透過您的圖書館登入
IP:3.144.17.45
  • 學位論文

鈀(磷)與純鈀金屬薄膜用於晶片接點之界面反應機構比較

Comparative Study between Palladium-Phosphorus and Pure Palladium Thin Film in the Reaction between Tin-Silver-Copper Alloy and Gold/Palladium/Nickel Tri-Layer

指導教授 : 何政恩

摘要


Au/Pd/Ni(P) (electroless nickel/electroless palladium/immersion gold,ENEPIG)表面處理技術,因具有下列優勢: (1)抗黑墊(black pads); (2)適於打線(wire-bonding); (3)成本低,故近來已廣泛取代傳統之化鎳浸金市場。其中Pd 層因鍍製還原劑的不同,使得Pd 層可分成Pd(P)與pure Pd兩種。本研究將建立Ni2SnP 的生長機制及Pd(P)與pure Pd 界面反應機構之比較。本實驗特用Sn-3Ag-0.5Cu 銲料與Au/Pd(P)/Ni 金屬銲墊來進行反應。其中Pd(P)膜的厚度為0.43 μm,Ni 膜為電鍍(electroplating)型。反應溫度固定在240 °C。在經過15 s 反應後,最表層的Au 已從界面上被移除,但仍有部份的Pd(P)薄膜殘存於界面上。此階段的反應 生成物分別為PdSn3 和Pd3P。當反應時間增長至30 s 時,界面上的Pd(P)已全被耗盡。相較於15 s 的反應,另有三種介金屬(Pd-Sn-P、Pd15P2、和Pd6P)會於界面上產生。若反應時間進一步增長至120 s 時,Pd(P)則會完全被轉換成PdSn4,並散佈於鄰近界面的銲料中。在此反應階段,界面上的主要生成物將轉變為(Cu,Ni)6Sn5。在(Cu,Ni)6Sn5 中,另有一具奈米結構之Ni2SnP 層產生。上述結果說明了雖然Pd(P)膜僅是次微米厚,且其P 含量相當低(2–5 %),但少量的P 仍在銲接反應中占極重要的地位。另外,在比較Pd(P)與pure Pd 用於ENEPIG [Au/Pd/Ni(P)]的晶片接點之界面反應機構比較。結果發現Pure Pd 反應系統在經過老化反應500 h,於界面生成另一介金屬(Ni,Cu)3Sn4。(Ni,Cu)3Sn4 的生長予否,更導致了銲點強度大幅下滑的主要原因。相較兩系統[pure Pd &Pd(P)]的界面反應結果,Pd(P)反應系統有著顯著的Ni2SnP 生成。Ni2SnP是一個良好Ni 的擴散障礙層(diffusion barrier)。因此,有效抑制(Ni,Cu)3Sn4 的生成。

並列摘要


Recently, the Pd (or Pd-P) film had been deposited over the Ni(P)metallization pads for high-end packaging applications. This was because an adequate Pd layer can prevent the Ni(P) from the galvanic attack derived from the immersion Au plating process; thus a hyper-corrosion in the Ni(P)(generally termed as “black pad”) was avoided. The different Pd film, whether it contained the phosphorus or not, was deposited by using the various reducing agent. This study established the mechanism of Ni2SnP growth from the Pd(P) thin film and provided a comparison Pd(P) and pure(P) on reactions during aging reaction with the Sn-3Ag-0.5Cu alloy. Reaction mechanism between liquid Sn-3Ag-0.5Cu solder and solid Au/Pd(P)/electrolytic-Ni films was examined using field-emission transmission electron microscope (FE-TEM) at different exposure times (15−600 s). After 15 s of exposure, the uppermost layer of Au was removed from the interface, and a portion of the Pd(P) film remained. At this stage of the reaction, the predominant products were PdSn3 and Pd3P.After 30 s of exposure, Pd(P) was completely exhausted, and four additional intermetallic species, including Pd-Sn-P, Pd15P2, and Pd6P,nucleated. After 120 s of exposure, the aforementioned species were destroyed, and Cu and Ni were involved in the reaction. The Pd existed primarily in the form of PdSn4 and was dispersed in the solder adjacent to the interface. The predominant product became (Cu,Ni)6Sn5, and the nucleation of a nanocrystalline Ni2SnP layer in the midst of (Cu,Ni)6Sn5 was observed. These results suggest that Pd and P play a vital role in the soldering reaction, even though the Pd(P) film is only a few submicrons (thickness) and its P content is quite low (2–5 %). Furthermore, in the Comparative Study between Pd(P) and pure Pd thin film, the different result of interfacial reaction between this two cases was one addition layer of (Ni,Cu)3Sn4 form between (Cu,Ni)6Sn5 and Ni(P) in the pure Pd case after aging at 180 °C for 500 h. Furthermore, mechanical response of the bonding interfaces of the two various film structures (i.e., pure Pd and Pd-P) was evaluated through a high-speed-ball shear (HSBS) test. In contrast with the result of interfacial reaction, the marked Ni2SnP formed at the interface in the Pd(P) case. Ni2SnP layer acted as a diffusion barrier to Ni from the Ni(P) layer during the aging reaction and had significant influences to retard the formation of (Ni,Cu)3Sn4.

並列關鍵字

nanocrystalline Ni2SnP ENEPIG Sn-Ag-Cu Pd-P pure Pd

參考文獻


[22] 林勝偉碩士論文,元智大學化學工程與材料科學學系研究所,2010
[26] S. P. Peng, W. H. Wu, C. E. Ho, Y. M. Huang, Journal of Alloys and Compounds, 493, p.431, 2010.
[1] K. N. Tu, Microelectronics Reliability, 51, p.518, 2011.
[2] A. Rahn (ed.), The Basics of Soldering, John Wiely & Sons, New York,1993.
[12] Z. Mei, P. Callery, D. Fisher, F. Hua and J. Glazer, Advances in Electronic Packaging, 2, p.1543, 1997.

被引用紀錄


王建菘(2012)。胃癌手術之住院日與醫療費用評估研究〔碩士論文,國立虎尾科技大學〕。華藝線上圖書館。https://doi.org/10.6827/NFU.2012.00189
林桂枝(2009)。影響兒科急診病患72小時再返之相關因素-以2005-2007年北部某醫院為例〔碩士論文,臺北醫學大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0007-2107200909202300
胡淑月(2010)。服用statins降血脂藥急診就醫者處方型態分析〔碩士論文,亞洲大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0118-1511201215464895
林昱吟(2013)。建構老年病患非計劃性重返急診之預測模式:以全民健保資料庫為例〔碩士論文,國立中正大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0033-2110201613553661
吳秀琴(2013)。建構骨關節炎患者於不同用藥下 之健保資源耗用模式 -以NSAIDs與專一性COX-2抑制劑為例〔碩士論文,國立中正大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0033-2110201613553558

延伸閱讀