痕的衝擊。(3) 線寬尺寸: 65 nm在銅/介電層中，銅的比例比90 nm的高，導致整體BOAC的強度較90 nm的強。(4) 低介電材料: 此因素對整體BOAC強度的影響最大，從實驗結果及斷面的觀察中可知其影響因素應為低彈性係數及低介面附著力所導致整體機械強度的弱化。總括而言，我們可以(1) 在銅/介電層中，增加銅的比例，(2) 使用彈性係數及介面附著力較佳的低介電材料，(3) 在BOAC結構的上層增加強度較佳的保護層以增強BOAC結構的機械強度。 本研究並使用兩種不同膜/基材的系統: (1) 鋁/均勻基材(軟膜/硬基材)和(2) oxide/均勻基材(硬膜/軟基材)來比較King’s方程式量化的結果在此兩種系統中之適用性與差異性。實驗結果發現不同BOAC結構機械強度的大小趨勢在此兩種系統中是相同的，但是受到基材硬度的影響，量化的值在鋁/均勻基材系統中是被高估，而在Oxide/均勻基材系統中是被低估。甚至，我們在Oxide/均勻基材系統中發現基材的影響範圍是會隨著壓痕的深度而持續往下的。 吾人利用P/S2這項函數將硬度的效應去除，反推BOAC結構真實的機械強度變化，發現BOAC基材並不是呈現一個單一的均勻基材，而是呈現出一個隨著壓痕深度持續改變的多層基材。利用此實驗方法所估算出oxide下基材的機械強度與利用複合材料理論計算出的值大小趨勢是一致的，且與King’s方程式的量化值相比，P/S2估算出的值與複合材料理論的值更為吻合。利用此方法更能了解BOAC結構機械強度的變化。總括以上的討論，我們成功的以簡化之King’s方程式與P/S2方法估算BOAC的機械強度，而在King’s方程式量化的值中，吾人推論其BOAC的複合彈性模數至少應具70 GPa方可確保通過接合性的測試。

The aim of this thesis was to establish a novel methodology using nanoindentation, which stimulated wire bonding and probing, to distinguish the mechanical resistance of various bonding pad-over-active-circuits (BOAC) structures by means of substrate effect. Such a quick turn-around methodology was also intended for predicting the bondability without a full array of reliability tests to reduce R&D cycle time. The mechanical stiffness of BOAC structures can be quantified using King’s model to fit the nanoindentation results assuming a uniform two-layered Al or oxide/substrate system. The parameters affecting the mechanical strength of BOAC such as bond pad types, copper density, line width/pitch (technology nodes) and low-k dielectric materials were investigated in this thesis to identify general design rules for BOAC layouts and structures. From nanoindentation results, normal pad was found to be much stronger than BOAC pad because normal pad had full stack of trench/via dummification. In addition, the top copper metal’s density had little influence on mechanical strength of BOAC structures because the top layers such as Al and oxide layer absorbed the majority of impact force; thus provided good protection for the structures underneath Al/oxide layers. When the metal line widths and pitches scaled from 90 nm to 65 nm process node, the BOAC became stronger because 65nm had higher copper fraction in Cu/low-k layer. The type of low-k materials was found to have great influence in the mechanical strength of BOAC structures. In summary, BOAC structures can be strengthened by (1) increasing aspect ratio (AR) in Cu/low-k layer, (2) using low-k materials with better modulus and interfacial adhesion, and (3) adding a stronger buffer layer such as oxide layer on the top of BOAC structures. In this study, two different film/substrate systems, i.e. (1) Al/composite substrate (soft film/hard substrate system) and (2) oxide/composite substrate (hard film/soft substrate system) were also examined to assess the applicability of fitting method using King’s model. The tendency of mechanical strength in different BOAC structures was found to be the same in these two different film/substrate systems. However, the quantified values were overestimated in Al/composite substrate, but underestimated in oxide/composite substrate due to the substrate hardness effect. Then, P/S2 term was used to eliminate the substrate hardness effect. The results truthfully showed a varying multilayered substrate with increasing indenter depth. This method could be used to analyze the changing mechanical strength of multilayer in BOAC structures as a function of indenter depth. The P/S2 results showed not only the same tendency of mechanical strength as that of theoretical calculation using the equation of composite materials, but also smaller deviation from theoretical calculation, as compared to values obtained from King’s model fitting. Overall, a novel methodology based on nanoindentation has been successfully established to distinguish the mechanical strength of various BOAC structures through a composite modulus values using a simplified film/uniform substrate model or P/S2 model. In addition, a modulus of 70 GPa in composite substrate of bond pad structure can be considered as a sufficient condition for new BOAC layouts passing the bondability tests.