促進競爭是文獻中解釋分購的理由之一，競爭的程度由參與供應商的數目決定，以供應商的家數作參數。我建立一個兩期的模型，用「從做中學」的效果，將競爭的程度用成本結構的差異大小來表現，成本結構差異越大，競爭越激烈，透過模型讓競爭的程度由購買者內生控制。希望透過這一個簡單的模型檢視當廠商數目固定時，購買者會否在考量控制供應商的成本相似下採用分購策略。雖然發現在我模型的設定中，購買者考量兩期之下一定選擇不採用分購策略，但是，仍能發現只考量第二期的購買價格下促進競爭的確是購買者採用分購的原因之一。

近年來，許多企業希望透過不同領域的專家彼此合作，為組織解決日益複雜的難題。然而，跨專業團隊的合作過程充滿挑戰，過去文獻多專注於如何提升知識分享的效果，卻忽略了許多關鍵新知識必須由跨界團隊合作探索而來。在面對陌生情境下，這樣的探索尤其重要，例如：開發新產品、啟用新製程，但我們至今仍對跨專業社群如何合作探索的過程所知甚少。 為了探究這個議題，本研究採用民族誌的方式，追蹤半導體晶圓廠工程師如何在複雜的生產系統中解決跨專業的維修難題。過程中，工作者常常必須面臨因果錯綜複雜、責任歸屬不清、跨界溝通不良所導致的合作困境。研究發現要解決這樣的困境，工作者必須運用疆界物件引發三種工作實務：解讀現象背後的因果、改變合作關係、整合跨界知識，才能找出問題核心並對症下藥。 根據本研究發現，如果我們能了解工作者在特定情境中的學習過程，便可提升既有的知識管理與疆界物件理論。在實務議題上，本研究的發現也能幫助現有員工培訓與跨專業溝通方式。 / Recently, more and more companies are gathering different types of specialist in order to solve increasingly complex problems. But the efforts paid in the process of cooperation are challenging for enterprises. Previous researchers had focused on the transfer and share of cross-disciplinary knowledge. However, they neglected the fact that some critical knowledge must be learned by collaborative exploring in terms of particular situation. To understand this issue, an ethnography study was used to examine the process of trouble shooting undertaken by engineers who encountered complex problems in the fabrication of semiconductor wafers. This thesis addressed three primary challenges faced by engineers. First, the nature of the problem may not be defined appropriately at the beginning. Second, responsibility may not be clearly attributed by cross specialist team who is in charge of investigation. Third, without comprehensive contexts of the practice, communication between engineers is problematic itself. To overcome the problem, engineers must be able to go beyond standard operating procedures so that they can find a new path of solution. This thesis argues that the use of boundary objects is an effective trigger of problem solving. The use of a boundary object is then described as a means of decoding the contexts behind the objects, reforming the relationship of cooperation, and integrating knowledge systemically. These findings suggest that theories of knowledge management and boundary objects could be improved organically by considering what people do and how people learn in practice. Furthermore, these findings bring us practical implications of employee training and cross-disciplinary collaboration.