| 摘要: | 本論文的研究主軸為塑膠球矩陣構裝中銲點金濃度對其剪力強度之影響。塑膠球矩陣構裝基板為目前系統晶片組及繪圖晶片組所使用的構裝基板。本研究中構裝基板之墊層為電鍍0.4 μm Au / 8 μm Ni表面處理的型式，墊層直徑為430 μm ，而錫球銲點的組成為目前業界所使用的63Sn-37Pb( wt.％)。迴銲參數之迴銲頂溫為225℃，而迴銲時間為90秒。本研究藉著銲點剪力強度的量測來評估銲點的可靠度。實驗內容分為兩部分：一為銲點於160℃固態熱處理後之剪力強度量測，錫球銲點於迴銲前的尺寸分別為直徑750 μm 、600 μm 及450 μm ，而迴銲後之銲點中金濃度分別為0.07 wt.％、0.13 wt.％及0.31 wt.％。二為銲點於重複迴銲/熱處理( RA500h160℃RA250h160℃RA250h160℃R )流程中之剪力強度量測，錫球銲點於迴銲前的尺寸分別為直徑600 μm 及450 μm 。此外，亦使用SEM及EPMA針對錫球銲點進行斷裂面及橫截面分析。 銲點於160℃進行固態熱處理，而熱處理時間為0∼4318小時不等。三種尺寸的錫球銲點之平均剪力強度皆有隨著固態熱處理時間增加而下降的趨勢，而三種尺寸的錫球銲點之主要差異在於剪力強度驟降的區間完全不相同。直徑750 μm 錫球銲點的平均剪力強度並沒有發生剪力強度驟降，而其平均剪力強度緩緩下降，熱處理2500小時後之平均剪力強度才降至初始平均剪力強度( 897g )的86％。直徑600 μm 錫球銲點平均剪力強度於熱處理48∼96小時的區間發生剪力強度驟降，熱處理96小時後之銲點平均剪力強度降至初始平均剪力強度( 895g )的85％。而直徑450 μm 錫球銲點平均剪力強度於熱處理0∼24小時的區間發生剪力強度驟降，熱處理24小時後之銲點平均剪力強度降至初始平均剪力強度( 860g )的90％。由此可知，銲點中Au濃度在0.07 wt.％∼0.13 wt.％的區間會使其平均剪力強度降幅的趨勢顯著改變，且銲點中Au濃度在0.13 wt.％∼0.31 wt.％的區間會使其初始平均剪力強度出現轉折變化。由錫球銲點的墊層部分斷裂面分析結果，發現有Ni墊層、Ni3Sn4晶粒、(AuxNi1-x)Sn4及銲料等斷裂面型態的存在，其中Ni3Sn4晶粒的斷裂面型態為沿晶斷裂模式，而(AuxNi1-x)Sn4的斷裂面型態為片狀之穿晶斷裂模式。由450 μm 錫球銲點的整體斷裂面觀測，發現大多數的斷裂面都是由銲料內部發生斷裂，此乃由於450 μm 銲點的最初剪力作用之橫截面積與墊層之面積接近。然而，600 μm 及750 μm 錫球銲點的墊層部分斷裂面則是發生於Ni墊層、Ni3Sn4晶粒、(AuxNi1-x)Sn4及銲料處，脆性斷裂面出現的機率變大。隨著Ni墊層參與斷裂面的比例增加，其銲點剪力強度有下降的趨勢，而所有銲點中剪力強度最差的斷裂面即由Ni墊層處發生斷裂。由剛迴銲後之橫截面觀測結果，界面處有一層Ni3Sn4( 2∼3 μm )。由固態熱處理1500小時後之橫截面觀測結果，(AuxNi1-x)Sn4三元化合物會回到界面處，其x值約為0.34∼0.37。此外，銲點中Au濃度的確為(AuxNi1-x)Sn4三元化合物回到界面處的驅動力，且銲料中的微結構粗化，這也是使其銲點平均剪力強度變弱的原因之一。 由直徑600 μm 及450 μm 錫球銲點於重複迴銲/熱處理流程中剪力強度之變化情形，我們可確立再次的迴銲程序的確可使之前固態熱處理所失去的銲點剪力強度予以強化，而其強化的程度無法使其立即回升至初始平均剪力強度，但經過數次的迴銲及固態熱處理程序後的確可以使其平均剪力強度回升至剛迴銲完後的初始平均剪力強度，甚至高於初始平均剪力強度。 The focus of the research work in this thesis was to think of the effect of gold concentration on shear strength of solder joints in the PBGA packaging. The PBGA substrates are used in core logic chipsets and graphic chipsets at present. The contact pads for solder balls on the PBGA substrates used in this study have the 0.4μm-Au / 8μm-Ni surface finish by electroplating. The diameter of the pads is 430 μm . The composition of the solder joints is 63Sn-37Pb ( wt.％) used in the present electronic industry. For the reflow, the peak reflow temperature was 225℃, and the reflow time was 90 s. The research was to evaluate the reliability of the solder joints by measuring the shear strength of the solder joints on the PBGA substrates. The experiment was divided into two parts. The first part was the measurement of shear strength for solder joints aged at 160℃. The sizes of these solder joints before reflow are 750 μm ﹐600 μm ﹐and 450 μm in diameter, respectively. Therefore, the gold concentration for three kinds of solder joints is 0.07 wt.％﹐0.13 wt.％﹐and 0.31 wt.％ respectively. The second part was the measurement of shear strength for solder joints during the process of RA500h160℃RA250h160℃RA250h160℃R. The sizes of these solder joints before reflow are 600 μm and 450 μm in diameter. Besides, we analyzed the fracture surface and cross-section view by SEM and EPMA. After aging at 160℃ for 0 to 4318 hours, the shear strength for the three kinds of solder joints decreased. Among the three kinds of solder joints, the major difference was the aging time at which a sharp decrease in shear strength occurred. The sharply decreasing shear strength for 750-μm solder joints did not occur after a long term aging at 160℃. The average shear strength decreased slightly right after aging at 160℃. After 2500 hrs of aging at 160℃, the average shear strength decreased to 86％ of the initial average shear strength ( 897g ). The aging time of sharply decreasing shear strength for 600-μm solder joints was from 48 to 96 hours. After 96 hrs of aging at 160℃, the average shear strength decreased to 85％ of the initial average shear strength ( 895g ). However, the aging time of sharply decreasing shear strength for 450-μm solder joints was from 0 to 24 hours. After 24 hrs of aging at 160℃, the average shear strength decreased to 90％ of the initial average shear strength ( 860g ). To sum up, the gold concentration between 0.07 wt.％ and 0.13 wt.％ would make the decreasing trend of the average shear strength change sharply. In addition, the gold concentration between 0.13 wt.％ and 0.31 wt.％ would make the initial average shear strength as a turning point. From the analytical results of the fracture surface, Ni, Ni3Sn4, (AuxNi1-x)Sn4, and solder existed in the fracture surface on the pad side. The fracture model of Ni3Sn4 was intergranular. But, the fracture model of (AuxNi1-x)Sn4 was transgranular. From the overall view of the fracture surface for 450-μm solder joints, the major fracture surface occurred inside the solder. This was because the cross-section area exerted by the initial shear force for 450-μm solder joints was close to the cross-section area of the pads. Nevertheless, the fracture surfaces for 600-μm and 750-μm solder joints occurred at Ni, Ni3Sn4, (AuxNi1-x)Sn4, and solder. There was a good possibility that the brittle fracture surface occurred. The proportion of Ni on the fracture surface increased, the shear strength of solder joints had the decreasing trend. Among the solder joints, the fracture surface on which the poorest shear strength occurred was made of Ni. The cross-section view right after reflow showed a thin layer Ni3Sn4 ( 2∼3 μm ) at the interface. From the analytical results of the cross-section view aged at 160℃ for 1500 hours, (AuxNi1-x)Sn4 would relocate from inside the solder joints to the interface, and the value of x for (AuxNi1-x)Sn4 at the interface was about 0.34 to 0.37. Moreover, the gold concentration inside the solder joints was the driving force for (AuxNi1-x)Sn4 to come back to the interface. The microstructure of solder coarsened, and this was one cause of the weak average shear strength for solder joints. During the process of RA500h160℃RA250h160℃RA250h160℃R, we assured that further reflow could strengthen the average shear strength of solder joints lost by aging at 160℃. But, the degree of strengthening would not at once make the average shear strength regain the initial average shear strength. However, after several reflow and aging cycles, the average shear strength could indeed come back to the initial value, and even be higher than the initial value. |
