經濟部中央標準局員工消費合作社印製 A7 五、發明説明（ 參考以往的申請案 本申请案已經在美國提出專利申請，其專利申請號碼爲 08/966,000 。 發明領域 / 本發明是有關於一種半導體裝置與形成該半導體裝置的 製私，更特別的是，具有包含阻障層與導電層的互連線結 構的的半導體裝置。 發明背景 知fla %路（1C)工業持續不斷的要製造出能改善可靠度與 性flb的金屬互連線。從底部到頂部的互連線的金屬層結構 包含鈦/氮化鈦/鋁或鋁合金/氮化鈦。Ti/TiN/A1/TiN。Ti 疋用來改善較低TiN層與底下隔絕層之間的黏著性，其中 較低ΉΝ層是當作擴散阻障層，而較高TiN層是當抗反射 層。銘合金可以包含銅或類似的元素。 一奴疋用鎢插塞來做A1互連線之間的連接。如果底下的 A1互連線具有TiN的抗反射層，則該TiN層通常會在後續 的介層蝕刻製程中被蝕刻掉，曝露出A1層。通常不會直接 把鎢沉積在含A1層上，因爲從WFs而來的氟會與鋁起反應 。所以，導電插塞（接觸插塞或介層插塞）通常具有—金屬 層結構，包含Tl/TiN/W。該Ti與TiN的功能與鋁互連線中 的相同。該Ti與TlN通常是在二獨立的物理氣相沉積步驟 中形成的，或是先沉積一厚Ti層，再將該層中的一部分轉 換成TiN。該Ti與TiN能避免氟在鎢的沉積製程時接觸到 鋁。 (請先閱讀背面之注意事項再填寫本頁) -裝· 訂 經濟部中央標準局員工消費合作社印製 A7 . , B7 五、發明説明（2 ) 導電插塞已經變得很普遍，有些給互連線用的鋁金屬層 結構包含Al/Ti/TiN。再次，A1可以是鋁或是鋁合金。在 本例中，Ti是在A1之上，但是這種金屬’層結構在界面的地 丨一:、方比較會形成AJ_3..Ti。許多熟練的專家都想避免掉Al3Ti / 因爲Al3Ti電阻很高。Al3Ti還有一個缺點是，铜（在Al-Cu 合金）會沿著Al3Ti界面擴散，結果影響到該結構的接觸窗/ 介層電阻。 另一種金屬層結構包含Al/TiN，其中A1是鋁或鋁合金， 而TiN是在A1之上。有些形成TiN的方法是使用含氮氣體 。該含氮氣體能與A1起反應，形成薄且高電阻値的氮化鋁 (A1N)層。該薄且高電阻値的A1N層相反的會影響到該結構 的接觸窗/介層電阻。 還有一種金屬層結構從底部到頂部包含Al/TiN/Ti/TiN 。 TiN/Ti/TiN —般是以三個分立薄膜層的方..式來形成 的。可以使用二個不同的濺鍍靶，但是會有粒子化的問題 。此外，濺鍍系統的濺鍍靶數目通常是有限的，一般是要 避免多餘的濺鍍革巴。如果使用其中一種濺鍍勒，則可以使 用鈦濺鍍靶來形成該三個分立薄膜層。通常，濺鍍擋板會 在濺鍍Ti之前，就先覆蓋住晶圓，以去除殘留在濺鍍靶上 的TiN，而且在後續的TiN層形成之前，再次以Ti覆蓋住 ，並在濺鍍TiN之前，先將鈦濺鍍靶的表面轉換成TiN。 該製程浪費掉大部分的濺鍍靶，因爲是被濺鍍到擋板上而 到非晶圓上。 此外，有些已知的互連線結構使用氮氧化鈦層。在該實 -5 - 本紙張尺度適用中國國家標準（CNS ) A4規格（'210 X W7公釐） I 訂 ^ 線 .. J i - (請先閱讀背面之注意事項再填寫本頁) A7B7 經濟部中央擦準局員工消費合作社印製 五、發明説明（3 例中，沉積出一鈦層後沉積一氮化鈦層，再沉積一含鈦較 多的氮化鈦層。在沉積出含鈦較多的氮化鈦層後，使用氧 電漿法將該層轉化成氮氧化鈦層。不幸的是，氧氧化鈦層 的電阻太大，不適合高性能的積體電路。雖然可以用鈕叙 取代氮氧化鈦中的鈦，但是高電阻還是一個問題。 雖然對含鈦層的努力很多，在互連線結構中還是希望使 用鈕以及與钽相關的化合物。例如.，使用純鈕層或氮化钽 層當作阻障層/黏著層，供铜互連線用。然而，很難用研磨 万法去除掉鈕，同時氮化钽與某些銅層具有黏著性的問題。 圖式的簡單説明 圖工是以三度空間的方式顯示依據本發明實施例的新互 連線結構。 圖2-4是以剖示圖的方式顯示料_ i 使 程室隨時間變化的情形。 ^ ㈣圖二是以XY圖顯示三種不同的實施例，其中每個實施 例都具有不同的氮成分。 圖8是以χγ圖顯示給阻障層110用的 電阻對時間的變化圖。 冑材枓的增加片 圖9是以條狀圖顯示習用裝置的應力 結構的應力k據。 數據，並比較圖i 圖10是以條、狀圖顯示不同習用裝置的片心 _ θ t /L 々电阻數據，並比 車父圖1結構的片電阻數據。 圖11是以對數χγ圖顯示已改善的圖1社 電阻。 口 1、、·。構的電子漂移 6 良紙張尺度適用中國國家襟準（CNS ) A4規格（21QX297公慶） 裝 訂 ^ 線 - (請先閲讀背面之注意事項再填寫本頁) A7 B7 經濟部中央標準局員工消費合作社印製 五、發明説明（4 圖12,與13是以XY圖分別顯示習用結構與圖1結構在！g 界面上的化學成分.。 圖14是以上視圖方式顯示一組沉積工具。 圖15是以剖示圖的方式顯示在形成互連線溝槽與接翁 開口後的一部分半導體裝置基底層結構。. 圖f 16是以剖示圖的方式顯示在形成氮化智層後，圖1 $ 的基底層結構。 圖17是以剖示圖的方式顯示在形成含鋰多的氮化銓層 之後，圖1 6的基底層結構。 圖18顯示在形成含妞多的氮化钽層之後，不同元素隨著 基底層深度的濃度變化。 圖19是以剖TF圖的方式顯7F在形成銅種子層之後，圖1 7 的基底層結構。 圖20是以剖示圖的方式顯示在種子層上電鍍出銅層； 後，圖19的基底層結構。 圖3 1是以剖示圖的方式顯示，研磨掉該基底層以去除掉 覆盍在互連線溝槽與接觸區外含短多的氮化短層上的銅居 之後，圖20的基底層結構。 圖22是以剖示圖的方式顯示，研磨含钽多的氮化起層p 及氮化钽層，形成供半導體裝置用的連接線之後，圖2^、 基底層結構。 圖23是以剖示圖的方式顯示已完成製作的半導體裝置 要注意的是，爲了簡化與方便説明起見，圖4由 „ 固八〒的構成 單元並沒有依據實際大小來繪製。例如相對於立& _ 本紙張尺度適用中國國家標準（CNS ) A4規格（210X297公釐） ----------^------訂---J----it W (請先聞讀背面之注意事項再填、寫本頁} 經濟部中央櫺準局員工消費合作社印製 A7 ,, B7 五、發明説明（5 ) 某些元件的尺寸被放太，以便能看得更清楚。此外，在圖 式中合適的地方，相同或相類似的構成單元是以相同的參 考數號來表示。 圖$的説明 产 本發明可以使用在很多不同的實施例中。在許多的實施 例中，使用一連續漸變部分的第一導電層，比如阻障層， 在例如互連線與導電插塞的導電結構中。該連續漸變部分 包含一第一元素與第二元素，比如耐高溫金屬與氮。該連 續漸變具有變化濃度的第一元素（如耐高溫金屬），更接近 具有比第一導電層更高導電性的一第二導電層（如鋁與銅 等）。特定實施例中的第一導電層具有鈦與氮，而第二導電 層大部分包含鋁，或是第一導電層具有妲與氮，而第二導 電層大邵分包含銅。 另一實施例是，半導體裝置包含一第一導電層..與一第二 導電層，大部分是銅所構成。該第一導電層具有一第一區 ，一第二區，以及一第三區，其中第二區是介於第一區與 第三區之間，而第三區是最接近該第二導電薄層。第二區 的氮濃度比第一與第三區的還要高。 形成具有該結構的半導體裝置的製程也同時被揭露出來 。本發明是由申請專利範圍所定義，而且在説明底下實施 例後會更爲容易了解。 在第一组實施例中，導電層被具有連續漸變部分的阻障 層所覆蓋。通常，該導電層以及連續漸變阻障層是相同互 連線的一部分。半導體晶圓安置在製程室内，其中鋁或鋁 -8- 本紙張尺度適用中國國家標準（CNS ) A4規格（210 X 297公釐） I 裝 訂 ^ 線 - I -- (請先閩讀背面之注意事項再填寫本頁) 合金沉積在該晶圓的頂部表面上。然後將晶圓放到另一製 程罜中，其中包含鈦濺鍍靶，而在濺鍍開始之前，鈦濺鍍 靶上是塗佈上一層薄的TiN層，接著利用氬氣（Ar)電漿環 境，將鈦濺鍍靶濺鍍到一部分或全部的丁iN層上，形成身. 有連續變化的阻障層部分。該部分具有TiNX成分，其中接 近鋁的氮濃度高於遠離鋁界面的氮濃度。這種特殊的氮分 佈/梯度會在以下的説明中做解釋。 在形成這種具連續變化部分的阻障層後，將含氮氣體（ 比如NO曝露到氬氣（Ar)電漿中，以產生氮/氬電漿，並在 濺射钦濺鍍靶中，產生接近化學計量比的TiN及/或化學計 量比的TiN。該接近化學計量比的TiN及/或化學計量比的 TiN评分沉積在具連續變化部分的阻障層的頂部。當形成 該接近化學計量比的TiN及/或化學計量比的TiN時，曝露 到氮氣電漿中也會改善要被濺鍍到晶圓的鈦濺鍍靶上的薄 TiN層。最後的結構是Al/TiNx/TiN複合式互連線結構，會 改善以下將説明的習用技術。 上述該複合式互連線結構比習用技術具有一些優點。第 一’在該互連線結構上形成，以便與該該互連線結構做電 性接觸的介層/接觸開口，是不需要以較大的介層/接觸窗 形狀比沉積出Ti或ήν阻障層。只需要一獨立鈦濺鍍靶， 形成TiNx/TiN即可。因爲有一個獨立鈦濺鍍靶沒有使用到 ’在TiN製程室中的粒子計數會降低、而阻障層的氮濃度 可以做更精確的控制。此外，最後的互連線結構包含有少 量曝露出來的鋁，或是沒有鋁曝露出來，所以與氟化鸽 經濟部中央標準局員工消費合作社印製 A7 --------B7 1 五、發明説明（7 ) (WF6)的逆反應會被降低或去除掉。 這種互連線結構要形成氮化鋁（A1N)的可能性會被降低 ’所以能改善介層/接觸窗電阻。此外，該互連線結構要形 成銘化妓（ALTi)的可能性冬會降低。如實驗數據所示的，，. 可以改善電子漂移3倍。此外，既然第—實施例的製程只 需要從氮化鈦靶做濺鍍以及從鋁靶濺鍍做濺鍍（亦即整個 複合互連線結構只需要二個製程室.），所以對晶圓的控制可 以減少，而晶圓缺陷也會比較少。另外，利用較少的製程 室，可以達到比習用方法還高的沉積系統產量。數據分析 也顯tf ’互連結構的薄膜應力比起習用技術來説已有改善 ’而且也改吾互連結構的介層/接觸窗電阻。因此.，該互連 結構的比起習用技術來説是一個有改進的解決方法。 現在要注思的到特定.的實施例。圖1顯示藉後續圖2 _ 4 中製程所形成的最後結構。圖j顯示，鋁（A1)層42先沉積 在介電層或阻障層（未顯示）上。乂層42最好是用鋁合金， 比如铭銅（Al-Cu)層，其中銅含量爲〇 5〇/〇到2 〇0/〇。在另— 形式中’ A1層42可以是銘銅石夕（Al_Cu-Si)層，鋁石夕（Al-Si) 屬’或相類似的金屬材料，或這些材料的組合。Ai層42 通系疋 >几積到厚度約4 0 0 - 70 0 nm，而5 5 0 nm是較佳的厚度 。铭可以沉積在先前已形成的阻障層材料上，比如Ti及/ 或TiN。可以用其它，的高溫金屬，比如艇（丁&)，來取代τι。 形成A1層42後，再形成阻障層中的連續漸變部分44。 以下將配合圖5-7，詳細説明漸變部分44的化學成分。— 般，漸變部分44會在A1層42的界面上包含較高的氮濃度 I 裝— 訂 I ^ ~"線 - - i- (請先閲讀背面之注意事項再填寫本頁)Printed A7 by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs. 5. Description of the invention (refer to previous applications. This application has been filed in the United States with a patent application number of 08 / 966,000. Field of the Invention / The invention relates to a semiconductor The device and the manufacturing of the semiconductor device, and more particularly, a semiconductor device having an interconnect structure including a barrier layer and a conductive layer. BACKGROUND OF THE INVENTION The fla% 1 (1C) industry continues to manufacture energy Improved reliability and reliability of metal interconnection lines. The metal layer structure of the interconnection lines from the bottom to the top includes titanium / titanium nitride / aluminum or aluminum alloy / titanium nitride. Ti / TiN / A1 / TiN. Ti 疋It is used to improve the adhesion between the lower TiN layer and the underlying insulation layer. The lower 层 N layer is used as a diffusion barrier layer, and the higher TiN layer is used as an anti-reflection layer. The alloy can contain copper or similar elements. A slave uses tungsten plugs to make connections between A1 interconnects. If the underlying A1 interconnect has a TiN anti-reflection layer, the TiN layer is usually etched away in the subsequent interlayer etching process , Exposing the A1 layer Tungsten is usually not deposited directly on the A1-containing layer, because fluorine from WFs will react with aluminum. Therefore, conductive plugs (contact plugs or interposer plugs) usually have a metal layer structure, including Tl / TiN / W. The functions of the Ti and TiN are the same as those in the aluminum interconnect. The Ti and TlN are usually formed in two separate physical vapor deposition steps, or a thick Ti layer is deposited first, and then Part of this layer is converted into TiN. The Ti and TiN can prevent fluorine from coming into contact with aluminum during the tungsten deposition process. (Please read the precautions on the back before filling this page) Printed by Consumer Cooperatives A7, B7 V. Description of the Invention (2) Conductive plugs have become very common, and some aluminum metal layer structures for interconnects include Al / Ti / TiN. Again, A1 can be aluminum or Aluminum alloy. In this example, Ti is above A1, but this metal 'layer structure is at the interface. A: The side will form AJ_3..Ti. Many skilled experts want to avoid Al3Ti / because Al3Ti has high resistance. Another disadvantage of Al3Ti is that copper (in Al-Cu alloy) will The Al3Ti interface diffuses, which affects the contact / interlayer resistance of the structure. Another metal layer structure includes Al / TiN, where A1 is aluminum or aluminum alloy, and TiN is above A1. Some methods of forming TiN are Use a nitrogen-containing gas. The nitrogen-containing gas can react with A1 to form a thin and high-resistance rhenium aluminum nitride (A1N) layer. The thin and high-resistance rhenium A1N layer will adversely affect the contact window / media of the structure Layer resistance. There is also a metal layer structure containing Al / TiN / Ti / TiN from the bottom to the top. TiN / Ti / TiN-generally formed in the form of three discrete film layers. Two different sputtering targets can be used, but there will be problems with particle formation. In addition, the number of sputtering targets for sputtering systems is usually limited, and it is generally necessary to avoid excessive sputtering. If one of the sputters is used, the three discrete thin film layers can be formed using a titanium sputter target. Generally, the sputtering barrier covers the wafer before sputtering Ti to remove the TiN remaining on the sputtering target, and before the subsequent formation of the TiN layer, it is covered with Ti again, and is sputtered. Prior to TiN, the surface of the titanium sputtering target was converted to TiN. This process wastes most of the sputtering target because it is sputtered onto the baffle and onto non-wafer. In addition, some known interconnect structures use a titanium oxynitride layer. In the actual -5-This paper size applies the Chinese National Standard (CNS) A4 specification ('210 X W7 mm) I Order ^ Line: J i-(Please read the precautions on the back before filling this page) A7B7 Economy Printed by the Consumers' Cooperative of the Ministry of Standards and Accreditation Service of the People's Republic of China. 5. Description of the Invention (In 3 cases, a titanium layer was deposited, a titanium nitride layer was deposited, and then a titanium nitride layer containing more titanium was deposited. After a large number of titanium nitride layers, the layer is converted into a titanium oxynitride layer using an oxygen plasma method. Unfortunately, the resistance of the titanium oxyoxide layer is too large to be suitable for high-performance integrated circuits. Instead of titanium in titanium oxynitride, high resistance is still a problem. Although there are many efforts for titanium-containing layers, it is still desirable to use buttons and tantalum-related compounds in interconnect structures. For example, using pure button layers or nitrogen The tantalum layer is used as a barrier layer / adhesive layer for copper interconnects. However, it is difficult to remove the buttons by grinding and the tantalum nitride has adhesion problems with some copper layers. The diagram is simple Explains that the graphics are displayed in a three-dimensional manner according to the present invention The new interconnect structure of the embodiment. Figure 2-4 shows the situation where the material chamber changes with time in a cross-sectional view. ^ ㈣ Figure 2 shows three different embodiments in an XY diagram, each of which Each embodiment has a different nitrogen composition. Fig. 8 is a graph showing the change of resistance to time for the barrier layer 110 in a χγ diagram. An increase sheet of cymbals is shown in Fig. 9 and the stress structure of a conventional device is shown in a bar graph. The data of the stress k is compared with the figure i. Figure 10 shows the chip core _ θ t / L 是以 resistance data of different conventional devices in a bar graph, which is compared with the chip resistance data of the structure of the car parent figure 1. Figure 11 is The logarithmic χγ diagram shows the improved resistance of Figure 1. The electronic drift of the structure 1, 1, .. 6 Good paper size is applicable to China National Standard (CNS) A4 specification (21QX297 public holiday) Binding ^-(Please read first Note on the back, please fill out this page again) A7 B7 Printed by the Consumers' Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs 5. Description of the invention (4 Figures 12 and 13 show the conventional structure and the structure of Figure 1 on the! G interface in XY diagrams, respectively. Chemical composition ... Figure 14 shows a set of views above Fig. 15 is a cross-sectional view showing a part of a semiconductor device base layer structure after forming interconnection trenches and abutment openings. Fig. F 16 is a cross-sectional view showing the formation of a nitride device. Figure 1 shows the structure of the base layer. Figure 17 shows the structure of the base layer of Figure 16 after forming a lithium-rich hafnium nitride layer in a cross-sectional view. Figure 18 shows the structure of the base layer containing many girls. After the tantalum nitride layer, the concentration of different elements varies with the depth of the base layer. Figure 19 shows the structure of the base layer of Figure 17 after the copper seed layer is formed by 7F in the form of a cross-section TF chart. Figure 20 is a cross-sectional view The method shows that a copper layer is electroplated on the seed layer; after that, the base layer structure of FIG. 19 is shown. FIG. 31 is a cross-sectional view showing that the base layer of FIG. 20 is ground after removing the copper layer overlying a short polynitride short layer outside the interconnect trench and the contact area by removing the base layer. Layer structure. 22 is a cross-sectional view showing the structure of the base layer after grinding the nitrided layer p and tantalum nitride layer containing tantalum and forming a connection line for a semiconductor device. FIG. 23 is a cross-sectional view showing the completed semiconductor device. It should be noted that, for the sake of simplicity and convenience of explanation, FIG. 4 is not drawn based on actual size.立 & _ This paper size applies to China National Standard (CNS) A4 specification (210X297 mm) ---------- ^ ------ Order --- J ---- it W ( Please read the notes on the back before filling out and write this page} Printed by A7, B7 of the Consumer Cooperatives of the Central Procurement Bureau of the Ministry of Economic Affairs 5. Description of the invention (5) The dimensions of some components are too large so that they can be seen More clearly. In addition, where appropriate in the drawings, the same or similar constituent units are denoted by the same reference numerals. The description of Figure $ The invention can be used in many different embodiments. In many In the embodiment, a first conductive layer, such as a barrier layer, is used in a continuous gradient portion in a conductive structure such as an interconnect and a conductive plug. The continuous gradient portion includes a first element and a second element, such as High temperature metals and nitrogen. This continuous gradient has a An element (such as a refractory metal) is closer to a second conductive layer (such as aluminum and copper) having higher conductivity than the first conductive layer. The first conductive layer in a specific embodiment has titanium and nitrogen, and Most of the second conductive layer contains aluminum, or the first conductive layer has hafnium and nitrogen, and the second conductive layer contains copper. Another embodiment is that the semiconductor device includes a first conductive layer .. and a first The two conductive layers are mostly made of copper. The first conductive layer has a first region, a second region, and a third region. The second region is between the first region and the third region. The third region is closest to the second conductive thin layer. The nitrogen concentration of the second region is higher than that of the first and third regions. The process of forming a semiconductor device having this structure is also revealed at the same time. The present invention is It is defined by the scope of patent application, and it will be easier to understand after explaining the following embodiments. In the first group of embodiments, the conductive layer is covered by a barrier layer having a continuous gradient portion. Generally, the conductive layer and the continuous gradient Barrier layer is one of the same interconnect The semiconductor wafer is placed in the process chamber, where aluminum or aluminum is used. 8- This paper size applies to the Chinese National Standard (CNS) A4 specification (210 X 297 mm). I Binding ^ Line-I-(Please read the back first (Please note again on this page) The alloy is deposited on the top surface of the wafer. Then the wafer is placed in another process, which contains a titanium sputtering target, and before the sputtering starts, the titanium sputtering target It is coated with a thin TiN layer, and then a titanium sputtering target is sputtered onto a part or all of the butadiene layer by using an argon (Ar) plasma environment to form a continuously changing barrier layer portion. This part has a TiNX composition, where the nitrogen concentration near aluminum is higher than the nitrogen concentration far from the aluminum interface. This special nitrogen distribution / gradient will be explained in the following description. After forming such a barrier layer with a continuously changing portion, a nitrogen-containing gas (such as NO is exposed to an argon (Ar) plasma to generate a nitrogen / argon plasma, and is sputtered to a sputtering target. TiN near stoichiometry and / or TiN near stoichiometry is generated. TiN near stoichiometric and / or TiN score near stoichiometry is deposited on top of the barrier layer with a continuously changing portion. When the near chemistry is formed With stoichiometric TiN and / or stoichiometric TiN, exposure to a nitrogen plasma also improves the thin TiN layer to be sputtered onto the titanium sputtering target of the wafer. The final structure is Al / TiNx / TiN The composite interconnect structure will improve the conventional technology that will be described below. The composite interconnect structure described above has some advantages over the conventional technology. First, it is formed on the interconnect structure to communicate with the interconnect. The structure of the dielectric layer / contact opening for electrical contact does not need to deposit a Ti or Val barrier layer with a larger dielectric layer / contact window shape ratio. Only a separate titanium sputtering target is required to form TiNx / TiN. .Because there is a separate titanium sputtering target not used in 'TiN The particle count in the process chamber will be reduced, and the nitrogen concentration of the barrier layer can be controlled more accurately. In addition, the final interconnect structure contains a small amount of exposed aluminum or no aluminum is exposed, so it is related to fluorination. Printed A7 by the Consumer Cooperative of the Central Standards Bureau of the Ministry of Pigeon Economy -------- B7 1 V. The reverse reaction of the invention description (7) (WF6) will be reduced or removed. This interconnect structure must form nitrogen The possibility of aluminum (A1N) will be reduced, so the interlayer / contact window resistance can be improved. In addition, the possibility that the interconnect structure will form ALTi will decrease in winter. As shown by experimental data , Can improve the electron drift by 3 times. In addition, since the manufacturing process of the first embodiment only requires sputtering from a titanium nitride target and sputtering from an aluminum target (that is, the entire composite interconnect structure only requires two Process room.), So the control of the wafer can be reduced, and the wafer defects will be less. In addition, using fewer process rooms, you can achieve a higher yield of the deposition system than conventional methods. Data analysis also shows tf 'Thin film stress ratio for interconnect structures The starting technology has been improved, and the interlayer / contact window resistance of the interconnection structure has also been improved. Therefore, the interconnection structure is an improved solution compared to the conventional technology. Now we must pay attention To specific embodiments. Figure 1 shows the final structure formed by the subsequent processes in Figures 2_4. Figure j shows that an aluminum (A1) layer 42 is deposited on a dielectric or barrier layer (not shown) first. The ytterbium layer 42 is preferably an aluminum alloy, such as a copper-plated (Al-Cu) layer, in which the copper content is from 050/00 to 2000 / 〇. In another form, the 'A1 layer 42 may be a copper-plated Shi Xi (Al_Cu-Si) layer, bauxite (Al-Si) metal or similar metal materials, or a combination of these materials. Ai layer 42 through system 疋 > thickness to about 4 0 0-70 0 nm, and 5 50 nm is the preferred thickness. Inscriptions can be deposited on previously formed barrier materials such as Ti and / or TiN. You can replace τι with other, high-temperature metals, such as boats (ding &). After the A1 layer 42 is formed, a continuous gradation portion 44 in the barrier layer is formed. The chemical composition of the gradient portion 44 will be described in detail below with reference to FIGS. 5-7. — Generally, the gradient portion 44 will contain a higher nitrogen concentration on the interface of the A1 layer 42. I — — I ^ ~ " Line--i- (Please read the precautions on the back before filling this page)
A7 B7 五、發明説明（8 ，而在接近頂部的地方含較低的m。這種穿過漸變部 刀44的氮刀佈/梯度是用目2-4所示的方法來達成的。通常 :漸變部分44在化學性質上可以是ΤιΝχ。漸變部分“的 厚度通常是在2-i5 nm之間，低於8 nm是比較好的。， /成漸變部分44後，接近化學計量比的⑽及/或化學 «十里比的ΤιΝ吨分46會沉積到漸變部分料上。丁以部分 6的厚度在60-100 nm之間，粗略的説⑽是比較好的 。ΤιΝ部为46被當成後續形成介層/接觸窗的蝕刻阻止層 ，以當成微影製程φ Μ 4上d » 才中的杬反射層（ARC)。阻障層包含漸變 部分44與TiN部分46的組合。 圖2_4顯示在相同的沉積/濺鍍室内，戟® 1中連續漸 夂#刀44以及TiN邵分46的特定製程。在圖2_4中，阻障 層116類似圖1的TiNx漸變部分44，而TiN層11 8類似圖 勺TiN層46在圖2-4所示的沉積步驟中，，連續對晶圓 曝光’而不使用擋·板或多餘晶圓。 在圖2中，顯示—沉積製程室1〇〇。沉積室製程1〇〇包含 :背板1〇1 :固定到鈇職艘乾102上。因爲對之前的晶圓 進仃處理，薄TiNxf 1〇4會在鈇錢鍍把1()2的曝露表面上 。沉積室製程100也包含—基座112，用來支撑半導體晶 圓114。晶圓114可以被箝住或放開到該基座112上，同時 阳圓114包含在其上形成的鋁層。此外，基座可以包 含加熱及/或冷卻裝置，用來控制在沉積/濺鍍時晶圓【Μ 的溫度。 機械手臂（圖2-4未顯示）將晶圓U4安置在基座n2上。 -11 - 本紙張尺度賴巾g|ϋ家標準（CNS ) Α4規格（2ΐ〇χ_297公釐 (請先閱讀背面之注意事項再填寫本頁) 裝. 訂 經濟部中夬椟準局員工消費合作衽印製A7 B7 V. Description of the invention (8, and the lower m near the top. This nitrogen knife cloth / gradient passing through the gradient knife 44 is achieved by the method shown in mesh 2-4. Usually : The gradient portion 44 may be TiNχ in chemical properties. The thickness of the gradient portion "is usually between 2 and 5 nm, and it is better to be lower than 8 nm. After the gradient portion 44 is formed, it is close to the stoichiometric ratio. And / or chemical «ten li ton of 46 tons will be deposited on the gradient part of the material. The thickness of the part 6 is between 60-100 nm, roughly speaking, it is better. The 46 part of the TiN is regarded as the follow-up An etch stop layer for the via / contact window is formed to serve as the 杬 reflection layer (ARC) on the photolithography process φ M 4. The barrier layer includes a combination of the gradation portion 44 and the TiN portion 46. Figure 2_4 shows in In the same deposition / sputtering chamber, the specific process of continuous gradual #knives 44 and TiN Shao Fen 46 in Halo® 1 is used. In FIG. 2_4, the barrier layer 116 is similar to the TiNx gradient portion 44 of FIG. 1, and the TiN layer 11 8 Similarly to the spoon TiN layer 46 in the deposition step shown in Figure 2-4, the wafer is continuously exposed without using a baffle plate. Excessive wafers. In Figure 2, it is shown-the deposition process chamber 100. The deposition chamber process 100 includes: the back plate 101: fixed to the shippin 102. Because the previous wafer is processed The thin TiNxf 104 will be on the exposed surface of the coin plating plate 1 () 2. The deposition chamber process 100 also includes a pedestal 112 to support the semiconductor wafer 114. The wafer 114 can be clamped or released Onto the pedestal 112, while the male circle 114 includes an aluminum layer formed thereon. In addition, the pedestal may include heating and / or cooling devices for controlling the temperature of the wafer during deposition / sputtering. Mechanical The arm (not shown in Figure 2-4) sets the wafer U4 on the base n2. -11-The paper size is g | 巾 家 standard (CNS) A4 size (2ΐ〇χ_297 mm (please read the back (Please fill in this page for matters needing attention)
I 經濟部中央標準局員工消費合作社印製 κι —----— -__Β7 , , 五、發明説明（9 ) ~ ' '— -- 在使用機械手臂安置晶圓後，會在沉積製程室ι〇〇產生氬 氣電漿106。氬氣電漿106包含受激的氬離子（Ar + )i〇8以= 党激的電子（e-)ll〇。使用電源來在鈦濺鍍靶ι〇2與氬氣電 聚1〇6之間產生電壓。與氬離子108受激狀態相類似的ς 壓差，會造成要從TiN層104濺鍍出去的材料，沉積在晶 圓I44頂部上，而形成阻障層工16。 w 圖3顯示濺鍍製程持續一段時間.（例如圖8與説明中所提 的時間），直到目標丁丨>^層104的材料由目標表面被去除掉 並沉積到晶圓114上形成阻障層116。因此，濺鍍製程最 後會導致阻障層116在晶圓上形成。如圖3所二的:氨 氣（Ar)電漿106會產生要從丁以層104去除掉的材料，所2 鈦濺鍍靶102上的TiN層104會變薄或整個從鈦濺鍍靶1〇2 上被去除掉。當TiN層104從鈦濺鍍靶ι〇2上被去"除掉時 ’阻障層116的组成會變得含更多的鈦或含更少的氮。亦 即’阻障層116的底部比頂部具有較高的氮濃度，'因爲τ道 層104的侵蝕是濺鍍時間的函數。在預設的時間過後，阻 障層116會達到預設的厚度，製程再繼續進行，如圖* TF 〇 圖4顯示出’提供最好是氮氣（Ν'2)的反應性氣體，终製 程室100内的處理環境用。因爲有提供氮氣，所2圖:的 氬氣電漿106會改變到圖4的氬-氮電漿1〇7。因此，氯 氮電漿107不只是包含受激的氬離子（Ar + )，而且還包含1 激的氮離子（N + )。與氬氣電漿106相同，氬-氮電浆1〇^ = 包含受激的電子（eJllO。 -12- 本紙張尺度適用中國國家標隼（CNS ) A4規格（210X297公釐） I¾------ir----^--- --*- (請先聞讀背面之注意事項再填寫本頁) 經濟部中央標準局負工消費合作社印製 A7 --------- Β7 , 五、發明説明（10 ) ~ - 〜- 在圖4氬-氮電漿107中的氮離子1〇9，會導致沉積到晶 圓1 14上的後續材料是化學計量比的TiN或接近化學計= 比的丁ιΝ因此，TlN層11 8被持續的沉積到阻障層丄丄6 。如圖4所示的，氬-氮電释1〇7中氮離子1〇9的出現會淨 生成ΤιΝ層104到鈇（Ti)濺鍍靶1〇2上。因此，具有 Α1/ΤιΝχ/ΤιΝ互連線'層的晶圓上14，可以用機械手臂方式， 從製程室100移開。移開該晶圓後，新的晶圓被安置到製 程室内，所以對圖2中的新晶圓，可以再次進行處理，同 時以重複的方式，對每個新晶圓持續進行到圖4。 圖5-7顯示三種不同化學組成的實施例，是會在圖4所示 曰η圓的各層114-1 1 8發生。圖5-7顯示圖4晶圓從TiN層118 頂邵到晶圓114頂部的垂直剖面中的氮濃度，該晶圓1工4 包含有銘層。 圖5顯示一特殊的實施例。位於晶圓i丨4頂部的銘層，. 幾乎沒有氮濃度或根本沒有氮濃度。缺乏相當量的氮濃度 是用線段120表示。在鋁114與阻障層116的界面上，氮濃 度會升高到接近化學計量比線段122。亦即，當圖2的沉 積/濺鍍剛開始時，從TiN層104所濺鍍出來的材料是接近 化學計量比的氮化鈦（TiN)，.使得圖3中阻障層116的底部 具有較高的氮原子（N)濃度。該高氮原子濃度是以圖5的線 段122來表示。 當氬氣電漿1 06繼續從鈦濺鍍靶1 〇2的TiN層1 04濺鍍出 材料時，TiN層104會變薄。當圖3的TiN層1〇4開始變薄 ，而且圖3的阻障層116開始變厚時，阻障層116的氮濃度 _： - 13- 本紙張尺度適用中國國家標牟（CMS ) A4規格（210X297公釐 . I . 裝 訂 線 ' J J - (請先聞讀背面之注意事項再填寫本頁) 經濟部中央標隼局員工消费合作社印製I Printed by the Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs κι —----— -__ Β7,, V. Description of the invention (9) ~ '' —-After the wafer is placed with a robot arm, it will be placed in the deposition process room. 〇〇Argon plasma 106 was generated. The argon plasma 106 contains stimulated argon ions (Ar +) i 08 = = party excited electrons (e-) 110. A power source was used to generate a voltage between the titanium sputtering target 102 and the argon polymer 106. A differential pressure similar to the excited state of the argon ion 108 will cause the material to be sputtered from the TiN layer 104 to be deposited on top of the wafer I44, forming a barrier layer process16. w Figure 3 shows that the sputtering process lasts for a period of time (for example, the time mentioned in Figure 8 and the description) until the material of the target layer ^ > ^ is removed from the target surface and deposited on the wafer 114 to form a resist Barrier layer 116. Therefore, the sputtering process eventually results in the formation of the barrier layer 116 on the wafer. As shown in Figure 3: The ammonia gas (Ar) plasma 106 will produce material to be removed from the silicon layer 104, and the TiN layer 104 on the titanium sputtering target 102 will become thin or the titanium sputtering target will be completely thin. 〇2 was removed. When the TiN layer 104 is removed from the titanium sputtering target 102, the composition of the barrier layer 116 becomes more titanium or less nitrogen. That is, 'the bottom of the barrier layer 116 has a higher nitrogen concentration than the top,' because the erosion of the τ channel layer 104 is a function of the sputtering time. After the preset time has elapsed, the barrier layer 116 will reach the preset thickness, and the process will continue, as shown in Figure * TF 〇 Figure 4 shows' provide a reactive gas that is preferably nitrogen (N'2), the final process For the processing environment in the chamber 100. Because nitrogen is provided, the argon plasma 106 in FIG. 2 is changed to the argon-nitrogen plasma 107 in FIG. 4. Therefore, the chlorine-nitrogen plasma 107 contains not only excited argon ions (Ar +), but also excited nitrogen ions (N +). Same as Argon Plasma 106, Argon-Nitrogen Plasma 1〇 ^ = Contains stimulated electrons (eJllO. -12- This paper size applies to China National Standard (CNS) A4 specification (210X297 mm) I¾ --- --- ir ---- ^ ----*-(Please read the precautions on the back before filling out this page) Printed by A7 Consumer Work Cooperative, Central Standards Bureau, Ministry of Economic Affairs -------- -Β7, V. Description of the invention (10) ~-~-The nitrogen ion 10 in the argon-nitrogen plasma 107 in FIG. 4 will cause the subsequent material deposited on the wafer 1 14 to be a stoichiometric TiN or Proximity to stoichiometry = ratio of butyl nitrate. Therefore, the TlN layer 118 is continuously deposited on the barrier layer 丄 丄 6. As shown in FIG. 4, the appearance of nitrogen ion 109 in the argon-nitrogen electrolysis 107 will occur. The TiN layer 104 is net generated on the Ti sputtering target 102. Therefore, the wafer 14 having the A1 / TιNχ / TιN interconnect line 14 can be removed from the process chamber 100 by a robotic arm. After the wafer is removed, the new wafer is placed in the process chamber, so the new wafer in FIG. 2 can be processed again, while each new wafer continues to FIG. 4 in a repeated manner. Figure 5-7 shows three Examples of different chemical compositions will occur in the layers 114-1 18 of the circle η shown in Figure 4. Figures 5-7 show the vertical cross-section of the wafer of Figure 4 from the TiN layer 118 to the top of the wafer 114 Figure 5 shows a special embodiment. The layer on the top of the wafer, there is almost no nitrogen concentration or no nitrogen concentration at all. It lacks a considerable amount of nitrogen. The concentration is represented by line segment 120. At the interface of aluminum 114 and barrier layer 116, the nitrogen concentration will rise to near the stoichiometric line segment 122. That is, when the deposition / sputtering of FIG. The material sputtered from 104 is titanium nitride (TiN) close to the stoichiometric ratio, so that the bottom of the barrier layer 116 in FIG. 3 has a higher nitrogen atom (N) concentration. The high nitrogen atom concentration is shown in the figure It is indicated by line segment 122 of 5. When the argon plasma 106 continues to sputter the material from the TiN layer 104 of the titanium sputtering target 10, the TiN layer 104 becomes thinner. When the TiN layer 104 of FIG. 3 When it begins to thin and the barrier layer 116 in FIG. 3 starts to thicken, the nitrogen concentration of the barrier layer 116 _:-13- This paper standard applies to Chinese national standards CMS) A4 size (210X297 mm I stapling line 'J J -.. (Please read smell precautions to fill out the back of this page) Ministry of Economic Affairs Central Office staff standard Falcon consumer cooperatives printed
A 7 }V 五、發明説明) 會開始降低，有更多從鈦濺鍍靶102的Ti，持續的被消耗 掉。氮濃度以及阻障層116含多Ti區的降低程度是由圖5 的線段124代表。圖5很清楚的顯示出，圖4中加入氮的步 驟是在整個鈦濺鍍靶102的TiN層1 04被去除掉之前便賜 始的。所以，圖5顯示出，沒有任何地方是只有純鈥從鈥 濺鍍靶102濺鍍出來，而沉積形成一部分的阻障層丨16(不 像圖7中所顯示的另一實施例）。在圖5中，鈇賤鍍鞋1 〇2 的TiN層104被完全去除掉前，會產生圖4的氬-氮電漿1〇7 ，以便沉積出具有比線段124還高氮濃度的接近化學計量 比TiN層或化學計量比TiN層，如圖5所示_。通常，線段 122與126幾乎是相同的。因此，沉積出具有起始氮濃渡 122且接近化學計量比的阻障層，使得氮濃度會隨阻障層 1 16厚度的增加而降低。同時，在另一製程中，從多餘晶 圓或製程室擋板，利用再氮化該濺鍍靶的晶圓濺鍍後，濺 鍍靶上的氮化鈦會再度形成。 圖6顯示另一實施例。圖6仍顯示出，晶圓114頂部的鋁 層包含很少量的氮12 0。圖6顯示，開始於圖2而冗成於圖 3的沉積製程，是在很短的時間内進行的。因爲圖6的阻障 層沉積時間很短，所以穿過阻障層116的氮濃度漸變分佈 變化在圖6中較不嚴重，而圖3中阻障層116的厚度較薄（ 約低於5 nm)然後氫-氮電漿會產生，藉此具有氮濃度126 的薄膜1 1 8會如圖6左手邊部分所示的來形成。 當圖4的氬-氮電漿107開始產生時，氬-氮電漿107會與 阻障層1 16表面上的原子鈦（Ti)反應。所以，氮離子109實 際上會增加阻障層116的氮濃度，進而使接近阻障層116 ~ 14 - 本紙張尺度適用中國國家標準（CNS ) A4規格（210X 297公舱） ----------^------ti------.ii - (請先閱讀背面之注*意事項界填转本頁) A 7 A 7 經濟部中央標準局員工消費合作社印製 五、發明説明（12 上表面的氮濃度隨著時間而增加。實際上，例如圖6的阻 障層116可能會整個轉換成化學計量比的TiN或接近化學 計量比的TiN，因爲在圖6的界面12讣上有氮離子ι〇9 = 出現，即使阻障層116剛開始是以非化學計量比的形式多 積。此外，當半導體裝置（晶圓）在進行金屬退火處理時， 如同習用的半導體技術，含氮少的區域與含氮多的區域間 的界面會變得不明顯。然而，分析結果顯示，當阻障層U6 的歲鍍時間較長（大於約2 0 -3 0 kW-秒）時，某些含欽多的區 域通常會在接近阻障層116與TiN層U8界面的地方被债 測出。 圖7顯示再另一個阻障層的實施例。圖7再次顯示出在 基底上的鋁層沒有包含或包含很少量的氮，線段〗2 〇。如 果圖3的沉積時間很長時（在Applied Materials Endura中約 大於1 kW下20-30秒）及/或氬氣電漿i〇6的功率較高時（在 Applied Materials Endura 中約大於 5 kW)，貝ij 圖 3 的 TiN 層 104很可能會整個被去除掉，而且純鈦會開始沉積，當成 一部分的阻障層116。這個現象在圖7中有顯示，所以阻 障層116較低部分的氮濃度較高，如圖中線段〗22所示， 而圖7中線段124清楚的顯示出沒有氮出現。所以，在圖7 中線段124上，純鈦從濺鍍靶1 〇2沉積出來，當作圖3中阻 障層116的上面部分。在圖7中線段124中，圖3的TiN層 104被整個被去除掉，而且純鈦從濺鍍靶102沉積出來。接 著圖7顯示，產生氬-氮電漿1〇7，所以TiN層1 18會因爲 氬-氮電漿而在阻障層116上形成。要注意的是，某些圖7 的含鈥多的區域會被出現在圖4氬-氮電漿1 07中的氮離子 15- 本紙張用中關家縣（CNs ) M规格（21QX 297公骚） ------ 种衣 訂 線 > f (請先閱讀背面之注·意事項#-填本頁j A7 ιυ 五、發明説明（13 所消耗掉。因此，圖5-7鞀- +他# π 不，阻障層116的不同分佈及/ 或农度可以利用改變濺鍍時間 ,π认你拉处兄卡七 又呼間與風氣電漿106，氬-氮電漿 107的錢嫂成量來達成。 圖8顯示濺鍍時間對濺鏟A 7} V V. Description of the invention) will begin to decrease, and more Ti from the titanium sputtering target 102 will be continuously consumed. The reduction in the nitrogen concentration and the multi-Ti-containing region of the barrier layer 116 is represented by the line segment 124 in FIG. 5. Fig. 5 clearly shows that the step of adding nitrogen in Fig. 4 was initiated before the entire TiN layer 104 of the titanium sputtering target 102 was removed. Therefore, FIG. 5 shows that there is nowhere other than pure “sputtering” from the sputtering target 102, and a portion of the barrier layer 16 is deposited (unlike the other embodiment shown in FIG. 7). In FIG. 5, before the TiN layer 104 of the base plated shoe 102 is completely removed, the argon-nitrogen plasma 107 shown in FIG. 4 is generated in order to deposit a close chemical with a higher nitrogen concentration than the line segment 124. The stoichiometric TiN layer or the stoichiometric TiN layer is shown in FIG. 5. Generally, the line segments 122 and 126 are almost the same. Therefore, a barrier layer having an initial nitrogen concentration of 122 and close to the stoichiometric ratio is deposited, so that the nitrogen concentration decreases as the thickness of the barrier layer 116 increases. At the same time, in another process, from the excess wafer or the process chamber baffle, after the wafer is re-nitrided with the sputtering target, the titanium nitride on the sputtering target is formed again. Figure 6 shows another embodiment. Figure 6 still shows that the aluminum layer on top of wafer 114 contains a small amount of nitrogen 120. FIG. 6 shows that the deposition process starting from FIG. 2 and redundant from FIG. 3 is performed in a short time. Because the barrier layer deposition time of FIG. 6 is short, the gradual change in the nitrogen concentration through the barrier layer 116 is less serious in FIG. 6, and the thickness of the barrier layer 116 in FIG. 3 is thinner (about 5 nm) Then a hydrogen-nitrogen plasma is generated, whereby a thin film 1 1 8 having a nitrogen concentration of 126 is formed as shown in the left-hand part of FIG. 6. When the argon-nitrogen plasma 107 of FIG. 4 starts to be produced, the argon-nitrogen plasma 107 will react with atomic titanium (Ti) on the surface of the barrier layer 116. Therefore, the nitrogen ion 109 actually increases the nitrogen concentration of the barrier layer 116, thereby bringing it closer to the barrier layer 116 ~ 14-This paper size applies the Chinese National Standard (CNS) A4 specification (210X 297 cabin) ----- ----- ^ ------ ti ------. ii-(Please read the note on the back * Notes and fill in this page) A 7 A 7 Staff Consumer Cooperatives, Central Standards Bureau, Ministry of Economic Affairs 5. The description of the invention (12 The nitrogen concentration on the upper surface increases with time. In fact, for example, the barrier layer 116 of FIG. 6 may be converted into a stoichiometric TiN or a near stoichiometric TiN, because Nitrogen ions on the interface 12 讣 in FIG. 6 = appears, even if the barrier layer 116 is initially accumulated in a non-stoichiometric ratio. In addition, when a semiconductor device (wafer) is subjected to a metal annealing process, As with conventional semiconductor technology, the interface between regions with less nitrogen and regions with more nitrogen will become inconspicuous. However, the analysis results show that when the barrier layer U6 has a longer plating time (greater than about 20 -3) 0 kW-seconds), some regions containing Chintau are usually close to the interface between the barrier layer 116 and the TiN layer U8 Detected. Figure 7 shows yet another embodiment of the barrier layer. Figure 7 shows again that the aluminum layer on the substrate does not contain or contains a small amount of nitrogen, line segment 2 0. If the deposition time of Figure 3 For a long time (20-30 seconds at more than 1 kW in Applied Materials Endura) and / or the power of argon plasma i06 is high (approximately more than 5 kW in Applied Materials Endura). The TiN layer 104 is likely to be completely removed, and pure titanium will begin to be deposited as a part of the barrier layer 116. This phenomenon is shown in Figure 7, so the lower part of the barrier layer 116 has a higher nitrogen concentration. As shown by line segment 22 in the figure, and line segment 124 in FIG. 7 clearly shows that no nitrogen is present. Therefore, on line segment 124 in FIG. 7, pure titanium is deposited from the sputtering target 102, as shown in FIG. The upper part of the barrier layer 116. In the line segment 124 in Fig. 7, the TiN layer 104 of Fig. 3 is completely removed, and pure titanium is deposited from the sputtering target 102. Then Fig. 7 shows that an argon-nitrogen plasma is generated. 107, so the TiN layer 118 will be formed on the barrier layer 116 due to the argon-nitrogen plasma. Note that Yes, some of the regions containing Figure 7 will appear as nitrogen ions in the Argon-Nitrogen Plasma 1 07 of Figure 4 15- Zhongguanjia County (CNs) M Specification (21QX 297) for this paper- ----- Threads for clothing > f (Please read the notes and notes on the back # -Fill this page j A7 ιυ 5. Invention description (13 is consumed. Therefore, Figure 5-7 鼗-+ 他# π No, the different distribution and / or agronomy of the barrier layer 116 can be used to change the sputtering time. πIdentify that you pull the sibling card Qi and Huo and the wind and gas plasma 106, argon-nitrogen plasma 107 money Amount to reach. Figure 8 shows sputtering time vs. shovel
__ 材料的増加片電阻的變化圖V 圖8清楚的顯示，在圖5-6中沾、& r… ' 中的濺鍍製程中剛開始的5秒所 沉積出來的材料，比鈦的電卩 、 %丨且性更间。斫即，圖8中0-5 秒所沉積出來的材料是本氣具古 口氣里间的鈦（與純鈦作比較），或__ The change of the chip resistance of the material is shown in Figure V. Figure 8 clearly shows that the material deposited in the first 5 seconds of the sputtering process in the dip, & r ... 'in Figure 5-6 is more efficient than that of titanium.卩,% 丨 and more sexual. That is, the material deposited in 0-5 seconds in Figure 8 is the titanium in the ancient tone of the gas (compared with pure titanium), or
是接近化學計量比的TiN。A h•著時間增加到圖8的1 〇 _工5 秒間，圖3的TiN層104被進—牛 . 延步蝕掉，所以沉積到阻 障層116上材料的氮濃度奋隆 展反g降低氮濃度。随著濺鍍時間增 加的這種氮濃度的降低（亦gp 土& 门 ( 牛低α即方除掉圖3的TiN層104),在 圖8中是很清楚的會增加爆兩卜士 〇 , 印瑨加争私性。圖8中右邊的導電性增 加是MTiN層104變薄，而且有更純Μ持續的沉積出 來。取後，在Applied Materials Endura的i謂下經過約 20秒的、麟後，所有的或接近所有的⑽I⑽在圖3中 會被去除掉，而如圖7中所清楚顯示的，純鈦（τι)會隨著阻 障層116的上面部分而開始沉積。圖8包含所有在約【 功车操作下，從Applied Materials Endura pvD系統中 得的數據。 所以，如果要使用圖3的製程進行濺鍍5杪時，圖8顯示 該結構具有相對應於圖6所形成的組成。圖8接著顯= 如果要在圖3的製程室中進行濺鍍1〇_12秒時，該結構Z有 相對應於圖5所形成的組成。最後，圖8顯示如果要在圖3 的製程限制中進行濺鍍30秒時，該結構具有相對應於圖7 -16 本紙張尺度 ^-、1T^ (請先閱讀背面之注"意事項豕填寫本頁) 經濟部中央標準局員工消費合作社印製 )A4規格（210X 297公浇） 經濟部中央標準局員工消費合作社印製 A7 ''' 1 ~'――———__B7 五、發明説明（μ ) 所形成的組成。 當賤鍍功率增加時，完全消耗掉圖3的τιΝ@ 1〇4的時 間會減少。相反的’在較低的功率下，完全侵蝕掉圖3的 TlN層104離開賤鍍革巴1〇2的時間就會更長。一般，圖5身 顯示的實施例可以利用功率*時間等於5,〇〇〇 kw_sec〇nds到 仏⑼^豕⑽他㈣〜⑷之間的條件來形^因此，^ 不只疋顯7F出要達到圖5_7結果所需的濺鍍量或燒灼時間 ，而且還顯示出功率與時間要如何控制，以便在使用圖2_ 4製程所形成的阻障層116中產生不同鈦與氮分佈。 圖9包含條狀圖，當與習用A1/Ti/TiN的實施例做比較時 ，顯π圖1互連線具有較低的薄膜應力。圖9的左邊部分 顯示退火處理前與退火處理後Α1/Τι/ΤιΝ結構的應力程度 。圖9的右邊邵分顯示退火處理前與退火處理後圖i中 Α1/ΤιΝχ/ΤιΝ結構的應力程度。從圖9可以很清楚的看出， 利用圖1 - 8所示的製程能降低與互連線層相關的薄膜應力 程度。 “ 圖10是以條狀圖顯示不同習用互連線結構的片電阻與 圖1新的互連線結構做比較。圖i 〇的左邊部分表示退火處 理前與退火處理後包含有Ti/A1/Ti/TiN的結構都具有較高 片電阻。另外，圖1 〇的中間部分表示退火處理前與退火處 理後包含有Al/Ti/TiN的結構。Al/Ti/TiN結構在退火處理 前與退火處理後的片電阻分別低於Ti/Al/Ti/TiN結構的片 電阻。圖1 0的右邊部分表示退火處理前與退火處理後，圖 1新Al/TiNx/TiN結構的片電阻。從圖10可以看出來，圖i _____ 17- 本紙張尺度適―關家辟（CNS )》4祕~ '----- ----------Μ------、玎---^---ΙΜ i ~ (請先閱讀背面之注意事項再填寫本頁} 經濟部中央標準局員工消費合作社印製 A7 1 1 B7 五、發明説明（15 ) 新Al/TiNx/TiN金屬互連線的片電阻已比其它的結構有了 改進，尤其是退火後的量測數據。 圖1 1顯示電子漂移數據302與300。Al/Ti/TiN結構的電 子漂移數據302通常比Ti/Al/Ti/TiN結構與Al/TiN/Ti/Ti识 結構較好。圖1 1的電子漂移數據3 00是針對圖1新的 Al/TiNx/TiN結構。可以從曲線302與300的斜率大小看出 來，圖1互連線的電子漂移比起Al/Ti/TiN的結構，已有大 幅的改善。實驗顯示，至少有3倍的電子漂移的改善已經 利用與習用結構不同的圖1結構來達成。Al/Ti/TiN結構與 圖1中Al/TiNx/TiN結構，在相同的條件下，會對Al/Ti/TiN 結構造成標準方差爲0.45約31小時的平均誤差時間（MTF) ，而對圖1中心Al/TiNx/TiN結構的平均誤差時間（MTF)會 大於或等於標準方差爲0.32的63小時。Al/TiNx/TiN結構 電子漂移特性的最大電流約爲242 μΑ，而用相同幾何形狀 在相同應力條件下所形成的Al/Ti/TiN 結構，會產生最大 約943 μΑ的電子漂移電流。所以，摘要的説，圖1的互連 線結構比起習用技術，已有大幅的改進。 圖12與13顯示圖1的結構不會有氮化鋁（Α1Ν)的形成， 與Al/TiN/Ti/TiN的結構不同。圖12顯示出在Al/TiN/Ti/TiN 結構中鋁與較低層TiN界面的能量分散光譜（EDS)數據。圖 12清楚的顯示在界面上有氮出現，而且還有A1N形成。相 反的，圖13顯示出在圖1中鋁42與阻障層的較低層部分 44的EDS數據。圖13清楚的顯示在圖1的界面上沒有A1N 形成。 -18 - 本紙張尺度適用中國國家標準（CNS ) A4規格U10 X 297公釐） I 裝 訂 ~線 一 ~ (請先閱讀背面之注意事項再填寫本頁) 經濟部中央標準局員工消费合作社印製 A7 B7 五、發明説明（16 ) 圖 14顯示一組沉積工具 4〇〇,如 AppliedMaterialsEndura PVD濺鍍系統。將晶圓經由製程室入口 4〇6，放入晶圓傳 送A 404。一旦晶圓在晶圓傳送室4〇4内，機械手臂會將 晶圓安置在铭（A1)濺鍍室401或402中。在其中一個賤緩备 401或402被用來沉積出圖i的鋁層a後，該晶圓會該機 械手臂經由傳送室404 ,而被傳送到圖4中製程室1〇〇a與 io〇b妁其中之一。一旦晶圓被安置在製程室1〇〇a或l〇〇b 内，便進行圖2-4的製程，在鋁層42的表面形成漸變部分 44與TiN部分46。一旦圖2-4的製程在製程室100&或1〇〇b 内几成’便從該組沉積工具4〇〇中將該晶圓經由晶圓傳送 罜404與製程室入口 406移出，同時可以對下個晶圓進行 處理。 圖1 - 14的結構與製私可以大幅的改善習用技術。利用圖 1的結構，可以不需阻障層便能形成覆蓋的鎢介層。然而 ，如果鎢芫全覆盍住隔絕層時，仍需要鈥及/或氮化鈇當作 黏著層。然而，並不太需要沉積這些側壁黏著層，來形成 圖1介層/接觸窗底下的阻障層。因此，可以進步降低後續 的介層直徑，而不會降低黏著的步階覆蓋能力。 另外，相對於圖2-4的製程是使用鈦濺鍍靶1〇2。所以， 避免複合型的氮化鈦（ΊΊΝ)靶，造成較少的粒子數，改善沉 積氮化鈦層的氮濃度，以及較高的產量。 如圖12-13所示的，圖1的互連線結構會完全去除掉或減 少氮化鋁（A1N)的形成。氮化鋁的減少是有利的，因爲氮化 鋁是較高電阻層，因爲AIN的出現會影響到片電阻以及介 -19- 本纸張尺度適用中國國家標準（CNS ) A4規格（210X297公整） 裝 訂 ^ ~線 一 „ (請先閲讀背面之注意事項再填寫本頁} A7 B7 經濟部中央標準局員工消費合作社印製 五、發明説明Γ17 層/接觸窗電阻。另外，既然鈥並不直接接觸到圖 任何一點上的鋁，所以鋁化鈦（A^i)的形 = 去除掉。此外，圖i裳置的電子漂移特 心皮降低或被 ，構的裝置，有大幅.的改善至少丄到3倍(見圖。 另外，不需要圖2_4中要分別形成TiN，Η， 、， 需的擋板製程或多餘晶圓製程。所以，使用圖 1N層所 會改善產量。此外，還會改進製造成本與產量，^製程 習用技術的實例來，用來形成圖i互連線結構 與製程室數目會減少。數據顯示，使用圖““Π 量可以增加約5〇%。另夕卜，較少控制晶圓會使得晶：材料 内的缺陷減少，而圖14的沉積系統包含多餘的製 比 如沉積系統中二個幾乎相同的A1製程室，以及二個幾乎 同的ΤϊΝ阻障層製程室）。多餘的製程室是有利的， 然製程室H)〇a與402可以用來進行製程處理，而製：： 100b與401可以進行保護性的保養或升級，而不需要二^ 的停止該系統400。如圖9-10中清楚顯示的，圖:結^全 薄膜應力比起習用技術的結構，被大幅的降低，而圖 勺 料的片電阻與介層/接觸窗電阻也比起習用技術來有改盖1。材 現在將目標轉移當第二實施例上，其中阻障層是在 包含銅或類似材料的導電層之前所形成的。再：：形：： 續漸變的阻障層 '然而，含多金屬的部分會接觸到導 本身。半導體裝置包含基底，在基底上的第一遒哈^包、. 及在第-導電層上的第二導電層。第—導電層：：；溫： 屬與氮。.第-導電層具有接近基底的第—區，以及離基底 -20- 表紙張尺度適用中國國家標準（CNS ) Α4規格（210Χ297公釐） I. 裝 訂 ^ Γ線 - - (請先閲讀背面之注意事項再填寫本頁) A7 B7 經濟部中央標準局員工消費合作社印製 五、發明説明（18 ) :爾二區。第二區的氮含量比第-區的低。第二導電 邵分是鋼。在第一導電層内的所有區域提供良好 ;政阻障層(呆—區)並與第二導電層具有良好黏著性(第二 區）0 •現在要注意到特定的實施例上。圖15包含半導體裝置^ 展1100，場隔絕區1102與摻雜區1104。如該ς定實=例 所使用的，半導體裝置基底1100包含單晶半導體晶圓，半 導體在絕緣體上晶圓或其它用來形成半導體裝置的基底。 閘介電層U06在半導时置基底丨⑽上形成，接著是碎 層1107與氮化矽層1108。定義出氮化矽層11〇8與矽層1107 圖案，以形成圖15所示的閘極。側壁隔層i 1〇9在鄰接包 含矽層1107與氮化矽層1108的該閘極處形成。雖然未顯 不出，.但是氮化矽區可以在摻雜區i丨〇4上。 第一層間介電層（ILD)IOU在基底1100與閘極上。該第 一 ILD 1011包含第一蝕刻阻止層m〇，第—平坦化隔絕 層1112，第二蝕刻阻止層1114，與第二平坦化隔絕層ιιΐ6 。所有以上各層111〇_1116通常都是絕緣體。在一特殊的 實施例中，蝕刻阻止層包含如氮化矽的氮化層，而平坦化 隔絕層1112與U16包含氧化層。雖然未顯示，可以形成 包含多矽的氮氧化矽抗反射層，當作第一 ILE) 1011的一部 分’而且在第二平坦化隔絕層i i 16形成，但未顯示。第一 ILD 1011被定義出圖案，形成開口 1〇12，包含互連溝槽區 1122與接觸區1124。互連溝槽區1122在第二平坦化隔絕 層1116内形成，而且通常是比接觸區H24寬，延伸到第 -21 - 本纸伕尺度適用中國國家標準（CNS ) A4規格（21〇χ297公釐 . 裝 、可---^---Ί 0 - ~ (請先閱讀背面之注意事項再填寫本頁) 經濟部中央標準局員工消費合作社印製 A7 B7 五、發明説明（19 ) ： — 一蝕刻阻止層πιο，第二蝕刻阻止層1114，第一平坦化 隔絕層1U2。到目前爲止的結構是用習用方法形成的。 接著氮化妲層1022沉積在基底上，並在開口 1〇12内。 氮化钽層1022通常包含33:50原子比率的氮，與鈕平衡v 該層是如圖16所示的形成並接觸到摻雜區n〇4。進一步 的處理該基底，在氮化鉅層1022上形成含钽多的氮化鈕層 1032，如圖17所示。氮化钽層1022與含鈕多的氮化鈕層 1032的結合便是第一導電層，是一種供形成連接線用的黏 著/阻障層。該第一導電層是在150 nm的範園内，通常是 在10-30 nm的範圍内。 形成氮化麵層1022與含钽多的氮化鈕層ι〇32的過程如 下所述。基底被安置在製程室的晶座内，比如沉積裝置。 該晶座被加熱到約50-250〇C的溫度範園。沉積該二層時的 壓力一般是在約15-40毫米托的範圍内。該壓力是與所要 沉積結構的形狀比與幾何形狀有關的。用來濺鍍材料的直 流電流功率通常是在0.5-3 kW的範園内，尤其是，通常在 1.2- 1 ·8 kW的範圍内。在濺鍍室内用來產生電漿的無線電 波（RF)功率通常是在1_2 kW的範圍内，尤其是，通常在 1.3- 1.7 kW。 通常疋在單一眞空循環中一次形成氮化赵廣1022與含 叙多的氮化鈕層ι〇23,在沉積第一區時，其中氮化麵層1〇22 接近化學计里比（TaN) ’沒有偏壓該基底。此時，含氮氣體 與貴氣，如氬，被導入濺鍍靶上。含氮氣體包含氮，氨或 其它類似的氣體。形成含短多的氮化纽層1〇32時，含氮氣 22- 表紙張尺度適用中國國家標準（CNS ) A4規格（210 X 297公瘦） ----------种衣------1T---'---1^ - - (請先聞讀背面之注意事項再填寫本頁) 經濟部中央襟準局員工消費合作社印製 A7 B7 五、發明説明（2〇 ) to會停止，而貴氣繼續流動，基底會變成偏壓，約有負乃 到負80伏。 在濺鍍沉積時，當氮原子比率如圖1 8所示的減少時，層 内的钽原子比率會增加。圖18包含濃度圖（原子比率），是 從曝露表面到第一 ILDl〇n距離的函數。含鉅多的氮化鈕 層具有約0-30原子比率範園的氮。在該實施例中，含鈕多 的氮化钽層1〇32的上表面是純鈕，而沒有氧原子。在上表 面上的較低氮原子比率通常會對幾乎是含銅的薄層有較佳 的黏著性。在上表面，如果銅的黏著性是問题時，钽原子 比率可以至少爲95%，而氮原子比率可以低於5%。在另一 實例中，含氮氣體與貴氣都可以停止，而製程室要在貴氣 流動之前抽眞空即可。將會形成具獨立區的第—導電層。 在一特殊的實施例中，從含氮氣體流出（氮化妲層 且偏壓關閉（含紐多的氮化妲層i 〇3 2)的時間，約等於含氮 氣體流出終止且偏壓打開的時間。然而，時間是不一樣的 。在某些實施例中，氮化鋰層1022是比含銓多的氮化鈕層 1032還厚。氮化钽層1〇22與含妲多的氮化銓層1〇32的總 厚度是在約1-50 nm的範圍内，通常是在約i 丄…j u nm白勺聋巳 圍内。氮化鈕層1〇22與含鈕多的氮化妲層1〇32是阻障層 的一部分。 銅種層1054沉積在含赵多的氮化垣層上，如园 所示。該銅種層1054可以用不同的方法沉積出來，包含物 理氣相沉積，化學氣相沉積或其它類似的方法。如果i用 金屬有機化學氣相沉積，飼的前驅質可以是^ < μ卜的任一種 -23- Ί紙張尺度適用中國國家標準（CNS ) Α4規格（2丨〇><297公釐） ' ~~_ . _ 裝 訂 ^ "線 i - (請先閱讀背面之注意事項再填寫本頁) 經濟部中央榇準局員工消費合作社印製 五、發明説明（21 ) 或多種：六氟醋酸乙晞三甲石々，卜 π — T石夕烷铜（VTMS);六氟醋酸3_ 乙決銅；或其它類似的化合物。另外可以使用銅降_二酉同 .化合物。在銅種層1054沉積後，在整個基底表面上形成電 鍍銅層1064，如圖20所示。兩祐加政 , ·呢艘銅層1064的厚度有足多’ 厚’可比填滿開口 1 〇 12的万途的，致拖· ·、 7立連線溝槽部分。銅種層1 054 與電鍍銅層1064形成第二導雷屑，士却八θm ^ 子私增，大邵分是銅，用來形成 連接線用。 . 進行第一化學機械研磨步.锻，去除掉覆蓋住含叙多的氮 化鈀層1032的所有銅種層1〇54與電鍍銅層1〇64，如圖21 所。當研磨到第二導電層（銅種層1〇54與電鍍銅層1〇64) 時，第一導電層（氮化钽層1022與覆蓋住含鋰多的氮化鋰 層1032)是研磨阻止層。進行第二化學機械研磨步驟，去除 掉在開口 12外第二平坦化隔絕層1116上的含鈕多的氮化 钽層1032與氮化钽層1022 ,如圖22所示。連接線1〇84包 含一互連線與接觸到其中一個摻雜區！丨〇4的一接觸區。雖 然未顯示出，但是互連線1082是使用接觸區做電性連接的 ’圖22未顯示。 繼續進行製程步驟，形成圖23所示的完工裝置。此時， 第三蚀刻阻止層191〇在連接線1〇82與1〇84上形成，第三 干坦化卩兩，纟巴層19 12在第三蚀刻阻.止層1 9 1 〇上形成，第四 蝕刻阻止層1 9 14在第三平坦化隔絕層χ 9 12上形成。雖然 未顯示出，但也有形成第四平坦化隔絕層。包含第三蚀刻 阻止層1910，第三平坦化隔絕層1912與第四蝕刻阻止層 1 9 14的第二ILD層會被定義出圖案’會在需要使用習用製 -24 本紙張尺度適用中國國家標準（CNS ) A4規格（210'乂297公釐） 神衣------.訂----^--- * ~ (請先閱讀背面之注意事項再填寫本頁) A7 B7 經濟部中夬祿準局屬工消費合作社印製 五、發明説明（22 程的地方，形成互連線溝槽與接觸開口。 形成開口後，沉積出氮化艇層】，接著是含艇多的氮 化鋰層1932，銅種層1954與電鍍銅層1964。進行適當的 研磨處理步驟後，形成鈍化層1〇98，以便完成整個裝置。产 雖然未顯示，但是可以製作出額外的層間介電層以及其它 %性連接線’圖23未顯示。 本發明逼有其它的優點。除了使用赵外.，還可以使用.其 它南溫金屬，來形成阻障層/黏著層。可以用如鎢，鉬或相 類似的高溫金屬來取代妲。再另一實施例中，含钽多的氮 化钽層1〇32與氮化钽層1022可以結合不同的半導體原子 來構成。例如，可以形成氮化矽妲以及含钽多的氮化矽鈕 二此外^可以出現鍺原子或矽與鍺原子。形成高溫金屬_ 半導體-氮的化合物時，通常濺鍍所使用的濺鍍靶是包含高 溫金屬與半導體原子。 在特定實施例中，可以利用矽化鈕，而且可以配合氬氣 與氮氣，濺鍍矽化妲靶來形成氮化矽钽。含鈕多的氮化矽 鈕層可以用類似形成含鈕多的氮化鈕層的方法來形成，其 中氮氣停止，而將氬導入濺鍍靶中，從之前的濺鍍中，去 除掉殘留在歲鍍乾表面上的的氮化碎叙。 在另一實施例中，純鈕層，钽多的氮化鈕層或氮化矽钽 層，可以在形成氮化銓層1022前便形成。在其中的實施例 中二可以對鋰靶進行短時間的濺鍍，以形成起始的鈕層， 接著氮化钽層與含鈕多的氮化钽層。該實施例能形成較佳 的接觸窗給基底與矽化物，因爲钽層會與出現在表面上的 (請先閲讀背面之注意事項再填寫本頁) 裝. -25- A7 ______B7 1 > 五、發明説明（23 ) 原始氧化物起反應。 在其它的結構中，可能會有單一無效連接線形成。此時 ，氮化钽層1022與氮化钽層1032的厚度與圖ι5_23所示的 雙無效結構比起來是會增加。 又另一實施例中，可以用化學氣相沉積或物理氣相沉積 所形成的單一銅層，來取代鋼種層1〇54與電鍍銅層1〇64 的結合。當形成此層時，可以加熱基底，使得銅能流到所 需的開口中。此時，基底或晶座溫度通常是在约350_450 1的範圍内。要非常小心進行，因爲如果溫度太高的話， 氮化Μ阻障層的完整性會被破壞掉。沉積時的溫度要高到 足以使銅流動，但是還不會高到讓氮化鋰阻障層的完整性 會被破壞掉。這些銅流動的實施例能達到如降低研磨時間 與凹陷的優點。 如上所述的實施例提供習用技術所沒有的優點。氮化妲 層很好的阻障層，但是使用含妲多的氮化妲層連接氮化鉦 層會提供較佳的黏著性，尤其是化學氣相沉積的銅層。相 =如果，化鈕層1032的氮含量太高，會有氰化物形成，影 經濟部中央標準局員工消費合作社印製 (諳先閱讀背面之注意事項异壤寫本頁} 會到黏著性。保持接觸到銅的表面氮濃度到很低含量，會 達到較佳的黏著性。 q 改善黏著性的另一優點也會達到較佳接觸電阻。此外， 本發明的實施例會降低電子漂移的影 ，例可以使用現有的設備，而不用特丄：或；展 邊際製程。 上述説明已經參照特定的實施例對本發明内容做了描述 ---26- 本紙張~~—--—_ 五 、發明説明（24 A7 B7 。然而，對 以下本發明 與改變。所· 非限定性的 圍内。已經 而，會發生 與任何要件 本的專利範 具有一般技術程度的人士來說， t _ . ^在不偏雜 申4專利範圍所述的範圍下，做 風出不同的修 以，説明内容與圖式只能視爲説明性的參考而 範圍，而且任何的修改都應包括在本發明的每 參照特定的實施例説明其好處與其它優點。然 任何好處與優點或變成更爲顯著的好處，優點 ，都不會被解釋成任何嚴格的，必需的，或基 園特性或構成要件。 I---------抽衣-------、玎----^---A -,Γ (請先閲讀背面之注意事項再填寫本頁) 經濟部中央標準局貞工消費合作社印製 -27- 本纸張尺度適用中國國家標準（CNS ) A4規格（210X297公釐）TiN is close to the stoichiometric ratio. A h • The time is increased to 10 ～ 5 seconds in FIG. 8, and the TiN layer 104 in FIG. 3 is advanced—newly etched away. Therefore, the nitrogen concentration of the material deposited on the barrier layer 116 is strong. Reduce nitrogen concentration. This decrease in nitrogen concentration with increasing sputtering time (also gp soil & gate (lower α means removing the TiN layer 104 in Figure 3), it is clear in Figure 8 that it will increase the burst 〇, Neem adds privacy. The increased conductivity on the right in Figure 8 is that the MTiN layer 104 becomes thinner and more pure M is continuously deposited. After taking it, it takes about 20 seconds under the i term of Applied Materials Endura. After this, all or nearly all of ⑽I 会 will be removed in Figure 3, and as clearly shown in Figure 7, pure titanium (τι) will begin to deposit with the upper part of the barrier layer 116. Figure 8 contains all the data obtained from the Applied Materials Endura pvD system under the operation of the bicycle. Therefore, if the process of FIG. 3 is to be used for sputtering 5 杪, FIG. 8 shows that the structure has a structure corresponding to that shown in FIG. 6. The formed composition is shown in FIG. 8. Then, if sputtering is performed for 10-12 seconds in the process chamber of FIG. 3, the structure Z has a composition corresponding to that of FIG. 5. Finally, FIG. 8 shows that if the When sputtering is performed for 30 seconds in the process limit of FIG. 3, the structure has a paper rule corresponding to that of FIGS. 7-16. Degrees ^-, 1T ^ (Please read the note on the back " Issue 豕 Fill this page first) Printed by the Consumers 'Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs) A4 size (210X 297 pouring) Printed by the Consumers' Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs System A7 '' '1 ~' ——————__ B7 V. Composition of the invention description (μ). When the base plating power is increased, the time to completely consume τιN @ 104 in Fig. 3 will decrease. On the contrary, at a lower power, the TlN layer 104 in FIG. 3 completely erodes away from the base plater 102 for a longer time. In general, the embodiment shown in FIG. 5 can be formed by using a condition between power * time equal to 5,000 kw_sec nds to 仏 ⑼ ^ 豕 ⑽ ㈣ 因此 ~ ^ ^ Therefore, not only does the 7F output reach The amount of sputtering or burning time required for the results of FIG. 5_7, and also shows how the power and time should be controlled in order to produce different titanium and nitrogen distributions in the barrier layer 116 formed using the process of FIG. 2_4. Figure 9 contains a bar graph. When compared with the conventional A1 / Ti / TiN embodiment, it is shown that the interconnect of Figure 1 has a lower film stress. The left part of FIG. 9 shows the stress degree of the A1 / Ti / TιN structure before and after the annealing treatment. The right side of Fig. 9 shows the stress levels of the A1 / TιNχ / TιN structure in Fig. I before and after annealing. It can be clearly seen from FIG. 9 that the process shown in FIGS. 1-8 can reduce the stress level of the thin film associated with the interconnect layer. "Figure 10 shows the bar resistance of different conventional interconnect structures in a bar graph compared with the new interconnect structure of Figure 1. The left part of Figure i 〇 shows that Ti / A1 / is included before and after annealing. The structures of Ti / TiN all have high sheet resistance. In addition, the middle part of Figure 10 shows the structure containing Al / Ti / TiN before and after annealing. The structure of Al / Ti / TiN before and after annealing The sheet resistance after the treatment is lower than that of the Ti / Al / Ti / TiN structure. The right part of Fig. 10 shows the sheet resistance of the new Al / TiNx / TiN structure in Fig. 1 before and after the annealing treatment. From the figure 10 can be seen, Figure i _____ 17- This paper is suitable for scale-Guan Jia Pi (CNS) "4 secrets ~ '----- ---------- M ------ 、 玎--- ^ --- ΙΜ i ~ (Please read the notes on the back before filling out this page} Printed by the Consumers' Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs A7 1 1 B7 V. Description of the invention (15) New Al / TiNx / TiN The sheet resistance of metal interconnects has been improved compared to other structures, especially the measured data after annealing. Figure 11 shows the electron drift data 302 and 300. The electricity of the Al / Ti / TiN structure The drift data 302 is generally better than the Ti / Al / Ti / TiN structure and the Al / TiN / Ti / Ti recognition structure. The electronic drift data 3 00 of FIG. 11 is for the new Al / TiNx / TiN structure of FIG. 1. It can be obtained from It can be seen from the slopes of the curves 302 and 300 that the electronic drift of the interconnection in Figure 1 has been greatly improved compared to the structure of Al / Ti / TiN. Experiments have shown that at least 3 times the improvement of the electronic drift has been utilized and used The structure of Figure 1 is different. The Al / Ti / TiN structure is the same as the Al / TiNx / TiN structure in Figure 1. Under the same conditions, the standard deviation of the Al / Ti / TiN structure is 0.45 and the average is about 31 hours. Error time (MTF), and the average error time (MTF) for the Al / TiNx / TiN structure in the center of Figure 1 will be greater than or equal to 63 hours with a standard deviation of 0.32. The maximum current for the electronic drift characteristics of the Al / TiNx / TiN structure is approximately 242 μΑ, and the Al / Ti / TiN structure formed with the same geometry and the same stress conditions will produce an electron drift current of up to about 943 μΑ. So, in summary, the interconnect structure in Figure 1 is more than conventional Technology has been greatly improved. Figures 12 and 13 show that the structure of Figure 1 is free of nitrogen. The formation of aluminum (Α1Ν) is different from the structure of Al / TiN / Ti / TiN. Figure 12 shows the energy dispersive spectrum (EDS) data of the interface between aluminum and the lower TiN in the Al / TiN / Ti / TiN structure. 12 clearly shows the presence of nitrogen at the interface and the formation of A1N. In contrast, FIG. 13 shows EDS data of the aluminum 42 and the lower layer portion 44 of the barrier layer in FIG. FIG. 13 clearly shows that no A1N is formed on the interface of FIG. 1. -18-This paper size applies to Chinese National Standard (CNS) A4 specification U10 X 297 mm) I Binding ~ Line 1 ~ (Please read the precautions on the back before filling this page) Printed by the Consumers Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs A7 B7 V. Description of the invention (16) Figure 14 shows a set of deposition tools 400, such as the AppliedMaterialsEndura PVD sputtering system. The wafer is placed in the wafer transfer A 404 through the process chamber entrance 406. Once the wafer is in the wafer transfer chamber 400, the robotic arm will place the wafer in the sputtering chamber 401 or 402 of A1. After one of the base buffers 401 or 402 is used to deposit the aluminum layer a of FIG. I, the wafer will be transferred by the robot arm to the process rooms 100a and io of FIG. 4 through the transfer chamber 404. b 妁 one of them. Once the wafer is placed in the process chamber 100a or 100b, the process of FIG. 2-4 is performed, and a graded portion 44 and a TiN portion 46 are formed on the surface of the aluminum layer 42. Once the process of Figure 2-4 is within a few minutes of the process room 100 & 100b, the wafer is removed from the set of deposition tools 400 through the wafer transfer 罜 404 and the process room entrance 406, and at the same time can be Process the next wafer. The structure and manufacturing of Figure 1-14 can greatly improve customary techniques. With the structure of FIG. 1, a covered tungsten interlayer can be formed without a barrier layer. However, if tungsten and rhenium cover the insulation layer, rhenium and / or rhenium nitride are still required as the adhesion layer. However, it is not necessary to deposit these sidewall adhesive layers to form the barrier layer underneath the interposer / contact window. Therefore, it is possible to progressively reduce the subsequent interlayer diameter without reducing the step coverage ability of the adhesion. In addition, compared with the process of FIGS. 2-4, a titanium sputtering target 102 is used. Therefore, avoiding a composite titanium nitride () N) target, resulting in a smaller number of particles, improving the nitrogen concentration of the deposited titanium nitride layer, and a higher yield. As shown in Figure 12-13, the interconnect structure of Figure 1 completely removes or reduces the formation of aluminum nitride (A1N). The reduction of aluminum nitride is advantageous, because aluminum nitride is a higher resistance layer, and the appearance of AIN will affect the sheet resistance and dielectrics. This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) ) Binding ^ ~ Line 1 „(Please read the notes on the back before filling in this page} A7 B7 Printed by the Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs 5. Description of the invention Γ17 layer / contact window resistance. In addition, since“ is not direct Contact with aluminum at any point in the picture, so the shape of titanium aluminide (A ^ i) = removed. In addition, the electron drift of the figure is reduced or covered, and the structure of the device has been greatly improved. At least丄 to 3 times (see figure. In addition, it is not necessary to form TiN, Η,, or the required baffle process or excess wafer process in Figure 2_4. Therefore, using the layer of Figure 1N will improve the yield. In addition, it will also improve the yield. Improving manufacturing costs and yields. Examples of conventional manufacturing process techniques will reduce the number of interconnect structures and process chambers used to form Figure i. The data shows that the use of the figure "" can increase the amount by about 50%. In addition, Less control of the wafer will make the crystal : Defects in the material are reduced, and the deposition system of FIG. 14 contains redundant processes such as two almost identical A1 process chambers in the deposition system, and two nearly identical TNO barrier process chambers). The redundant process chamber is advantageous Of course, the process chambers H0a and 402 can be used for process processing, and the system: 100b and 401 can be protected for maintenance or upgrade without stopping the system 400. See Figure 9-10 It is clearly shown in the figure that the stress of the full film is greatly reduced compared with the structure of the conventional technology, and the sheet resistance and interlayer / contact window resistance of the material are also changed compared to the conventional technology. Now the target is shifted to the second embodiment, in which the barrier layer is formed before the conductive layer containing copper or a similar material. Re :: Shape :: Continuously changing barrier layer 'However, the polymetal-containing portion It will come into contact with the conductor itself. The semiconductor device includes a substrate, a first package on the substrate, and a second conductive layer on the first conductive layer. The first conductive layer ::; temperature: belongs to nitrogen .. The first conductive layer has a first region close to the substrate, And from the base -20- sheet paper size applies Chinese National Standards (CNS) A4 specifications (210 × 297 mm) I. Binding ^ Γ line--(Please read the precautions on the back before filling this page) A7 B7 Central Standard of the Ministry of Economic Affairs Printed by the Bureau ’s Consumer Cooperatives. 5. Description of the invention (18): District II. The nitrogen content of the second zone is lower than that of the first zone. The second conductive material is steel. All areas in the first conductive layer provide good ; The political barrier layer (dead-region) and has good adhesion to the second conductive layer (second region) 0 • Now notice the specific embodiment. Figure 15 contains the semiconductor device ^ 1100, field isolation region 1102 And doped region 1104. As used in this example, the semiconductor device substrate 1100 includes a single crystal semiconductor wafer, a semiconductor wafer on an insulator, or other substrate used to form a semiconductor device. The gate dielectric layer U06 is formed on a semi-conducting substrate, followed by a chipping layer 1107 and a silicon nitride layer 1108. The patterns of the silicon nitride layer 1108 and the silicon layer 1107 are defined to form the gate shown in FIG. 15. A sidewall spacer i109 is formed adjacent to the gate including the silicon-containing layer 1107 and the silicon nitride layer 1108. Although not shown, the silicon nitride region may be on the doped region i. A first interlayer dielectric layer (ILD) IOU is on the substrate 1100 and the gate. The first ILD 1011 includes a first etch stop layer m0, a first-planarization isolation layer 1112, a second etch-stop layer 1114, and a second planarization isolation layer ιιι6. All the above layers 1110-1116 are usually insulators. In a particular embodiment, the etch stop layer includes a nitride layer such as silicon nitride, and the planarization isolation layers 1112 and U16 include an oxide layer. Although not shown, a polysilicon oxynitride anti-reflection layer may be formed as a part of the first ILE) 1011 'and formed on the second planarization barrier layer i i 16, but it is not shown. The first ILD 1011 is patterned to form an opening 1012, which includes an interconnect trench region 1122 and a contact region 1124. The interconnect trench region 1122 is formed in the second planarization isolation layer 1116, and is generally wider than the contact region H24, and extends to the -21th.-The size of this paper is applicable to the Chinese National Standard (CNS) A4 specification (21〇297297mm) . 装 、 可 --- ^ --- Ί 0-~ (Please read the notes on the back before filling out this page) Printed by the Consumers' Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs A7 B7 V. Invention Description (19): — An etch stop layer πο, a second etch stop layer 1114, and a first planarization barrier layer 1U2. The structure so far has been formed by a conventional method. Then a hafnium nitride layer 1022 is deposited on the substrate and an opening 1012 is formed. The tantalum nitride layer 1022 usually contains 33:50 atomic ratio of nitrogen, which is in equilibrium with the button v. This layer is formed as shown in FIG. 16 and contacts the doped region n04. Further processing the substrate, the nitride A tantalum-containing nitride button layer 1032 is formed on the giant layer 1022, as shown in Fig. 17. The combination of the tantalum nitride layer 1022 and the button-containing nitride layer 1032 is the first conductive layer, which is used to form a connection. Adhesive / barrier layer for wire. The first conductive layer is in the 150 nm range, usually In the range of 10-30 nm. The process of forming the nitrided surface layer 1022 and the tantalum-containing nitride button layer om32 is as follows. The substrate is placed in a wafer seat of a process chamber, such as a deposition device. The crystal The seat is heated to a temperature of about 50-250 ° C. The pressure when depositing the two layers is generally in the range of about 15-40 mm Torr. The pressure is related to the shape ratio of the structure to be deposited and the geometry. The DC current power used for sputtering materials is usually in the range of 0.5-3 kW, especially in the range of 1.2-1.8 kW. Radio waves (RF) used to generate plasma in the sputtering room ) The power is usually in the range of 1_2 kW, in particular, usually 1.3- 1.7 kW. Usually, in a single hollow cycle, nitriding Zhao Guang 1022 and the nitride-containing button layer ι〇23 are deposited at one time. In the first zone, where the nitrided surface layer 1022 is close to the chemical ratio (TaN), the substrate is not biased. At this time, a nitrogen-containing gas and noble gas, such as argon, are introduced onto the sputtering target. Nitrogen-containing The gas contains nitrogen, ammonia, or other similar gases. When forming a short and more nitrided layer 1032, nitrogen contains 22- The paper size is applicable to Chinese National Standard (CNS) A4 (210 X 297 male thin) ---------- seed coat ----- 1T ---'--- 1 ^- -(Please read the notes on the back before filling out this page) A7 B7 printed by the Employees' Cooperatives of the Central Procurement Bureau of the Ministry of Economic Affairs 5. Description of the invention (2) to will stop, and the noble gas will continue to flow, and the substrate will become biased Pressure, it is about negative to even 80 volts. During the sputter deposition, when the nitrogen atomic ratio decreases as shown in Figure 18, the tantalum atomic ratio in the layer will increase. Figure 18 contains a concentration map (atomic ratio) as a function of distance from the exposed surface to the first ILD10n. The huge nitride-containing layer has nitrogen in the range of about 0 to 30 atomic ratios. In this embodiment, the upper surface of the button-containing tantalum nitride layer 1032 is a pure button without oxygen atoms. Lower nitrogen atomic ratios on the upper surface usually give better adhesion to thin layers that are almost copper-containing. On the top surface, if copper adhesion is a problem, the tantalum atomic ratio can be at least 95% and the nitrogen atomic ratio can be less than 5%. In another example, both the nitrogen-containing gas and the noble gas can be stopped, and the process chamber can be evacuated before the noble gas flows. A first conductive layer with separate regions will be formed. In a special embodiment, the time when the nitrogen-containing gas flows out (the hafnium nitride layer and the bias voltage is turned off (the Nitrogen-containing hafnium nitride layer i 032)) is about equal to the time when the nitrogen-containing gas flow ends and the bias voltage is turned on. However, the time is not the same. In some embodiments, the lithium nitride layer 1022 is thicker than the hafnium-containing nitride button layer 1032. The tantalum nitride layer 1022 and the hafnium-containing nitrogen The total thickness of the hafnium layer 1032 is in the range of about 1-50 nm, usually within the range of about 丄 ... ju nm. The nitrided button layer 1022 and the nitride containing the button are The hafnium layer 1032 is part of the barrier layer. A copper seed layer 1054 is deposited on a nitrided layer containing Zhaoduo, as shown in the garden. The copper seed layer 1054 can be deposited by different methods, including a physical vapor phase. Deposition, chemical vapor deposition or other similar methods. If i is metal organic chemical vapor deposition, the feed precursor can be any of ^ < μ -23- Ί paper size applicable to Chinese National Standard (CNS) Α4 Specifications (2 丨 〇 > < 297mm) '~~ _. _ Binding ^ " Thread i-(Please read the precautions on the back before filling this page) Printed by the Consumer Cooperatives of the Central Bureau of Standards, Ministry of Economic Affairs of the People's Republic of China. 5. Description of Invention (21) or more: Acetyl Hexafluoroacetate, Trimethyllithium Acetate, Bu π — T Siloxane Copper (VTMS); 3_ Ethyl Acetate; Or other similar compounds. In addition, copper compounds can be used. After the copper seed layer 1054 is deposited, an electroplated copper layer 1064 is formed on the entire surface of the substrate, as shown in Figure 20. The thickness of the copper layer 1064 is enough to be 'thick' comparable to filling the opening 1012 million, which leads to the dragging of the grooves. The copper seed layer 1 054 and the electroplated copper layer 1064 form a second conductor. Lightning scraps, but θm θm ^ increase privately, Da Shao is copper, used to form the connection line.. The first chemical mechanical grinding step. Forging, remove the palladium nitride layer 1032 covering the polysilicon. All copper seed layers 1054 and electroplated copper layers 1064, as shown in Figure 21. When the second conductive layer (copper seed layer 1054 and electroplated copper layer 1064) is ground, the first conductive layer (nitrogen The tantalum layer 1022 and the lithium nitride layer 1032) covering the lithium-containing layer are grinding prevention layers. A second chemical mechanical polishing step is performed to remove The button-containing tantalum nitride layer 1032 and tantalum nitride layer 1022, which are dropped on the second planarization insulation layer 1116 outside the opening 12, are shown in Fig. 22. The connection line 1084 includes an interconnection line and contacts therein. A doped region! A contact region of 〇04. Although not shown, the interconnect line 1082 is electrically connected using the contact region. 'FIG. 22 is not shown. Continue the process steps to form the completed structure shown in FIG. At this time, the third etch stop layer 1910 is formed on the connection lines 1082 and 1084, the third is dry and the second layer 1912 is formed in the third etch stop layer 1911. The fourth etch stop layer 1 9 14 is formed on the third planarization isolation layer χ 9 12. Although not shown, a fourth planarizing insulating layer is also formed. The second ILD layer containing the third etch stop layer 1910, the third planarization barrier layer 1912, and the fourth etch stop layer 1 9 14 will be defined with a pattern 'will be used in customary systems-24 This paper standard applies Chinese national standards (CNS) A4 specification (210 '乂 297mm) Shen Yi --------. Order ---- ^ --- * ~ (Please read the notes on the back before filling this page) A7 B7 Ministry of Economy Printed by Zhongluo Luzhun Bureau under the Industrial and Consumer Cooperatives. 5. Description of the invention (where the 22nd line is formed, interconnecting grooves and contact openings are formed. After the openings are formed, a nitrided boat layer is deposited], followed by a large amount of nitrogen containing the boat. Lithium layer 1932, copper seed layer 1954, and electroplated copper layer 1964. After appropriate polishing treatment steps, a passivation layer 1098 is formed to complete the entire device. Although not shown, additional interlayer dielectric layers can be fabricated And other% sexual connection lines' are not shown in Figure 23. The present invention has other advantages. In addition to using Zhao, other South temperature metals can also be used to form the barrier layer / adhesive layer. For example, tungsten, molybdenum can be used. Or a similar high-temperature metal to replace rhenium. In yet another embodiment, The tantalum nitride tantalum nitride layer 1032 and the tantalum nitride layer 1022 can be formed by combining different semiconductor atoms. For example, silicon nitride and silicon nitride buttons containing tantalum can be formed. In addition, germanium atoms or Silicon and germanium atoms. When forming high-temperature metal _ semiconductor-nitrogen compounds, the sputtering targets usually used for sputtering include high-temperature metal and semiconductor atoms. In a specific embodiment, a silicide button can be used, and argon and Nitrogen is sputtered to form a silicon tantalum nitride by sputtering a hafnium silicide target. A silicon nitride button layer containing a poly button can be formed in a similar manner to a button nitride layer containing a poly button, in which the nitrogen is stopped and argon is introduced into the sputtering. In the target, the nitrided residue remaining on the dried surface is removed from the previous sputtering. In another embodiment, a pure button layer, a tantalum-rich nitride button layer, or a silicon nitride tantalum layer It can be formed before the hafnium nitride layer 1022 is formed. In the embodiment, the lithium target can be sputtered for a short time to form the initial button layer, and then the tantalum nitride layer and the button-containing nitrogen Tantalum layer. This embodiment can form a better The contact window gives the substrate and silicide, because the tantalum layer will appear on the surface (please read the precautions on the back before filling this page). -25- A7 ______B7 1 > V. Description of the invention (23) Original oxidation In other structures, a single invalid connection line may be formed. At this time, the thickness of the tantalum nitride layer 1022 and the tantalum nitride layer 1032 may increase compared with the double invalid structure shown in FIG. 5-23. In still another embodiment, a single copper layer formed by chemical vapor deposition or physical vapor deposition can be used to replace the combination of the steel seed layer 1054 and the electroplated copper layer 1064. When this layer is formed, heating can be performed The substrate allows copper to flow into the required openings. At this time, the substrate or wafer temperature is usually in the range of about 350-450 1. Be very careful because if the temperature is too high, the integrity of the M nitride barrier layer will be destroyed. The temperature during deposition is high enough to allow the copper to flow, but not so high that the integrity of the lithium nitride barrier layer is destroyed. These copper flowing embodiments can achieve advantages such as reduced milling time and depression. The embodiments described above provide advantages not found in conventional techniques. The hafnium nitride layer is a good barrier layer, but the use of hafnium-containing hafnium nitride layers to connect the hafnium nitride layer will provide better adhesion, especially for chemical vapor deposited copper layers. Phase = If the nitrogen content of the chemical layer 1032 is too high, cyanide will be formed. It is printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economy and Economics. Keeping the nitrogen concentration on the surface in contact with copper to a very low level will achieve better adhesion. Q Another advantage of improving adhesion will also achieve better contact resistance. In addition, the embodiment of the present invention will reduce the effect of electron drift, For example, the existing equipment can be used instead of special: or; marginal process. The above description has been described with reference to specific embodiments of the present invention --- 26-this paper ~ ~ --- --- 5. Description of the invention (24 A7 B7. However, to the following inventions and changes. Therefore, it is not limited. It has already happened, and it will happen to anyone who has a general technical level with the patent scope of any element, t _. ^ Is not biased Under the scope of the Miscellaneous Patent No. 4 patent, different revisions can be made, and the description and drawings can only be regarded as illustrative references. Any modification should be included in each reference of the present invention. Specific embodiments illustrate its benefits and other advantages. However, any benefits and advantages or become more significant benefits, advantages, will not be interpreted as any strict, required, or basic characteristics or constitutive elements. I-- ------- Drawing -------, 玎 ---- ^ --- A-, Γ (Please read the notes on the back before filling this page) Printed by Consumer Cooperatives -27- This paper size applies to Chinese National Standard (CNS) A4 (210X297 mm)