201121043 六、發明說明： 【發明所屬之技術領域】 尤指一種金氧 本發明係關於一種金屬閘極結構及其製作方法 半導體電晶體之金屬閉極結構。 【先前技術】 春 斜導體產業中，由於多晶f材料具有抗熱性質，因此在製作 傳統金屬氧化物半導體_s)電晶料通常會細多晶賴料來作 為電晶體_極電極，使其源極與汲極區域得以在高溫下一起進行 退火。其次，由於多晶石夕能夠阻擋以離子佈植所推雜之原子進入^ 道區域，因此在__化之魏容魏形成自行解的源極與沒 極區域。 然而，隨著半導體元件的尺寸觸微縮，傳統M〇s電晶體的 結構開始面臨到新的考驗。首先，與大多數金屬材料相比，多晶石夕 間極是以較高電阻值的半導體材料所形成，因此多晶㈣極所提供 的操作速率會比金屬閘極為低。此外，多晶㈣極容易產生空乏效 應(d—effect)。由於摻雜濃度上的限制，當多晶石夕間極受到偏 壓時丄缺乏載子’使靠近多晶·極與閘極介電層的介面上就容易 產生空乏區。此空乏效應除了會使等效的閘極介電層厚度增加，又 同時造成閘極電容值下降，進而導致元件驅動能力衰退等困境。故 目前便有新的間極材料被研製生產，例如利用功函數(鑛心㈣ 201121043 金屬來取代傳統的多晶矽閘極。 其-人，隨著半導體元件的尺寸持續微縮，傳統河〇;5電晶體的 間，介電層厚度也隨之漸薄。然而，微_二氧化料或氮氧化石夕 層合易導致電子的穿遂效應(她⑼叩effect)，因而產生漏電流過大 的^理限制。為了有效延展邏輯元件的世代演進，MOS電晶體的閘 極介電層開始採用«介電常數(以下簡稱為High-K)材料，以降低物 理極限厚度’並且在侧的等錄化厚度(equivalent oxide thickness ’以下簡稱為Ε〇τ)下，有效降低漏電流，並達成等效電容 以控制通道開關等優點。 功函數金屬閘極一方面需要與電晶體搭配，另一方面則 品/、PMOS電晶體搭配，因此使得相關元件的整合技術以及製程控 制更形複雜，且各材料的厚度與成分控制要求亦更形嚴苛。在這個 厭可的製程環境下，如何製作良好之功函數金屬閘極，以提升m〇s 電晶體之運作效果，仍為現今一重要課題。 【發明内容】 有鑑於此，本發明提供一種金屬閘極結構及其製作方法，使金 屬閘極結構所含之氮/鈦元素比例會隨著厚度由下而上減少 ，可減少 M〇S 電晶體的等效氧化厚度（equivalent oxide thickness，EOT)，並 且維持有效功函數（effective work function，EWF )。 201121043 根據本發明之較佳實施例，本發明提供一種金屬閘極結構，包 括一閘極介電層、一設置於閘極介電層上之第一功函數金屬層，以 及一設置於第一功函數金屬層上之第二功函數金屬層。其中，第一 與第一功函數金屬層本質上均包括（being substantially composed by) —第一元素與一第二元素，且第一功函數金屬層具有至少一種 與第二功函數金屬層不同之物理性質。 φ 根據本發明之較佳實施例，本發明另提供一種金屬閘極結構， 包括一閘極介電層，以及一設置於閘極介電層上之功函數金屬層。 其中，功函數金屬層本質上包括-第—元素與―第二元素，且功函 數金屬層具有至少一種隨著厚度改變之物理性質。 根據本發明之較佳實施例，本發明另提供一種形成金屬閘極結 構之方法，包括形成-開極介電層，以及進行一製程以於間極介電 層上形成-功函數金屬。其+，形成該功函數金屬之該製程包括改 • 變至少一製程參數》 據此，本發明可減少HK/MG⑴㈣娜她㈣她丨购） 型MOS電晶體的E0T ’並且維持平帶電壓（flatband她㈣不 致增加，進而提供良好之有效功函數。 【實施方式】 下文依本發明之金屬閘極結構及其製作方法，特舉實施例配合 201121043 所附圖式作詳細說明，但所提供之實施例並非用 ， 蓋的範圍，椒流卿___ ^ 方法步驟重新組合之執行流程，所產生具有均等功效的^任= ^㈣所涵蓋的細。其帽式僅以說明為目的，並未依照原= 本，明之金屬閉極結構可以為包含多晶石夕與金屬之201121043 VI. Description of the Invention: [Technical Field] The invention relates to a metal gate structure and a manufacturing method thereof. The metal closed-pole structure of a semiconductor transistor. [Prior Art] In the spring oblique conductor industry, since the polycrystalline f material has heat resistance, the conventional metal oxide semiconductor _s) is usually finely crystallized as a transistor _ pole electrode. The source and drain regions are annealed together at high temperatures. Secondly, since the polycrystalline stone can block the atoms impregnated by the ion implantation into the ^ region, the Wei Rongwei in the __ is formed into the source and the annihilation region of the self-solving. However, as the size of semiconductor components has shrunk, the structure of conventional M〇s transistors has begun to face new challenges. First, the polycrystalline whisker is formed of a higher resistance semiconductor material than most metallic materials, so the polycrystalline (tetra) pole provides an operating rate that is much lower than that of the metal gate. In addition, polycrystalline (tetra) is highly prone to d-effect. Due to the limitation of the doping concentration, when the polycrystalline spine is subjected to a bias voltage, the germanium lacks a carrier, so that a depletion region is easily generated on the interface close to the polycrystalline electrode and the gate dielectric layer. This depletion effect not only increases the thickness of the equivalent gate dielectric layer, but also causes the gate capacitance value to decrease, which leads to the dilemma of component drive capability degradation. Therefore, new inter-polar materials have been developed and produced, for example, by using the work function (minexin (4) 201121043 metal to replace the traditional polysilicon gate. Its - human, as the size of semiconductor components continues to shrink, the traditional rivers; 5 electricity Between the crystals, the thickness of the dielectric layer is also gradually thinned. However, the micro-dioxide or the oxynitride layer is easy to cause the electron-piercing effect (her (9) 叩 effect), thus causing excessive leakage current. Limitation. In order to effectively extend the evolution of logic components, the gate dielectric layer of MOS transistors begins to use «dielectric constant (hereinafter referred to as High-K) material to reduce the physical limit thickness' and the thickness of the recording on the side. (Equivalent oxide thickness 'hereinafter referred to as Ε〇τ), effectively reducing the leakage current, and achieving the equivalent capacitance to control the channel switch, etc. The work function metal gate needs to be matched with the transistor on the one hand, and the product on the other hand / The PMOS transistor is matched, which makes the integration technology and process control of related components more complicated, and the thickness and composition control requirements of each material are more stringent. In this disgusting process Under the circumstances, how to make a good work function metal gate to improve the operation effect of the m〇s transistor is still an important issue today. [Invention] In view of this, the present invention provides a metal gate structure and its fabrication. In this way, the proportion of nitrogen/titanium element contained in the metal gate structure is reduced from bottom to top, which can reduce the equivalent oxide thickness (EOT) of the M〇S transistor and maintain an effective work function. According to a preferred embodiment of the present invention, the present invention provides a metal gate structure including a gate dielectric layer and a first work function metal disposed on the gate dielectric layer. a layer, and a second work function metal layer disposed on the first work function metal layer, wherein the first and first work function metal layers substantially include: a first element and a second An element, and the first work function metal layer has at least one physical property different from the second work function metal layer. φ According to a preferred embodiment of the present invention, the present invention A metal gate structure includes a gate dielectric layer and a work function metal layer disposed on the gate dielectric layer, wherein the work function metal layer essentially includes a -th element and a "second element", and The work function metal layer has at least one physical property that varies with thickness. According to a preferred embodiment of the present invention, the present invention further provides a method of forming a metal gate structure, including forming an open dielectric layer, and performing a process Forming a work function metal on the interpolar dielectric layer. The process of forming the work function metal includes changing at least one process parameter. According to the present invention, the present invention can reduce HK/MG(1)(4) Na (4) her purchase ) MOS transistor E0T 'and maintain the flat band voltage (flatband she (4) does not increase, and thus provide a good effective work function. [Embodiment] Hereinafter, the metal gate structure and the manufacturing method thereof according to the present invention, the specific embodiment is described in detail in conjunction with the drawings of 201121043, but the embodiment provided is not used, the scope of the cover, the pepper flow ___ ^ Method steps recombine the execution process, resulting in a uniform effect of ^ Ren = ^ (d) covered by the fine. The cap type is for illustrative purposes only, and is not in accordance with the original = this, the metal closed-end structure of the Ming may be composed of polycrystalline stone and metal
,(~a，^ gate) ? , ,tJ 可以先沉積顧數金勒'、4之域方式 从金^雜。或者，先娜崎卿賴置晴體 完成電晶體之後，侧刻方式掏空虛置開極導 -後，域再填人所需的金屬，以製作金屬閑極。 請參照第】圖至第6圖，第】圖至第6圖為本發明第一較佳實 施例製作具有金屬·66之電晶體的示賴，其中第—較佳實施例 係利用虛置閘極之方式形成金屬閘極.圖式中相同的元件或部位 沿用相同的符號來表示’且圖式僅以說明為目的，並未依照原尺寸 作圖。、如第1圖所示，首先提供一基底12，例如-石夕基底、含石夕基 底、或-絕緣層上覆石夕(sincon_〇n_insulat〇r ; s〇l)基底等。在基底Η 中可定義至少-電晶體區14,並形成至少—淺溝隔離㈣ 她-，以下簡稱為抑結構18。電晶體區14用以形成pM〇s 電晶體、NMOS電晶體或CM〇s電晶體，而阳結構Μ用以隔離 201121043 電晶體區14。接著在基底12表面形成一閘極介電層22，並於問極 介電層22上形成一選擇性之遮蓋層24。在本實施例中，閘極介電 層22可包含矽酸铪氧化合物(HfSi〇)、矽酸給氮氧化合物 (HfSlON)、氧化铪(HfO)、氧化鑭(LaO)、鋁酸鑭(LaAlO)、氧化鍅 (Zr〇)、矽酸鍅氧化合物(ZrSiO)、鍅酸铪(HfZrO)等高介電常數介電 層或其組合。而於其他實施例中，閘極介電層22也可包含二氧化 矽層或氮氧化矽層等介電層。 —請繼續參閱第2a®與第2b圖，其後進行一製程，例如進 行物理氣相沉積(physical v叩㈣仍出〇n，製程化 二轧相’儿積（chemical vap〇r dep〇siti〇n，CVD)製程或原子層 沉積（at〇mic iayerdep〇siti〇n，ald)製程以於間極介電層u ^形成—伽力純金駿合層26。於職功函數金聽合層26之 發明可改變至少—製程參數，以使本發明之功函數金屬 声可為多層結構。亦即於本實施例中，形成功函數金屬複合 個實際上可包括複數個製程階段（_fashi〇n)，各製程 ^彼程參數Μ上維持但相鄰之各製程階段之製程 %之製程中停止供應製程氣體、 參數_進：體率^ 屬複合層26實際上包括複數個材料層，相鄰之==不函= 201121043 物理性質’林材料層本身可具有均—之物理性質不會隨著厚度之 變化而改變。 更詳細地說，本發明可如第2a圖所示先形成第-功函數 金屬層26a設置於間極介電層Μ上，之後如第％圖所示，形成 第二功函數金屬層26b設置於第一功函數金屬層26a上，藉此本實 施例一之功函數金屬複合層26即可包括一第一功函數金屬層細與 -第二功函數金屬層挪。其中，本發明之第一與第二功函數金屬 層26a 26b本質上均包括（being substantia办⑺叫) 一第— 兀素與-第一tl素’其中第一元素較佳為鈦G—)或组 (tantalum) ’❿第二元素較佳為氮(nitr〇gen)。亦#，本發明之功函 數金屬複合層26較佳可&域倾(TiN)錢化師㈣，且尤以 化鈦為佳。 氣 第-與第二功函數金屬層26a、26b之主要不同之處在於，由於 形成第-功函數金屬層加之製程參數與形成第二功函數金屬層 2你之製程參數不同，因此第一功函數金屬層—可以具有至少— 種與第二功函數金屬層26b不同之物理性質，例如分子式（譬如元 素成分比例）、密度、電阻係數（resistivity)或晶體排列方向。例如 於本實施例中，形成第一功函數金屬層26&之製程參數如下：直节 功率較佳是約介於8GG瓦至謂瓦，交流辨她是約介於州& 瓦至960瓦’氬氣流量較佳是約介於16 sccm (standard, (~a, ^ gate) ? , , tJ can be deposited first in the way of the number of Jinle ', 4 domain from the gold ^ miscellaneous. Or, after Nagasaki Qing Lai Qing Qing completed the crystal, the side engraving method was hollowed out to open the pole guide - after that, the domain was filled with the required metal to make the metal idle pole. Referring to FIG. 6 to FIG. 6 , FIG. 6 to FIG. 6 are diagrams showing the fabrication of a transistor having a metal 66 in accordance with a first preferred embodiment of the present invention, wherein the first preferred embodiment utilizes a dummy gate. The metal gate is formed in the extreme manner. The same elements or parts in the drawings are denoted by the same reference numerals, and the drawings are for illustrative purposes only and are not drawn according to the original dimensions. As shown in Fig. 1, a substrate 12 is first provided, for example, a stone base, a stone base, or an insulating layer, a sincon_〇n_insulat〇r; s〇l substrate. At least - the transistor region 14 may be defined in the substrate , and at least - shallow trench isolation (four) she-, hereinafter referred to as the suppression structure 18, may be formed. The transistor region 14 is used to form a pM〇s transistor, an NMOS transistor or a CM〇s transistor, and the anode structure is used to isolate the 201121043 transistor region 14. A gate dielectric layer 22 is then formed on the surface of the substrate 12, and a selective mask layer 24 is formed on the dielectric layer 22. In the present embodiment, the gate dielectric layer 22 may include niobium oxynitride (HfSi〇), niobium oxynitride (HfSlON), hafnium oxide (HfO), lanthanum oxide (LaO), lanthanum aluminate ( A high-k dielectric layer such as LaAlO), yttrium oxide (Zr〇), yttrium oxynitride (ZrSiO), yttrium ruthenate (HfZrO), or a combination thereof. In other embodiments, the gate dielectric layer 22 may also include a dielectric layer such as a hafnium oxide layer or a hafnium oxynitride layer. - Please continue to refer to Figures 2a and 2b, followed by a process such as physical vapor deposition (physical v叩 (4) still 〇n, process two-rolled phase 'chemical product (chemical vap〇r dep〇siti) 〇n, CVD) process or atomic layer deposition (at〇mic iayerdep〇siti〇n, ald) process for the inter-electrode layer u ^ formation - Galileo pure gold Jun layer 26. The job function function gold listening layer The invention of 26 can change at least the process parameters such that the work function metal sound of the present invention can be a multi-layer structure. That is, in the present embodiment, the shape success function metal composite can actually include a plurality of process stages (_fashi〇n The processing of the process gas is stopped in the process of maintaining the process parameters of each of the adjacent process stages, and the parameter _in: body rate ^ the composite layer 26 actually includes a plurality of material layers, adjacent == 不函 = 201121043 Physical Properties 'The material layer of the forest material itself can have a physical property that does not change with the thickness. In more detail, the present invention can form the first work as shown in Figure 2a. The functional metal layer 26a is disposed on the inter-electrode dielectric layer As shown in the % diagram, the second work function metal layer 26b is formed on the first work function metal layer 26a, whereby the work function metal composite layer 26 of the first embodiment can include a first work function metal layer. And a second work function metal layer, wherein the first and second work function metal layers 26a to 26b of the present invention essentially include (being substantia (7) called) a first - a halogen and - a first one The first element is preferably titanium G—) or tantalum '❿. The second element is preferably nitrogen (nitr〇gen). Also, the work function metal composite layer 26 of the present invention is preferably <RTI ID=0.0>> The main difference between the gas-first and second work function metal layers 26a, 26b is that since the process parameters for forming the first work function metal layer and the process parameters for forming the second work function metal layer 2 are different, the first work is performed. The functional metal layer—which may have at least one physical property different from the second work function metal layer 26b, such as a molecular formula (such as an elemental composition ratio), a density, a resistivity, or a crystal alignment direction. For example, in the present embodiment, the process parameters for forming the first work function metal layer 26 & are as follows: the straight node power is preferably about 8 GG watts to the watt, and the exchange is about between about state & watts to 960 watts. 'The flow rate of argon is preferably about 16 sccm (standard
Centimetersperminute，每分鐘標準毫升）至24sccm，氮氣流e量較佳 201121043 是約介於32 seem至48 SCCm，其中氮氣流量/氬氣流量之比值大體 上介於1.6至2.4之間’使得第一功函數金屬層26a可以具有較高之 氮/鈦元素比例（N-rich，亦稱為富氮），更明碟地說，第一功函數金 屬之鈦/氮元素比例大體上大於〇 8。再者，形成第二功函數金屬層 26b之製程參數如下：直流功率較佳是約介於8〇〇 .瓦至12〇〇瓦交 流功率較佳是約介於640瓦至960瓦，氬氣流量較佳是約介於16 seem至24 seem，氮氣流量較佳是約介於16 sccm至24 sccm，其中 φ氮氣流量/氬氣流量之比值大體上介於0.8至1.2之間，使得第二功 函數金屬層26b可以具有較高之鈦/氮元素比例（Ti_rich，亦稱為富 .鈦），更明確地說’第二功函數金屬之鈦/氮元素比例大體上小於〇9。 此處所謂之Ti-rich係表示功函數金屬層呈現金屬狀態（metalUc mode)，亦即功函數金屬層表面之主要元素仍為鈦元素；而所謂之 N_rich係表示功函數金屬層呈現化合狀態（poison mode)，亦即功函 數金屬層表面已被氮化合為氮化鈦。因此，隨著製程氣體氮氣之氣 _體流量逐漸增加’所形成之功函數金屬層的氮元素含量也會逐漸增 加’因此氮氣流量越大越容易形成化合狀態之功函數金屬層，反之 亦然。 針對第一與第二功函數金屬層26a、26b之材料選擇’於製作 CM〇S電晶體元件等具有NMOS電晶體與PMOS電晶體之狀況 下’金屬閘極之費米能位（fermi level)較佳是接近矽的中間能階 (mid-gap)’以便於調整電晶體與PM〇s電晶體的臨界電壓 201121043 (Vth) ’使nm〇s電晶體與脱⑺電晶體的臨界電壓能相匹配。 此本發明之金屬閘極材料較佳是具備良好熱穩定性、阻播性與 附著性，使_材料本林錄人基減介電种造成污染，不易 讓雜質穿透擴散，且不易祕1合上述條件之材料均可作為第一 與第二功函數金屬層26a、26b之材料。例如於其他實施例中，功函 數金屬複合層26亦可包括其他金紐料層，像是氮化鶴(丽)。 一般而言，PMOS適合之功函數數值約5 2，而適合之 功函數數值約4.1卜根據本發明之研究，具有金屬·之腿^電 晶體較佳使时狀氮化鈦作為中間能階之材料，而本發明所形成 =金屬閘極將可適·ΡΜΟδ電晶體、NM〇s電晶體或cm〇s電 晶體’尤其本發明使用了 N_rich之氮化鈦可以提供pM〇s較適合之 功函數數值。然而需注意的是，N_rich之氮化鈦同時也會導致較大 的等效氧化厚度（equivalent oxide thickness，Ε〇τ)。這是因為當氮元 素或氧it料料電物質㈣MOS之功絲金屬（丽k㈣齡 metal)時，會導致M0S呈現iE〇T增加，偏離預定值。因此，本 發明於Ν-nch之氮化鈦上再形成-層Ti_rich之氮化鈦，藉由Ti_rich 之氮化鈦可以提供縮減EOT之效果。 針對製程參數之部分，本發明研究發現，隨著製程氣體氮氣之 氣體流量逐漸增加，所形成之功函數金屬層的物理性質，例如元素 成分比例、密度、電阻係數、晶體排列方向與缺陷密度（defect density) ’均可能受到影響。其中，製程氣體氮氣之氣體流量接近 201121043 20 seem時，功函數金屬層的物理性質會具有非線性之轉變，稱之為 轉換點（turning point)。轉換點產生之一可能原因在於，當氮氣之、 流量接近例如2〇seem時，晶格排列方向會傾向於(1，丨，丨)與(2,〇,〇) 之間作轉換。當製減航氣之氣财量小於2Gseem時，功函數 金屬層的電阻係數與EOT均係隨著氮氣流量之增加而逐漸降低，且 功函數金屬層的平帶電壓之增加量係隨著氣氣流量之增加而漸增； 當製程氣體氮氟之氣體流量大於2〇 scem時，功函數金屬層的電阻 籲係數與EOT均係隨著氮氣流量之增加而逐漸增加，且功函數金屬層 的平帶電壓之增加量係隨著氮氣流量之增加而趨緩。所以，當製程 氣體氣氣之氣體流量接近20 sccm時（接近轉換點時），所形成之 Ti-rich之氮化鈦對於薄化Ε〇τ可以提供更佳之效果。 此外，由於CVD製程所形成之氮化鈦通常較容易具有雜質， 而高功率PVD製程之電漿可能會影響到High_K材料本身因此尤 以低功率之PVD製程所形成之氮化鈦可具有更佳之品質。 按卜來如弟3圖所示，於功函數金屬複合層26上形成一多晶 層28以及—遮罩層3G。多晶料制來做為—犧牲層，可^ T具有任何摻質的多祕材料、具有摻質的多晶石夕材料 ^曰曰石夕或其他材料所構成；而鮮層3Q射包含二氧切陶 氮化石夕_、碳切(Sic)或氣氧化鄉趣）。接著，形成一圖案 先阻層(圖絲)麵罩㈣上，並個 一圖案轉雜程，絲科_罩層M，再叫次侧 201121043 步驟去除部分的多晶石夕層28 22，並剝除此圖案化光阻層， 、功函數麵複合層％及問極介電層 以於電晶體區14形成_虛置閘極D。 =後如4 4圖所示，在電晶體區14選擇性進行一輕推雜製程， 以械所需_摻_與祕。舉啦說，本發日柯先覆蓋一 化光阻層(圖未利在電晶體區14以相輯，然光Centimetersperminute, standard milliliters per minute to 24sccm, the amount of nitrogen flow e is better 201121043 is about 32 seem to 48 SCCm, where the ratio of nitrogen flow / argon flow is generally between 1.6 and 2.4' The functional metal layer 26a may have a higher nitrogen/titanium element ratio (N-rich, also referred to as nitrogen-rich), and more specifically, the titanium/nitrogen element ratio of the first work function metal is substantially greater than 〇8. Furthermore, the process parameters for forming the second work function metal layer 26b are as follows: the DC power is preferably about 8 〇〇. watt to 12 〇〇 watts, preferably about 640 watts to 960 watts, of argon. Preferably, the flow rate is between about 16 seem and 24 seem, and the nitrogen flow rate is preferably between about 16 sccm and 24 sccm, wherein the ratio of the φ nitrogen flow rate to the argon flow rate is substantially between 0.8 and 1.2, such that the second The work function metal layer 26b may have a higher titanium/nitrogen element ratio (Ti_rich, also referred to as rich titanium), more specifically 'the second work function metal titanium/nitrogen element ratio is substantially less than 〇9. The Ti-rich system here means that the work function metal layer exhibits a metal state (metalUc mode), that is, the main element of the work function metal layer surface is still a titanium element; and the so-called N_rich system indicates that the work function metal layer exhibits a compound state ( Poison mode), that is, the surface of the work function metal layer has been nitrided into titanium nitride. Therefore, as the process gas nitrogen gas gradually increases, the nitrogen element content of the work function metal layer is gradually increased. Therefore, the larger the nitrogen gas flow rate, the easier it is to form the work function metal layer in the combined state, and vice versa. For the materials of the first and second work function metal layers 26a, 26b, the fermi level of the metal gate is selected in the case of fabricating a CM〇S transistor or the like having an NMOS transistor and a PMOS transistor. It is preferable to be close to the mid-gap of the crucible to facilitate adjusting the threshold voltage of the transistor and the PM〇s transistor 201121043 (Vth) 'to make the threshold voltage phase of the nm〇s transistor and the de(7) transistor match. The metal gate material of the present invention preferably has good thermal stability, blocking property and adhesion, so that the material of the material is reduced by the dielectric type, and the impurity is not easily penetrated and spread, and it is not easy to secret. Materials constituting the above conditions can be used as the material of the first and second work function metal layers 26a, 26b. For example, in other embodiments, the work function metal composite layer 26 may also include other gold metal layers, such as a nitride crane. In general, PMOS is suitable for a work function value of about 52, and a suitable work function value is about 4.1. According to the study of the present invention, a metal having a leg is preferably used as an intermediate level. Material, and formed by the invention = metal gate will be suitable for ΡΜΟ δ crystal, NM 〇 s transistor or cm 〇 s transistor 'In particular, the invention uses N_rich titanium nitride to provide pM 〇s more suitable work Function value. However, it should be noted that N_rich titanium nitride also leads to a large equivalent oxide thickness (Ε〇τ). This is because when the nitrogen element or the oxygen-it material (4) MOS is a wire metal (L), it causes the MOS to exhibit an increase in iE〇T, which deviates from the predetermined value. Therefore, the present invention further forms a Ti_rich titanium nitride on the tantalum-nickel titanium nitride, and the Ti_rich titanium nitride can provide an effect of reducing EOT. For the part of the process parameters, the present inventors have found that as the gas flow rate of the process gas nitrogen is gradually increased, the physical properties of the formed work function metal layer, such as elemental composition ratio, density, resistivity, crystal alignment direction and defect density ( Defect density) 'may be affected. Among them, when the gas flow rate of the process gas nitrogen is close to 201121043 20 seem, the physical properties of the work function metal layer will have a nonlinear transformation, which is called a turning point. One of the possible reasons for the conversion point generation is that when the flow rate of nitrogen is close to, for example, 2 〇 seem, the lattice alignment direction tends to be converted between (1, 丨, 丨) and (2, 〇, 〇). When the gas yield of the gas reduction is less than 2Gseem, the resistivity and EOT of the work function metal layer gradually decrease with the increase of the nitrogen flow rate, and the increase of the flat band voltage of the work function metal layer is accompanied by the gas. When the gas flow rate of the process gas nitrogen fluoride is greater than 2〇scem, the resistance coefficient and the EOT of the work function metal layer gradually increase with the increase of the nitrogen flow rate, and the work function metal layer The increase in the flat band voltage tends to decrease as the nitrogen flow rate increases. Therefore, when the gas flow rate of the process gas is close to 20 sccm (close to the switching point), the Ti-rich titanium nitride formed can provide a better effect for thinning Ε〇τ. In addition, since the titanium nitride formed by the CVD process is generally more likely to have impurities, the plasma of the high-power PVD process may affect the High_K material itself, so the titanium nitride formed by the low-power PVD process may be better. quality. As shown in Figure 3, a polycrystalline layer 28 and a mask layer 3G are formed on the work function metal composite layer 26. The polycrystalline material is made as a sacrificial layer, and can be composed of a multi-secret material having any dopant, a polycrystalline stone material having a dopant, or a stone material or other materials; and the fresh layer 3Q is composed of two Oxygen cut pottery nitride 夕, carbon cut (Sic) or gas oxidation of the fun). Next, a pattern of the first resist layer (picture) mask (four) is formed, and a pattern is transferred to the pattern, the silk layer _ cover layer M, and then the secondary side 201121043 step removes part of the polycrystalline layer 28 22, and The patterned photoresist layer, the work function surface recombination layer %, and the interposer dielectric layer are stripped to form a dummy gate D in the transistor region 14. After the picture is shown in Fig. 4, a lightly pushing process is selectively performed in the transistor region 14, which is required for the device. To put it bluntly, this hair is first covered with a photoresist layer (Fig. unprofitd in the transistor area 14 to match the series, then light
阻f作縣進行—離子佈觀程，糾雜質植人電晶體=4 之虛置閘極32 _的基底12中，以於形成職〇s電晶體之輕推雜 源極與錄34。或者，將p型摻質植人電晶體區14之虛置問㈣ 兩側的基底12中，以戦P腦電晶體之輕摻雜源極與沒極別。 另外，在形成輕摻雜源極與汲極34之前還可於虛置閉極％之 側壁先开^成襯墊層（Spacer，圖未示）C 隨後進行側壁子製程，例如先氧化多晶石夕層28的表面或以沉積 的方式形成-氧化石夕層％，接著再沉積一氮化石夕層4〇並利用敍刻 方式形成由氧化石夕層38與氮化石夕層4〇所構成的側壁子，設置於虛癱 置閘極32的側壁周圍。針對不同導電類型之嶋8電晶體，另可以 選擇性形成應力覆蓋層（stressed cap iayer)或進行一選擇性蟲晶成 長(selective ePitaxial growth，SEG)製程，以對PM〇s電晶體的通道區 域施加壓縮應力(COmpressivestrain)，或腿〇s電曰曰曰體的通道區域施 加拉伸應力(tensile strain)。 心後進订-重摻雜離子佈植製程，以分卿成所需的源極/汲極The resistance f is made in the county—the ion cloth observation process, and the impurity implanted in the human crystal = 4 of the dummy gate 32 _ in the substrate 12, so as to form the light source and the source of the transistor. Alternatively, the p-type dopant implanted in the human crystal region 14 is placed in the substrate 12 on both sides of the substrate, and the lightly doped source of the 戦P brain transistor is indistinguishable. In addition, before forming the lightly doped source and the drain 34, a sidewall layer (Spacer, not shown) may be opened on the sidewall of the dummy gate, and then a sidewall process, such as first oxidizing polycrystal, is performed. The surface of the stone layer 28 is formed by depositing - the oxidized stone layer %, and then depositing a nitriding layer 4 〇 and forming a layer consisting of the oxidized stone layer 38 and the nitride layer 4 The sidewalls are disposed around the sidewalls of the dummy gate 32. For different conductivity types of 嶋8 transistors, a stress cap layer or a selective ePitaxial growth (SEG) process can be selectively formed to pass the channel region of the PM〇s transistor. A compressive stress is applied, or a tensile strain is applied to the channel region of the leg 〇s. After the heart is ordered - heavy doping ion implantation process to separate the required source / bungee
12 201121043 48 ° 48 作法，惟源極她區域48之離子佈植製程之摻質濃度較高，故不 再贅述。另需注意的是’上述源極/沒極區域的形成製程亦可利 擇性蟲晶成長製程來達成，且側壁子的數目並不限於此。接著 擇性形成-氮切層54在虛置_32、氮化魏㈣基底12表 面氮化⑪層54主要做為後續進行平坦化時之—侧停止層亦可 形成較厚之氮化梦層兼作為應力覆蓋層。然後形成—層間介電層 摩_~rdielectric)56並覆蓋氮化石夕層54。此層間介電層可包^氮 化物、氧化物、碳化物、低介電係數材料其中之一或其組合。 如第5圖所示’進行—化學機械研磨㈣ pohshmg’CMP)製程或-乾餘刻製程，以去除部分的層間介電層 56、氮化韻54及遮罩層3()，並使多㈣層28頂部約略切齊於層 間介電層56表面而受到裸露。接著，進行一選擇性之乾侧或濕: 亥沒私’例如利用氨水(amm〇nium hydr〇xide，卿〇幵)或氫氧化四甲 敍(Tetramethylammonium Hydroxide，TMAH)等蝕刻溶液來去除電晶 體區14中的多晶石夕層28但不餘刻詹間介電層兄，以在電晶體區μ 形成-開口 58 ’同時暴露出設於各開口 58底部的功函數金屬複合 層26。此處CMP製程或乾餘刻製程可以設定一預定之進行時間， 藉以直接暴露出多晶石夕層28。或者，也可以根據不同材料層之研磨 狀況而將CMP製程區分為翅階段，例如先以遮罩層3()表面作為 第一階段之研磨終止點，依序研磨層間介電層56與氮化矽層“直 至遮罩層30表面’此時可偵測出研磨之材料層變化而停止第一階段 201121043 之研磨，之後再進行第二階段向τ研磨遮罩層3q直至多祕層^ 表面又之研磨可以設定是持續_預定時間或者是可以伯 測研磨之材㈣變化作為研祕德。 、 於其他實關t ’也可以先進行CMp触聽侧製程層間 電層56魅切層54直至遮罩層3G表面，再彻濕侧製 除遮罩層30。 接著如第6圖所不，填入一由低電阻材料所構成的導電層64 在電晶體區14的功函數金屬複合層26上並填關口 58。在本實施 幻中導電層64可她、鶴、鈦|g合金(TiAi)紐觸化物(咖沾 tungsten phosphide ’ CGWp)等低電阻材料所構成。之後，進行另一 化學機械研磨製程，去除部分的導電層64，⑽成具有金屬問極結 構66的M0S電晶體。 值％·υ疋，本發明之金屬閘極結構可以應用於閘極前置 (_彻)製程，換言之，不需定義虛置閉極，而是直接使用包 含有不同功函數金屬之轉式結構作為閘極^明確地說，另一較 佳實施例可根據第i圖之步驟，先提供基底12、奶結構Μ、間極 介電層22與選擇性之遮蓋層24。縣，可如第&騎示，先形成 第-功函數金屬層如設置於閘極介電層Μ上，之後如第沘圖所 P形成第.二功函數金屬層挪設置於第一功函數金騎加上， 而不蚊祕置_，即可完成金屬_結構之製作。 201121043 此外’上述第1圖至第6圖的製程進行順序均可依製程需求改 變或調整’例如於其他實施例中，第一與第二功函數金屬層26a、 26b亦可於去除多晶石夕層28之後再形成於遮蓋層上。換言之， 於其他實施财，形成虛㈣_並不沉積功函數金屬層，而在將 虛置閘極之多晶石夕層移除後，才沉積功函數金屬層，故在此一作法 中功函數金屬層亦會存在於閘極側壁。 據此，前述金屬閘極結構66可以作為pM〇s電晶體、刪⑽ 電晶體或C廳S電晶體之閘極。此外，為了更符合不_電晶體12 201121043 48 ° 48 practice, but the concentration of the dopant in the ion implantation process of her region is higher, so I will not repeat them. It should also be noted that the forming process of the source/drain region described above can also be achieved by a selective worm growth process, and the number of sidewalls is not limited thereto. Then, the selective formation-nitrification layer 54 is nitrided on the surface of the dummy _32 and the nitrided (four) substrate 12, and the layer 94 is mainly used for subsequent planarization. The side stop layer can also form a thicker nitride layer. Also serves as a stress coating. Then, an interlayer dielectric layer _~rdielectric 56 is formed and covers the nitride layer 54. The interlayer dielectric layer may comprise one or a combination of a nitride, an oxide, a carbide, a low dielectric constant material. As shown in Fig. 5, the process is performed - chemical mechanical polishing (4) pohshmg 'CMP) process or dry process to remove part of the interlayer dielectric layer 56, nitride 54 and mask layer 3 (), and more (4) The top of layer 28 is approximately flush with the surface of interlayer dielectric layer 56 and exposed. Next, a selective dry side or wet is performed: for example, an etching solution such as ammonia (amm〇nium hydr〇xide, 〇幵 〇幵) or Tetramethylammonium Hydroxide (TMAH) is used to remove the crystal. The polycrystalline layer 28 in the region 14 but not the inter-Jun dielectric layer brother, exposes the work function metal composite layer 26 provided at the bottom of each opening 58 while forming an opening 58' in the transistor region μ. Here, the CMP process or the dry process can be set for a predetermined time to directly expose the polycrystalline layer 28. Alternatively, the CMP process may be divided into wing stages according to the grinding condition of different material layers. For example, the surface of the mask layer 3 (1) is used as the polishing end point of the first stage, and the interlayer dielectric layer 56 and the nitride are sequentially polished. The enamel layer "up to the surface of the mask layer 30" can detect the change of the layer of the polished material and stop the grinding of the first stage 201121043, and then perform the second stage of grinding the mask layer 3q to the τ layer until the surface of the layer is The grinding can be set to be continuous _ predetermined time or can be changed by the material of the test (four) as the research secret. In other real-time t ', the CMp-touching process can also be performed first. On the surface of the cover layer 3G, the mask layer 30 is formed on the wet side. Next, as shown in Fig. 6, a conductive layer 64 composed of a low-resistance material is filled in the work function metal composite layer 26 of the transistor region 14. And fill in the gateway 58. In this embodiment, the conductive layer 64 can be composed of a low-resistance material such as her, crane, titanium, g alloy (TiAi) neotide (Cungbeng phosphide 'CGWp), and then another chemical machine. Grinding process to remove part of the conductive The layer 64, (10) is a MOS transistor having a metal interrogation structure 66. The value of % υ疋, the metal gate structure of the present invention can be applied to the gate pre- (_) process, in other words, it is not necessary to define a dummy The pole is directly used as a gate structure comprising a metal having a different work function. Specifically, another preferred embodiment can provide the substrate 12, the milk structure, and the inter-electrode according to the steps of the first step. The electric layer 22 and the selective covering layer 24. The county may, as in the & riding, first form a first work function metal layer, such as disposed on the gate dielectric layer, and then form a first step as shown in FIG. The two-function metal layer is placed in the first work function Jinjia plus, and the mosquito-free _ can be used to complete the metal _ structure. 201121043 In addition, the process of the above 1st to 6th processes can be performed in sequence. The first and second work function metal layers 26a, 26b may be formed on the cover layer after removing the polycrystalline layer 28, for example, in other embodiments. For example, in other implementations , forming a virtual (four) _ does not deposit a work function metal layer, but will be dummy After the removal of the polycrystalline layer, the work function metal layer is deposited. Therefore, in this method, the work function metal layer is also present on the gate sidewall. Accordingly, the metal gate structure 66 can be used as pM〇s. The gate of the transistor, the (10) transistor or the C- Hall S transistor. In addition, in order to better match the non-electrode
广處理、改變製程條件，或是增減功函數金屬複合 層之材料雜目。修，本發日柯崎對所“PM -或第二功函數金屬層上另行佈值= ❿ 增加功函數數值。 再者，本發明更可於前述之功函數金 功函數金屬層，例如另覆蓋Nn層6上另覆盍其他 第7 摘之功函數金屬層。請參照第7圖， 較佳實施例製作具有金屬閑極脱之電晶體的 ’本實施例可根據第 12、sti結構18、嶋電層22軸性樓/Γ而第* 佳實施例與第一較佳實施例之 a *第一較 成功函數金屬複合層126之製 ’第二較佳實施例形 貫際上可包括至少三健程階段， 201121043 分別用以形成第一、第二與第三功函數金屬層26a、26b、26c。亦 即，利用帛2a ®至第2b圖之步驟形成第一與第二功函數金屬層 26a、26b之後，再改變製程參數而形成第三功函數金屬層26<^其 中，第二功函數金屬層26c與相鄰之第二功函數金屬層26b之至少 一物理性質不同。舉例而言，第三功函數金屬層26c較佳可以為 N-rich之氮化鈦，例如第三功函數金屬層可以與第一功函數金 屬層26a之物雌質與製程參數相同。其後，再利用前述第3圖至 第6圖之步驟形成具有金屬閘極结構166的M〇s電晶體。 據此，第二實_讀佳實施祕係使肖pvD製轉成功函數 金屬複合層126 ’其巾第—製程階段之氮氣流量本質上較佳是約介 於32 seem至48 seem ’製程功率較佳是約介於4〇〇瓦至6〇〇瓦，第 -功函數金屬層26a之實際材料層厚度較佳是約介於招埃 Ung_m)至72埃；第二製概段之氮氣流量較佳是約介於μ 至24 sccm ’製程功率較佳是約介於㈣瓦至12〇〇瓦，第二功 函數金屬㈣b之實際材·厚度較佳是約介於㈣（a%伽⑷ 至96埃；第三製程階段之氮氣流量較佳是約介於^ _至48 露功率較佳是約介於瓦至㈣瓦，第三功函數 加之實際材料層厚度較佳是約介於48埃（angstr〇m)至72埃^ 此一來，本實細可轉供約14_19奈米（nan。输）之e〇t、約 -0.565伏特之平帶電壓，與約4 58之功函數數值。 、 於其他實施例中，形成功函數金屬複合層之製程實際上亦可不 201121043 必被區二，複數個製程階段’而是於單—製程中逐漸調整製程參 ，曰使传IUt鈇複合結構或單—I化鈦層所含之請^元素比例會隨 著厚度由下而上減少。請參照第8圖與第9圖，第請與第9圖為 本發明第二與第四較佳實施例製作具有金屬閘極跡挪之電晶體 的不意圖。 如第8圖所不，第三較佳實施例與第一較佳實施例之一不同之 φ處在於，第二較佳實施例形成功函數金屬複合層226之製程實際上 係逐漸調整製程參數，使功函數金屬複合層226可以僅包含單一瓦 化鈦層，且功函數金屬複合層226所含之至少一種物理性質，例如 氮/鈦元素比例，會隨著厚度由下而上逐漸減少。功函數金屬複合層 226之具體形成方式例如可以逐漸調整製程之氮氣之氣體流量或製 程功率’使得氮氣之氣體流量從40 seem至20 seem逐漸減少。 或者如第9圖所示’第四較佳實施例與第三較佳實施例之一不 鲁同之處在於，第四較佳實施例於逐漸改變製程參數之步驟中，包括 一抽氣降壓（pumpingdown)之步驟。如此一來，第一功函數金屬 層326a本身所含之至少一種物理性質，例如氮/鈦元素比例，會隨 者厚度由下而上逐漸減少；第二功函數金屬層326b本身所含之至少 一種物理性質，例如氮/鈦元素比例，也會隨著厚度由下而上逐漸減 少；但是功函數金屬複合層326之物理性質並非隨著厚度的改變而 連續變化，而是於進行抽氣降壓步驟的前後會有不連續之物理性質 變化；也就是說第一與第二功函數金屬層26a、26b交界處之物理性 17 201121043 質並非連續變化。 於其他實施例中，本發明亦可不_虛置閘極的作法，亦即不 去除多靖28及填充導電層64，而是直接使用多妙層％作為 閘極結構的-雜，或是就於韻數金顧合層m 226、326上形成導 μ來取代多祕層28，以職金屬間極。 綜上所述，本發明之金屬間極結構包含有功函數金屬複合層， 而功函數金屬複合叙物雌f猶著厚纽變。例如，功函數金 屬複合層仰縫顿複合結構，而肢鈦複合結構所含之氛/欽元 素比例會縣賴由下社減少。具體而言，減職合結構可包 含單-氮化鈦層’域化鈦相含之氮/鈦元素_會隨著厚度由下 而上逐漸減少；或者’氮化鈦複合結構可包含複數躲疊之氮化欽 層，其:各氮化鈦層本身所含之氮/鈦元素比例維持固定，而位於最 下層之氮化鈦層的氮/鈦元素比例會比相鄰之氮化鈦層的氮/欽元素 比例更1¾，又或者，氮化鈦複合結構可包含複數層堆疊之氮化欽層， 其中各氮化鈦層本搞含之氮/鈦元纽例會隨著厚度由下而上逐 漸減少’而位於最下層之氮化鈦層的氮/鈦元素比例又會比相鄰之氮 化鈦層的氮/鈦元素比例更高。 藉由上述作法，本發明可具有下列優點。首先，金屬閘極結構 具有低電_無空乏效鮮優點，可tb傳歸晶㈣極具有更好的 几件ϋ動能力與速度。此外於位魏化鈦複合結構最下層之氤 201121043 化鈦層會具有較高的氮/鈦元素比例（N-rich )，因此可以提供pm〇S 較適合之功函數數值。再者，氮化鈦複合結構中同時會具有鈦/氮元 素比例較高（Ti-rich)的氮化鈦層’因此可有效避免氮元素或氧元 素專非導電物質穿過金屬閘極結構之功函數金屬，進而避免EOT偏 離預定值，且較低氮/鈦元素比例的氮化鈦層（製程氣體氮氣之氣體 流量約介於16 seem至24 Sccm)也可以提供較薄之Ε〇τ效果。據 此，本發明可提升MOS電晶體的整體效能。 以上所賴為本發明讀佳實關，凡依本㈣申請專利 所做之均㈣化與修飾，皆闕本發明之涵蓋範圍。 【圖式簡單說明】 第 實把例製作具有金屬閘極之電晶 1圖至第6圖為本發明第一較佳 體的示意圖。Widely process, change process conditions, or increase or decrease the material miscellaneous of the work function metal composite layer. Repair, this day, Keqi on the "PM - or the second work function metal layer on the additional value = ❿ increase the work function value. Moreover, the present invention can be more in the aforementioned work function gold work function metal layer, for example another Covering the Nn layer 6 and additionally covering the other seventh function metal layer. Referring to FIG. 7, the preferred embodiment is used to fabricate a transistor having a metal idle pole. This embodiment can be based on the 12th, sti structure. The second preferred embodiment of the first embodiment of the present invention is the same as the first preferred embodiment of the a* first successful function metal composite layer 126. The second preferred embodiment may include At least three training stages, 201121043 are respectively used to form the first, second and third work function metal layers 26a, 26b, 26c. That is, the first and second work functions are formed by the steps of 帛2a® to 2b. After the metal layers 26a, 26b, the process parameters are changed to form a third work function metal layer 26, wherein the second work function metal layer 26c is different from at least one physical property of the adjacent second work function metal layer 26b. In terms of the third work function metal layer 26c, preferably N-rich nitrogen Titanium, for example, a third work function metal layer may be the same as the process element of the first work function metal layer 26a. Thereafter, the steps of FIGS. 3 through 6 are used to form a metal gate structure 166. According to this, the second real-time implementation of the secret system makes the Xiao pvD conversion success function metal composite layer 126 'the nitrogen flow rate of the towel-process stage is preferably about 32 seem to 48 seem 'the process power is preferably between about 4 watts and 6 watts, and the actual material layer thickness of the first work function metal layer 26a is preferably about 招 Un Ung_m) to 72 angstroms; The nitrogen flow rate in the approximate section is preferably between about μ and 24 sccm. The process power is preferably about (four) watts to 12 watts, and the second work function metal (four) b is preferably about (4) (a% gamma (4) to 96 angstroms; the nitrogen flow rate in the third process stage is preferably between about _ and 48. The dew power is preferably from about watt to four watts, and the third work function is preferably the actual material layer thickness. It is about 48 angstroms (angstr〇m) to 72 angstroms ^ This one can be transferred to about 14_19 nm (nan. lose) E〇t, a flat-band voltage of about -0.565 volts, and a work function value of about 4 58. In other embodiments, the process of forming a metal layer of a success function may not actually be 201121043, and must be zone 2, a plurality of processes In the stage, the process parameters are gradually adjusted in the single-process, so that the proportion of the elements contained in the IUt鈇 composite structure or the single-titanium layer will decrease from the bottom to the bottom. Please refer to Figure 8. 9 and 9, FIG. 9 and FIG. 9 are schematic views showing the fabrication of a transistor having a metal gate track according to the second and fourth preferred embodiments of the present invention. As shown in FIG. 8, the third preferred embodiment The difference from the first preferred embodiment is that the process of the second preferred embodiment of the success-function metal composite layer 226 is actually to gradually adjust the process parameters so that the work function metal composite layer 226 can contain only a single watt. The titanium layer, and at least one physical property contained in the work function metal composite layer 226, such as a nitrogen/titanium element ratio, gradually decreases from bottom to top. The specific formation of the work function metal composite layer 226 can, for example, gradually adjust the gas flow rate of the process or the process power of the process so that the gas flow rate of the nitrogen gas gradually decreases from 40 seem to 20 seem. Or, as shown in FIG. 9, the fourth preferred embodiment is different from the third preferred embodiment in that the fourth preferred embodiment includes a pumping down step in the step of gradually changing the process parameters. The step of pumping down. As a result, the first work function metal layer 326a itself contains at least one physical property, such as a nitrogen/titanium element ratio, which gradually decreases from bottom to top; the second work function metal layer 326b itself contains at least A physical property, such as a nitrogen/titanium element ratio, also gradually decreases with thickness from bottom to top; however, the physical properties of the work function metal composite layer 326 do not continuously change with thickness, but are instead pumped down. There is a discontinuous physical property change before and after the pressing step; that is, the physical property of the first and second work function metal layers 26a, 26b at the junction 17 201121043 is not continuously changed. In other embodiments, the present invention may also not use a dummy gate, that is, without removing the Duo 28 and filling the conductive layer 64, but directly using the multi-layer % as the gate structure - or On the rhyme number of the gold layer 226, 326 formed a guide μ to replace the multi-layer 28, the occupational metal interpole. In summary, the intermetallic structure of the present invention comprises a work function metal composite layer, and the work function metal composite narrative is still thick. For example, the work function metal composite layer is an inverted seam composite structure, and the ratio of the atmosphere/Qin element contained in the limb titanium composite structure is reduced by the lower county. Specifically, the reduced-reduction structure may include a single-titanium nitride layer 'the domain-formed titanium phase containing nitrogen/titanium element _ will gradually decrease with thickness from bottom to top; or 'the titanium nitride composite structure may include multiple hiding a layer of nitrided nitride layer, wherein the proportion of nitrogen/titanium element contained in each titanium nitride layer is maintained constant, and the proportion of nitrogen/titanium element in the lowermost layer of titanium nitride is higher than that of adjacent titanium nitride layer The nitrogen/zinc element ratio is more than 13⁄4, or alternatively, the titanium nitride composite structure may comprise a plurality of layers of nitrided layers, wherein each of the titanium nitride layers has a nitrogen/titanium element which will follow the thickness The nitrogen/titanium element ratio in the lowermost titanium nitride layer is higher than the nitrogen/titanium element ratio of the adjacent titanium nitride layer. By the above method, the present invention can have the following advantages. First of all, the metal gate structure has the advantages of low power _ no space and no effect, and can be transferred to the crystal (4) pole to have better turbulence and speed. In addition, in the lowermost layer of the Weiwei titanium composite structure, the 201121043 titanium layer will have a higher nitrogen/titanium ratio (N-rich), thus providing a suitable work function value of pm〇S. Furthermore, the titanium nitride composite structure has a titanium-nitride layer with a high titanium/nitrogen ratio (Ti-rich), thereby effectively preventing nitrogen or oxygen-specific non-conductive substances from passing through the metal gate structure. The work function metal, in order to avoid the EOT deviation from the predetermined value, and the lower nitrogen/titanium ratio of the titanium nitride layer (the gas flow rate of the process gas nitrogen is about 16 seem to 24 Sccm) can also provide a thinner Ε〇 effect. . Accordingly, the present invention can improve the overall performance of the MOS transistor. The above is based on the fact that the invention is well-received, and all the four (four) patents and modifications made by this (4) patent are covered by the present invention. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 6 is a schematic view showing a first preferred embodiment of the present invention.
7圖為本發明第二較佳實 圖。 · 施例製作具有金屬閘極之電晶體的示意 第8=為本發明第三錄實施㈣作具有金仙極之電晶體的示意 第9=為本發明第四較佳實施例製作具有金屬_之電晶體的示意 14 電晶體區 【主要元件符號說明】 12 基底 201121043 18 淺溝隔離結構 22 閘極介電層 24 遮蓋層 26 功函數金屬複合層 26a 第一功函數金屬層 26b 第二功函數金屬層 26c 第三功函數金屬層 28 多晶矽層 30 遮罩層 32 虛置閘極 34 輕摻雜源極與汲極 38 氧化矽層 40 氮化矽層 48 源極/汲極區域 54 氮化矽層 56 層間介電層 58 開口 64 導電層 66 金屬閘極 126 功函數金屬複合層 166 金屬閘極 226 功函數金屬複合層 266 金屬閘極 326 功函數金屬複合層 326a 第一功函數金屬層 326b 第二功函數金屬層 366 金屬閘極Figure 7 is a second preferred embodiment of the present invention. · Example of making a transistor having a metal gate 8th = the third embodiment of the present invention (4) is a schematic diagram of a transistor having a gold fairy pole. 9 = a fourth preferred embodiment of the present invention has a metal _ Schematic diagram of the transistor 14 transistor area [main component symbol description] 12 substrate 201121043 18 shallow trench isolation structure 22 gate dielectric layer 24 mask layer 26 work function metal composite layer 26a first work function metal layer 26b second work function Metal layer 26c third work function metal layer 28 polysilicon layer 30 mask layer 32 dummy gate 34 lightly doped source and drain 38 tantalum oxide layer 40 tantalum nitride layer 48 source/drain region 54 tantalum nitride Layer 56 Interlayer dielectric layer 58 Opening 64 Conductive layer 66 Metal gate 126 Work function metal composite layer 166 Metal gate 226 Work function metal composite layer 266 Metal gate 326 Work function metal composite layer 326a First work function metal layer 326b Two work function metal layer 366 metal gate