1286325 17116pif.doc 九、發明說明: 本申請案主張於2004年6月3日向韓國智慧財產局提出 申請之韓國專利申請案第P2004-40542號和在2004年9月17日 申請案第P2004-74730號的優先權之益處,該專利申請案所揭 露之内容結合於本說明書中。 【發明所屬之技術領域】 本發明是有關於動態隨機存取記憶體(D RA Μ)半導體元 件,且特別是有關於覆蓋於該元件上,在圖案化金屬層禮, 路由(routing)電源與訊號跡線之方法和裝置。 【先前技術】 DRAM元件包括一個記憶陣列、存取記憶陣列的電路、 和與外部元件一起控制DRAM操作及通訊的周邊電路。習知 的記憶陣列是由一重複格式的次記憶細胞陣列所組成,散佈 在用來給存取記憶陣列的電路部分區域,剩餘的存取電路則 通常位在記憶陣列邊緣的行解碼器與列解碼器裡面。 圖1繪示一習知記憶體1〇〇配置,它包括一個記憶陣列 10、一個行解碼器20、和一個列解碼器30。記憶陣列10被配 置成很像西洋棋盤;該次記憶細胞陣列(SMCAs)被次字元線 ,動器(SWDs)垂直地分隔,並被給用在記憶細胞的感測放大 器(SAs)水平地分隔。每一個次記憶細胞陣列包括多數個記憶 細胞(MC),每一個記憶細胞是由一條次字元線(SWL)帶動的 存取電晶體及儲存資料的電容器所組成。SAs是被連接區(CJs) 垂直地为隔,該CJs包含給SAs用的控制訊號產生電路。 7 1286325 17116pif.doc /于解碼器20在行選擇線(CSLs)上產生訊號,以便依照一 個提供的行位址(CA)去選擇_裡_至乡行來做讀寫。 列解碼器30藉著選取多婁U条主要字元線(NWE)中的-條線和字7〇線選擇(ρχ)訊號,來向一個提供的列位址反應, 以便啟動在陣列裡成一列之記憶細胞。 圖1的其他方面將與繪示在圖2裡記憶陣列10的一部分 的更詳細圖示來合併解釋。兩個記憶細胞M c丨和M c 2分別顯 示在SMCA1和SMCA2裡面。每一個記憶細胞包括一個電容 器C,s亥電谷器連接一個細胞板電壓(Vp)和一個存取電晶體ν 的源極。通常,Vp是一半的電源電壓。每一個存取電晶體(N) 的閘極是被一相對應的次字元線(s w L)以S W L1控制M C1的 存取電晶體和SWL2控制MC2的存取電晶體的方式來控制。 每一個存取電晶體的汲極連接至一相對應的位線(BL), 比方説BL1對MCI和BL2對MC2。每一個位線亦是以存取電 晶體(未繪示)連接至SWLs的方式,連接至各別SMCAs裡的其 他記憶細胞(未繪示)。一個感測放大器區塊是處在Smcai和 SMCA2之間。請參考SMCA1,BL1和BL1B連接至一個在SA1 裡的預先充電電路PRE1 ’並且通過一個位元絕緣閘iso 1以連 接至一對感測位線SBL和SBLB。至於SMCA2,BL2和BL2B 則連接至在SA1裡的預先充電電路PRE2,並且通過一個位元 絕緣閘IS02以連接至一對感測位線SBL和SBLB。一個位元感 測放大器BLSA和一個資料輸出入閘極I0G也連接至感測位 線SBL和 SBLB。 1286325 17116pif.doc 位元感測放大器放大介於在記憶細胞MCI的BL1和 BL1B之間的電壓差,比方說,在如下的順序中,記憶細胞代 表兩個邏輯狀態(多重狀態記憶細胞亦存在和習知使用更複 雜的感測放大器電路)中的一個。絕緣閘IS01連接BL1至 SBL,亦連接BL1B至SBLB。預先充電電路PRE1給BL1和 BL1B充電到一個電壓中途點,那是介於一個放電電容器c(比 方說,代表邏輯0)的電壓和一個充電電容器C(在同例裡,代 表邏輯1)的電壓之間。SWL1被通電,以便將MCI内的記憶細 胞電容器耦合至BL1。當細胞電容器放電時,電荷分享效應 會導致在BL1上的電壓,相對於BL1B減少些;而當細胞電容 器充電時,電荷分享效應會導致在BL1上的電壓,相對於 BL1B增加些。在電荷分享效應完成後,絕緣閘18〇1開始生 效,以致於介於位線BL1/BL1B之間的一微小電壓差被轉移至 感測位線SBL1/SBL1B。而在任何一個情況下,位元感測放 大器BLSA在預設時段期間是啟動著,以便去感測和放大介 於位線BL1/BL1B之間的微小電壓差。 當輸出入閘IOG啟動時,會耦合SBL和SBLB至一對區域 輸出入線LIO和LIOB,該線也會連接至在SA1上方和下方之 其他SA地區(未繪示)裡的其他之輸出入閘i〇G。在此處,輸 出入閘IOG是啟動著,以對行選擇線CSL(未繪示)反應。當LIO 和LIOB有效時,一個區域整體輸出入閘LGIOG伺服給選擇性 地耦合LIO和LIOB至一對整體輸出入線GI〇和gi〇b。因此感 測的記憶細胞狀態是與一個週邊輸出入電路耦合在一起。 9 1286325 17116pif.doc 由圖1和圖2可了解到大量的導線路由過記憶陣列1〇。 NWE線經由·^ $己憶細胞陣列上面,而垂直地路由越過記憶陣 列,並且PX、LIO、和LIOB線經由連接區與感測放大器區上 面,而垂直地路由越過記憶陣列。CSL、GIO、和GIOB線經 由次§己憶細胞陣列上面,而垂直地路由越過記憶陣列。未繪 不的疋電源線,该線也必須路由越過記憶陣列,以提供電源 I 給在SA、CJ、和SWD區塊裡的電路。 、圖3顯示記憶陣列1〇的一塊區域,該圖省略在下方的詳 細電路,而以覆蓋其上方的金屬跡線來解釋。在第一金屬層 上,LIO、PX、和NWE跡線與第一條電源線pi隔開著,該電 源線提供記憶陣列電路所需不同等位的電壓。有些第一條電 源線P1可包括接地電壓線(VSS)和電源線(VCC)。其他的第一 條電源線P1可包括一條參考電壓線(Vref)、一條負 (VBB) ^ 〇 cl 和GIO跡線與第二條電源線P2隔開著,該電源線提供不同等 _ 位的電壓。有些第二條電源線P2亦可包括接地電壓線(vSS) 和電源線(VCC)。其他的第二條電源線P2可包括一條參考電 壓線(Vref)、一條負電源線(VBB)、一條增強電壓線(vpp)等 等。在同等位電壓的P2跡線覆肪跡線地方,該兩條跡線相 互連接在一起而產生一個柵極。!>2跡線連接至位在dram元 件的記憶陣列區外面的電源供應器。 '.圖4顯示圖1裡列解碼器3〇的一個簡化方塊圖。列解碼器 30包括-個列位址解碼器區3(M和一個列位址預解碼器區 1286325 17116pif.doc 30 2在列位址解碼器區3〇_丨内,每一個繪示的第一解碼器 區RD1產生一個字元線選擇訊號?又和每一個繪示的第二解 碼器區RD2產生一個主要字元線訊號:^^¥]£,以對應於依序被 列位址預解碼器區3〇-2產生的列位址ra和預解碼位址DRA。 舅5繪示列解碼器3〇的一部分區域,該圖省略在下方的 詳細電路,而以覆蓋其上方的金屬跡線來解釋。在第一金屬 層上’覆蓋一個第一解碼器區rD1,訊號線81(例如:ρχ線) 是在第一電源線PVINT1和PVSS1側面。在第一金屬層上,覆 蓋一個第二解碼器區RD2,訊號線S1(例如:NWE線)是在額 外的第一電源線PVINT1和PVSS1側面。 第二金屬層包含訊號線S2(例如:RA線和DRA線)和第二 電源線PVINT2和PVSS2。PVINT2連接至PVINT1並且相重 疊,而PVSS2連接至PVSS1並且相重疊。PVINT2和PVSS2跡 線連接至位在DRAM元件的記憶陣列區外面的電源供應器。 在此情況下,電源線不能設計成較寬的線而毫不增加晶片面 積。 【發明内容】 當DRAM元件依比例縮成較小細胞尺寸及或增加在記憶 陣列的細胞數量時,每單位面積裡,會有更多的訊號線路由 過記憶陣列和列解碼器;而此單位面積,在本質上是先前伺 服較小數量訊號線的相同面積。電源線的寬度因此依比例減 小,以便適用於較稠密的記憶陣列。然而減小電源線的寬度 是不好的;當減小電源線的寬度時,會導致對電流造成較大 11 1286325 17116pif.doc 的電阻、較大的電壓降與電源、;肖耗、 計納-個::==設 U雙θ至助以她,會㈣增進訊 【實施方式】 口。接下來的實施例使用在記憶陣列、列解碼器、及或行解 碼盗上之三層金屬層。這些實施例裡,較寬的電源線通常是 可能的,並可改善電源的分配與穩定性。實施例裡的不同優. 點可從以下所提出的圖示描述裡明顯看到。 圖6解釋使用三層金屬層在訊號線與電源線路由過記憶 陣列的第一實施例。第一金屬層包含NWE、pχ、U0訊號線 和P1電源線,這與先前技術類似。第二金屬層包含CSL和GIO 矾號線,但沒有電源線。第三金屬層包含P3電源線,該線與 形成在第一金屬層上的P1電源線垂直。因為CSL線和GIO線 不與覆蓋於記憶陣列上的金屬層3競爭,故P3電源線可做的 比用先前技術形成於第二金屬層上的P2電源線寬些。雖然為 了清楚的目的,圖6裡並未繪示P3電源線的特徵,該電源線 的部分地方甚至可覆蓋在CSL和GIO線上。在相同電壓下, 12 1286325 17116pif.doc1286325 17116pif.doc IX. Invention Description: This application claims Korean Patent Application No. P2004-40542 filed on June 3, 2004 with the Korea Intellectual Property Office and P2004-74730 on September 17, 2004. The benefit of the priority of the number is disclosed in this specification. BACKGROUND OF THE INVENTION 1. Field of the Invention This invention relates to a dynamic random access memory (D RA Μ) semiconductor device, and more particularly to overlying the device, in a patterned metal layer, routing power and Method and apparatus for signal traces. [Prior Art] A DRAM device includes a memory array, a circuit for accessing the memory array, and peripheral circuits for controlling DRAM operation and communication together with external components. A conventional memory array consists of a repeating array of secondary memory cells interspersed in a portion of the circuit used to access the memory array, and the remaining access circuits are typically located at the edge of the memory array. Inside the decoder. 1 illustrates a conventional memory configuration that includes a memory array 10, a row decoder 20, and a column decoder 30. The memory array 10 is configured much like a checkerboard; the secondary memory cell arrays (SMCAs) are vertically separated by sub-word lines, actuators (SWDs) and are applied horizontally to sense amplifiers (SAs) of memory cells. Separate. Each sub-memory cell array includes a plurality of memory cells (MC), each of which is composed of a secondary word line (SWL)-driven access transistor and a capacitor for storing data. The SAs are vertically separated by the connected areas (CJs), which contain control signal generating circuits for the SAs. 7 1286325 17116pif.doc / The decoder 20 generates signals on the row select lines (CSLs) to select ___to the township line for reading and writing according to a provided row address (CA). The column decoder 30 reacts to a provided column address by selecting a -line and a word 7-line selection signal in a plurality of U main character lines (NWE) to initiate a column in the array. Memory cells. Other aspects of Figure 1 will be explained in conjunction with a more detailed illustration of a portion of memory array 10 illustrated in Figure 2. Two memory cells, M c丨 and M c 2 , were shown in SMCA1 and SMCA2, respectively. Each memory cell includes a capacitor C, which is connected to a cell plate voltage (Vp) and a source for accessing the transistor ν. Typically, Vp is half the supply voltage. The gate of each access transistor (N) is controlled by a corresponding sub-word line (sw L) with SW L1 controlling the access transistor of M C1 and SWL2 controlling the access transistor of MC2. . The drain of each access transistor is connected to a corresponding bit line (BL), such as BL1 to MCI and BL2 to MC2. Each bit line is also connected to other memory cells (not shown) in the respective SMCAs by means of access transistors (not shown) connected to the SWLs. One sense amplifier block is between Smcai and SMCA2. Referring to SMCA1, BL1 and BL1B are connected to a precharge circuit PRE1' in SA1 and connected to a pair of sense bit lines SBL and SBLB through a bit insulation gate iso1. As for SMCA2, BL2 and BL2B are connected to the precharge circuit PRE2 in SA1, and are connected to a pair of sense bit lines SBL and SBLB through a bit insulation gate IS02. A bit sense amplifier BLSA and a data output gate I0G are also connected to sense bit lines SBL and SBLB. 1286325 17116pif.doc The bit sense amplifier amplifies the voltage difference between BL1 and BL1B in memory cell MCI. For example, in the following sequence, memory cells represent two logical states (multiple state memory cells also exist and One of the more sophisticated sense amplifier circuits is known. The insulating gate IS01 is connected to BL1 to SBL, and is also connected to BL1B to SBLB. The precharge circuit PRE1 charges BL1 and BL1B to a voltage midway point, which is a voltage between a discharge capacitor c (say, logic 0) and a charge capacitor C (in the same example, representing logic 1). between. SWL1 is energized to couple the memory cell capacitor within the MCI to BL1. When the cell capacitor is discharged, the charge sharing effect causes the voltage on BL1 to decrease relative to BL1B. When the cell capacitor is charged, the charge sharing effect causes the voltage on BL1 to increase relative to BL1B. After the charge sharing effect is completed, the insulating gate 18〇1 starts to be effective, so that a slight voltage difference between the bit lines BL1/BL1B is transferred to the sensing bit lines SBL1/SBL1B. In either case, the bit sense amplifier BLSA is activated during the preset period to sense and amplify the small voltage difference between the bit lines BL1/BL1B. When the output IOG is started, SBL and SBLB are coupled to a pair of regional output lines LIO and LIOB, which are also connected to other output gates in other SA areas (not shown) above and below SA1. i〇G. Here, the output gate IOG is activated to react to the row selection line CSL (not shown). When LIO and LIOB are active, a region's overall output gate LGIOG servo selectively couples LIO and LIOB to a pair of integral output lines GI〇 and gi〇b. Thus the sensed memory cell state is coupled to a peripheral output-in circuit. 9 1286325 17116pif.doc It can be seen from Figures 1 and 2 that a large number of wires are routed through the memory array. The NWE line passes over the cell array and is vertically routed across the memory array, and the PX, LIO, and LIOB lines are routed vertically across the memory array via the connection region and the sense amplifier region. The CSL, GIO, and GIOB lines are routed vertically across the memory array via the secondary § cell array. The power line is not drawn, and the line must also be routed across the memory array to provide power I to the circuits in the SA, CJ, and SWD blocks. Figure 3 shows an area of the memory array 1 ,, which omits the detailed circuit below, and is explained by the metal traces overlying it. On the first metal layer, the LIO, PX, and NWE traces are separated from the first power line pi, which provides different equipotential voltages required by the memory array circuitry. Some of the first power lines P1 may include a ground voltage line (VSS) and a power line (VCC). The other first power line P1 may include a reference voltage line (Vref), a negative (VBB) ^ 〇cl and a GIO trace separated from the second power line P2, the power line providing different equal bits Voltage. Some of the second power lines P2 may also include a ground voltage line (vSS) and a power line (VCC). The other second power line P2 may include a reference voltage line (Vref), a negative power line (VBB), an enhanced voltage line (vpp), and the like. Where the P2 traces of the same bit voltage are overlaid, the two traces are connected together to create a gate. ! The > 2 trace is connected to a power supply located outside the memory array area of the dram element. '. Figure 4 shows a simplified block diagram of the column decoder 3 of Figure 1. The column decoder 30 includes a column address decoder region 3 (M and a column address predecoder region 1286325 17116pif.doc 30 2 in the column address decoder region 3〇_丨, each of which is depicted A decoder area RD1 generates a word line selection signal, and each of the second decoder areas RD2 is shown to generate a main word line signal: ^^¥]£, corresponding to the sequentially listed address. The column address ra and the pre-decoded address DRA generated by the decoder area 3〇-2. 舅5 shows a part of the area of the column decoder 3〇, which omits the detailed circuit below, but covers the metal traces above it. The line is explained. On the first metal layer, 'covering a first decoder region rD1, the signal line 81 (for example, ρχ line) is on the side of the first power lines PVINT1 and PVSS1. On the first metal layer, covering a first The second decoder area RD2, the signal line S1 (for example, the NWE line) is on the side of the additional first power lines PVINT1 and PVSS1. The second metal layer includes the signal line S2 (for example, the RA line and the DRA line) and the second power line. PVINT2 and PVSS2. PVINT2 is connected to PVINT1 and overlaps, and PVSS2 is connected to PVSS1 and overlaps. The PVINT2 and PVSS2 traces are connected to a power supply located outside the memory array area of the DRAM component. In this case, the power supply line cannot be designed as a wider line without increasing the chip area. When the ratio is reduced to a smaller cell size and or the number of cells in the memory array is increased, more signal lines per unit area are passed through the memory array and the column decoder; and this unit area is essentially the previous servo. The same area of a smaller number of signal lines. The width of the power line is therefore proportionally reduced to suit a denser memory array. However, reducing the width of the power line is not good; when reducing the width of the power line, Causes a large resistance to the current 11 1286325 17116pif.doc, a large voltage drop and power supply; Xiao consumption, counting - a:: == set U double θ to help her, will (four) enhance the message [implementation The following embodiment uses a three-layer metal layer on the memory array, column decoder, and or row decoder. In these embodiments, a wider power line is usually possible and can improve the power. The distribution and stability of the embodiments. The differences between the embodiments can be clearly seen from the following description of the illustrations. Figure 6 illustrates the first implementation of a memory array using a three-layer metal layer on the signal line and power line. For example, the first metal layer includes NWE, pχ, U0 signal lines, and P1 power lines, which is similar to the prior art. The second metal layer includes CSL and GIO 矾 lines, but no power lines. The third metal layer includes P3 power lines. The line is perpendicular to the P1 power line formed on the first metal layer. Since the CSL line and the GIO line do not compete with the metal layer 3 overlying the memory array, the P3 power line can be formed by using the prior art. The P2 power line on the two metal layers is wider. Although for clarity, Figure 6 does not show the characteristics of the P3 power line, some parts of the power line can even cover the CSL and GIO lines. At the same voltage, 12 1286325 17116pif.doc
P3電,線覆蓋P1電源線地方,ρι電源線的連接處,存在一間 隙問題,可用-連通柱接合(第三金屬層與第—金屬層之間的 直接連接)或用一媒介物P2焊接點來連接至金屬層1。p3電源 線因此可以較少電阻與改善電源分配的方式路由過。在CSL 與GIO之間的間距也由於少了p2跡線、降低了串擾、和加強 了訊號傳遞速度而改善。 Φ 圖7解釋使用三層金屬層在訊號線與電源線路由過記憶 陣列的第二實施例。在這實施例裡,P1電源線不在金屬層i 上,而且P2電源線平行於金屬層2上的CSL與GIO,並分配電 源至記憶陣列電路。P3電源線配置在金屬層3上,垂直於p2 電源線,並連接至P3電源線與^電源線有同等位電壓交叉的 P2電源線位置。當P3電源線可做的寬些,以有效地傳送電流 至周邊所需地區時,P2電源線可維持相對較細些。 圖^解釋使用三層金屬層在訊號線與電源線路由過記憶 陣列的第三實施例。在這實施例裡,細的ρι電源線與細的 • 電源線交又。同等位電壓的P1和P2電源線在相交叉地方連 接:寬的P3電源線平行地路由過P2電源線,且通常是重疊在 ,等位電壓的P2電源線上。當P3與P2電源線沿著它們長^重 疊時,兩條線之間的連接可做在一長形的管道裡,或經g使 用的小型連通柱裡。當P3/P2與CSL和GIO共同分享金屬層, 卻佔有非常少的空間之時,P3/P2結構是每單位長度仍有&低 的電阻。 一 13P3 power, the line covers the P1 power line, the connection of the power line, there is a gap problem, the joint can be connected (the direct connection between the third metal layer and the first metal layer) or welded with a medium P2 Click to connect to metal layer 1. The p3 power line can therefore be routed with less resistance and improved power distribution. The spacing between CSL and GIO is also improved by the lack of p2 traces, reduced crosstalk, and enhanced signal transfer speed. Φ Figure 7 illustrates a second embodiment of a memory array using a three-layer metal layer over the signal line and power line. In this embodiment, the P1 power line is not on metal layer i, and the P2 power line is parallel to CSL and GIO on metal layer 2 and distributes power to the memory array circuitry. The P3 power cable is placed on the metal layer 3, perpendicular to the p2 power line, and connected to the P2 power line where the P3 power line and the power line have the same voltage crossing. The P2 power line can be kept relatively thin when the P3 power line can be made wider to effectively deliver current to the desired area. Figure 2 illustrates a third embodiment of a memory array using a three-layer metal layer over the signal line and power line. In this embodiment, the thin power cord is connected to the thin • power cord. The P1 and P2 power lines of the same voltage are connected at the intersection: the wide P3 power line is routed in parallel through the P2 power line, and is usually superimposed on the P2 power line of the equipotential voltage. When the P3 and P2 power lines are stacked along their length, the connection between the two lines can be made in an elongated pipe or in a small connecting column used by g. When P3/P2 shares the metal layer with CSL and GIO, but has very little space, the P3/P2 structure still has a low resistance per unit length. One 13
1286325 17116pif.doc 圖9解釋使用三層金屬層在訊號線與電源線路由過記憶 陣列的第四實施例。在這實施例裡,金屬層1包含細的?1電 源線,該線平行地路由過NWE線。金屬層2包含細的P2電源 線,該線垂直地路由過P1電源線,並與CSL和GIO線平行。 在同等位電壓之P2電源線與Pi電源線相交叉地方,此兩條線 連接在一起。金屬層3包含相對較寬的P3電源線,並與?1電 源線平行,而較佳路由是重疊一起,並置同等位電壓的Η電 源線在下面。在同等位電壓之P3電源線與?2電源線相交叉地 方,此兩條線連接在一起。 圖10解釋使用三層金屬層在訊號線與電源線路由過記 憶陣列的第五實施例。這實施例與第三實施例類似(圖8),但 GIO線是路由過金屬層3,而不是金屬層2。這可是一個很吸 引的,代方式,因為重疊的P2和p3電源線可當作一個具低電 阻的單一導線一起作用,以容許P3電源線無需太寬 ,並且留 工間、、七至屬層3上的汛號線。因此,介於csls線間的線間 距可大些,以讓耦合噪音減少。 、 為了路由訊號線與電源線覆蓋於列解碼器之上,較好但 非必要地與之别任—實施例結合的不同種類的實施例也在 此提出。® 11解釋-個第―列解碼器的實關。在第一金屬 層上提供相對較細的電源線PVINT]^PVSS1, 2方的列解碼器電路。例如,灣 成由上在下,朝向列解石馬器區順的外側運作,並留下 部區段覆蓋於RD1上,以運作在第—金屬裡的訊號線S卜其 1286325 17116pif.doc 他的列解碼器訊號線S2則形成在第二金屬上,以垂直於 PVINT1、PVSS1、和S1線的方式運行。在第三金屬層上,相 對較寬的電源線PVINT3和PVSS3,以平行於幻線的方式運 行,該電源線PVINT3和PVSS3的每一條是與一條或多條訊號 線S2重疊。在PVINT3和PVSS3重疊而不是與幻重疊地方,一 個連接點在此兩條電源線間形成。同樣地,在pvg;s3和PVSS1 0 重疊而不是與S2重疊地方,一個連接點在此兩條電源線間形 成。在這實施例裡,連接可涉及一個部分充滿金屬層2的連 通柱,但沒有連續金屬層2的電源線存在。連接可直接由金 屬層3和金屬層1之間達成(連通一個接點)。這配置好處是容 吕午金屬層2有多餘空間,以去擴散或增加S2線的數量,並且 也透過比先前技術的金屬層2上電源線更大橫截面之金屬層 3,來提供電源分配。 圖12解釋與圖11類似的一個第二列解碼器之實施例,但 是在金屬層2上使用額外的電源線PVINT2和PVSS2,該電源 # 線與外側訊號線S2平行地運行。在PVINT2和PVINT1重疊地 方,一個連接點在此兩條電源線間形成;而在pVSS2和PVSS1 之間亦做類似的連接。PVINT3和PVINT2重疊(並也可與一條 或多條訊號線S2重疊),並在pVINT3和PVINT2兩條電源線重 疊之間地方,做成連接。此連接可以是一狹長通道,或是沿 著PVINT3和PVINT2長度,分隔著一連串甚多的小型連通柱 裡。相同的配置與連接亦在PVSS3和PVSS2之間存在。 15 1286325 17116pif.doc 獨Π解釋與圖11類似的一個第三列解碼器之實施例。 PVINT1和PVSS1放在中間,但是在列解碼器區RD1上方,訊 號線si則位在pvinti和pvssi的外侧。在此處,PVINT2和 PVSS2不存在第二金屬層上面。 圖14解釋與圖12類似的一個第四列解碼器之實施例。 PVINT1和PVSS1放在中間,但是在列解碼器區RD1上方,訊 號線S1位在PVINT1和PVSS1的外側。在此處,pviNT2和 • PVSS2連同訊號線S2存在第二金屬層上面。 為了路由訊號線與電源線覆蓋於行解碼器之上,較好但 非必要地與之前任一實施例結合的不同種類的實施例也在 此提出。圖15解釋一個第一行解碼器的實施例,例如,與在 金屬層3上具有GIO線的圖10實施例一起使用。行解碼器2〇, 使用訊號線S1,而電源線PVINT1和PVSS1位在金屬層1上, 並且訊5虎線S2、電源線PVINT2和PVSS2位在金屬層1上方的 金屬層2上。但是在金屬層3上,金屬層3的GIO線(和可選擇 Φ 地用金屬層3上的電源線,未顯示,提供電力至記憶陣列)覆 蓋於圮憶陣列上,並且連續直接地橫過行解碼器,朝向周邊 的輸出入電路(未顯示)。 圖16解釋與圖15類似的一個第二行解碼器之實施例,其 令該GIO線路由過金屬層3上的行解碼器。但是,在剛通過行 解碼器處,每一條GIO線透過一個連通柱連接至一個Gi〇 線,並連續地橫過,比方說,有如圖6-9所描述的金屬層2上 的記憶陣列。 16 1286325 17116pif.doc 任何熟習此技藝人士當了解, 想像是落在所描述的實施例之一 f夕其他的路由排列可被 間隔並未討論到,因為這些 ^構内。絕對的線寬度與 數。些微的修改和詳細的精進t件和處理程序需求的函 並當視為本申請專利範圍。I3在本發明的實施例裡, 先前的實施例只是範例。雖鈇 書可能引用「一」、「一個、、;:在只麵例一些地方,說明 這不必要意指每-如此參考L t,」、或「一些」’ 只限應用在單-的實施例。H °貫施例,歧指此特徵 定本=本ΓΓ已哺佳實施编絲上,财麵用以限 圍内\可^1熟習此技藝者,在不脫離本發明之精神和範 视後附:申請===為以此本發明之保護範圍當 【圖式簡單說明】 記憶元件裡一般先前技術的記憶陣列及 外電裡記憶陣列的一部分放大圓示,詳細顯示額 圖3也繪示圖丨裡記憶陣列的一部分放大圖示,此圖特別 路ίΐίίί憶陣列上之雙層金屬層之訊號與電源跡線的 圖4繪示圖m列解碼器的一部分放大圖示,詳 外電路與訊號線。 Λ 17 1286325 17116pif.doc 圖5也繪示圖1裡列解碼器的一部分放大圖示,此圖特別 針對覆蓋於列解碼器上之雙層金屬層之訊號與電源跡線的 路由選擇設計。 圖6-10繪示多個實施例,顯示路由過記憶陣列之三層金 屬層的訊號線與電源線。 圖11-14繪示多個實施例,顯示路由過列解碼器之三層金 屬層的訊號線與電源線。 圖15-16繪示多個實施例,顯示路由過行解碼器之三層金 屬層的訊號線與電源線。 【主要元件符號說明】 10 ·•記憶陣列 20 :行解碼器 20’ :行解碼器 30-1 :列位址解碼器區 30-2 :列位址預解碼器區 30 :列解碼器 100 :記憶體配置 BL (bit line):位線 BL1B (bit line):位線 BL2B (bit line):位線 BLSA (bit line sense amplifier):位元感測放大器 C (capacitor):電容器 CA (cell array):細胞陣列 1286325 17116pif.doc CA (column address):行位址 CJ (conjunction region):連接區 CSL (column select line) ·•行選擇線 DRA (pre-decoding row address):預解碼位址 GIO (global input/output line):整體輸出入線 IOG (input/output gate):輸出入閘 ISO (bit isolation gate):位元絕緣閘 LGIOG (local global input/output gate):區域整體輸出入 閘 LIO (local input/output line):區域輸出入線 LIOB (local input/output line):區域輸出入線 MC (memory cell):記憶細胞 N (access transistor):存取電晶體 NWE (main word line):主要字元線 PI (first power line):第一電源線 P2 (second power line) ··第二電源線 P3 (third power line) ··第三電源線 PRE (precharge circuit):預先充電電路 PVINT 1 (first power line):第一電源線 PVINT2 (second power line):第二電源線 PVINT3 (third power line):第三電源線 P VS SI (first power line) ··第一電源線 PVSS2 (second power line):第二電源線 19 1286325 17116pif.doc P VS S3 (third power line):第三電源線 PX (word line select):字元線選擇 RA (row address):列位址 RD1 (first decoder area):第一解碼器區 RD2 (second decoder area):第二解碼器區 51 (signal line):訊號線 52 (signal line):訊號線 SA (sense amplifier) ··感測放大器 SBL (sensing bit line):感測位線 SBLB (sensing bit line) ··感測位線 SMC A (sub memory cell array) ··記憶細胞陣列 SWD (sub word line driver):次字元線驅動器 SWL (sub word line):次字元線1286325 17116pif.doc Figure 9 illustrates a fourth embodiment of a memory array using a three-layer metal layer over the signal line and power line. In this embodiment, the metal layer 1 is fine? 1 power line, which is routed through the NWE line in parallel. Metal layer 2 contains a thin P2 power line that is routed vertically through the P1 power line and parallel to the CSL and GIO lines. Where the P2 power line of the same voltage crosses the Pi power line, the two lines are connected together. Metal layer 3 contains a relatively wide P3 power line, and with? 1 The power lines are parallel, and the preferred routes are overlapped together, and the Η power line with the same bit voltage is below. In the P3 power line with the same bit voltage? 2 The power lines cross each other and the two wires are connected together. Figure 10 illustrates a fifth embodiment of a memory array using a three-layer metal layer over a signal line and a power supply line. This embodiment is similar to the third embodiment (Fig. 8), but the GIO line is routed through the metal layer 3 instead of the metal layer 2. This is an attractive, generational approach because the overlapping P2 and p3 power lines can act as a single wire with low resistance to allow the P3 power line to be not too wide, and the work room, seven to the genus The nickname line on 3. Therefore, the line spacing between the csls lines can be larger to reduce the coupling noise. In order to route the signal line and the power line over the column decoder, it is preferred, but not necessary, to carry out the different types of embodiments in combination with the embodiments. ® 11 explains the realization of a column-column decoder. A relatively thin power supply line PVINT]^PVSS1, a 2-sided column decoder circuit is provided on the first metal layer. For example, the bay is operated from the top to the bottom, facing the outside of the calcite region, and leaves a section covering the RD1 to operate the signal line in the first metal. Its 1286325 17116pif.doc his column The decoder signal line S2 is formed on the second metal to operate perpendicular to the PVINT1, PVSS1, and S1 lines. On the third metal layer, relatively wide power supply lines PVINT3 and PVSS3 operate in parallel with the phantom line, each of which is overlapped with one or more signal lines S2. Where PVINT3 and PVSS3 overlap rather than overlap with the phantom, a connection point is formed between the two power lines. Similarly, where pvg;s3 and PVSS1 0 overlap rather than overlap with S2, a connection point is formed between the two power lines. In this embodiment, the connection may involve a via post partially filled with metal layer 2, but no power line of continuous metal layer 2 is present. The connection can be made directly between the metal layer 3 and the metal layer 1 (connecting a contact). The advantage of this configuration is that the Rong Luwu metal layer 2 has extra space to despread or increase the number of S2 lines, and also provides power distribution through the metal layer 3 of a larger cross section than the power line on the metal layer 2 of the prior art. . Figure 12 illustrates an embodiment of a second column decoder similar to that of Figure 11, except that additional power lines PVINT2 and PVSS2 are used on metal layer 2, which operates in parallel with the outer signal line S2. In the overlap of PVINT2 and PVINT1, a connection point is formed between the two power lines; a similar connection is made between pVSS2 and PVSS1. PVINT3 and PVINT2 overlap (and can also overlap with one or more signal lines S2) and are connected between the two power lines of pVINT3 and PVINT2. This connection can be a narrow channel or a series of small connected columns along the length of PVINT3 and PVINT2. The same configuration and connections also exist between PVSS3 and PVSS2. 15 1286325 17116pif.doc An embodiment of a third column decoder similar to that of FIG. 11 is explained. PVINT1 and PVSS1 are placed in the middle, but above the column decoder area RD1, the signal line si is located outside the pvinti and pvssi. Here, PVINT2 and PVSS2 are not present on the second metal layer. Figure 14 illustrates an embodiment of a fourth column decoder similar to that of Figure 12. PVINT1 and PVSS1 are placed in the middle, but above the column decoder area RD1, the signal line S1 is located outside of PVINT1 and PVSS1. Here, pviNT2 and PVSS2 are present along with the signal line S2 on the second metal layer. In order to route the signal and power lines over the row decoder, a different type of embodiment that is preferably, but not necessarily, combined with any of the previous embodiments is also presented herein. Figure 15 illustrates an embodiment of a first row decoder, for example, for use with the Figure 10 embodiment having a GIO line on metal layer 3. The row decoder 2 〇 uses the signal line S1, and the power lines PVINT1 and PVSS1 are located on the metal layer 1, and the bit line S2, the power lines PVINT2 and PVSS2 are located on the metal layer 2 above the metal layer 1. However, on the metal layer 3, the GIO lines of the metal layer 3 (and the power lines on the metal layer 3 that can be selected for Φ, not shown, providing power to the memory array) are overlaid on the memory array and continuously traversed directly The row decoder is an output-in circuit (not shown) that faces the periphery. Figure 16 illustrates an embodiment of a second row decoder similar to that of Figure 15 which causes the GIO line to pass through a row decoder on metal layer 3. However, just after passing through the row decoder, each GIO line is connected to a Gi〇 line through a connecting post and continuously traversed, for example, a memory array on metal layer 2 as depicted in Figures 6-9. 16 1286325 17116pif.doc Anyone skilled in the art will understand that the imagination is one of the described embodiments. Other routing arrangements can be separated by intervals, as these are within the structure. Absolute line width and number. Minor modifications and detailed functional requirements for processing and processing requirements are considered to be within the scope of this patent application. I3 In the embodiments of the present invention, the previous embodiments are merely examples. Although the script may quote "one", "one, and; in some places only, it means that this does not necessarily mean that every reference to Lt," or "some" is only applicable to the implementation of the single- example. H ° application example, the finger refers to this feature book = this book has been fed on the implementation of the knitting wire, the financial face is used to limit the inside \ can be familiar with this skill, without leaving the spirit and scope of the invention: Application === is the scope of protection of the present invention. [Simplified description of the drawing] A generalized memory array of the memory element and a part of the external memory array are enlarged and displayed in detail, and the detailed display of the figure 3 is also shown in the figure. A portion of the memory array is shown in an enlarged view. This figure specifically illustrates the signal of the double-layer metal layer on the array and the power trace. FIG. 4 shows a partial enlarged view of the m-column decoder, and the external circuit and the signal line. Λ 17 1286325 17116pif.doc Figure 5 also shows a partial enlarged view of the column decoder of Figure 1, which is specifically designed for the routing of signal and power traces over a two-layer metal layer over a column decoder. 6-10 illustrate various embodiments showing signal lines and power lines routed through a three-layer metal layer of a memory array. Figures 11-14 illustrate various embodiments showing the signal and power lines of a three-layer metal layer routing a column decoder. Figures 15-16 illustrate various embodiments showing the signal lines and power lines of the three-layer metal layer of the routing pass decoder. [Main component symbol description] 10 • Memory array 20: Row decoder 20': Row decoder 30-1: Column address decoder area 30-2: Column address predecoder area 30: Column decoder 100: Memory configuration BL (bit line): Bit line BL1B (bit line): Bit line BL2 (bit line): Bit line sense amplifier: Bit sense amplifier C (capacitor): Capacitor CA (cell array ): Cell array 1286325 17116pif.doc CA (column address): row address CJ (conjunction region): link area CSL (column select line) · • row select line DRA (pre-decoding row address): pre-decode address GIO (global input/output line): IOG (input/output gate): output isolation gate ISO (bit isolation gate): bit global isolation gate LGIOO (local global input / output gate): regional overall output gate LIO ( Local input/output line): local input/output line: local area input line MC (memory cell): memory transistor N (access transistor): access transistor NWE (main word line): main character Line PI (first power line): The first power line P2 (second power line) ·· Second power line P3 (precharge circuit): precharge circuit PVINT 1 (first power line): first power line PVINT2 (second power line): second power line PVINT3 (third Power line): third power line P VS SI (first power line) ··first power line PVSS2 (second power line): second power line 19 1286325 17116pif.doc P VS S3 (third power line): third power supply Line line select: RA (row address): RD1 (first decoder area): first decoder area RD2 (second decoder area): second decoder area 51 (signal line) Signal line (signal line): signal line SA (sensing bit line): sensing bit line SBLB (sensing bit line) · sensing bit line SMC A (sub memory cell array) · memory cell array SWD (sub word line driver): sub-word line driver SWL (sub word line): sub-word line
2020