TW200832411A - System and method for programming non-volatile memory with improved boosting - Google Patents

System and method for programming non-volatile memory with improved boosting Download PDF

Info

Publication number
TW200832411A
TW200832411A TW96131530A TW96131530A TW200832411A TW 200832411 A TW200832411 A TW 200832411A TW 96131530 A TW96131530 A TW 96131530A TW 96131530 A TW96131530 A TW 96131530A TW 200832411 A TW200832411 A TW 200832411A
Authority
TW
Taiwan
Prior art keywords
voltage
boost
storage element
memory
channel
Prior art date
Application number
TW96131530A
Other languages
Chinese (zh)
Other versions
TWI349286B (en
Inventor
Fumitoshi Ito
Original Assignee
Sandisk Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US11/516,976 external-priority patent/US7440326B2/en
Application filed by Sandisk Corp filed Critical Sandisk Corp
Publication of TW200832411A publication Critical patent/TW200832411A/en
Application granted granted Critical
Publication of TWI349286B publication Critical patent/TWI349286B/en

Links

Classifications

    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C16/00Erasable programmable read-only memories
    • G11C16/02Erasable programmable read-only memories electrically programmable
    • G11C16/04Erasable programmable read-only memories electrically programmable using variable threshold transistors, e.g. FAMOS
    • G11C16/0483Erasable programmable read-only memories electrically programmable using variable threshold transistors, e.g. FAMOS comprising cells having several storage transistors connected in series
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C11/00Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor
    • G11C11/56Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using storage elements with more than two stable states represented by steps, e.g. of voltage, current, phase, frequency
    • G11C11/5621Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using storage elements with more than two stable states represented by steps, e.g. of voltage, current, phase, frequency using charge storage in a floating gate
    • G11C11/5628Programming or writing circuits; Data input circuits
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C16/00Erasable programmable read-only memories
    • G11C16/02Erasable programmable read-only memories electrically programmable
    • G11C16/06Auxiliary circuits, e.g. for writing into memory
    • G11C16/10Programming or data input circuits
    • G11C16/12Programming voltage switching circuits
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C16/00Erasable programmable read-only memories
    • G11C16/02Erasable programmable read-only memories electrically programmable
    • G11C16/06Auxiliary circuits, e.g. for writing into memory
    • G11C16/34Determination of programming status, e.g. threshold voltage, overprogramming or underprogramming, retention
    • G11C16/3418Disturbance prevention or evaluation; Refreshing of disturbed memory data
    • G11C16/3427Circuits or methods to prevent or reduce disturbance of the state of a memory cell when neighbouring cells are read or written
    • GPHYSICS
    • G11INFORMATION STORAGE
    • G11CSTATIC STORES
    • G11C16/00Erasable programmable read-only memories
    • G11C16/02Erasable programmable read-only memories electrically programmable
    • G11C16/06Auxiliary circuits, e.g. for writing into memory
    • G11C16/34Determination of programming status, e.g. threshold voltage, overprogramming or underprogramming, retention
    • G11C16/3418Disturbance prevention or evaluation; Refreshing of disturbed memory data

Abstract

Non-volatile storage elements are programmed in a manner that reduces program disturb, particularly at the edges storage elements strings, by using modified pass voltages. In particular, during the programming of a selected storage element, an isolation voltage is applied to a storage element proximate to the selected storage element thereby electrically dividing the channel associated with the storage elements into two isolated areas. Additional isolated areas are formed remotely from the selected storage element by applying the isolation voltage to other remote storage elements. The isolated channel regions associated with the storage elements are then boosted with different pass voltages in order to alleviate the effects of program disturb. Thus, a standard pass voltage is applied to storage elements immediately adjacent to the selected storage element, and a lower pass voltage is applied to storage elements remote from the selected storage element. In one preferred embodiment, a higher pass voltage is applied to storage elements immediately adjacent the selected storage element on the side having previously programmed storage elements. These techniques reduce the leakage of charge from adjacent boosted channel regions caused by gate induced drain leakage at the source select line and the drain select line, as well as from isolation word lines, thereby reducing program disturb effects.

Description

200832411 九、發明說明: 【發明所屬之技術領域】 本揭示内容係關於程式化非揮發記憶體,且更明確地 說,係關於在程式化期間降低程式化干擾效應的技術。 【先前技術】 於各種電子裝置。例 半導體記憶體已越來越普遍地用200832411 IX. INSTRUCTIONS: TECHNICAL FIELD OF THE INVENTION The present disclosure relates to stylized non-volatile memory and, more specifically, to techniques for reducing stylized interference effects during stylization. [Prior Art] Various electronic devices. Example Semiconductor memory has become more and more popular

如,非揮發半導體記憶體係用於蜂巢式電話、數位相機、 個人數位助理、行動計算裝置、非行動計算裝置及其它裝 置中。電τ抹除可程式化唯讀記憶體(eeprom)及快閃記 憶體係屬於最普遍的非揮發半導體記憶體。與傳統、且有 完全特徵的EEPROM相比 EEPROM類型)則可在單一 吕己憶體之一部分的内容。 ’採用快閃記憶體(其同樣係一種 步驟中抹除整個記憶體陣列或該 EEPRQM與快閃記憶體兩者會運用—種具有浮動閉極的 電晶體結構,該㈣問極係位於—半導體基板中的通道區 上方並且與該通道區絕緣,並且位於該等源極區與汲極區 之間。—控制閘極會被設置在該浮動閘極上方並與該浮動 。《藉由保留在该浮動閘極上的電荷之數量可控制 =此=形成的電晶體之臨界。也就是,在接通電晶 :Ί員在違控制閘極處施加最小數量之電壓,以允 :猎由該浮動閘極上之電荷的位準來控制其源極與汲極間 ,每一個 已抹除狀 Λ動閘極可用來儲存兩個範圍的電荷,所 言己憶體元件均可1 t 9 了 /、有兩個可能狀態,舉例來說 I23781.doc 200832411 態與已程式化狀態。此快閃記憶體裝置有時稱為二兵 :記憶體裝置’因為每一記憶體元件均可儲存一位元的資For example, non-volatile semiconductor memory systems are used in cellular phones, digital cameras, personal digital assistants, mobile computing devices, inactive computing devices, and other devices. Electrical τ erasing programmable epexes and flash memory systems are among the most common non-volatile semiconductor memories. The EEPROM type can be compared to a traditional and fully featured EEPROM. 'Using flash memory (which is also used in one step to erase the entire memory array or the EEPRQM and flash memory will be used - a transistor structure with floating closed poles, the (4) questioning system is located in - semiconductor Above and insulating the channel region in the substrate, and between the source region and the drain region. - The control gate is disposed above the floating gate and floats with the floating gate. The amount of charge on the floating gate can be controlled = this = the criticality of the formed transistor. That is, when the transistor is turned on: the employee applies a minimum amount of voltage at the gate of the control gate to allow: hunting by the float The level of the charge on the gate controls the source and the drain, and each erased gate can be used to store two ranges of charge, which can be 1 t 9 /, There are two possible states, for example, I23781.doc 200832411 state and stylized state. This flash memory device is sometimes called two soldiers: memory device 'because each memory component can store one bit Capital

A _-多狀態或多位準快閃記憶體裝置係藉由識別多 經^許/有效的已程式化臨界電壓範圍來施行。每—個不 同臨界電壓範圍會對應於記憶體裝置中已編碼的資料位— 組的-職數值。舉例來說,#該元件可被放置在對應= :個不同臨界電壓範圍的四個離散電荷帶的其中—者:中 守母儲存元件便可儲存兩位元的資料。 一般來說,在-程式化程序期間會以幅度會隨時間而辦 加的系列脈衝的方式施加—程式化電壓w 體元件的控制閘極。在-種可行的方式中,該等脈衝之; 度會隨著每個連續脈衝而增加預定的步進 A 七、 v Pgni,举 例“’0.2 W。在該等程式化脈衝間的週期中, 會進行驗證操作。也就是,要被平行程式化的—群元件中 的每-個元件的程式化位準會在連續的程式化脈衝之間被 讀取並且會與—驗證位準作比較,該驗證位準表示的係兮 凡:要被程式化的狀態。對於多狀態快閃記憶體元件陣歹: 而:可針對一兀件之每一狀態來實施一驗證步驟以判斷 X几件疋否已到達其與貧料相關聯的驗證位準。舉例來 能夠在四種狀態中儲存資料的一多狀態記憶體元件可 此需要針對三個比較點來實施驗證操作。 再者田私式化—EEPR0M或快閃記憶體裝置(例如 NAND串中的NAND快閃記憶體裝置)時,通常會將^施 123781.doc 200832411 加於控制閘極並對該位元線進行接地,從而使來自該記憶 體元件之通道的電子會被注入該浮動閘極之中。當電子累 積在該浮動閘極之中時,該浮動閘極會變成帶負電並且該 記憶體兀件之臨界電壓會升高,以便將該記憶體元件視為 處於已程式化狀態。在標題為”S〇UrCe Side Self BoostingA _- multi-state or multi-bit quasi-flash memory device is implemented by identifying a multi-mode/validized programmed threshold voltage range. Each of the different threshold voltage ranges will correspond to the coded data bits in the memory device - the job value of the group. For example, # this component can be placed in the corresponding =: a range of four discrete charge bands of different threshold voltages - in which: the mother storage component can store two bits of data. In general, during the -programming process, the control gate of the voltage component is programmed as a series of pulses whose amplitude will be applied over time. In a feasible manner, the pulses will increase by a predetermined step A, v Pgni with each successive pulse, for example "'0.2 W. During the period between the stylized pulses, Verification is performed. That is, the stylized levels of each of the components to be parallelized are read between successive stylized pulses and compared to the verify level. The verification level indicates the state to be stylized. For a multi-state flash memory component array: and: a verification step can be performed for each state of a component to determine a number of Xs. Whether the verification level associated with the poor material has been reached. For example, a multi-state memory component capable of storing data in four states may need to perform verification operations for three comparison points. - EEPR0M or a flash memory device (such as a NAND flash memory device in a NAND string), usually applying 123781.doc 200832411 to the control gate and grounding the bit line, thereby Electricity of the channel of the memory component a sub-portion is injected into the floating gate. When electrons accumulate in the floating gate, the floating gate becomes negatively charged and the threshold voltage of the memory element rises to facilitate the memory element Considered to be in a stylized state. Under the heading "S〇UrCe Side Self Boosting

Technique For Non-Volatile Memory” 之美國專利案 6,859,397及在標題為’’Detecting 〇ver programmed Mem〇ry" 之美國專利申請公告案2〇〇5/〇〇24939中可找到關於此類程 式化的更多資訊,兩者均以引用的方式全部被併入本文之 中〇 在程式化一選定記憶體元件期間,可能會在被稱為程式 化干擾的一程序中不慎地程式化鄰近的記憶體元件。舉例 來說,當將vpgm施加於一字元線時,可能會不慎地程式化 -並不希望被喊化但是位在與被選定要進行帛式化的一 記憶體元件相同的纟元線上的記憶體元#。可運用數項技 術來防止程式化干擾。舉例來說,採用自昇壓(SB),可電 隔離與未被選擇之位元線相關聯的通道,並且將一傳導電 壓(例如Η) V)施加於”未被選擇之字元線。該等未被選 擇之字元線會麵合至與該等未被選擇之位元線相關聯之通 道’從而使-電壓(例如8V)存在於該等未被選擇之位元線 的通道中,其會趨向於降低該程式化干擾效應。因此,自 昇壓技術會讓電壓昇壓存在於通道内,此趨向於降低橫跨 該穿随氧化物之電壓微分並且會因而降低該程式化干擾效 應。再者’其它經修正技術(例如局部自昇壓(lsb)、修正 123781.doc 200832411 式局邛自昇壓(RLSB)、已抹除區域自昇壓(EASB)、以及 修正式已抹除區域自昇壓(REASB))則會藉由隔離先前已程 式化元件的通道及要被禁止程式化元件的通道以試圖降低 程式化干擾。舉例來說,REASB技術會藉由利用一非常低 的隔離電壓(舉例來說,0 V至i V)來驅動該已程式化側上 已選定字元線附近的一字元線,並且利用一中等電壓(約3 V)來驅動正好位於該已程式化側上已選定字元線旁邊的一 字凡線,來隔離該已程式化側與該已抹除側。所有其它未 被選擇的字元線則會利用一傳導電壓來驅動。因此,便會 產生兩條分離的昇壓通道。由於該等已程式化記憶體元件 之臨界電壓的效應的關係,該已程式化側上之已昇壓通道 電壓通《會比該已抹除側上之已昇壓通道電壓低1 V至2 V,不過,卻仍足以提昇該隔離字元線的臨界電壓。 當記憶體元件中的通道長度持續縮短時,用以降低程式 化干擾的習知通道昇壓技術的能力便會持續地下降。明確 地說’記憶體元件之通道長度可能變得太短以致不能充份 地隔離所選擇之字元線之汲極與源極側處的兩個分離昇壓 的通道區。因此,便可能會降低昇壓通道電壓,從而會導 致程式化干擾。此外,高電場效應(例如帶至帶穿隧 (BTBT)及閘極引發汲極洩漏(GIDL))亦可能會造成該昇壓 通道被放電,及/或因高電場所產生的熱載子可能會被注 入該等穿隧氧化物之中或是該等記憶體元件的浮動閘極之 中。因此’仍舊需要解決程式化干擾問題的改良技術。 【發明内容】 123781.doc 200832411 本發明揭*在程式化非揮發記憶體時用於降低程式化干 擾效應的技術。明確地說,會使用多個隔離區來將和一电 儲存元件相關聯的通道分成多個昇麼區。接著便可利用合 宜的傳導電壓來對該等多個昇屡區中的每一條字元線進行 偏屋’以幫助減低因源極選擇線與汲極選擇線及隔離字元 線上的閘極引發汲㈣漏的關係而從旁邊的昇《區處所流 出的茂漏電流。該些技術特別適用於防止(例如)在一長串 NAND記憶體的較高字元線與較低字元線處的一記憶體區 的邊緣處發生嚴重的程式化干擾效應。 Μ中-具體實_中’會選擇—儲存元件來進行程式 化。與該組儲存元件相關聯的通道會藉由施加—隔離電壓 至該選定的儲存元件近端的其中—儲存元件並且施加至兮 =的健Μ件遠離的另—儲存元件而被分成至少三個: 離通暹區。因此,會將一第一盥一 f,、, /、弟一通道區設置在彼此 ==一部位於該選定儲存元件的相反側上。第三通道區則 :::°亥選:儲存70件。第三通道區的電位會利用-低於 忒等第,第一通道區的昇壓電壓來提昇。 _、 於-較佳的具體實施例中,該隔離電壓會被 該選定儲存元件的第三儲存元件處,因此 = 選定儲存元件處產生-第四通道區。進-步言之,;= =可被施加至多個儲以件,以便創造多個隔離;;= 於-特佳的具體實施例中,會將__標準的 至位於和該選定蝕六_、 书&化加 儲存70件近端的通道區相對應的儲存元件 123781.doc -10- 200832411 處,並且會將一較低的傳導電壓施加至位於和遠離該選定 元件的通道區相對應的儲存㈣處。⑤—特佳的具體實施 例中,被施加至位於該選定儲存元件近端的通道區之中的 傳導電壓會高於該標準的傳導電壓。 參考下文的詳細說明與附圖便會更瞭解本發明的特點與 優點,其中,附圖提出運用本發明原理的解釋性具體實施 例。 【實施方式】 本揭示内容係關於用於降低程式化干擾的技術,尤其是 在一NAND記憶體串邊緣處的字元線處。用於降低程式化 干擾的基本概念係沿著該NAND記憶體串中提供額外的隔 離區,以便降低被施加至該νανε^£憶體串邊緣處之字元 線的傳導電壓。 適合於施行本發明的非揮發記憶體系統之一範例係使用 NAND快閃記憶體結構,其中會將多個電晶體串聯配置在 一NAND串中的兩個選擇閘極之間。圖“系顯示一 的一俯視圖。圖2係其等效電路。圖丨與2中所描述的Nand 串包含四個電晶體1 〇〇、1 〇2、1 〇4以及1 〇6,它們會串聯並 且被夾設在一第一選擇閘極12〇與一第二選擇閘極122之 間。選擇閘極120與122會分別將該NAND串連接至位元線 接點126與源極線接點128。藉由將適當電壓分別施加於控 制閘極120CG與122CG便會控制選擇閘極12〇與122。電晶 體1〇〇、102、104及1〇6中每一者均具有一控制閘極與一浮 動閘極。電晶體100具有控制閘極1〇〇CG與浮動閘極 123781.doc 200832411 1〇〇FG °電晶體102包含控制閘極102CG與浮動閑極 1〇2FG °電晶體1〇4包含控制閘極i〇4CG與浮動閘極 4 F G 氣晶體1 〇 6包含一控制閘極1 〇 6 C G與浮動閑極 106FG。控制閘極1〇〇(^、i〇2CG、1〇4CG及 1〇6CG會分別 被連接至字元線WL3、WL2、WL1及WL0。在一可行的設 什中,電晶體1〇〇、1〇2、104及106各為記憶體元件或儲存 元件。在其它的設計中,該等記憶體元件可能包含多個電 晶體或者可能不同於圖丨與二中所描述者。選擇閘極12〇會 被連接至汲極選擇線SGD,而選擇閘極122則會被連接至 源極選擇線SGS。 睛注意,儘管圖1至2在NAND串之中顯示出四個記憶體 元件,但使用四個電晶體僅係作為一範例。配合本文所述 之技術來使用的一 NAND串亦可使用四個以下的記憶體元 件或是四個以上的記憶體元件。舉例來說,特定的nand 串將包3八、十六、二十二、六十四個或更多記憶體元 件.。本文之討論並不限於NAND串中的任何特定數量之記 憶體元件。 一般而言,本發明可用於藉由福勒-諾爾德海姆(F〇wier_ Nordheim)牙隧作用來程式化與抹除的的裝置。本發明還 可應用於使用一多層介電堆疊的裝置,例如一由氧化矽、 氮化石夕、與氧化矽(ΟΝΟ)所形成的介電質,或是一由氧化 矽、高k層、與氧化矽所形成的介電質。一多層介電質會 被夾設在-導電控制閑極與該記憶體元件通道上面的一半 導體基板的一表面之間。舉例來說,本發明還可應用於使 123781.doc -12· 200832411 用小型導電材料島(例如奈米晶體)並且以一電荷陷捕層(例 如氮化矽)取代浮動閘極作為電荷儲存區的裝置。可以採 用與以浮動閘極為主的NAND快閃裝置雷同的方式來程式 化並抹除此類記憶體裝置。 圖3係描述三個NAND串的一電路圖。使用NAND結構的 快閃記憶體系統的一典型架構包含數個NAND串。舉例來 說,圖中在一具有更多NAND串的記憶體陣列中顯示出三 個NAND串201、203及205。該等NAND串中每一者均包含 二個選擇電晶體與四個記憶體元件。舉例來說,NAND串 201包含選擇電晶體220與230,以及記憶體元件222、 224、226以及228。NAND串203包含選擇電晶體240與 250,以及記憶體元件242、244、246以及248。NAND串 205包含選擇電晶體260與270,以及記憶體元件262、 264、266以及268。每一個NAND串均會藉由其選擇電晶體 (例如,選擇電晶體230、250或270)被連接至該源極線。一 選擇線SGS係用來控制該等源極側選擇閘極。藉由汲極選 擇線SGD所控制的選擇電晶體220、240、260等可將各 NAND串201、203及205連接至個別的位元線202、204及 206。在其它具體實施例中,該等選擇線並不一定必須為 共用。字元線WL3會被連接至記憶體元件222、242以及 262的控制閘極。字元線WL2會被連接至記憶體元件224、 244以及264的控制閘極。字元線WL1會被連接至記憶體元 件226、246以及266的控制閘極。字元線WL0會被連接至 記憶體元件228、248以及268的控制閘極。從圖中可以看 123781.doc -13- 200832411 出’每一條位70線與該個別nand串包括該記憶體元件陣 列或組中的行。該等字元線(WL3、WL2、WL1及WL0)包 括該陣列或組中的列。每個字元線會連接該列中每個記憶 體元件之控制閘極。舉例來說,字元線WL2會被連接至記 憶體元件224、244及264之控制閘極。 每個記憶體元件均可儲存資料。舉例來說,當儲存一位 兀的數字貪料時’該記憶體元件之可行臨界電壓的範圍會 刀成兩個範圍’該等範圍會被指派邏輯資料值"1,,及,,〇,,。 在NAND型&閃記憶體的一範例巾,在抹除該記憶體元件 之後的臨界電壓係負值並且會被定義為邏輯"丨"。在一程 式刼作後之臨界電壓為正並且會被定義為邏輯,,〇"。當臨 界電壓係負並且嘗試讀取時,該記憶體元件將會開啟以指 示正在儲存邏輯’ 1"。當臨界電壓係正並且嘗試讀取操作 時,該記憶體元件將不會開啟,此指示會儲存邏輯,,〇,,。 一記憶體元件亦可儲存多個位準的資訊,舉例來說,多位 元的數字資料。在此情況下,臨界電壓之範圍會被分成資 料之位準的數量。舉例來說,倘若儲存四個資訊位準的 話,便會指派四個臨界電壓範圍給資料值"u,,、q〇π、 ”〇1”及”00”。在一NAND型記憶體的範例中,在抹除操作 之後的臨界電壓係負值並且會被定義為”丨丨”。正臨界電壓 係用於"1〇"、,,〇1”及"00”等狀態。被程式化於該記憶體元 件之中的資料與該元件之臨界電壓範圍間的特定關係取決 於該等記憶體元件所採用的資料編碼方案。舉例來說,標 題為 ’’Novel Multi-State Memory·’之美國專利案第 6 222 762 123781.doc -14- 200832411 號及標題為"Tracking Cells For A Memory System”之美國 專利申請案第2004/0255090號兩案便說明用於多狀態快閃 記憶體元件之各種資料編碼方案,兩案之全部内容均以引 用的方式併入本文之中。 NAND型快閃記憶體的相關範例及其操作已提供在下面 的參考專利案之中:美國專利案第5,386,422號、美國專利 案第5,522,5 80號、美國專利案第5,57〇,315號、美國專利案 第5,774,397號、美國專利案第6,〇46,935號、美國專利案第 6,45 6,528號、以及美國專利案第6,522,58〇號;本文以引用 的方式將每一案併入。 當程式化一快閃記憶體元件時,會將一程式化電壓施加 於該元件之控制閘極,且與該元件相關聯之位元線會被接 地。來自該通道的電子會被注入該浮動閘極之中。當電子 累積在浮動閘極之中時,該浮動閘極會變為帶負電而且該 元件之臨界電壓會升高。為將該程式化電壓施加於正在被 私式化的元件的控制閘極,該程式化電壓會被施加於適當 的字元線上。如上面的討論,該字元線還會被連接至共享 同一字元線的其它NAND串中每一串中的其中一個元件。 舉例來說,當程式化圖3的元件224時,程式化電壓Vpgnj^ 會被施加至元件224、244及264的控制閘極。當希望程式 化一字元線上的其中一元件而不程式化被連接至相同字元 線的其它元件時,舉例來說,當希望程式化元件224而不 程式化元件244時,便會出現問題。因為該程式化電壓會 被施加至與一字元線連接的所有元件,所以,該字元線上 123781.doc •15· 200832411 未被選擇的元件(也就是,未被選擇要進行程式化的元 件)便可能會在一通常被稱為程式化干擾的程序中不慎地 被程式化。舉例來說,當程式化元件224時,可能要擔心 的鄰近元件244或264會不慎被程式化。請注意,程式化干 擾最可能會出現在被選擇要進行程式化的一字元線上的未 被選擇之記憶體元件上。然而,在特定情況下,程式化 干擾還可能出現在該被選擇字元線以外的記憶體元件之 中〇 可以運用數項技術來防止程式化干擾。採用先前說明的 自昇壓(SB),便可電隔離與該等未被選擇之位元線相關聯 之通道,並且在程式化期間將一傳導電壓(舉例來說,ι〇 V)施加於該等未被選擇之字元線。該等未被選擇之字元線 會耦合至和該等未被選擇之位元線相關聯的通道,從而讓 一電壓(舉例來說,8 V)存在於該等未被選擇之位元線的通 道中,此會趨向於降低程式化干擾。因此,自昇壓便會讓 電壓昇壓存在於該通道内,此會趨向於降低橫跨該穿隧氧 化物之電壓並且因而降低程式化干擾。 一 NAND串通常會對應於字元線從WL〇至WL3的依序配 置,從該源極侧至該汲極側被程式化,舉例來說,從記憶 體元件228至圯憶體元件222。作為一範例,假設相鄰的 NAND串203中所有或大部分的記憶體元件均已經在ΝΑΝβ 串201中所有或大部分的記憶體元件之前被程式化。因 此,當該程式化程序準備程式化該NAND串2〇1中後面數個 記憶體元件時,在該些前面被程式化的記憶體元件的浮動 123781.doc -16· 200832411 閘極之上便會有負電荷。因此,使用標準SB技術的昇壓電 位在部分的NAND串203之中可能不夠大,而無法在nand 串203之中和後面數條字元線相關聯的元件上避免發生程 式化干擾效應。舉例來說,當程式化NAND串201上的元件 222時,倘若先前已程式化的NAND串203上的元件248、 246及244的話’那麼該些電晶體(244、24 6及248)中每一者 便會在它們的浮動閘極上具有一負電荷,此將會限制該自 昇壓程序的昇壓位準並且可能會在元件242之上產生程式 化干擾。 局部自昇壓(LSB)及已抹除區域自昇壓(EASB)均係嘗試 藉由將先前被程式化之元件的通道與被禁止之元件的通道 隔離來解決習知自昇壓之缺點的修正技術。舉例來說,倘 若正在程式化圖3的元件224的話,那麼LSB及EASB便會嘗 試藉由將元件244之通道與先前被程式化的元件(246與248) 隔離來禁止元件244中的程式化。對SB、EASB、及LSB昇 壓方法或是該些昇壓方法的變化例來說,要被程式化的元 件的位元線會被連接至接地或另一低電壓(通常係〇 V至^ v),而具有要被禁止之元件的NAND串的位元線則係處於 Vdd,其通常係處於h5 乂至3 v的範圍之中。該程式化電壓 VPgm(舉例來說,20 V)會被連接至該被選擇的字元線。於 LSB昇壓模式的情況中,位於該等被選擇的字元線旁邊的 字元線會被設定為〇 V或接近0 V,而其餘未被選擇的字元 線則έ被$又疋為傳導電壓vpass。舉例來說,位元線2〇2係 處於〇 V而位元線2〇4則係處於vdd。汲極選擇線sgd係處 123781.doc -17- 200832411 於Vsgd(通常係2.5 V至4.5 V)而源極選擇線SGS則係處於〇 v °被選擇的字元線WL2(用於程式化元件224)則係處於 vPgm。鄰近的字元線WL1及WL3係處於〇v,而其它字元線 (舉例來說,WLO)則係處於V_s。 LSB模式之缺點係,被選擇之字元線下面的已昇壓通道 電壓可此非常高,因為該通道中一部分會與未被選擇之字 元線下面的其它通道區隔離,且因此主要係藉由該高程式 化電壓vpgm來決定該昇壓電壓。由於該高昇壓的關係, BTBT或GIDL可能會出現在被偏壓於〇 v的字元線附近,而 且所產生的熱載子可能會被注入未被選擇NAND串中的該 等被選擇的字元線之中。通道昇壓的數量可利用該easb 方法而受到良好控制。EASB和SB雷同,唯一的例外係, 該被選擇的字元線鄰近的源極侧係被偏壓在〇 V處。因 此,在戒被選擇子元線下面的通道區及在該等被選擇元件 的汲極侧處的通道區係相連的,且因此,該通道昇壓會由 被施加至具有一已抹除狀態的該等未被選擇字元線的 電壓來穩疋地決定’因為昇壓不會受到前面已經被程式化 的資料的影響。倘若Vpass太低的話,那麼該通道中的昇壓 將不足以防止程式化干擾。倘若vpass太高的話,則可能會 程式化一被選擇之NAND串中未被選擇之字元線(該位元線 上為Ο V)’或者可能會因GIDL而干擾未被選擇的NAND串 中被選定的子元線。舉例來說,WL1會處於〇 v而非 Vpass,而WL3則會處於Vpass。在一具體實施例中,Vp_係 7 V至 10 V。 123781.doc -18- 200832411 雖然LSB與EASB會提供改良之自昇 存在一項問題,該問題會取決於該源極侧上的鄰近元件 (元件246係元件244之源極側鄰近元件)究竟係要被程式化 或被抹除。倘若該源極側鄰近元件係處於已程式化狀態, 那麼在其浮動閘極上便會一負電荷。再者,將〇 V施加於 控制閘極,並且與帶負電閘極下面的高度反向偏壓接面 (由於昇壓的關係所造成)結合,便可能會引起閘極引發汲 極/¾漏(GIDL) ’其中電子會洩漏至該昇壓通道之中。 GIDL會伴隨著由於該昇壓的關係而在該等記憶體元件之 /及極/源極區上的南電壓所導致的該接面上的大額偏壓以 及-低的或負值的閘極電壓而出$見,此正好係當該源極側 邠近兀件被耘式化且該汲極接面被昇壓時的情況。 可能會造成該昇壓電壓過早地浅漏掉,從而產生一程式化 干k並且胃P現著突然且高度摻雜的接面(當元件尺寸縮 小時便有需要)而更加嚴重。論_流夠高的話,該 通道區中的昇壓電位便可能會減小,從而造成程式化干 擾。再者,正在被程式化的字元線離該没極越近,出現在 該昇壓通道區之φ沾带》μ人t 、 、电何便w越少。因此,該昇壓通道之 中的電壓將會快速下降 ^而弓丨起程式化干擾。GIDL之 另一可能的副作用係可能合 H 』此曰產生熱載子(電子與電洞兩 者)此荨熱載子可祐、、t人& κ、,^ -# ^ - .. /牙隧乳化物區之中或被注入鄰 近5己fe體凡件之浮動閘極之 科。 干並且因而會引起程式化干 因⑽L所造成的相同種類的程式化干擾會發生在該 123781.doc 200832411 NAND串之邊緣處的複數條字元線上,例如一 32_nand串 之中的WL0、WL1、WL3G、以及WL31。因為源極與沒極 側選擇閘極均被偏壓在低電壓處(對源極側來說通常係〇 V,而對汲極側來說通常係2·5 v),所以GIDL會發生在選 — 擇閘極汲極邊緣,其會導致昇壓電壓被放電。對較長的 ’ NAND串來說,此問題會變得更嚴重。 倘若抹除源極侧鄰近記憶體元件的話,那麼便會在浮動 φ 閘極上存在一正電荷,而且該電晶體之臨界電壓將很可能 係負的。所以,即使施加〇 v於該字元線,該電晶體仍可 能不會關閉。倘若該記憶體元件係開啟的,那麼該NAND 串便不會運作在EASB模式之中。更確切地說,該1^八1^〇串 係運作在自昇壓模式之中,此可能會具有如以上說明的昇 壓不足之問題。倘若程式化其它源極側元件最可能會出現 此情形,此會限制源極侧昇壓。此問題係較短通道長度的 最大問題,因為更可能會出現洩漏的問題。 • 圖4為一可用以施行本揭示内容之一或多個具體實施例 的快閃記憶體系統之一具體實施例的方塊圖。亦可使用其 它系統與施行方式。記憶體元件陣列302係受控於行控制 • 包路304、列控制電路3、c源極控制電路3 1 〇、以及p井 ^ 控制電路308。行控制電路3〇4會被連接至記憶體元件陣列 3〇2的位兀線,用以讀取被儲存在該等記憶體元件之中的 2料,用以決定該等記憶體元件在一程式化操作期間的狀 忍’以及用以控制該等位元線的電位位準以促成或禁止進 订程式化與抹除。列控制電路306會被連接至該等字元線 123781.doc -20- 200832411 以選擇該等字元線中的其中一者、施加讀取電壓、施加與 :行控制電路3〇4控制的位元線電位位準組合之程式化電 ί二V:除電壓。C源極控制電路31°會控制與該 等圯憶體兀件連接的一丘用、、盾托始,卢面 ^ /、用源極線(在圖4中標記為"C源 極)。Ρ井控制電路3 〇 8會控制ρ井電塵。 被儲存在料記憶體元件之巾的諸會由該行控制電路 3〇4讀出並且會經由資料輸入/輸出緩衝器312被輸出至外 部⑽線。要被儲存在該等記憶體元件之中的程式資料合 經由外部亂線被輸入至資料輸入/輸出緩衝器312,並且合 輸至行控制電路304。該等外部1/〇線會被連接至控: 益 3 1 8。 /於控制該快閃記憶體裝置的命令資料會被輸入至控制 σ中4等°卩令貧料會通知該快閃記憶體需要進行 何種刼作。該輸入命令會被傳輸至狀態機,苴係控制 =控制電路3〇6、C源極控制電路31〇、P井控制電路308及 貝料輸入/輸出緩衝器3 12。狀態機3 16亦可輸出該快閃記 憶體的狀態資料,如準備/忙綠或通過/失敗。 器318會被連接至或可連接至一主機系統,例如個 *婁文位相冑、個人數位助理、或是類似的主機系 广^制裔3 1 8會與該主機相通以接收命令(例如將資料儲 次P己體陣列302之中或是從該記憶體陣列302處讀取 务=並且W提供或接收此類資料。控制器3 1 8會將該等 , 、成卩7 ^號並且在控制線上發送它們,以便為控 123781.doc 21 200832411 制電路系統315之部分的命令電路314來解譯與執行。命令 电路3 14 W與狀態機3 1 6進行通信。控制器3 1 8通常含有緩 衝^ 體,供要被寫入該記憶體元件陣列302之中或從該 屺fe體70件陣列3〇2中被讀取的使用者資料使用。 一不範性記憶體系統包括一積體電路,其包含控制器 • 318 ;以及一或多個積體電路晶片,每一者均包含一記憶 體陣列及相關聯的控制器、輸入/輸出及狀態機電路。趨 φ 冑係將4等記憶體陣列與-系統的控制器電路一起整合於 -或多個積體電路晶片之上。該記憶體系統可被嵌入而作 為該主機系統的部分,或者可將其包含於一可以抽取的方 斗式被插入該等主機系統之中的記憶卡(或其它封幻内。此 =卡可能包含整個記憶體系統(舉例來說,其包含該控制 的)或僅包含與周邊電路相關聯的記憶體陣列(將控制器或 控制功能嵌入主機中)。因此,該控制器可被嵌入該主機 内或包含於該可抽取的記憶體系統内。 • 在特定的施行方式中,可組合圖4中的特定組件。進一 步言之,在各種設計中,除了記憶體元件陣列302之外’ 圖4中的一或多個組件均可被視為係一管理電路。舉例來 . 言兒’ -或多個管理電路可能包含控制電路系統、一命令電 • 路、一狀態機、一列控制電路、一行控制電路、一井控制 電路、一源極控制電路、以及一資料1/()電路中任一者或 組合。 圖5提供圖4之記憶體元件陣列3〇2的一範例結構。作為 一範例,本文所述的NAND快閃EEpR〇M會被分成^^個 12378l.doc •22- 200832411 區塊。在一抹除操作中,會同時抹除被儲存於每個區塊中 的資料在項設計中,區塊係會被同時抹除的最小元件 單元。在此範例中,於各區塊中存在8,512行,它們會分 成偶數行與奇數行。該等位元線還會被分成偶數位元線 (BLe)與可數位元線(BL〇)。圖中所示的四個記憶體元件會 串聯連接以形成-NAND串。儘f圖中顯示出在每一 NAND串之中包含四個元件,但是,仍可以使用多於或少Techniques for Non-Volatile Memory, U.S. Patent No. 6,859,397, and U.S. Patent Application Publication No. 2,5/24,939, entitled " Detecting 〇ver programmed Mem〇ry" More information, both of which are incorporated herein by reference. During the programming of a selected memory component, adjacent memory may be inadvertently programmed in a program called stylized interference. For example, when vpgm is applied to a word line, it may be inadvertently stylized - it does not want to be shouted but is the same as a memory element selected for simplification. The memory element on the meta-line can use several techniques to prevent stylized interference. For example, self-boosting (SB) can electrically isolate the channel associated with the unselected bit line and will The conduction voltage (eg Η) V) is applied to the "unselected word line. The unselected word lines meet to the channel associated with the unselected bit lines' such that a voltage (eg, 8V) is present in the channels of the unselected bit lines It will tend to reduce this stylized interference effect. Therefore, the self-boosting technique causes a voltage boost to be present in the channel, which tends to reduce the voltage differential across the pass-through oxide and thus reduce the stylized interference effect. In addition, 'other modified techniques (such as local auto-boost (lsb), modified 123781.doc 200832411-style self-boosting (RLSB), erased area self-boosting (EASB), and modified erased areas Self-boosting (REASB) attempts to reduce stylized interference by isolating the channels of previously programmed components and the channels from which the stylized components are to be disabled. For example, REASB technology uses a very low isolation voltage (for example, 0 V to i V) to drive a word line near the selected word line on the programmed side, and utilizes a A medium voltage (about 3 V) is used to drive a line of words next to the selected word line on the programmed side to isolate the programmed side from the erased side. All other unselected word lines are driven with a conduction voltage. Therefore, two separate boost channels are created. Due to the effect of the threshold voltage of the programmed memory elements, the boosted channel voltage on the programmed side will be 1 V to 2 lower than the boosted channel voltage on the erased side. V, however, is still sufficient to increase the threshold voltage of the isolated word line. As the length of the channel in the memory component continues to decrease, the ability of conventional channel boosting techniques to reduce stylized interference continues to decrease. Specifically, the channel length of the 'memory element' may become too short to adequately isolate the two separate boosted channel regions at the drain and source sides of the selected word line. As a result, the boost channel voltage can be reduced, resulting in stylized interference. In addition, high electric field effects (such as band-to-band tunneling (BTBT) and gate-induced drain leakage (GIDL)) may also cause the booster channel to be discharged, and/or hot carriers due to high-voltage places may Will be injected into the tunneling oxides or into the floating gates of the memory components. Therefore, there is still a need to improve the techniques for stylizing interference problems. SUMMARY OF THE INVENTION 123781.doc 200832411 The present invention discloses a technique for reducing stylized interference effects when staging non-volatile memory. In particular, multiple isolation regions are used to divide a channel associated with a storage element into a plurality of zones. A suitable conduction voltage can then be used to bias each of the plurality of binning regions to help reduce the gates caused by the source select line and the drain select line and the isolated word line.汲 (4) Leakage relationship and the leakage current flowing from the side of the area. These techniques are particularly useful for preventing severe stylized interference effects such as, for example, at the edges of a memory region at the higher and lower word lines of a long string of NAND memories. Μ中- 实实_中' will select—storage components for programming. The channel associated with the set of storage elements is divided into at least three by applying an isolation voltage to the storage element of the proximal end of the selected storage element and to the other storage element that is remote from the health element. : From the Siam area. Therefore, a first one, f, /, and a channel region are disposed on each other == one portion on the opposite side of the selected storage element. The third channel area is :::°Hai: Store 70 pieces. The potential of the third channel region is boosted by the boost voltage of the first channel region, which is lower than 忒. In a preferred embodiment, the isolation voltage is generated by the third storage element of the selected storage element, thus = the selected fourth generation channel region is generated. In-step, ; = = can be applied to multiple storages in order to create multiple isolations;; = in the specific embodiment, the __ standard to the location and the selected eclipse _ , book & storage and storage of 70 proximal channel regions corresponding to the storage element 123781.doc -10- 200832411, and will apply a lower conduction voltage to the channel region located away from the selected component Storage (four). In a preferred embodiment, the conduction voltage applied to the channel region at the proximal end of the selected storage element will be higher than the standard conduction voltage. The features and advantages of the present invention will become more apparent from the detailed description and the appended claims. [Embodiment] The present disclosure relates to techniques for reducing stylized interference, particularly at word lines at the edges of a NAND memory string. The basic concept for reducing stylized interference is to provide an additional isolation region along the NAND memory string in order to reduce the conduction voltage applied to the word line at the edge of the νανε^ memory string. An example of a non-volatile memory system suitable for practicing the present invention is the use of a NAND flash memory structure in which a plurality of transistors are arranged in series between two select gates in a NAND string. The figure "shows a top view of Figure 1. Figure 2 is the equivalent circuit. The Nand string described in Figure 2 and Figure 2 contains four transistors 1 〇〇, 1 〇 2, 1 〇 4, and 1 〇6, which will Connected in series and sandwiched between a first selection gate 12A and a second selection gate 122. The selection gates 120 and 122 respectively connect the NAND string to the bit line contact 126 and the source line Point 128. Select gates 12 and 122 are controlled by applying appropriate voltages to control gates 120CG and 122CG, respectively. Each of transistors 1 , 102 , 104 , and 1 6 has a control gate The pole 100 has a floating gate. The transistor 100 has a control gate 1 〇〇 CG and a floating gate 123781.doc 200832411 1 〇〇 FG ° transistor 102 includes a control gate 102CG and a floating idle pole 1 〇 2FG ° transistor 1 〇4 includes control gate i〇4CG and floating gate 4 FG gas crystal 1 〇6 contains a control gate 1 〇6 CG and floating idler 106FG. Control gate 1〇〇(^, i〇2CG, 1〇 4CG and 1〇6CG are respectively connected to the word lines WL3, WL2, WL1 and WL0. In a feasible setting, the transistors 1〇〇, 1〇2, 104 and 106 are respectively Memory elements or storage elements. In other designs, the memory elements may contain multiple transistors or may differ from those described in Figures 2 and 2. Select gate 12〇 will be connected to the drain Line SGD, and select gate 122 is connected to source select line SGS. Note that although Figures 1 through 2 show four memory elements in the NAND string, four transistors are used as one Example. A NAND string used in conjunction with the techniques described herein may also use four or fewer memory elements or more than four memory elements. For example, a particular nand string will be packaged in three eight or sixteen. Twenty-two, sixty-four or more memory elements. The discussion herein is not limited to any particular number of memory elements in a NAND string. In general, the invention can be used by Fowler-Nord A device for staging and erasing the dental tunnel. The invention is also applicable to devices using a multilayer dielectric stack, such as yttrium oxide, cerium nitride, and yttrium oxide ( ΟΝΟ) the dielectric formed, Or a dielectric formed of yttrium oxide, a high-k layer, and yttrium oxide. A multilayer dielectric is sandwiched between a conductive control idler and a semiconductor substrate above the memory device channel. Between the surfaces. For example, the invention can also be applied to make 123781.doc -12· 200832411 use a small conductive material island (such as nanocrystals) and replace the floating gate with a charge trapping layer (such as tantalum nitride). As a device for the charge storage area, such a memory device can be programmed and erased in the same manner as a NAND flash device that is very dominant in floating gates. Figure 3 is a circuit diagram depicting three NAND strings. A typical architecture for a flash memory system using a NAND structure includes several NAND strings. For example, three NAND strings 201, 203, and 205 are shown in a memory array with more NAND strings. Each of the NAND strings includes two select transistors and four memory elements. For example, NAND string 201 includes select transistors 220 and 230, and memory elements 222, 224, 226, and 228. NAND string 203 includes select transistors 240 and 250, and memory elements 242, 244, 246, and 248. NAND string 205 includes select transistors 260 and 270, and memory elements 262, 264, 266, and 268. Each NAND string is connected to the source line by its selection transistor (e.g., selection transistor 230, 250 or 270). A select line SGS is used to control the source side select gates. Each of the NAND strings 201, 203, and 205 can be connected to individual bit lines 202, 204, and 206 by select transistors 220, 240, 260, etc., controlled by the drain select line SGD. In other embodiments, the select lines do not have to be shared. Word line WL3 is coupled to the control gates of memory elements 222, 242, and 262. Word line WL2 is coupled to the control gates of memory elements 224, 244, and 264. Word line WL1 is coupled to the control gates of memory elements 226, 246, and 266. Word line WL0 is coupled to the control gates of memory elements 228, 248, and 268. From the figure, you can see 123781.doc -13- 200832411 out of each of the 70 lines and the individual nand string including the line in the memory element array or group. The word lines (WL3, WL2, WL1, and WL0) include columns in the array or group. Each word line connects the control gate of each memory element in the column. For example, word line WL2 will be coupled to the control gates of memory elements 224, 244, and 264. Each memory component can store data. For example, when storing a digit of a digital cradle, 'the range of feasible threshold voltages of the memory component will be cut into two ranges'. These ranges will be assigned logical data values "1,,,,,〇 ,,. In an example of a NAND type & flash memory, the threshold voltage after erasing the memory element is a negative value and is defined as a logical "丨". The threshold voltage after a one-shot operation is positive and will be defined as logic, 〇". When the critical voltage is negative and an attempt is made to read, the memory component will be turned on to indicate that the logic '1" is being stored. When the threshold voltage is positive and a read operation is attempted, the memory component will not turn on, and this indication will store the logic, 〇, ,. A memory component can also store multiple levels of information, for example, multi-bit digital data. In this case, the range of the threshold voltage is divided into the number of levels of the material. For example, if four information levels are stored, four threshold voltage ranges are assigned to the data values "u,,,q〇π, "〇1" and "00". In an example of a NAND type memory, the threshold voltage after the erase operation is a negative value and is defined as "丨丨". The positive threshold voltage is used for the states ""1〇",,,〇1", and "00". The particular relationship between the data programmed into the memory component and the threshold voltage range of the component depends on the data encoding scheme employed by the memory components. For example, US Patent No. 6 222 762 123781.doc -14-200832411 entitled "'Novel Multi-State Memory," and US Patent Application No. 2004 entitled "Tracking Cells For A Memory System" The two cases of /0255090 illustrate various data encoding schemes for multi-state flash memory components. The entire contents of both cases are incorporated herein by reference. Related Examples and Operation of NAND Flash Memory It is provided in the following reference patents: U.S. Patent No. 5,386,422, U.S. Patent No. 5,522,580, U.S. Patent No. 5,57,315, U.S. Patent No. 5,774,397, U.S. Patent. No. 6, pp. 46, 935, U.S. Patent No. 6, 45, 528, and U.S. Patent No. 6,522,58; each of which is incorporated herein by reference in its entirety. A stylized voltage is applied to the control gate of the component, and the bit line associated with the component is grounded. Electrons from the channel are injected into the floating gate. When electrons accumulate in the floating brake In the middle of the pole, the floating gate becomes negatively charged and the threshold voltage of the component rises. To apply the stylized voltage to the control gate of the component being privateized, the stylized voltage is Applied to the appropriate word line. As discussed above, the word line is also connected to one of each of the other NAND strings that share the same word line. For example, when stylizing Figure 3 When element 224, the stylized voltage Vpgnj^ is applied to the control gates of elements 224, 244, and 264. When it is desired to program one of the elements on a word line without stylizing the other connected to the same word line In the case of a component, for example, when it is desired to program the component 224 without programming the component 244, a problem arises because the stylized voltage is applied to all components connected to a word line, so the character Online 123781.doc •15· 200832411 Unselected components (that is, components that are not selected for stylization) may be inadvertently programmed in a program commonly referred to as stylized interference. Come When stylizing component 224, adjacent components 244 or 264 that may be of concern may be inadvertently stylized. Note that stylized interference is most likely not present on a character line selected for stylization. On the memory component. However, in certain cases, stylized interference may also occur in memory components other than the selected word line, and several techniques can be used to prevent stylized interference. Boost (SB), which electrically isolates the channels associated with the unselected bit lines, and applies a conduction voltage (eg, 〇V) to the unselected during stylization The word line. The unselected word lines are coupled to channels associated with the unselected bit lines such that a voltage (eg, 8 V) is present in the unselected bit lines In the channel, this tends to reduce stylized interference. Thus, self-boosting causes voltage boost to be present in the channel, which tends to reduce the voltage across the tunneling oxide and thereby reduce stylized interference. A NAND string will typically correspond to the sequential configuration of the word lines from WL 〇 to WL 3 from which the source side to the drain side are programmed, for example, from memory element 228 to memory element 222. As an example, assume that all or most of the memory elements in adjacent NAND strings 203 have been programmed before all or most of the memory elements in the ΝΑΝβ string 201. Therefore, when the stylized program is ready to program the next few memory elements in the NAND string 2〇1, the floating of the previously programmed memory elements above the gate of 123781.doc -16·200832411 There will be a negative charge. Therefore, the boost potential using standard SB techniques may not be large enough in portions of the NAND string 203 to avoid the effects of programmed interference on the elements associated with the subsequent number of word lines in the nand string 203. For example, when the component 222 on the NAND string 201 is programmed, if the components 248, 246, and 244 on the previously programmed NAND string 203 are then 'each of the transistors (244, 24 6 and 248) One will have a negative charge on their floating gates, which will limit the boost level of the self-boosting process and may cause stylized interference on component 242. Both local self-boosting (LSB) and erased area self-boosting (EASB) attempt to solve the shortcomings of conventional self-boosting by isolating the channels of previously programmed components from the channels of the disabled components. Fix the technology. For example, if element 224 of Figure 3 is being programmed, then LSB and EASB will attempt to disable stylization in element 244 by isolating the channel of element 244 from previously programmed elements (246 and 248). . For SB, EASB, and LSB boosting methods or variations of these boosting methods, the bit line of the component to be programmed will be connected to ground or another low voltage (usually 〇V to ^ v), while the bit line of the NAND string with the elements to be disabled is at Vdd, which is typically in the range of h5 乂 to 3 v. The stylized voltage VPgm (for example, 20 V) is connected to the selected word line. In the case of the LSB boost mode, the word line next to the selected word line is set to 〇V or close to 0 V, while the remaining unselected word lines are $ 疋Conduction voltage vpass. For example, bit line 2〇2 is at 〇V and bit line 2〇4 is at vdd. The drain selection line sgd is located at 123781.doc -17- 200832411 at Vsgd (usually 2.5 V to 4.5 V) and the source select line SGS is at 〇v ° selected word line WL2 (for stylized components) 224) is at vPgm. The adjacent word lines WL1 and WL3 are at 〇v, while the other word lines (for example, WLO) are at V_s. The disadvantage of the LSB mode is that the boosted channel voltage below the selected word line can be very high, since some of the channel will be isolated from other channel areas below the unselected word line, and therefore the main The boost voltage is determined by the high program voltage vpgm. Due to this high boost relationship, BTBT or GIDL may appear near the word line biased to 〇v, and the resulting hot carriers may be injected into the selected words in the unselected NAND string. In the middle of the line. The amount of channel boost can be well controlled using this easb method. The EASB and SB are identical, with the only exception that the source side of the selected word line is biased at 〇 V. Therefore, the channel region below the selected sub-element and the channel region at the drain side of the selected component are connected, and therefore, the channel boosting is applied to have an erased state. The voltages of these unselected word lines are steadily determined 'because the boost is not affected by the data that has been previously programmed. If Vpass is too low, the boost in that channel will not be sufficient to prevent stylized interference. If the vpass is too high, it may be programmed to select an unselected word line in the selected NAND string (the bit line is Ο V)' or may interfere with the unselected NAND string due to GIDL. Selected child line. For example, WL1 would be at 〇v instead of Vpass, while WL3 would be at Vpass. In a specific embodiment, Vp_ is 7 V to 10 V. 123781.doc -18- 200832411 Although there is a problem with LSB and EASB providing improved boost, the problem will depend on the proximity of the source on the source side (the source 246 is the source side of the component 244). To be stylized or erased. If the source side adjacent component is in a programmed state, it will have a negative charge on its floating gate. Furthermore, applying 〇V to the control gate and combining it with a highly reverse bias junction under the negative gate (due to the boosting relationship) may cause the gate to induce a drain/3⁄4 drain. (GIDL) 'The electrons leak into the boost channel. GIDL is accompanied by a large bias voltage on the junction and a low or negative gate caused by the south voltage on the //pole/source regions of the memory elements due to the boost relationship. The pole voltage is $ see, which is the case when the source side is clamped and the drain junction is boosted. This boost voltage may be prematurely leaked out, resulting in a stylized dry k and the stomach P is now more severe with a sudden and highly doped junction (which is needed when the component size is reduced). If the _ stream is high enough, the boost potential in the channel region may be reduced, causing stylized interference. Moreover, the closer the word line being programmed is to the immersed pole, the φ smear that appears in the boost channel region, and the less the power, the less. Therefore, the voltage in the boost channel will drop rapidly and the staggered stylized interference. Another possible side effect of GIDL is that H can produce hot carriers (both electrons and holes). This hot carrier can be blessed, t human & κ,, ^ -# ^ - .. / In the dental tunnel emulsion zone or in the vicinity of the floating gate of the adjacent body. The same kind of stylized interference caused by the (10)L will occur on the complex word lines at the edge of the 123781.doc 200832411 NAND string, such as WL0, WL1 in a 32_nand string. WL3G, and WL31. Since the source and the gateless selection gate are both biased at a low voltage (usually 〇V for the source side and 2·5 v for the drain side), GIDL will occur at Select - Select the gate pole edge, which will cause the boost voltage to be discharged. For longer 'NAND strings, this problem becomes more serious. If the source side adjacent memory element is erased, then a positive charge will be present on the floating φ gate and the threshold voltage of the transistor will likely be negative. Therefore, even if 〇 v is applied to the word line, the transistor may not be turned off. If the memory component is turned on, the NAND string will not operate in the EASB mode. More specifically, the 1^81 system operates in the self-boost mode, which may have the problem of insufficient boost as explained above. This is the most likely case for stylized other source side components, which limits the source side boost. This problem is the biggest problem with shorter channel lengths because leaks are more likely to occur. Figure 4 is a block diagram of one embodiment of a flash memory system that can be used to implement one or more embodiments of the present disclosure. Other systems and implementation methods are also available. The memory element array 302 is controlled by row control • packet 304, column control circuit 3, c source control circuit 3 1 〇, and p well ^ control circuit 308. The row control circuit 3〇4 is connected to the bit line of the memory device array 3〇2 for reading the two materials stored in the memory elements for determining the memory elements in a The shape and duration of the stylized operation and the potential level used to control the bit lines to facilitate or prohibit the ordering and erasing of the order. Column control circuit 306 is coupled to the word line 123781.doc -20-200832411 to select one of the word lines, apply a read voltage, apply a bit with the control of row control circuit 3〇4 The stylized combination of the potential of the potential of the line is 二V: the voltage is removed. C source control circuit 31 ° will control the connection with the memory element, the shield, the Lu, the source line (marked as "C source in Figure 4) . The well control circuit 3 〇 8 will control the ρ well electric dust. The portions of the towel stored in the memory device are read by the row control circuit 3〇4 and output to the outer (10) line via the data input/output buffer 312. The program data to be stored in the memory elements is input to the data input/output buffer 312 via external scramble lines, and is output to the line control circuit 304. These external 1/〇 lines will be connected to the control: Benefit 3 1 8. / The command data for controlling the flash memory device is input to the control σ, and the poor material informs the flash memory what kind of action is required. The input command is transmitted to the state machine, which is controlled by the control circuit 3〇6, the C source control circuit 31〇, the P well control circuit 308, and the billiple input/output buffer 312. The state machine 3 16 can also output status data of the flash memory, such as ready/busy green or pass/fail. The device 318 can be connected to or can be connected to a host system, such as a digital device, a personal digital assistant, or a similar host system. The 3 3 8 will communicate with the host to receive commands (eg, The data storage sub-array array 302 is either read from the memory array 302 and provides or receives such data. The controller 3 1 8 will, for example, become a 7^ number and The control lines send them for interpretation and execution for the command circuit 314 that is part of the control circuit system 315. The command circuit 3 14 W communicates with the state machine 3 16 6. The controller 3 1 8 typically contains The buffer body is used for user data to be written into or read from the memory element array 302. The non-standard memory system includes an integrated body. a circuit comprising a controller 318; and one or more integrated circuit chips, each comprising a memory array and associated controller, input/output and state machine circuits. The memory array is integrated with the controller circuit of the system - or a plurality of integrated circuit chips. The memory system can be embedded as part of the host system or can be included in a removable card that is inserted into the host system (or other phantoms. This = card may contain the entire memory system (for example, it contains the control) or only the memory array associated with the peripheral circuitry (embed controller or control functions in the host) Therefore, the controller can be embedded in the host or included in the extractable memory system. • In a specific implementation, the specific components in Figure 4 can be combined. Further, in various designs, In addition to the memory element array 302, one or more of the components of Figure 4 can be considered as a management circuit. For example. - or multiple management circuits may contain control circuitry, a command circuit. A circuit, a state machine, a train of control circuits, a row of control circuits, a well control circuit, a source control circuit, and a data 1/() circuit. Figure 5 provides the memory of Figure 4. An example structure of the component array 3〇2. As an example, the NAND flash EEPR〇M described herein will be divided into 12128l.doc • 22-200832411 blocks. In an erase operation, it will be erased at the same time. The data stored in each block is in the item design, and the block is the smallest component unit that will be erased at the same time. In this example, there are 8,512 rows in each block, which are divided into even rows and odd rows. The bit lines are also divided into even bit lines (BLe) and count bit lines (BL〇). The four memory elements shown in the figure are connected in series to form a -NAND string. Shows that there are four components in each NAND string, but you can still use more or less

於四個記憶體元件。實際上,參考圖14至17便明確地涵蓋 較長串的具體實施例。該NAND串的其中一個端子會經由 -選擇電晶體SGD被連接至一對應的位元線,@另一個端 子會經由一第二選擇電晶體SGS被連接至該^源極線。 在項取及耘式化操作之一組態期間,會同時選擇6 個記憶體兀件。書亥等被選擇的記憶體元件會具有同一字元 線及同一種類的位元線(例如偶數位元線或奇數位元線 所以’可同時讀取或程式化會形成_邏輯頁的阳個資料 位7G組,亚且該記憶體中的一區塊可以儲存至少八個邏輯 頁(四條字元線,每條字元線各具有奇數頁與偶數頁)。對 多狀態記憶體元件而言,當每個記憶體元件儲存兩個資料 位几時(其中將此等兩個位元中每一者會被儲存在不同頁 之中)’ -個區塊便可儲存十六個邏輯頁。本發明亦可使 用其它大小的區塊及頁。此外,亦可使用圖4及5以外的架 構來施行本發明。舉例來說,在—項設計中,料位元: 並不會被分成奇數與偶數位元線,俾使可同時(或不同护 程式化並讀取所有位元線。 ^ 123781.doc -23- 200832411 藉由將P井升高至抹除電壓(例如2〇 v)並將一被選擇的區 塊之字元線接地便可抹除記憶體元件。源極與位元線係浮 動的。亦可在整個記憶體陣列、分離區塊、或該記憶體裝 置之一部分的另一記憶體元件單位上來實施抹除。電子會 , 從該等記憶體單元之浮動閘極被傳輸至p井區,以便讓該 , 等記憶體元件之臨界電壓變為負的。 在讀取與驗證操作中,選擇閘極(SGD及SGS)會被連接 φ 至2·5 V至4·5 V範圍中的電壓,而該等未被選擇之字元線 (例如當WL1係被選擇的字元線時,便係WL〇、與 WL3)則會被升高至一讀取傳導電壓(通常係一落在4·5 v至 6 V範圍中的電壓)以使該等電晶體作為傳導閘極來運作。 被選擇之字元線WL1會被連接至一電壓,該電壓之位準係 針對每個讀取及驗證操作而指定,以便判斷相關記憶體元 件之臨界電壓究竟係高於還係低於此位準。舉例來說,在 一雙位準記憶體元件的讀取操作中,該被選擇的字元線 • WL1可能會被接地,所以便可判斷該臨界電壓高於〇 v。 舉例而言,在一雙位準記憶體元件的驗證操作中,該被選 擇的,元線WL1會被連接至〇·8 v,所以其便可驗證該臨界 . 電堡疋否已經抵達至少U V。源極與ρ井係處於〇 ¥處。該 等被選擇的位元線(假設係偶數位元線(BLe))會被預充電至 (例如)0·7 V的位準。偏若該臨界電塵高於該字元線上的讀 取或驗證位準的話,那麼由於該未導通記憶體元件的關 係,^關注的元件相關聯的位元線(BLe)的電位位準便會 保持问位準。另一方面,倘若該臨界電壓係低於讀取或驗 123781.doc -24 - 200832411 —妨*的a㈣因為該導電記憶體元件會對該位元線進 :位進的關係’相關位元線(BLe)之電位位準便會降低至 線舉例來說,小於G.5 V。因&,可藉由與該位元 接的-電墨比較器感測放大器來偵測該記憶體元件之 狀恶。 已述的抹除、讀取、収驗證操作係根據本技術中 0、技*來實施。因Λ,熟習本技術的人士便可改變上 面所解釋的眾多細節。亦可使用本技術中已知的其它抹 除、讀取、以及驗證操作。 /、 如上所述,可以將每個區塊劃分成數頁。在其中一種方 式中’ H程式化單位。在特定施行方式中,可將個 別頁劃分成複數個片斷並且該^斷可能包含在進行一基 本程式化操作會一次被寫入的最少數元件。通常會將一: 多個資料頁儲存在—列記憶體元件之中…頁可以儲存一 或多個區段。一區段包含使用者資料與管理資料(晴—ad data) ’例如已從㈣段之使用者㈣巾所算出的錯誤校正 碼(ECC)。當資料正在被程式化至該陣列之中冑,該控制 器之一部分便會計算該ECC,並且當從該陣列之中讀=資 料時還使用該ECC來檢查該資料。或者,該等Ecc及/或^ 它管理資料會被儲存在和它們相關的使用者資料不同的頁 或甚至係不同的區塊之中。在其它設計中,該記憶體裝置 的其它部分(例如狀態機)則可以計算ECC。 使用者資料的一區段通常係512個位元組,其會對應於 磁碟機中的一區段的大小。管理資料通常係額外的16至2〇 個位元組。龐大數量的頁會形成一區塊,無論在何處其均 123781.doc -25 · 200832411 包含8頁至(例如)最多32、64或更多頁。 圖6描述一示範性程式化脈衝波形。該程式化電壓v P gmFor four memory components. In fact, a particular embodiment of a longer string is explicitly covered with reference to Figures 14-17. One of the terminals of the NAND string is connected to a corresponding bit line via a select transistor SGD, and the other terminal is connected to the source line via a second select transistor SGS. During the configuration of one of the item selection and the simplification operation, 6 memory elements are selected at the same time. The selected memory elements such as Shuhai will have the same word line and the same type of bit line (for example, even bit lines or odd bit lines, so 'can be read or programmed simultaneously to form a positive page of _ logical pages Data bit 7G group, and a block in the memory can store at least eight logical pages (four word lines, each of which has odd and even pages). For multi-state memory components When each memory element stores two data bits (where each of these two bits is stored in a different page) - a block can store sixteen logical pages. Other sizes of blocks and pages may also be used in the present invention. In addition, the present invention may be implemented using architectures other than those of Figures 4 and 5. For example, in the item design, the material bits are not divided into odd numbers. With even bit lines, you can program (and program different bit lines simultaneously. ^ 123781.doc -23- 200832411 by raising the P well to the erase voltage (eg 2〇v) and The memory component can be erased by grounding the word line of a selected block. Floating with the bit line system. The erasing can also be performed on the entire memory array, the separation block, or another memory element unit of one of the memory devices. The electrons will float from the memory cells. The gate is transferred to the p-well region so that the threshold voltage of the memory component becomes negative. In the read and verify operations, the select gate (SGD and SGS) is connected to φ to 2·5 V. Voltages in the range of 4·5 V, and the unselected word lines (eg, when WL1 is selected as the word line, WL〇, and WL3) are raised to a read Conduction voltage (usually a voltage falling in the range of 4·5 v to 6 V) to operate the transistors as conductive gates. The selected word line WL1 is connected to a voltage, which is The level is specified for each read and verify operation to determine if the threshold voltage of the associated memory component is above or below this level. For example, reading a dual level memory component In operation, the selected word line • WL1 may be grounded, so it can be judged The threshold voltage is higher than 〇v. For example, in the verification operation of a two-level memory element, the selected element line WL1 is connected to 〇·8 v, so it can verify the threshold. The electric castle has arrived at least UV. The source and the ρ well are at 〇¥. The selected bit lines (assuming the even bit line (BLe)) are precharged to (for example) 0· Level of 7 V. If the critical dust is higher than the read or verify level on the word line, then due to the relationship of the non-conducting memory elements, the bit line associated with the element of interest (BLe) The potential level will remain at the level of the question. On the other hand, if the threshold voltage is lower than the reading or verification of 123781.doc -24 - 200832411 - a * (four) because the conductive memory component will be the bit Line advance: The relationship of the bit-in relationship of the potential bit line (BLe) will be reduced to the line, for example, less than G.5 V. Because &, the memory element can be detected by the --ink comparator sense amplifier connected to the bit. The erase, read, and verify operations described above are implemented in accordance with the techniques of the present technology. Because of this, those skilled in the art can change many of the details explained above. Other erase, read, and verify operations known in the art can also be used. /, As described above, each block can be divided into several pages. In one of the ways 'H stylized units. In a particular implementation, the individual pages can be divided into a plurality of segments and the breaks may contain the fewest components that are written once at a time for a basic stylization operation. Usually one will be: Multiple data pages are stored in the - column memory component... The page can store one or more segments. One section contains user data and management data (e.g., ad data), for example, an error correction code (ECC) calculated from the user (4) of the paragraph (4). When the data is being programmed into the array, the ECC is calculated in one part of the controller and is also used to check the data when reading the data from the array. Alternatively, the Ecc and/or ^ management data will be stored in a different page or even a different block from their associated user profile. In other designs, other parts of the memory device (e.g., state machines) can calculate ECC. A section of user data is typically 512 bytes, which would correspond to the size of a sector in the drive. Management data is usually an additional 16 to 2 bytes. A large number of pages form a block, wherever it is 123781.doc -25 · 200832411 contains 8 pages to, for example, up to 32, 64 or more pages. Figure 6 depicts an exemplary stylized pulse waveform. The stylized voltage v P gm

會被分成眾多脈衝。該等脈衝之幅度會隨著每一脈衝而增 加一預定的步進大小。於包含該等記憶體單元儲存一資料 位元的一具體實施例中,步進大小的範例係0 · 8伏特。於 包含該等記憶體單元儲存多個資料位元的一具體實施例 中,步進大小的範例係〇·2伏特。程式化電壓Vpgm2起始位 準的一範例係12 V。當嘗試禁止一單元被程式化時,便會 藉由將位元線電壓提升至Vdd來隔離該通道並且該通到會 藉由該傳導電壓(Vpass)來昇壓。於包含該等記憶體單元儲 存多個資料位元的特定具體實施例中,傳導電壓的振 幅可能不會向上步進。 在该等脈衝之間的週期中會進行驗證操作。也就是,正 在被平行程式化的每一單元的已程式化彳立準會在每一程式 化脈衝之間被讀取以判斷其究竟是等於還是大於正在被程 式化的驗證位準。舉例來說,倘若臨界電壓上升至2·5伏 特枯,那麼該驗證程序便會判斷臨界電壓是否至少為2· $ 伏特。倘若其判斷出_給定記憶體單元的臨界電遂已經超 過該驗證位準的話,那麼該單元的該NAND串的位元線電 尾便會k 〇 v被提升至Vdd。要被平行程式化的其它單元的 知式化操作會繼續進行,直到它們陸續抵達它們的驗證位 準為止。 圖7係說明用於程式化一 、匕°己隱體的一方法的一具體實施 例的流程圖。在其中一藉無 種只仃方式甲,會在程式化之前先 123781.doc •26- 200832411Will be divided into a number of pulses. The amplitude of the pulses increases by a predetermined step size with each pulse. In a specific embodiment in which the memory cells are stored in a data bit, an example of a step size is 0. 8 volts. In a specific embodiment comprising a plurality of data bits for storing the memory cells, the example of the step size is 〇 2 volts. An example of the starting position of the stylized voltage Vpgm2 is 12 V. When an attempt is made to disable a unit from being programmed, the channel is isolated by raising the bit line voltage to Vdd and the pass is boosted by the conduction voltage (Vpass). In a particular embodiment including the memory cells storing a plurality of data bits, the amplitude of the conducted voltage may not step up. A verify operation is performed during the period between the pulses. That is, the programmed finalization of each unit being parallelized is read between each stylized pulse to determine if it is equal to or greater than the verification level being programmed. For example, if the threshold voltage rises to 2.5 volts, the verification procedure determines if the threshold voltage is at least 2 volts. If it is judged that the critical power of the given memory cell has exceeded the verify level, then the bit line tail of the NAND string of the cell will be boosted to Vdd. The customisation of other units to be parallelized will continue until they reach their verification level. Figure 7 is a flow chart illustrating a specific embodiment of a method for stylizing a corpus. In one of them, there is no way to do it, it will be before the stylization 123781.doc •26- 200832411

抹除記憶體單元(以區塊或其它單位來進行)。在圖7的步驟 55〇t ’控制器318會發出"資料載入"命令,並且將其輸入 至貢料輸入/輸出緩衝器312之中。因為—命令鎖存信號(圖 中並未顯示)會被輸人至命令電路314之中,所以該輸入資 料會被認定係-命令並且會被狀態機316鎖存。在步驟W 中’會將表示該頁位址的位址資料從控制器318處輸入至 資料輸入/輸出緩衝器312。因為該位址鎖存信號會被輸入 至命令電路3 14之中,所以該輸入資料會被認定係頁位址 並且會被狀態機3丨6鎖存。在步驟554中,會將532個位元 組的程式化資料輸入至資料輸入/輸出緩衝器312之中。該 資料會被鎖存至該等被選擇位元線的資料儲存暫存器之 中。於特定的具體實施例中,還會使用該資料來進行驗證 操作。在步驟556中,控制器318會發出,,程式化”命令,並 且將其輸入至資料輸入/輸出緩衝器312之中。因為該命令 鎖存信號會被輸入至命令電路314之中,所以該命令會被 狀態機3 16鎖存。 被該”程式化”命令觸發之後,被鎖存在該等資料儲存暫 存為之中的資料便會使用圖6脈衝中所示的步進程式化而 被程式化至受控於狀態機3 16的該等被選擇記憶體單元之 中。在步驟558中,Vpgm會被初始化至起始脈衝(舉例來 說,12 V)且狀態機316所保留的程式化計數值Pc會被初始 化至0。在步驟560中,會將第一 Vpgm脈衝施加至該被選擇 的字元線。倘若邏輯被儲存在一特殊的資料儲存暫存 器之中的話,那麼對應的位元線便會被接地。在另一方 123781.doc -27- 200832411 倘右邏輯”1,,被儲存在 m ^ ^ ㈣廿W存器之中的每 "應的位元線便會被連接至Vdd’用以禁止 。 ▲在步驟562中,會驗證該等被選擇之記 :_。、 態。倘若偵測出__選定單元@ …早70的狀 準(舉例來說,H… 界電盤已達到適當位 t… 之邏輯”°',或-特殊狀態的程 式位準),那麼所儲存的資料便會被改變為邏輯"卜倘 右福測到該臨界電壓尚未達到該適當位準的Erasing the memory unit (in blocks or other units). At step 55 〇t ' of Fig. 7, the controller 318 issues a "data load" command and inputs it into the tribute input/output buffer 312. Because the command latch signal (not shown) will be entered into command circuit 314, the input data will be asserted and commanded and will be latched by state machine 316. In step W, the address data representing the page address is input from the controller 318 to the data input/output buffer 312. Since the address latch signal is input to the command circuit 314, the input data is asserted as a page address and is latched by the state machine 3丨6. In step 554, the 532 bytes of stylized data are entered into the data input/output buffer 312. The data is latched into the data storage registers of the selected bit lines. In certain embodiments, this material is also used for verification operations. In step 556, the controller 318 issues a "program" command and inputs it into the data input/output buffer 312. Since the command latch signal is input to the command circuit 314, The command will be latched by state machine 3 16. After being triggered by the "stylized" command, the data latched in the data store temporary will be stylized using the step shown in the pulse of Figure 6. Programmized into the selected memory cells controlled by state machine 3 16. In step 558, Vpgm is initialized to the start pulse (e.g., 12 V) and the program retained by state machine 316 The count value Pc will be initialized to 0. In step 560, a first Vpgm pulse will be applied to the selected word line. If the logic is stored in a particular data storage register, then the corresponding The bit line will be grounded. On the other side, 123781.doc -27- 200832411, if the right logic is "1," each bit stored in the m ^ ^ (four) buffer will be the bit line. Connected to Vdd' to disable. ▲ In step 562, the selected records are verified: _. State. If the status of __selected unit @ ... early 70 is detected (for example, H... the interface has reached the logic of the appropriate bit t...°, or - the program level of the special state), then stored The information will be changed to logic " If right, the threshold voltage has not yet reached the appropriate level.

會改變被儲存在該資料儲存暫存 么便不 飞什^之中的貧料。依此方 ’,’必程式化在其對應的資料儲存暫存器中儲存 田所有貝枓儲存暫存器均係儲存邏輯"厂 時’該狀態機便會知道已經程式化所有被選擇的單元。在 步驟564中,會檢查是否所有資料儲存暫存器均係储存邏 輯1 。右是’那麼該程式化程序便已完成且成功,因為 所有被選擇的記憶體單元皆已被程式化及驗證。在步驟 366中會報告一”通過”狀態。 倘若在步驟564中判斷出並非所有資料儲存暫存器均係 儲存邏輯Τ’的話,那麼便會繼續該程式化程序。在步驟 568中θ依^耘式化限制數值來檢查程式化計數值 PC。程式化限制數值的其中—個範例為2G。偽若該程式化 計數值PC不小於20的話,那麼該程式化程序便已經失敗, 並且會在步驟570中報告一”失敗,,狀態。倘若該程式化計 數值PC小於20的話,那麼便會在步驟572中將Vpp位準增 加該步進大小並且遞增該程式化計數值pc。在步驟572之 後,該程序便會返回步驟56〇,以便套用下一個v 123781.doc -28- 200832411 衝。 圖7中所描述的鞋皮 > — ,或是其變化例,可透過各種程式 化方法來貫現。舉 』人§兄,圖8 A說明當每個記憶體元件儲 存兩個資料位元砗兮“ t ^^ ^邊$憶體元件陣列之臨界電壓分佈。E 描述已抹除記恃髀 μ體凡件的第一臨界電壓分佈。A、B及c描 述已程式化記恃髀 %、餸70件的三個臨界電壓分佈。在一項設計 中’ Ε分佈中的臨界 I為負,而Α、Β及C分佈中的臨界 電壓為正。It will change the poor materials stored in the data storage temporary storage. According to this side, 'it' must be programmed in the corresponding data storage register to store all the data storage buffers are stored logic "factory time' will know that all the selected programs have been programmed unit. In step 564, it is checked if all data storage registers are storing logic 1 . Right is ' then the stylized program is completed and successful because all selected memory units have been programmed and verified. A "pass" status is reported in step 366. If it is determined in step 564 that not all data storage registers are stored logically, then the stylized program will continue. In step 568, θ is used to check the stylized count value PC. One of the examples of stylized limit values is 2G. If the stylized count value PC is not less than 20, then the stylized program has failed, and a "failure, state" is reported in step 570. If the stylized count value PC is less than 20, then The Vpp level is incremented by the step size in step 572 and the stylized count value pc is incremented. After step 572, the program returns to step 56, to apply the next v 123781.doc -28-200832411. The shoe cover described in Fig. 7 or a variation thereof can be realized by various stylized methods. Figure 8A shows that each memory element stores two data bits.砗兮 “t ^^ ^ 边 $ Recall the critical voltage distribution of the body element array. E Describes the first critical voltage distribution of the erased body. A, B, and c describe the three critical voltage distributions of the programmed 70% and 餸70 pieces. In one design, the critical I in the Ε distribution is negative, while the critical voltages in the Α, Β, and C distributions are positive.

母個不同的界f壓範圍均對應於該組資料位元的預定 婁值被&式化至該記憶體元件中的資料與該元件之臨界 電壓位準之間的特定關係會取決於該等元件所採用的資料 編碼方案。甘ψ ,— y .. _ /、 耗例會指派邏輯資料值”11,,給臨界電壓 靶圍E (狀恶E),指派,,1〇”給臨界電壓範圍A (狀態A),指 一 、口^界包壓範圍B (狀態B)以及指派"〇1”給臨界電壓 fc圍C (狀恶c)。然而,在其它設計中則可使用盆它方 案。 、 在圖7的步驟562中,使用三個讀取參考電壓ν^、v吮及 *來從記憶體元件讀取資料。藉由測試-給定儲存元件 之臨界電壓就竟高於還係低KVra、Vrb&Vrc,該系統便 能夠決定該記憶體元件的狀態。目中還表示出三個驗證參 考電壓Vva、Vvb及Vvc。#將記憶體元件程式化至狀態 A、B或C時,該系統將會在圖7的步驟中測試該些記憶 體X件之臨界電壓是否分別大於或等於Vva、Vvb或。 在被稱為全序列程式化的一種方法中,可以將記憶體元 123781.doc -29- 200832411 件彳之已抹除狀態E直接程式化至已程式化狀態A、B或c中 任一者(如圖中的曲線箭頭所示)。舉例來說,要被程式化 的一群记憶體元件可能會先被抹除,而使得該群之中的所 有尤fe體tl件均會處於已抹除狀態E之中。雖然特定的記 體元件係彳文狀恶E被程式化至狀態a,不過,其它記悚 體兀件則係從狀態E被程式化至狀態狀/或1狀態E被程式The specific relationship between the parental f-pressure ranges corresponding to the predetermined threshold of the set of data bits is & the specific relationship between the data in the memory component and the critical voltage level of the component depends on the The data encoding scheme used by the components. Ganzi, - y .. _ /, the consumption will assign the logical data value "11," to the threshold voltage target E (like E, assign,, 1 〇" to the critical voltage range A (state A), refer to The port boundary range B (state B) and the assignment "〇1" give the threshold voltage fc a C (like c). However, in other designs, the pot scheme can be used. In 562, three read reference voltages ν^, v吮, and * are used to read data from the memory device. By testing - the threshold voltage of a given storage element is higher than the lower KVra, Vrb & Vrc, The system is capable of determining the state of the memory component. Three verification reference voltages Vva, Vvb, and Vvc are also shown. When the memory component is programmed to state A, B, or C, the system will be in the figure. The step of 7 tests whether the threshold voltages of the memory X pieces are greater than or equal to Vva, Vvb or respectively. In a method called full sequence stylization, the memory elements 123781.doc -29-200832411 can be used. The erased state E is directly stylized to any of the programmed states A, B, or c (the curve in the figure) The arrow shows. For example, a group of memory elements to be stylized may be erased first, so that all the tl parts in the group will be in the erased state E. Although the specific body element is stylized to state a, other items are programmed from state E to state state or state 1 state.

、狀〜、C 7如圖8 A中所描述般地利用圖7的程序來程 式化該等記憶體單元。 圖8B %明用於程式化儲存二個不同頁(一上頁與一下頁) 之貝料的-多狀態記憶體元件的雙通技術的—範例。圖中 C(01)。對狀悲E來說,兩頁均儲存τ。對狀態a來說,下 頁儲存”〇,,而上頁儲存,,1Π。肖狀態B來說,兩頁皆儲存 〇 。對狀態C來說,下頁館存”^,而上頁儲存,,〇,,。應注音 的係’儘管已將特定位元圖案指派給每—個該等狀態^ 仍可指派不同的位元圖帛。兩次程式化均可利用圖7的程 序來施行。於第-次程式化中,會根據要被程式化至下邏 輯頁的位it來設定該元件的臨界電壓位準。倘若該位元係 邏輯”1”的話,則不會改變臨界電壓,因為其會因先前已 被抹除的關係而處於適當的狀態之中。然"若要被程 ^化的位元係邏輯T的話,那麼Μ件的臨界位準便會 〜加而變成狀態A ’如箭頭43晴示。其會結束該第一次程 式化操作。 在第二次程式化操作中,會依據正在被程式化至上邏輯 12378] .doc -30- 200832411 頁之中的位TL來設定該元件的臨界電壓位準。倘若上邏輯 頁位7L要儲存邏輯”丨”的話,那麼便不會進行彳壬何程式 化,因為該元件係處於狀態E或A其中-者之中,端視該 下頁+位元之程式化而定,二者皆會攜帶一上頁位元"1Π。 倘若該上頁位元欲為邏輯的話,那麼便會偏移該臨界 電[。俄右第一次造成該元件仍停留在已抹除狀態£之中 的話,那麼在第二次中,該元件便會被程式化,^使得該 臨界電難高而落在狀態α,如箭頭434所描述。倘若該 70件於第一次程式化之後便被程式化至狀態Α之中的話, 那麼在第二次中,該記憶體元件便會被進一步程式化,而 使得該臨界電壓提高而落在狀態B内,如箭頭432所描述。 第二次的結果會將該元件程式化W為上頁儲存邏輯 〇的狀態中,而不會改變下頁的資料。 i在其中一種方式中’若寫入足夠的資料以填滿一整頁的 治’那麼便可建立-系統來實施全序列寫入。倘若並未針 對-整頁寫入足夠資料的話,那麼該程式化程序便可利用 所接收到的資料來程式化下頁。當接收到後續資料時,那 Γ系統便會程式化上頁。在又—種方式中,該系統可能 έ在程式化下1的模式中開始寫人,並且若在後面接收到 足夠㈣以填滿—字元線之記憶體元件的全部或大部分的 話,則會轉換成全序列程式化模式。在美國專利公開案第 2006/0126390號中便揭示過此種方式的更多細節,該案的 ㈣為"Pipelined Pn)g_ming of Nonv〇iatiie usingEadyData”’本文以引用的方式將其完整併入。 123781.doc 31 200832411 序圖二至9°描述用於程式化非揮發記憶體的另一種程 ί動=對Γ特殊記憶體元件,藉由下列方式來降低 之極轉合··在寫入前頁的鄰近記憶體元件 對於一特殊頁來寫入該特殊記憶體元件。在一示範 性施订方式中,每一個該等非揮發記憶體 種繼編The patterns ~, C7 are programmed as described in Fig. 8A using the program of Fig. 7 to program the memory units. Figure 8B shows an example of a two-pass technique for multi-state memory components for stylized storage of two different pages (one upper page and one lower page). In the figure, C(01). For sadness E, both pages store τ. For state a, the next page stores "〇, and the previous page is stored, 1Π. For the state B, both pages are stored. For state C, the next page stores "^, and the previous page is stored. ,,〇,,. The system that should be phonetic can assign different bit maps even though a particular bit pattern has been assigned to each of these states. Both stylization can be performed using the procedure of Figure 7. In the first stylization, the critical voltage level of the component is set according to the bit it to be programmed to the next logical page. If the bit is logic "1", the threshold voltage will not be changed because it will be in an appropriate state due to the previously erased relationship. However, if the bit system to be processed is logic T, then the critical level of the element will be changed to become state A ’ as indicated by arrow 43. It will end the first programming operation. In the second stylization operation, the critical voltage level of the component is set according to the bit TL in the program being programmed to the top logic 12378] .doc -30- 200832411. If the logical page bit 7L is to store a logical "丨", then no stylization will be performed because the component is in the state E or A, and the program of the next page + bit is viewed. Depending on the situation, both will carry a previous page bit "1Π. If the upper page bit is to be logical, then the critical voltage will be shifted. The first time the Russian right causes the component to remain in the erased state, then in the second time, the component will be programmed, so that the critical power is high and falls in the state α, such as the arrow. Described in 434. If the 70 pieces are programmed into the state after the first stylization, then in the second time, the memory component will be further programmed, causing the threshold voltage to rise and fall into the state. Within B, as depicted by arrow 432. The second result will stylize the component to the state in which the last page stores the logic, without changing the data on the next page. i In one of the ways 'if enough data is written to fill a full page' then the system can be set up to implement a full sequence of writes. If sufficient data is not written for the full page, the stylized program can use the received data to program the next page. When subsequent data is received, the system will program the previous page. In yet another way, the system may start writing in the stylized mode 1 and if enough (4) is received later to fill all or most of the memory elements of the word line, then Will be converted to full sequence stylized mode. Further details of this approach are disclosed in U.S. Patent Publication No. 2006/0126390, the disclosure of which is incorporated herein by reference to "Pipelined Pn.g_ming of Nonv〇iatiie usingEadyData" 123781.doc 31 200832411 Sequence diagrams 2 to 9° describe another procedure for stylizing non-volatile memory. = For special memory components, reduce the polarity of the transition by the following methods. The adjacent memory elements of the previous page are written to the special memory element for a particular page. In an exemplary application mode, each of the non-volatile memory types is programmed.

係已抹除狀態而狀態A、BA ^ M t t 行次L係已私式化狀態。狀態£儲 =二態A儲!資料01,狀態_存資料1〇,而狀 二J子貝料00。14係非袼雷編碼(non-Gray coding)的 -耗例’因為兩個位元均會在相鄰狀態八與3之間來改 變:亦可使用物理資料狀態之其他資料編碼。每個記憶體 π件均會#1存來自兩個資料胃的位心基於參考的目的, 2等資料頁將被稱為上頁與下頁;’然而,亦可為其提供其 4§己。對狀態絲說’上頁會儲存位元0而下頁則儲存位 化對狀態Β來說’上頁會储存位以而下頁則儲存位元 1對狀恶c來祝’兩頁皆儲存位元資料〇。該程式化程序 2有兩個步驟。兩個步驟均可利用圖7的程序來施行。在 第 > 步驟中會程式化下頁。偶若下頁保留資料^的話,那 麼該記憶體元件狀態便會維持在狀態以。倘若該資料要 被程式化為0的話’那麼該記憶體元件的臨界電塵VTH便會 提高,而使得該記憶體元件被程式化為狀態Bl。所以 9A顯示出將記憶體元件從狀態E程式化至狀態B,,其代表 中間狀態B ;所以’在圖中描述為驗證點Vvb,,其係低於 圖9C中所描述的vvb。 123781.doc •32- 200832411 在一項設計中,於一記憶體元件從狀態E被程式化至狀 態之後,其一鄰近字元線上的鄰近記憶體元件便會相對 於其下頁而被程式化。於程式化鄰近記憶體元件之後,浮 動閘極至洋動閘極耦合效應將會升高所探討的記憶體元件 視I界私壓,該記憶體元件係處於狀態B,。此具有將狀 恶B’之臨界電壓分佈加寬至圖8B中的臨界電壓分佈45〇所 €者的放應。备私式化上頁時將會續正臨界電壓分佈之 此視加寬作用。The state has been erased and the state A, BA ^ M t t row L system has been privateized. State £ store = two-state A store! Data 01, state_storage data 1〇, and shape 2 J sub-bats material 00. 14-series non-Gray coding-consumption case because both bits will be in adjacent state eight and three Change between: You can also use other data encoding of the physical data status. Each memory π will be #1 stored from the two data stomach based on the purpose of the reference, the second data page will be called the upper page and the next page; 'However, it can also be provided for its 4 § . Say to the status wire, 'The previous page will store bit 0 and the next page will be stored. For the status Β, 'The last page will store the bit. The next page will store the bit 1 for the disgusting c. I wish both pages are stored. Bit data 〇. This stylized program 2 has two steps. Both steps can be performed using the procedure of Figure 7. The next page will be stylized in step >. Even if the next page retains the data ^, then the state of the memory component will remain in the state. If the data is to be stylized to 0, then the critical dust VTH of the memory component is increased, and the memory component is programmed into state Bl. So 9A shows that the memory element is stylized from state E to state B, which represents the intermediate state B; so 'depicted in the figure as the verification point Vvb, which is lower than the vvb described in Figure 9C. 123781.doc •32- 200832411 In one design, after a memory component is programmed from state E to state, adjacent memory components on an adjacent word line are stylized relative to its next page. . After stylizing adjacent memory components, the floating gate-to-ocean gate coupling effect will increase the memory component of the memory cell in question. The memory component is in state B. This has the effect of widening the critical voltage distribution of the B' of B' to the critical voltage distribution of 45 图 in Fig. 8B. The widening of the threshold voltage distribution will be continued when the previous page is privately placed.

圖9C^述私式化上頁的f呈彳。倘若該記憶體元件處於已 抹除狀E中’且該上頁要維持在i處的話,那麼該記憶體 兀件將會維持在狀態E之中。倘若該記憶體元件處於狀態E 中’且其上頁資料要被程式化至〇的話,那麼該記憶體元 件的臨界電壓將會被提升,俾使該記憶體元件處於狀態A 中a匕倘# 口亥5己j思體70件處於具有中間臨界電壓分佈450 二:態B’中,且該上頁資料要維持在1處的話,那麼該記 :體元件將會被程式化至最終狀態B。倘若該記憶體元件 -於具有中間臨界電麼分佈450的狀態B,中,且該 ^要變成資料0的話,那麼該記憶體元件的臨界電屢將會 被㈣,俾使該記憶體元件處於狀態c之中。圖9AK所 不的転序雷降低浮動閘極 有鄰近記憶體元件的上頁“=_合的效應,因為僅 的滿转灭” 頁釭式化會影響-給定記憶體元件 1時從八#代狀心、.扁碼的槌例係在該上頁資料為 夕至狀態c ’並且在該上頁資料為0時移至狀 恶雖然圖提供-相對於四個資料狀態與兩個資 12378l.doc -33- 200832411 料頁的範例,不過,其所教示的概念卻可套用至具有四個 以上或以下之狀態及兩頁以上或以下的其它施行方式。關 於各種程式化方案及浮動閘極至浮動閘極搞合的更多細節 可以在美國利申請案第11/099,133號中找到,其名稱為 * ^Compensating For Coupling During Read Operations Of ^ Non-Volatile Memory”,本文以引用的方式將其完整併 入0 圖9D描述用於套用一程式化脈衝作為圖7之步驟56〇之部 馨 分的一典型時序圖。信號1 500描述被套用至和目前正在被 程式化的儲存元件相關聯的一字元線的程式化波形; 而仏號1 5 1 0所描述的係被套用至其它字元線的傳導電壓 Vpass。仏號1520所描述的係當施加vpgm時禁止被程式化的 儲存元件的位元線電壓Vbl ;而信號153〇所描述的係當施 加Vpgm時允許被程式化的儲存元件的位元線電壓。信號 1540所描述的係一NAND串的汲極側選擇閘極電壓。 • 在^處,該汲極側選擇閘極會藉由施加一較高的電壓(舉 J來況3 V至4· 5 V)而開路。請注意,該源極側選擇閘極 會維持被偏壓在〇 V處。接著,在以,會施加該位元線電 . 壓Vbl用以程式化-儲存元件,於此情況中,Vbl係處於或 . 接近〇 V,或是處於0 ▽至1 v的範圍之中以供進行粗略/精 細驗證方法;或是用以藉由施加一電壓V心(通常係一從u “至3 v之電壓)來禁止該儲存元件進行程式化。當位元線 電壓VbL係〇 V或另一低電壓時(曲線1530),此電壓將會被 1要被私式化的儲存元件的通道區。倘若被施加一較 123781.doc -34- 200832411 鬲的位元線電壓乂虹時(曲線1520),該通道將會抵達較高 的電壓(在理想的情況中為Vdd)。在。處,倘若該位元線處 於Vdd的話,vSGD會下降以切斷該選擇閘極,同時會讓該 選擇閘極針對0 V至1 V範圍中較低的VBL仍保持在導電狀態 中。在U處,vpass會被施加至該被選擇的字元線,並且會被 施加至該NAND串中所有或幾乎所有未被選擇的字元線。 在處,尚程式化電壓Vpgm會被施加至該被選擇的字元 線,而且相依於該通道究竟係要被昇壓至一高電壓或是被 偏壓至一低電壓,該儲存元件將會分別被禁止程式化或允 許被程式化。在程式化電壓Vpgm已經提高至該固定振幅位 準之後,所有狀態的實際程式化主要會發生在。至卜。在卜 處’私式化電壓Vpgm會下% ;而在“處,%同樣會下 降,請注意,程式化電壓Vpgm可能會上升至其固定振幅及/ 或回降,而不會停止在程式化電壓^處。最後,在^ ^ VSGD與VBL也會被移㉟。接著,便可實施一或多項驗 證操作,基本上係讀取操作,用以驗證被選擇要進行程式 化的儲存元件是否已經抵達它們的目標臨界電壓狀態。可 以施加具有高振幅的額外程式化脈衝,直到所有或大部分 的儲存元件均已抵達它們所需的臨界電壓狀態為止。 如剛面所提,當程式化非揮發記憶體装置(例如nand快 閃記憶體裝置)時可使用各種昇壓方法。舉例來說,在二 處,會將-合宜的昇壓電壓施加至所有(或幾乎所有)未被 選擇的字元線,用以禁止或允許進行程式化。因此,當在 “、t7期間實際進打程式化時,便會存在所需的昇壓。圖 123781.doc -35- 200832411 10至17說明會達成不同結果的各種昇壓方法並且顯示出在 時間t6處特定字元線的狀態。自昇壓(SB)可用於二進制裝 置,因為該方法可允許按隨機順序來程式化一 NAND串中Figure 9C illustrates the f-presentation of the private page. If the memory component is in the erased state E and the upper page is to be maintained at i, then the memory component will remain in state E. If the memory component is in state E and the data on the previous page is to be stylized to 〇, then the threshold voltage of the memory component will be boosted, so that the memory component is in state A. 70 pieces of mouth hai 5 have a middle threshold voltage distribution 450: state B', and the previous page data should be maintained at 1, then the body component will be programmed to the final state B . If the memory component is in state B with an intermediate criticality distribution 450, and the information is to become data 0, then the critical component of the memory component will be (4), so that the memory component is State c. Figure 9AK does not reduce the frequency of the floating gate. The floating gate has the effect of the previous page "=_合合, because only the full turn-off" page will affect - given memory element 1 from eight #代状心,. The example of the flat code is on the previous page of the data to the state c 'and moved to the situation when the data on the previous page is 0. Although the figure provides - relative to the four data states and two assets 12378l.doc -33- 200832411 An example of a material page, however, the concepts taught can be applied to other implementations with four or more states and more than two pages or less. More details on various stylization schemes and floating gate-to-floating gates can be found in US Application No. 11/099,133, entitled *^Compensating For Coupling During Read Operations Of ^ Non-Volatile Memory "This article is incorporated by reference in its entirety. Figure 9D depicts a typical timing diagram for applying a stylized pulse as part of step 56 of Figure 7. Signal 1500 description is applied to and is currently being a stylized waveform of a word line associated with a stylized storage element; and the nickname 1 5 1 0 is applied to the conduction voltage Vpass of other word lines. The nickname 1520 is described when applied. Vpgm disables the bit line voltage Vbl of the stylized storage element; and signal 153 描述 describes the bit line voltage of the storage element that is allowed to be programmed when Vpgm is applied. The signal NAND is described as a NAND string. The drain side selects the gate voltage. • At ^, the drain side select gate is opened by applying a higher voltage (3 V to 4·5 V). Please note that Source side selection gate It is biased at 〇V. Then, the bit line is applied, and Vbl is used to program-store the component. In this case, Vbl is at or near 〇V, or at 0 ▽. Between the range of 1 v for rough/fine verification methods; or for applying a voltage V-core (usually a voltage from u "to 3 v") to disable the storage element from being programmed. When the bit line voltage VbL is 〇 V or another low voltage (curve 1530), this voltage will be the channel area of the storage element to be privateized. If a bit line voltage of 123781.doc -34-200832411 乂 is applied (curve 1520), the channel will reach a higher voltage (in the ideal case Vdd). in. At this point, if the bit line is at Vdd, vSGD will drop to turn off the select gate, and the select gate will remain in a conducting state for the lower VBL in the 0 V to 1 V range. At U, vpass will be applied to the selected word line and will be applied to all or nearly all of the unselected word lines in the NAND string. Wherein, the still stylized voltage Vpgm is applied to the selected word line, and depending on whether the channel is to be boosted to a high voltage or biased to a low voltage, the storage element will They are prohibited from being stylized or allowed to be stylized. After the programmed voltage Vpgm has been raised to this fixed amplitude level, the actual stylization of all states will mainly occur. To the end. In the case of 'private voltage Vpgm will be %; and at the same time, % will also drop, please note that the stylized voltage Vpgm may rise to its fixed amplitude and / or fall back, without stopping the stylization The voltage ^. Finally, the ^ ^ VSGD and VBL will also be shifted by 35. Then, one or more verification operations can be performed, basically a read operation to verify whether the storage element selected for programming has been Arriving at their target threshold voltage state. Additional stylized pulses with high amplitude can be applied until all or most of the storage elements have reached their desired threshold voltage state. As mentioned in the face, when stylized non-volatile Various boost methods can be used for memory devices (such as nand flash memory devices). For example, at two locations, a suitable boost voltage will be applied to all (or almost all) unselected characters. Line, which is used to prohibit or allow for stylization. Therefore, when the actual program is programmed during t7, there will be a required boost. Figure 123781.doc -35- 200832411 10 to 17 illustrates various boosting methods that will achieve different results and show the state of a particular word line at time t6. Self-boosting (SB) can be used in binary devices because this method allows for programming a NAND string in random order.

的字元線。然而,對多位準裝置而言,通常不會使用隨機 順序程式化。在此情況下,可以使用LSB與EASB或此等方 法之變化例。以LSB及EASB為主的方法的一優點係,通道 昇壓會更有效率,且因此可以減小程式化干擾。不過,當 記憶體元件的尺度縮小時,帶至帶穿隧或GIDL則可能會 出現在該被接地字元線的汲極附近。已昇壓通道可能會因 GIDL而被放電,從而會導致程式化干擾,或者可能會產 生熱載子,它們會被注入該等記憶體元件的穿隧氧化物或 該等浮動閘極之中。參考圖1 〇來說明該問題。 圖10說明當使用EASB昇壓模式時具有不平衡昇壓通道 區的NAND串1〇〇〇。該NANE^ 1000包含一源極側選擇閘 極1〇1〇、汲極側選擇閘極1055,以及記憶體元件1〇15、 1020、1025、1030、1035、1040、1045與 1050,該等記情 體7L件會分別被耦合至一 p井區i 〇〇5中的該等選擇閘極之 間的字元線WL0至WL7。雖然本圖中顯示出八個記憶體元 件,不過該配置僅具解釋意義,亦可使用其它組態。如前 面所述,⑨其巾-可能的方式中,程式化可能會始於源極 側記憶體元件(舉例來說,元件1〇1 5),並且每次會往前前 進一個記憶體元件,至汲極側記憶體元件1〇5〇。在本範例 中:記憶體元件购係目前正在被程式化之被選擇的H 體兀件,而且其會經由字元線WL5來接收程式化電堡 123781.doc -36- 200832411 pgm。一傳導電壓Vpass(其通常係位於5 乂至1〇 v的範圍之 中)έ、、二由匕們個別的字元線被施加至其餘的記憶體元 件,疋件1035除外,其會接收一低電壓(其通常係〇 ν或是 4於0 V至1 ν的範圍之中)。於此範例中,記憶體元件 1015 1020、1025、1030以及1〇35已經被程式化,而記憶 • 冑元件1G45與刪則尚未被程式化,或至少尚未抵達它們 的最終已程式化狀態。也就是,記憶體元件㈣並 • 未被程式化及/或被部分程式化。在特定情況中,如圖9中 所描述的私式化方案’該記憶體元件1〇45便可能係處於中 間已程式化狀態B,之中。再者,在圖9的程式化方案的情 況下,鄰近的記憶體元件1035亦可能會處於中間已程式化 狀態之中。於另一可能的程式化方式中,當該記憶體元件 1040正在被程式化時,鄰近的記憶體元件1〇35僅會部分被 程式化。 再者¥ 4專6己彳思體元件正在被程式化時,與NAND串 • 1〇00相關聯的一位元線接點可能會被接地,或是可能會被 耦合至一中間電壓(其通常係在〇·2 ¥至1 V的範圍内)用以 進行精細模式程式化。舉例來說,參見美國專利案第 • 6,888,758 5虎’其標題為,,Programming Non-VolatileThe word line. However, for multi-level devices, random order stylization is usually not used. In this case, LSB and EASB or variations of these methods can be used. An advantage of the LSB and EASB-based approach is that channel boosting is more efficient and therefore reduces stylized interference. However, when the size of the memory component is reduced, band-to-band tunneling or GIDL may occur near the drain of the grounded word line. The boosted channels may be discharged due to GIDL, which may cause stylized interference, or may generate hot carriers that are injected into the tunneling oxides of the memory components or the floating gates. Refer to Figure 1 for a description of this problem. Figure 10 illustrates a NAND string 1 具有 with an unbalanced boost channel region when using the EASB boost mode. The NANE 1000 includes a source side selection gate 1〇1〇, a drain side selection gate 1055, and memory elements 1〇15, 1020, 1025, 1030, 1035, 1040, 1045, and 1050. The modal 7L pieces are respectively coupled to the word lines WL0 to WL7 between the selection gates in a p-well area i 〇〇5. Although eight memory elements are shown in this figure, this configuration is only illustrative and other configurations can be used. As mentioned earlier, in a possible manner, stylization may begin with the source side memory component (for example, component 1〇1 5), and each time a memory component is advanced, To the 侧-side memory element 1〇5〇. In this example: the memory component is currently being programmed to be selected for the H component, and it will receive the stylized electric castle 123781.doc -36-200832411 pgm via word line WL5. a conduction voltage Vpass (which is usually in the range of 5 乂 to 1 〇 v) 、, and two are applied to the remaining memory elements by their individual word lines, except for the element 1035, which receives one Low voltage (which is usually 〇ν or 4 in the range of 0 V to 1 ν). In this example, memory components 1015 1020, 1025, 1030, and 1 〇 35 have been programmed, and memory 胄 component 1G45 and defragmentation have not yet been programmed, or at least have not yet reached their final programmed state. That is, the memory component (4) is not stylized and/or partially stylized. In a particular case, the memory element 1 〇 45, as depicted in Figure 9, may be in the intermediate programmed state B. Furthermore, in the case of the stylized scheme of Figure 9, adjacent memory elements 1035 may also be in an intermediate programmed state. In another possible stylized manner, when the memory component 1040 is being programmed, the adjacent memory components 1 〇 35 are only partially stylized. In addition, when a 4 component is being programmed, a bit line contact associated with the NAND string • 1〇00 may be grounded or may be coupled to an intermediate voltage (its It is usually in the range of ¥·2 ¥ to 1 V) for fine mode stylization. For example, see US Patent No. 6,888,758 5 Tiger's title, Programming Non-Volatile

Memory”,本文以引用的方式將其明確併入。在該nand • 串1000中該被選擇字元線上的記憶體元件1〇40已經被程式 化至所需狀態之後,便可以將一禁止電壓Vdd施加至該字 70線接點以禁止記憶體元件1〇4〇進一步程式化,直至被連 接至同一條被選擇之字元線的其它NAND串上之其它記憶 123781.doc -37· 200832411 體元件同樣被程式化至所需狀態為止。 由於施加該傳導電壓的關係,(例如)會在NAND串looo 的被選擇字元線的源極側上,先前已被程式化的記憶體元 件下方形成一低通道昇壓區;(例如)同時會在1^八^〇串 - 1G00的被選擇字元線的汲極側上,該被選擇的記憶體元件 ,卩及該等未被程式化及/或已被部分程式化的記憶體元件 下方形成一向通道昇壓區。圖1〇示意性說明該些昇壓區。 一般來說,已經被程式化至特定狀態的記憶體元件會導致 位於該些記憶體元件下方的一相關聯通道區的昇壓比較沒 有效用。再者,因為會從源極侧至汲極側程式化額外的記 憶體元件,所以不良昇壓區的大小將會增加,而未被程式 化及/或被部分程式化的記憶體元件之高昇壓區的大小則 會縮小。由於該等昇壓通道電位中的差異以及因不良昇壓 源極側區所造成的小基板偏壓效應的關係,電荷可能會從 該高昇壓通道區洩漏至該低昇壓通道區,從而會導致該高 鲁 幵壓通道區中的電位下降。因此,位於該被選擇字元線上 之未被選擇s己憶體元件的程式化干擾的可能性便會提高。 藉由增加已經被程式化區域中的昇壓通道電位,便可以防 止電荷從高昇壓區汽漏至低昇壓區。在一種方式中,此可 以藉由下列方式達到:針對和已經被程式化之記憶體元件 ’ 相關聯的字元線使用較高的Vpass數值,如結合圖11所說明 者。 圖11說明具有平衡昇壓通道區的一 N AND串11 〇 〇。該 NAND串1100包含一源極側選擇閘極丨丨10、汲極侧選擇閘 123781.doc -38 - 200832411 極 1155、以及記憶體元件 1115、1120、1125、1130、 1135、1140、1145與1150,該等記憶體元件會分別被耗合 至一 P井區1105中的該等選擇閘極之間的字元線WL〇至 WL7。於其中一可能的方式中,程式化係始於源極側記憶 體元件(舉例來說,元件1115),並且每次會往前前進一個Memory, which is explicitly incorporated herein by reference. After the memory element 1〇40 on the selected word line in the nand•string 1000 has been programmed to the desired state, a disable voltage can be applied. Vdd is applied to the word 70 line contact to prevent further memory element 1〇4〇 from being programmed until the other memory on the other NAND string connected to the same selected word line 123781.doc -37·200832411 The component is also programmed to the desired state. Due to the application of the conduction voltage, for example, it will form on the source side of the selected word line of the NAND string looo, under the previously programmed memory element. a low channel boosting region; (for example) simultaneously on the drain side of the selected word line of 1^8^〇 string-1G00, the selected memory component, and the unprogrammed and / or a partially channeled boost region is formed under the partially stylized memory component. Figure 1 is a schematic illustration of the boost regions. In general, memory components that have been programmed into a particular state will result in the Memory components The boost comparison of an associated channel region below has no effect. Furthermore, since the additional memory components are programmed from the source side to the drain side, the size of the bad boost region will increase without being programmed. The size of the high-boost region of the memory element that is partially and/or partially programmed is reduced. Due to the difference in the potential of the boosting channel and the small substrate bias effect caused by the poor boosting source side region Relationship, electric charge may leak from the high-boost channel region to the low-boost channel region, thereby causing a potential drop in the high-ruthenium channel region. Therefore, the selected word line is not selected. The possibility of stylized interference of the body element is increased. By increasing the potential of the boost channel in the already programmed region, it is possible to prevent charge from leaking from the high boost region to the low boost region. In one way, This can be achieved by using a higher Vpass value for the word line associated with the memory element that has been programmed, as illustrated in connection with Figure 11. Figure 11 illustrates a balanced rise An N AND string 11 通道 of the channel region. The NAND string 1100 includes a source side select gate 丨丨10, a drain side select gate 123781.doc -38 - 200832411 pole 1155, and memory elements 1115, 1120, 1125, 1130, 1135, 1140, 1145, and 1150, the memory elements are respectively subtracted to word lines WL 〇 WL WL WL7 between the select gates in a P well region 1105. In this way, the stylization begins with the source side memory component (for example, component 1115) and advances one at a time.

記憶體元件,至汲極側記憶體元件1丨50。在本範例中,記 憶體元件1140係目前正在被程式化之被選擇的記憶體元 件,而且其會經由字元線WL5來接收程式化電壓vpgm。一 車父南的傳導電壓Vpass會經由字元線WL0、WL1、WL2、以 及WL3分別被施加至先前已被程式化的記憶體元件,舉例 來說,記憶體元件1115、1120、1125、以及Π30,源極侧 記憶體元件1135除外,其會經由字元線WL4接收〇 v。一 較低的傳導電壓vpass2會分別經由字元線WL6與WL7被施加 至該被選擇記憶體元件之汲極側處未被程式化的記憶體元 件或尚未達到它們最終已程式化狀態的記憶體元件(舉例 來說,記憶體元件1145與1150)。因此,在此範例中, vpass〗與vpasS2會被施加至該NAND串中個別的記憶體元件 子集,其中,每個子集均包含位於當前正在被程式化的記 憶體凡件之相反側上的一或多個記憶體元件,但是未必為 當前正在被程式化的記憶體元件之相反側上的所有記憶體 元件。如前面所述’在使用圖9之程式化方案的情況下, 位於該被選擇記憶體元件旁邊的記憶體元件(在此範例中 為記憶體元件11 45)可能係步於 ,aa J J此係處於一中間已程式化狀態B,之 中。再者,在圖9之程戎介姑分-& 枉飞化技術的情況下,記憶體元件 123781.doc -39- 200832411 1135亦可能係處於中間已程式化狀態之中。 實際上,一低電壓(約為〇 VSi v)會經由字元線WL4被 施加至鄰近的源極側記憶體元件1135。因為藉由該汲極與 源極側上較均等的昇壓以及因改良的源極側昇壓所造成的 充份基板偏壓效應的關係會降低洩漏,所以,該昇壓電位 仍可能會夠高。在該相鄰字元線上施加一高於〇 V的偏壓 會降低GIDL,其可能係出現在該相鄰字元線上。Memory element, to the drain side memory element 1丨50. In this example, memory element 1140 is the selected memory element that is currently being programmed, and it receives the stylized voltage vpgm via word line WL5. The conduction voltage Vpass of a vehicle father south is applied to the memory elements that have been previously programmed, such as memory elements 1115, 1120, 1125, and Π30, via word lines WL0, WL1, WL2, and WL3, respectively. Except for the source side memory element 1135, which receives 〇v via the word line WL4. A lower conduction voltage vpass2 is applied to the memory elements that are not stylized at the drain side of the selected memory element or to the memory that has not yet reached their final programmed state via word lines WL6 and WL7, respectively. Elements (for example, memory elements 1145 and 1150). Thus, in this example, vpass and vpasS2 are applied to individual subsets of memory elements in the NAND string, where each subset contains the opposite side of the memory device currently being programmed. One or more memory elements, but not necessarily all of the memory elements on the opposite side of the memory element currently being programmed. As described above, in the case of using the stylized scheme of Figure 9, the memory component (in this example, the memory component 145) located next to the selected memory component may be stepped on, aa JJ. In an intermediate stylized state B, among them. Furthermore, in the case of the process of Fig. 9 , the memory component 123781.doc -39- 200832411 1135 may also be in the intermediate programmed state. In fact, a low voltage (about 〇 VSi v) is applied to the adjacent source side memory element 1135 via word line WL4. Since the leakage between the drain and the source side is equal, and the relationship between the substrate bias effect due to the improved source side boosting reduces leakage, the boost potential may still be High enough. Applying a bias above 〇 V on the adjacent word line reduces the GIDL, which may occur on the adjacent word line.

因此,與已經被程式化之記憶體元件對應的通道區的昇 壓便會獲得改良。明確地說,在先前已被程式化的記憶體 凡件下面以及在該被選擇之記憶體元件與該等未被程式化 及/或被部分程式化的記憶體元件下面會形成高通道昇壓 區。與先别被程式化的記憶體元件相關聯的通道區的較大 傳導電壓會補償由於該等記憶體元件處於已程式化狀態中 所引起的較低昇壓效應。確實的補償需要知道該等被程式 化的記憶體it件處於何種狀態。不過,被程式化的記憶體 几件的數篁及它們要被程式化至何種狀態卻會隨著每一條— NAND串而不fgj。一般來說,在和先前被程式化的記憶體 儿件相關聯的字元線上約2 ¥至3獨範圍中進行補償最佳 的係會補償隨機資料被寫入該等先前被程式化的記憶體元 件之中時的平均情況。也就是,應該要超過%2約2 V至3 V。此差異可藉由測試而針對特定的記憶體裝置來進 行最佳化。 應注意,上述方 LSB方法以及此等 法的應用並不限於EASB,亦可應用於 方法之變化例。一般來說,補償係提供 123781.doc 200832411 以降低該被選擇的記憶體元件的源極側上之通道因部分或 所有忒等圮憶體元件處於已程式化狀態中所造成的昇壓。 藉由提尚和已經被程式化的記憶體元件相關聯的字元線的 傳導電壓,以便降低或消除該等兩個昇壓區之間的電荷洩 漏,來提供補償。因此,在該被選擇字元線以及與該等未 被程式化及/或被部分程式化的記憶體元件相關聯的字元 線下面的昇壓通道電位將會比較高,並且幾乎會與正在被 程式化的字元線無關。所以,便會降低程式化干擾並且呈 現較小的字元線相依性。再者,由於該改良昇壓的關係, Vpass2亦可能會比較低。舉例來說,在一種可能的方式 中,Vpassl約10 V至11 V而Vpassj々8 v,而在遞增幅度的連 續脈衝中’ vpgm的範圍可能係從丨6 V至24 V。可以藉由測 試來決定特定記憶體裝置的Vpassi及νρ_2之最佳位準。還 預期會減低V至▼穿隧,因為被接地的字元線下面的橫向 電場將會因此昇壓技術而降低。可藉由配合圖丨2所討論的 昇壓方案來達到帶至帶穿隧之進一步降低的目的。 圖12說明在昇壓通道區之間具有一隔離區的一 nand串 1200。邊NAND串1200包含一源極侧選擇閘極121〇、汲極 側選擇閘極1255、以及記憶體元件1215、122〇、1225、 1230、1235、1240、1245與1250,該等記憶體元件會分別 被耦合至一 p井區1205中的該等選擇閘極之間的字元線 WL0至WL7。於此範例中,記憶體元件124〇被選為要進行 私式化且其會經由子元線WL5來接收該程式化電壓 VPgm。一較鬲的傳導電壓vpassl會經由字元線WL〇與WL1分 123781.doc -41 · 200832411 別被施加至先前已被程式化的記憶體元件中的一或多者, 舉例來說,記憶體元件1215與122〇,而一較低或相同的傳 導電壓Vpass2會經由字元線WL6與WL7分別被施加至未被程 式化及/或被部分程式化的記憶體元件,舉例來說,記憶 體兀件1245與1250。此外,較低的傳導電壓Vpass3、 及Vpass5會經由字元線WL4、WL3、與WL2分別被施加至先 前被程式化的記憶體元件1235、123〇及1225,它們係位於 攻被選擇的圮憶體元件1240與接收較高傳導電壓乂…“的 記憶體元件m5& 1220之間。νρ_3、νρ_4及Vpass5均會小 於 Vpass 1 〇 在一種方式中,Vpass4係小於Vpass3及Vpass5。Vpa⑴及 Vpass5可以彼此約略相等。或者,Vpass3及Vpaw不同。舉例 來說,vpass3及vpass5可能約為2 ¥至4 v ;而Vpass4約為〇 V 至1 V。如前,¥叫“可能約為10 ¥至11 V,Vpass2可能約為 8 V,而在連續脈衝中,Vpgm的範圍可能係從16 ¥至24 V。可藉由測試來決定特定記憶體裝置的最佳電壓。在此 方式中,所施加之電壓基本上會形成一槽形或隔離區,其 係集中在具有最低所施加之傳導電壓之記憶體元件(例如 記憶體元件1230)周圍,其中,該等傳導電壓會在該隔離 區之每側上對稱或不對稱地增加。該隔離區可能包含奇數 或偶數的記憶體元件。該隔離區係用於隔離兩個高度昇壓 的通道區並減低具有最低偏壓電壓的字元線(例如與記憶 體元件1230相關聯的字元線WL3)之汲極與源極區上的電 壓,以避免或減低該字元線下面的帶至帶穿隧。配合圖13 123781.doc -42- 200832411 會提供具有一替代隔離區的另一具體實施例。 圖13說明在昇壓通道區之間具有一替代隔離區的一 NAND串1300。該NAND串1300包含一源極侧選擇閘極 1310、汲極側選擇閘極1355、以及記憶體元件1315、 * 1320、1325、1330、1335、1340、1345 與 1350,該等記憶 • 體元件會分別被耦合至一 P井區1 305中的該等選擇閘極之 間的字元線WL0至WL7。於此範例中,記憶體元件1345被 選為要進行程式化,且其會經由字元線WL6來接收程式化 _ 電壓Vpgm。一較南的傳導電壓Vpassl會經由字元線WL0被施 加至先前已被程式化的記憶體元件中的一或多者,舉例來 說,記憶體元件13 1 5,而一較低或相同的傳導電壓Vpw 會經由字元線WL7被施加至未被程式化及/或被部分程式 化的記憶體元件之一或多者,舉例來說,記憶體元件 1350。此外,較低的傳導電壓Vpass3、、v_s5、 Vpass6 及 Vpass7 會經由字元線 WL5、WL4、WL3、WL2、與 參 WL1分別被施加至先前被程式化的記憶體元件1340、 1335、1330、1325及1320,它們係位於該被選擇的記憶體 元件1345與接收較鬲傳導電壓vpassl的記憶體元件13 15之 間。 ' 在一種方式中,vPass5 係小於 Vpass3、Vpass4、Vpass6& , Vpass7。此外,Vpass4 及 VpaSS6 可能小於 Vpass3 及 Vpass7。 vpass3、vpass4、vpass5、vpass6 及 Vpass_ 小於 Vpassi。ν_4 及vpass6可能彼此約略相等。或者,、川4及〜川6不同。同 樣地,Vpass3及Vpass7可能彼此約略相等。或者,及 123781.doc -43- 200832411Therefore, the boost of the channel region corresponding to the memory element that has been programmed can be improved. In particular, a high channel boost is formed under the memory component that has been previously programmed and under the selected memory component and the unprogrammed and/or partially programmed memory components. Area. The large conduction voltage of the channel region associated with the memory element that is not programmed first compensates for the lower boosting effect caused by the memory elements being in the programmed state. The exact compensation needs to know what state these programmed memory components are in. However, the number of pieces of stylized memory and the state to which they are to be programmed will follow each one—the NAND string is not fgj. In general, the best compensation in the range of about 2 ¥ to 3 in the character line associated with the previously programmed memory is compensated for the random data being written into the previously stylized memory. The average of the body components. That is, it should be more than 22 to about 2 V to 3 V. This difference can be optimized for a particular memory device by testing. It should be noted that the above-described LSB method and the application of such methods are not limited to EASB, and may be applied to variations of the method. In general, the compensation system provides 123781.doc 200832411 to reduce the boost on the source side of the selected memory component due to the partial or all of the memory elements being in the programmed state. Compensation is provided by raising the conduction voltage of the word line associated with the memory element that has been programmed to reduce or eliminate charge leakage between the two boost regions. Thus, the boost channel potential below the selected word line and the word line associated with the unprogrammed and/or partially programmed memory elements will be relatively high and will be almost It is irrelevant to the stylized character line. Therefore, stylized interference is reduced and smaller word line dependencies are presented. Furthermore, due to the improved boost relationship, Vpass2 may also be lower. For example, in one possible approach, Vpassl is approximately 10 V to 11 V and Vpassj 々 8 v, while the range of 'vpgm' may range from 丨6 V to 24 V in successive pulses of increasing amplitude. The optimum level of Vpassi and νρ_2 for a particular memory device can be determined by testing. It is also expected to reduce the V to ▼ tunneling because the lateral electric field below the grounded word line will therefore be reduced by the boosting technique. The further reduction of band-to-band tunneling can be achieved by the boosting scheme discussed in conjunction with Figure 2. Figure 12 illustrates a nand string 1200 having an isolation region between the boost channel regions. The edge NAND string 1200 includes a source side selection gate 121A, a drain side selection gate 1255, and memory elements 1215, 122A, 1225, 1230, 1235, 1240, 1245, and 1250, and the memory elements will Word lines WL0 through WL7 between the select gates in a p-well region 1205, respectively, are coupled. In this example, memory component 124 is selected to be privateized and will receive the stylized voltage VPgm via sub-line WL5. A relatively low conduction voltage vpassl will be applied to one or more of the previously programmed memory elements via the word line WL 〇 and WL1, 123781.doc -41 · 200832411, for example, memory Elements 1215 and 122A, and a lower or the same conduction voltage Vpass2 is applied to memory elements that are not programmed and/or partially programmed via word lines WL6 and WL7, for example, memory Parts 1245 and 1250. In addition, the lower conduction voltages Vpass3, and Vpass5 are applied to the previously programmed memory elements 1235, 123A, and 1225 via word lines WL4, WL3, and WL2, respectively, which are located in the selected memory. The body element 1240 is between the memory element m5 & 1220 that receives the higher conduction voltage 乂...". νρ_3, νρ_4, and Vpass5 are all smaller than Vpass 1 〇 In one mode, Vpass4 is less than Vpass3 and Vpass5. Vpa(1) and Vpass5 can be mutually About equal. Vpass3 and Vpaw are different. For example, vpass3 and vpass5 may be about 2 ¥ to 4 v; and Vpass4 is about 〇V to 1 V. As before, the call is "may be about 10 ¥ to 11 V. Vpass2 may be approximately 8 V, while in continuous pulses, Vpgm may range from 16 ¥ to 24 V. The optimum voltage for a particular memory device can be determined by testing. In this manner, the applied voltage substantially forms a channel or isolation region that is concentrated around a memory component (e.g., memory component 1230) having the lowest applied conduction voltage, wherein the conduction voltages It will increase symmetrically or asymmetrically on each side of the isolation zone. This isolation area may contain odd or even memory components. The isolation region is for isolating the two highly boosted channel regions and reducing the drain and source regions of the word line having the lowest bias voltage (eg, word line WL3 associated with memory element 1230) Voltage to avoid or reduce the band-to-band tunneling below the word line. Another specific embodiment with an alternative isolation zone will be provided in conjunction with Figure 13 123781.doc -42- 200832411. Figure 13 illustrates a NAND string 1300 having an alternate isolation region between boost channel regions. The NAND string 1300 includes a source side selection gate 1310, a drain side selection gate 1355, and memory elements 1315, * 1320, 1325, 1330, 1335, 1340, 1345, and 1350, and the memory elements are Word lines WL0 through WL7 between the select gates in a P well region 1 305 are coupled, respectively. In this example, memory component 1345 is selected to be programmed and it receives the stylized _ voltage Vpgm via word line WL6. A souther conducting voltage Vpass1 is applied via word line WL0 to one or more of the previously programmed memory elements, for example, memory element 13 1 5, and a lower or the same The conduction voltage Vpw is applied via word line WL7 to one or more of the unprogrammed and/or partially programmed memory elements, for example, memory element 1350. In addition, the lower conduction voltages Vpass3, v_s5, Vpass6, and Vpass7 are applied to the previously programmed memory elements 1340, 1335, 1330, 1325 via word lines WL5, WL4, WL3, WL2, and WL1, respectively. And 1320, which are located between the selected memory element 1345 and the memory element 13 15 that receives the lower conduction voltage vpassl. ' In one mode, vPass5 is less than Vpass3, Vpass4, Vpass6&, Vpass7. In addition, Vpass4 and VpaSS6 may be smaller than Vpass3 and Vpass7. Vpass3, vpass4, vpass5, vpass6, and Vpass_ are less than Vpassi. Ν_4 and vpass6 may be approximately equal to each other. Or, Chuan 4 and ~ Chuan 6 are different. Similarly, Vpass3 and Vpass7 may be approximately equal to each other. Or, and 123781.doc -43- 200832411

p 7不同舉例來說,Vpass3與Vpass7可能約為6 V至8V, 二㈣與V”6可能約為2 V至4v,而Vpass5可能約為〇 ^ 如引Vpassl可能約為10 ¥至u V,Vpass2可能約為8 V而在連績脈衝中,Vpgm的範圍可能係從16 V至Μ V。 可藉由测試來決定特定記憶體裝置的最佳電壓。於此方式 中所轭加之電壓會形成一延伸槽形或隔離區,其係集中 在具有最低所施加之傳導電壓之記憶體元件(例如記憶體 凡件133G)周目’且當該等傳導電壓在該隔離區之每側上 對稱或不對稱地增加時,其昇壓輪廓的斜率會以比較平緩 的方式遠離該隔離區。再者,該隔離區可能包含奇數或偶 數的記憶體元件。該隔離區係用於隔離兩個高度昇壓的通 道區亚減低具有最低偏壓電壓的字元線(例如與記憶體元 件1330相關聯的字元線貿。)之汲極與源極區上的電壓, 以避免或減低該字元線下面的帶至帶穿隧。藉由延長隔離 區的長度,便會進一步隔離該等兩個高度昇壓的通道區以 避免或減低該等兩個昇壓區之間的洩漏而且避免或減低該 字元線下面的帶至帶穿隨。 上面所述的程式化干擾降低技術適用於多位準與單位準 程式化兩者。多位準記憶體預期可獲得較大利益。對單位 準記憶體而言,在下列情況下預期會實現較大利益··相較 於以隨機順序來進行程式化,以預定的字元線順序從 N AND串之源極側至汲極側進行程式化。再者,該等技術 原理上可配合所有昇壓技術來使用。不過,多位準昇壓模 式(例如EASB及其變化例)預期會有最大利益。 123781.doc -44- 200832411 當裝置進一步縮小時,該些技術便可作進一步修正,以 解決程式化干擾效應與相關的崩潰。舉例來說,當從8_ NAND串按比例變成i6-NAND串及32-NAND串時,使用該 些昇壓技術便會實現明顯的利益。現在將焦點移至較長串 上’舉例來說,64_NAND串,其中,單一位元線會接觸M 個儲存元件。因此,程式化干擾在較長的串中會最糟,且 程式化干擾的持續時間會係32-NAND串的持續時間的兩 倍’ 16-NAND串的持續時間的四倍,以及8-NAnd串的持 績時間的八倍。 圖14A說明一具有64個記憶體元件的NAND串1400。於 此具體實施例中,會使用額外的隔離字元線來進一步分割 該等已程式化及/或已抹除區。該NAND串1400包含一源極 侧選擇閘極1470與一汲極侧選擇閘極148〇,以及記憶體元 件1401、402、…1464,它們係被配置在一 p井區149〇之 中。母一個記憶體元件均可經由一對應的字元線WL〇、 WL1 · · · WL63 來選擇。 假設字元線WLX被選為要進行程式化,那麼程式化電壓 Vpgm便會被施加至該字元線。在該被選擇的記憶體元件的 源極側上,會藉由施加一非常低的隔離電壓Vis。(約〇丨至! V)至字元線WLX·2而產生一隔離區,並且減輕閘極引發浅 漏GIDL,一中間閘極保護電壓(約3 v)vgp會被施加至字元 線WLxq。在與源極側(已程式化側)上該被選擇記憶體元件 相隔一特定的預定距離處,會藉由施加該隔離電壓vis。至 字元線WLZ而產生一第二隔離區。另外,在汲極側(已抹除 123781.doc -45- 200832411 側)上特疋的預定距離處,還會藉由施加該隔離電壓vis。至 字元線WLy而產生另一第二隔離區。於一較佳的具體實施 例中’反映在圖14B中,在該被選擇記憶體兩側上之介於 4第二隔離區與該被選擇記憶體之間的距離係相同的,換 口之為十六條子元線。亦可使用其它的組態,並且可透過 測武來決定不同裝置組態的最佳間隔。因此會產生多個隔 離區用以界定多個昇壓區,它們會被標示為區域A、區域 B、區域c、以及區域d。被施加在區域]8與c之中的較佳傳 &電壓係Vpass(約為8 V),而被施加在區域a與D之中的較 ^傳‘電壓係VpassL(約為6 v至7 V)。亦可使用其它電壓。 最重要的係,被施加在區域八與D之中的電壓低於被施加 在區域B與C之中的電壓。程式化干擾在該NAND串的任一 邊緣上的字元線處(也就是,WL0、 WL1、WL2以及 WL61、WL62、WL63)會比較嚴重,因此,被施加至該等 較同與較低字元線的較低傳導電壓vpassL會充足地昇壓通 道電壓,以便即使較長的串中的程式化干擾的持續時間較 長,仍可實質上減輕程式化干擾。因為該等較低與較高字 元線係由VpassL來昇壓,所以即使係較長的串,仍可實質 上減輕程式化干擾。 圖14B顯示使用在此修正技術的一範例中的偏壓條件範 例。最左邊的行顯示被選擇要進行程式化的字元線。該被 選擇子元線右邊的64行顯示被施加至(本範例中)前面二十 條子元線中每一條字元線的電壓。因此,第一列的 右邊)顯示被選擇要進行程式化的字元線WL〇以及被施加 123781.doc -46- 200832411 至字元線WLG的程式化電壓Vpgm。因為wu)係最靠近源極 側的第-字元線’所以’右邊的所有其它儲存元件均不會 被程式化’並且不會朝此第—儲存元件的源極側產生任何 隔離區。如上述’藉由施加隔離電壓U字元線和5便 會在該未被程式化的汲極侧上與該被選擇的字元線相隔十 六條字元線處產生一第二隔離區。如上面所提,昇壓電壓For example, Vpass3 and Vpass7 may be about 6 V to 8 V, two (four) and V”6 may be about 2 V to 4 V, and Vpass 5 may be about 〇^, such as Vpassl may be about 10 ¥ to u V Vpass2 may be approximately 8 V. In the sustain pulse, Vpgm may range from 16 V to Μ V. The optimum voltage for a particular memory device can be determined by testing. An extended trough or isolation region is formed which is concentrated on the memory element (e.g., memory device 133G) having the lowest applied conduction voltage and when the conduction voltage is on each side of the isolation region When symmetrically or asymmetrically increased, the slope of the boost profile will be away from the isolation region in a relatively gentle manner. Furthermore, the isolation region may contain odd or even memory components. The isolation region is used to isolate two The highly boosted channel region sub-subtracts the voltage on the drain and source regions of the word line with the lowest bias voltage (eg, the word line associated with memory element 1330) to avoid or reduce the word Belt to tunnel under the line. By extending the isolation The length of the zone further isolates the two highly stepped channel regions to avoid or reduce leakage between the two boost regions and to avoid or reduce band-to-band wear below the word line. The stylized interference reduction technique is applicable to both multi-level and unit quasi-stylization. Multi-level memory is expected to obtain greater benefits. For unit quasi-memory, it is expected to be larger under the following conditions. The benefits are stylized from the source side to the drain side of the N AND string in a predetermined word line order compared to programming in a random order. Again, these techniques can be combined with all boosts in principle. Technology is used. However, multi-bit quasi-boost mode (such as EASB and its variants) is expected to have the best benefit. 123781.doc -44- 200832411 When the device is further reduced, these techniques can be further corrected to solve Stylized interference effects and related crashes. For example, when scaling from 8_ NAND strings to i6-NAND strings and 32-NAND strings, the use of these boosting techniques will achieve significant benefits. Now move the focus to Longer On the string 'for example, 64_NAND string, where a single bit line will touch M storage elements. Therefore, stylized interference will be the worst in longer strings, and the duration of stylized interference will be 32-NAND The duration of the string is twice the duration of the 16-NAND string and eight times the duration of the 8-NAnd string. Figure 14A illustrates a NAND string 1400 with 64 memory elements. In an embodiment, additional isolated word lines are used to further divide the programmed and/or erased regions. The NAND string 1400 includes a source side select gate 1470 and a drain side select gate 148. 〇, and memory elements 1401, 402, ... 1464, which are disposed in a p-well 149. The parent memory element can be selected via a corresponding word line WL 〇 , WL1 · · · WL63. Assuming that the word line WLX is selected for programmaticization, the stylized voltage Vpgm is applied to the word line. On the source side of the selected memory element, a very low isolation voltage Vis is applied. (about ! to V) to the word line WLX·2 to generate an isolation region, and reduce the gate induced shallow drain GIDL, an intermediate gate protection voltage (about 3 v) vgp will be applied to the word line WLxq . The isolation voltage vis is applied by a predetermined predetermined distance from the selected memory element on the source side (programmed side). A second isolation region is created to the word line WLZ. In addition, the isolation voltage vis is also applied by a predetermined distance on the drain side (the side of the 123781.doc -45-200832411 has been erased). Another second isolation region is created to the word line WLy. In a preferred embodiment, 'reflected in FIG. 14B, the distance between the second isolation region and the selected memory on both sides of the selected memory is the same, and the replacement is It is a sixteen child line. Other configurations can also be used, and the best spacing for different device configurations can be determined by measurement. Thus, a plurality of isolation regions are created to define a plurality of boost regions, which are labeled as region A, region B, region c, and region d. The preferred pass & voltage system Vpass (approx. 8 V) applied to the regions 8 and c is applied to the voltage-passing voltage VpassL in the regions a and D (approximately 6 v to 7 V). Other voltages can also be used. The most important system is that the voltage applied to the regions 8 and D is lower than the voltage applied to the regions B and C. Stylized interference at the word line on either edge of the NAND string (ie, WL0, WL1, WL2, and WL61, WL62, WL63) is more severe and, therefore, applied to the same and lower words. The lower conduction voltage vpassL of the line will sufficiently boost the channel voltage to substantially alleviate stylized interference even if the duration of the stylized interference in the longer string is longer. Since these lower and higher word lines are boosted by VpassL, even with longer strings, the stylized interference can be substantially alleviated. Fig. 14B shows an example of a bias condition used in an example of the correction technique. The leftmost line shows the word line that was selected for stylization. The 64 lines to the right of the selected sub-line display the voltage applied to each of the first twenty sub-line lines (in this example). Therefore, the right side of the first column shows the word line WL 被 selected to be programmed and the stylized voltage Vpgm to which 123781.doc -46 - 200832411 to the word line WLG are applied. Since wu) is closest to the source-word line of the source side, all other storage elements on the right side of the 'n' are not programmed' and do not create any isolation regions towards the source side of the first storage element. A second isolation region is created by applying an isolation voltage U-character line and 5 as described above at a distance of sixteen character lines from the selected word line on the unstylized drain side. As mentioned above, boost voltage

Vpass會被施加至該被選擇字元線與該第二隔離點之間的所 有字元線(WL1至WL14) ’而-較低的昇壓電壓VpassL會被 施加至該第二隔離點以外朝向該汲極的所有字元線 至 WL63) 〇 舉例來說,當字元線WL19被選為要進行程式化,那麼 圖14B顯示被施加至WL19的程式化電壓Vpgm。進一步言 之,閘極保濩電壓vgp會被施加至源極側上就在該被選擇 ^元線旁邊的字元線WL18,而隔離電壓%。則會被施加至 字7G線WL17,從源極側上該被選擇字元線處移除一字元 線。此外,會藉由施加該隔離電壓Vis。至字元線wl4,與 該被選擇字元線相隔十六條字元線,而在該被選擇字元線 的源極側上產生-第二隔離區。圖14B中雖然並未顯示, 不過還會藉由施加該隔離電壓%。至字元線和4,與該被 選擇字元線相隔十六條字元線,而在該被選擇字元線的汲 極側上產生另一第二隔離區。亦可使用其它變化例。舉例 來說:閘極保護電壓Vgp可被施加至要施加該隔離電壓%。 的字元線兩側上的字元線。 圖15A.兒明田利用單一傳導電愿以典型的紅mb技術來 123781.doc -47- 200832411 程式化該等較高字元線時的一 64-NAND串1500的偏壓情 形。於此情況中,程式化電壓Vpgm會被施加至字元線 WL62,閘極保護電壓Vgp會被施加至WL6i,而隔離電壓 vis。則會被施加至字元線WL60。傳導電壓Vpass會被施加至 该已程式化區之中的其餘字元線(字元線Wl〇至WL5 9)及該 已抹除區之中的其餘字元線(字元線界以”二者。不過,在 該些偏壓條件下,倘若橫跨該隔離區的洩漏電流很大的 話,昇壓電壓的效應便會因該已程式化側上的大通道電容 及該已抹除側上的小通道電容而減弱。 圖1 5B說明一種經修正的偏壓技術,其有助於在程式化 較高(或較低)字元線時來限制該通道電容。程式化電壓 Vpgm會再度被施加至字元線WL62,閘極保護電壓會被 施加至WL61,而隔離電壓。則會被施加至字元線 WL60。藉由施加隔離電壓Vis◦至字元線WLz便會在該已程 式化侧上產生一第二隔離區,其係位於和該被選擇字元線 相隔特疋預设距離處,舉例來說,於其中一具體實施例中 會相隔16條字元線。這會有效地將該已程式化區分成兩個 隔離區。一傳導電壓Vpass(約8 ¥至1〇 V)會被施加至該被選 擇字元線兩侧上的字元線(舉例來說,冒^+1至wl59與 WL63),不過,兩個區域上的傳導電壓並不必相同。一較 低的傳導電壓VpassL(較佳的係、,比Wl v)會被施 加至該第二隔離區的源極侧上的其餘字元線(舉例來說, WL0至WLZ-1)。|亥已程式化側上的電容便會因而在該等隔 離區之間被分割,並且會限制浪漏對下—個區域的衝擊。 123781.doc -48- 200832411 圖1 6說明該修正偏壓技術的變化例。因為由於已程式化 臨界電壓的關係,該已程式化區之中的昇壓通道電壓通常 會比較低,所以,會使用略高的傳導電壓來補償區 域B之中的低通道電位。 程式化電壓Vpgm會被施加至該被選擇的字元線WLX。藉 由施加隔離電壓Vis。至字元線WLX_2且施加閘極保護電壓 Vgp至字元線WLxu便會產生一隔離區。藉由施加隔離電壓 Vis。至字元線界1^與WLy便會在該被選擇字元線的兩側上產 生第二隔離區。於此圖中,該等隔離昇壓區係被標示為區 域A、區域B、區域C、以及區域〇。標準的傳導電壓 Vpass(約8 V至10 V)會被施加在區域c之中。略高的傳導電 壓VpassH(約比Vpass高1 V至2 V)會被施加至源極側昇壓區B 之中的字元線(約6 V至7 V)。略低的傳導電壓VpassL(約比 Vpasd&l V至2 V)會被施加在其餘的區域A與D之中。此配 置會補償區域B中的低通道電位。 圖17說明該修正偏壓技術的又一變化例。這和圖14A中 所示的組態雷同,不同之處在於使用一條以上的字元線來 產生該等第二隔離區。因此,程式化電壓Vpgm會被施加至 該被選擇的字元線WLX。藉由施加隔離電壓Vis。至字元線 WLX·2且施加閘極保護電壓Vgp至字元線贾、」便會產生一 隔離區。藉由施加隔離電壓vis。至位於與該被選擇字元線 相隔特定距離的字元線WLZ與WL/E會在該被選擇字元線 的兩側上增強第二隔離區。於此具體實施例中,係藉由施 123781.doc -49- 200832411 加閘極保護電壓V $Α μ 一 私i ν印至予兀線WLZ+1與WLy”來產生該第二 離區。傳導電壓八…會被施加至區域^與C之中的字元 線而較低的傳導電壓VpassL會被施加至區域八與1)之中的 一、、、复此配置有助於經由區域B與D之中的隔離區及 GIDL來降低洩漏電流。 其它的變化例同樣可獲得好處。舉例來說,-間極保護 電2 vgp可被施加至要施加該隔離電壓◦的字元線兩側上 的子元線’舉例來說’偏若被施加至乳W與机z的 活,那麼Vgp便會被施加至WLx_3與WLw以及】诳 WLZ+1。 …、 ,另欠化例中,會將一傳導電壓施加至該該被選擇字 域之源極側上的字元線。因此,舉例而言,倘若^被 靶加至WL62的話,那麼便會被施加至wl^,丨⑼會被 施加至WL6〇,Vis。而會被施加至wl59。 本文前面已經提出本發明之詳細說明以達解釋與說明的 目的其並不希望包攬無遺或將本發明限制於所揭示的刻 板形式。根據以上教示内容’可進行許多修改及變更。本 文所況明的具體實施例均經過選擇以便最佳地解釋本發明 的原理及其實務應用,從而使熟習本技術的人士能夠在各 種具體實施例中運用本發明並且配合於所預期之特定用法 之各種修正例來運用本發明。本文希望本發明之範脅係由 隨附之申請專利範圍來定義。 【圖式簡單說明】 圖1係一非揮發記憶體NAND串的俯視圖。 123781.doc -50- 200832411 圖2係圖1的NAND串的一等效電路圖。 圖3係用於圖示三條NAND串的一電路圖。 圖4係一可用以施行本發明之一或多個具體實施例的快 閃記憶體系統的一具體實施例的方塊圖。 圖5說明一記憶體陣列之組織的範例。 圖6描述一程式化電壓信號的範例。 圖7係用於說明一程式化程序的具體實施例的流程圖。 圖8 A描述在從已抹除狀態直接程式化至已程式化狀態的 一多狀態裝置中的一組臨界電壓分佈範例。 圖8B顯示在從已抹除狀態雙通程式化至已程式化狀態的 一多狀態裝置中的一組臨界電壓分佈範例。 圖9A至9C所示的係各種臨界電壓分佈並且說明一種用 於程式化非揮發記憶體的程序。 圖9D係用於解釋一典型程式化程序的時序圖。 圖10說明當使用EASB昇壓模式時具有不平衡昇壓通道 區的一 NAND串。 圖11說明具有平衡昇壓通道區的一 NAND串。 圖12說明在昇壓通道區之間具有一隔離區的一 nand 串。 圖13說明在昇壓通道區之間具有一替代隔離區的一 NAND串。 圖14A說明利用經修正的rEASB|術來對使用額外隔離 區的一 64-NAND串進行偏壓。 圖14B係顯示使用圖14a的REASB技術之用於多條字元 123781.doc •51 - 200832411 線的偏壓條件表。 圖15A說明當利用習知的REASB技術來程式化該等較高 的字元線時的一 64-NAND串的偏壓情形。 圖15B說明當利用經修正的REASB技術來程式化該等較 高的字元線時的一 64-NAND串的偏壓情形。 圖16說明利用經修正的REASB技術的變化例來對一 64-N.AND串進行偏壓。 圖1 7說明利用經修正的REASB技術的另一變化例來對一 64-NAND串進行偏壓的示意圖。 【主要元件符號說明】 100 電晶體 100CG 控制閘極 100FG 浮動閘極 102 電晶體 102CG 控制閘極 102FG 浮動閘極 104 電晶體 104CG 控制閘極 104FG 浮動閘極 106 電晶體 106CG 控制閘極 106FG 浮動閘極 120 |選擇閘極 120CG 控制閘極 123781.doc -52- 200832411 122 122CG 126 128 201 202 203 204Vpass will be applied to all word lines (WL1 to WL14) between the selected word line and the second isolation point and - a lower boost voltage VpassL will be applied to the outside of the second isolation point All of the word lines of the drain to WL63) 〇 For example, when word line WL19 is selected for programming, then Figure 14B shows the stylized voltage Vpgm applied to WL19. Further, the gate guard voltage vgp is applied to the word line WL18 on the source side just below the selected element line, and the voltage % is isolated. It is then applied to word 7G line WL17, removing a word line from the selected word line on the source side. In addition, the isolation voltage Vis is applied. To the word line wl4, sixteen word lines are separated from the selected word line, and a second isolation area is generated on the source side of the selected word line. Although not shown in Fig. 14B, the isolation voltage % is also applied. To the word line and 4, a sixteen word line is separated from the selected word line, and another second isolation area is created on the negative side of the selected word line. Other variations can also be used. For example, the gate protection voltage Vgp can be applied to the % of isolation voltage to be applied. The word line on either side of the word line. Figure 15A. The use of a single conduction power by a child with a single conduction is intended to be a typical red mb technique. 123781.doc -47- 200832411 Stylizes the bias of a 64-NAND string 1500 when these higher word lines are programmed. In this case, the stylized voltage Vpgm is applied to the word line WL62, and the gate protection voltage Vgp is applied to the WL6i to isolate the voltage vis. It is then applied to the word line WL60. The conduction voltage Vpass is applied to the remaining word lines (word lines W1〇 to WL5 9) in the programmed area and the remaining word lines among the erased areas (the word line boundary is "2" However, under these bias conditions, if the leakage current across the isolation region is large, the effect of the boost voltage will be due to the large channel capacitance on the programmed side and the erased side. The small channel capacitance is attenuated. Figure 1 5B illustrates a modified bias technique that helps to limit the channel capacitance when stylizing higher (or lower) word lines. The stylized voltage Vpgm is again Applied to word line WL62, a gate protection voltage is applied to WL61, and an isolation voltage is applied to word line WL60. This is programmed by applying isolation voltage Vis◦ to word line WLz. A second isolation region is formed on the side, which is located at a predetermined distance from the selected word line. For example, in one embodiment, 16 word lines are separated. This will effectively The program is divided into two isolation zones. A conduction voltage Vpass (about 8 To 1 〇V) will be applied to the word lines on both sides of the selected word line (for example, ^+1 to wl59 and WL63), however, the conduction voltages on the two regions do not have to be the same. A lower conduction voltage VpassL (preferably, Wl v) is applied to the remaining word lines on the source side of the second isolation region (for example, WL0 to WLZ-1). The capacitance on the programmed side of the sea is thus split between the isolation regions and limits the impact of the leakage on the next region. 123781.doc -48- 200832411 Figure 1 illustrates the modified bias technique A variation of the boost channel voltage in the programmed region is usually lower because of the programmed threshold voltage, so a slightly higher conduction voltage is used to compensate for the low channel potential in region B. The stylized voltage Vpgm is applied to the selected word line WLX. An isolation region is generated by applying the isolation voltage Vis. to the word line WLX_2 and applying the gate protection voltage Vgp to the word line WLxu. By applying the isolation voltage Vis. to the word line boundary 1^ and WLy will be in the selected word A second isolation region is created on both sides of the line. In this figure, the isolation boost regions are labeled as region A, region B, region C, and region 〇. Standard conduction voltage Vpass (about 8 V to 10 V) will be applied in the region c. A slightly higher conduction voltage VpassH (about 1 V to 2 V higher than Vpass) will be applied to the word line in the source side boost region B (about 6 V to 7 V). A slightly lower conduction voltage, VpassL (approximately Vpasd & 1 V to 2 V), is applied to the remaining regions A and D. This configuration compensates for the low channel potential in region B. Figure 17 illustrates A further variation of the biasing technique is modified. This is similar to the configuration shown in Figure 14A, except that more than one word line is used to create the second isolation regions. Therefore, the stylized voltage Vpgm is applied to the selected word line WLX. By applying an isolation voltage Vis. An isolation region is created by the word line WLX·2 and the application of the gate protection voltage Vgp to the word line. By applying an isolation voltage vis. The word lines WLZ and WL/E located at a specific distance from the selected word line enhance the second isolation region on both sides of the selected word line. In this embodiment, the second isolation region is generated by applying a gate protection voltage V Α μ 一 私 至 兀 兀 WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL WL The conduction voltage VIII... is applied to the word line among the regions ^ and C and the lower conduction voltage VpassL is applied to one of the regions VIII and 1), and the configuration is facilitated via the region B. The isolation region and the GIDL in D are used to reduce the leakage current. Other variations can also benefit. For example, the interpole protection power 2 vgp can be applied to both sides of the word line to which the isolation voltage is applied. The upper sub-line 'for example' is applied to the milk W and the machine z, then Vgp will be applied to WLx_3 and WLw and 诳 WLZ+1. ..., in another example, Applying a conduction voltage to the word line on the source side of the selected word field. Thus, for example, if the target is added to WL62, then it will be applied to wl^, 丨(9) will be Applied to WL6〇, Vis. will be applied to wl59. The detailed description of the invention has been presented herein for explanation and explanation. The invention is not intended to be exhaustive or to limit the invention to the disclosed form. Many modifications and variations are possible in light of the above teachings. The specific embodiments set forth herein are selected to best explain the invention. The present invention may be applied to a person skilled in the art using the present invention in various embodiments and in various modifications to the specific usage contemplated. It is intended that the present invention be The scope of the patent application is defined. [Simplified Schematic] Figure 1 is a top view of a non-volatile memory NAND string. 123781.doc -50- 200832411 Figure 2 is an equivalent circuit diagram of the NAND string of Figure 1. A circuit diagram for illustrating three NAND strings. Figure 4 is a block diagram of a particular embodiment of a flash memory system that can be used to implement one or more embodiments of the present invention. Figure 5 illustrates a memory An example of the organization of an array. Figure 6 depicts an example of a stylized voltage signal.Figure 7 is a flow chart illustrating a specific embodiment of a stylized program. A set of threshold voltage distribution paradigms in a multi-state device that is directly programmed from the erased state to the programmed state. Figure 8B shows a multi-state in a double-pass programmed from the erased state to a programmed state. A set of threshold voltage distribution examples in the device. Figures 9A through 9C show various threshold voltage distributions and illustrate a program for stylizing non-volatile memory. Figure 9D is a timing diagram for explaining a typical stylized program. Figure 10 illustrates a NAND string with unbalanced boost channel regions when using the EASB boost mode.Figure 11 illustrates a NAND string with balanced boost channel regions. Figure 12 illustrates a nand string with an isolation region between the boost channel regions. Figure 13 illustrates a NAND string with an alternate isolation region between boost channel regions. Figure 14A illustrates the use of modified rEASB| to bias a 64-NAND string using additional isolation regions. Figure 14B is a table showing bias conditions for a plurality of characters 123781.doc • 51 - 200832411 lines using the REASB technique of Figure 14a. Figure 15A illustrates the biasing of a 64-NAND string when the higher word lines are programmed using conventional REASB techniques. Figure 15B illustrates the biasing of a 64-NAND string when the higher ideology lines are programmed using the modified REASB technique. Figure 16 illustrates the use of a modified version of the modified REASB technique to bias a 64-N. AND string. Figure 17 illustrates a schematic diagram of biasing a 64-NAND string using another variation of the modified REASB technique. [Main component symbol description] 100 transistor 100CG control gate 100FG floating gate 102 transistor 102CG control gate 102FG floating gate 104 transistor 104CG control gate 104FG floating gate 106 transistor 106CG control gate 106FG floating gate 120 |Select Gate 120CG Control Gate 123781.doc -52- 200832411 122 122CG 126 128 201 202 203 204

206 220 222 224 226 228 230 240 242 244 246 248 250 260 262 123781.doc 選擇閘極 控制閘極 位元線接點 源極線接點 NAND串 位元線 NAND串 位元線 NAND 串 位元線 選擇電晶體 記憶體元件 記憶體元件 記憶體元件 記憶體元件 選擇電晶體 選擇電晶體 記憶體元件 記憶體元件 記憶體元件 記憶體元件 選擇電晶體 選擇電晶體 記憶體元件 -53- 200832411 264 記憶體元件 266 記憶體元件 268 記憶體元件 270 選擇電晶體 302 記憶體元件陣列 304 行控制電路 306 列控制電路 308 P井控制電路 310 C源極控制電路 3 12 資料輸入/輸出緩衝器 314 命令電路 315 控制電路系統 316 狀態機 3 18 控制器 1000 NAND 串 1005 P井區 1010 選擇閘極 1015 記憶體元件 1020 記憶體元件 1025 記憶體元件 1030 記憶體元件 1035 記憶體元件 1040 記憶體元件 1045 記憶體元件 123781.doc -54· 200832411206 220 222 224 226 228 230 240 242 244 246 248 250 260 262 123781.doc Select gate control gate bit line contact source line contact NAND string bit line NAND string bit line NAND string bit line selection Transistor memory element memory element memory element memory element selection transistor selection transistor memory element memory element memory element memory element selection transistor selection transistor memory element -53- 200832411 264 memory element 266 Memory Element 268 Memory Element 270 Selective Crystal 302 Memory Element Array 304 Row Control Circuit 306 Column Control Circuit 308 P Well Control Circuit 310 C Source Control Circuit 3 12 Data Input/Output Buffer 314 Command Circuit 315 Control Circuit System 316 state machine 3 18 controller 1000 NAND string 1005 P well region 1010 selection gate 1015 memory component 1020 memory component 1025 memory component 1030 memory component 1035 memory component 1040 memory component 1045 memory component 123781.doc - 54· 200832411

1050 記憶體元件 1055 選擇閘極 1100 NAND 串 1105 p井區 1110 選擇閘極 1115 記憶體元件 1120 記憶體元件 1125 記憶體元件 1130 記憶體元件 1135 記憶體元件 1140 記憶體元件 1145 記憶體元件 1150 記憶體元件 1155 選擇閘極 1200 NAND 串 1205 P井區 1210 選擇閘極 1215 記憶體元件 1220 記憶體元件 1225 記憶體元件 1230 記憶體元件 1235 記憶體元件 1240 記憶體元件 1245 記憶體元件 -55- 123781.doc 200832411 1250 1255 1300 1305 • 1310 , 1315 1320 1325 # 1330 1335 1340 1345 1350 1355 1400 1401 φ 14021050 Memory Element 1055 Select Gate 1100 NAND String 1105 p Well Area 1110 Select Gate 1115 Memory Element 1120 Memory Element 1125 Memory Element 1130 Memory Element 1135 Memory Element 1140 Memory Element 1145 Memory Element 1150 Memory Component 1155 Select Gate 1200 NAND String 1205 P Well Area 1210 Select Gate 1215 Memory Element 1220 Memory Element 1225 Memory Element 1230 Memory Element 1235 Memory Element 1240 Memory Element 1245 Memory Element -55- 123781.doc 200832411 1250 1255 1300 1305 • 1310 , 1315 1320 1325 # 1330 1335 1340 1345 1350 1355 1400 1401 φ 1402

1464 1470 1480 1490 1500 SGD SGS WL0-WL63 記憶體元件 選擇閘極 NAND 串 ρ井區 選擇閘極 記憶體元件 記憶體元件 記憶體元件 記憶體元件 記憶體元件 記憶體元件 記憶體元件 記憶體元件 選擇閘極 NAND 串 記憶體元件 記憶體元件 記憶體元件 選擇閘極 選擇閘極 P井區 NAND 串 没極選擇線 源極選擇線 字元線 123781.doc -56-1464 1470 1480 1490 1500 SGD SGS WL0-WL63 Memory Element Selection Gate NAND String ρ Well Area Selection Gate Memory Element Memory Element Memory Element Memory Element Memory Element Memory Element Memory Element Memory Element Selection Gate Polar NAND string memory element memory element memory element selection gate selection gate P well area NAND string no pole selection line source selection line word line 123781.doc -56-

Claims (1)

200832411 十、申請專利範圍: 1’ —#心程m非揮發儲存元件 隔離和該等非揮發儲存 ㈣: 區’其中第-通道區與第二通道區=的至少三個通道 元件,而筮二、s、爸广 罪近—被選擇的儲存 而弟二通道區則遠離該被選 提昇每-個通道區的電位,Α中/储存元件; 係利用低於該等第一與第:二通運區的電位 提昇。 &之電位的昇壓電壓來 2·如請求们之方法,其中 係利用笛曰「 與弟二通道區之電位 利用箓-曰广 而忒弟二通道區的電位係 矛J用弟一昇壓電壓來提昇。 3·如請求項2之方法,直中 曰 昇壓電壓。 ”中4弟—外壓電壓略大於該第二 4.如請求項3 壓電壓大 >’,、中該第-昇壓電壓約比該第二昇 包I大一至二伏特。 5 ·如請求項1 、<万法,其中該等第一盥笙一 該被選擇的钱六 /、弟一通迢區係位於 的储存元件的相反側上。 6 ·如請求項$夕 俜利用第方法’其中該等第-與第二通道區之電位 利用第二^昇壓電墨來提昇,而該第三通道區的電位係 於’楚汁壓電壓來提昇,且其中該第一昇壓電壓略大 於3亥弟二昇壓電壓。 7·如請求項5 处 A决,其進一步包括: 4吏用第一 使用第〜昇壓電壓來提昇該第一通道區的電位, 外壓電壓來提昇該第二通道區的電位,以及 123781.doc 200832411 使用第二昇壓電壓來提昇該第三通道區的電位, 其中該第三昇壓電壓大於該第一昇壓電壓,且該第〆 昇壓電壓大於該第二昇壓電壓。 8·如睛求項7之方法,#中該第三昇壓電壓略大於該第〆 昇壓電壓。 9·如哨求項8之方法,其中該第三昇壓電壓約比該第一昇 壓電壓大一至二伏特。 10.如請求項7之方法,其進一步包括: 隔離—第四通道區,其中該第四通道區遠離該被選擇 的儲存元件並且係位於該第三通道區旁邊,且其中該第 二通道區係位於該第一通道區旁邊; 使用第=昇壓電壓來提昇該第四it道區的電位。 11·如請求項1〇之方法,其進一步包括: :吏用第三昇壓電壓來提昇該第一通道區的電位,其中 該第三昇壓電壓大於該第一昇壓電壓。 12.如請求項1之方法,其中該隔離步驟包含: ^ 隔離電壓至位於該被選擇的儲存元件近端的第 一儲存7C件且施加至位於該被選擇的儲存元件遠端的第 一儲存元件。 13·如明求項12之方法,其中該隔離步驟進-步包含: 閘極保護電壓至正好位於該第-儲存元件旁邊 的第三儲存it件且施加至正好位於該第二儲存元件旁邊 的第四儲存元件。 14.如請求項1 ^夕 、z芡方法,其中該隔離步驟進一步包含: 123781.doc 200832411 施加一閘極保護電壓至正好位於該第一儲存元件之相 反側上方邊的第二與第四儲存元件且施加至正好位於該 第二儲存元件之相反側上旁邊的第五與第六儲存元件。 15·如請求項12之方法,其中該等昇壓步驟包含: 施加戎第一昇壓電壓至位於和該等第一與第二通道區 相對應處的儲存元件,以及施加該第二昇壓電壓至位於 和忒第二通道區相對應處的儲存元件。 16· —種用於程式化一組非揮發儲存元件的系統,其包括·· 複數個非揮發儲存元件; 吕理電路,其會與該等儲存元件相通,其中該管理 電路會隔離和該等非揮發儲存元件相關聯的至少三個通 道區其中第一通道區與第二通道區靠近一被選擇的儲 存几件,而第三通道區則遠離該被選擇的儲存元件,並 且S提昇每一個通道區的電位,其中該第三通道區的電 位係利用低於该等第一與第二通道區之電位的昇壓電壓 來提昇。 17·如明求項16之系、、统,其中該等第一與第二通道區之電位 係由該管理電路利用第一昇壓電壓來提昇,而該第三通 道區的電位係利用第二昇壓電壓來提昇。 1 8 ·如請求項 <糸統,其中該第一昇壓電壓略大於該第二 昇壓電壓。 19·如明求項18之系統,其中該第一昇壓電壓約比該第二昇 壓電壓大一至二伏特,。 2 0 ·如請求項〗&夕i μ t 、之糸、、先,其中該等第一與第二通道區係位於 123781.doc 200832411 该被選擇儲存元件的相反側上。 士明求:20之系統,其中該等第一與第二通道區之電位 係由該管理電路利用第一昇壓電壓來提昇,而該第三通 . 這區的電位係由該管理電路利用第二昇壓電壓來提昇, * 且其中該第—昇壓電壓略大於該第二昇壓電壓。 ”月牧員20之系統,其中該管理電路會利用第三昇壓電 :來:昇該第—通道區的電位,利用第一昇壓電壓來提 • 什ό亥第二通道區的 ,—曰 ^ 1 扪用弟一幵壓電壓來提昇該第 三通道區的電位,1 φ 筮—曰 其中该弟二昇壓電壓大於該第一昇壓 電壓且該第—昇壓電壓大於該第二昇壓電壓。 23·如請求項22之系統,苴中 曰 甲4弟二歼壓電壓略大於該第一 昇壓電壓。 24·如請求項23之系統,1中 曰 ,、Τ邊弟二歼壓電壓約比該第一異 壓電壓大一至二伏特。 ^ 25.如請求項20之系統,1 其 ,、甲口亥g理黾路會進一步隔離一第 • 四:道區,並且使用第二昇麗電塵來提昇該第四通道區 的电位’八中4第四通道區遠離該被選擇的儲存元件並 且係位於該第三通道 -^ ^ 且其中該第三通道區係位 於該第一通道區旁邊。 26.如請求項25之系统,甘士 — . 、先其中该官理電路會進一步使用第三 昇遷電壓來提昇該裳_ 、、, 弟一通遏區的電位,其中該第三昇壓 電壓大於該第一昇壓電壓。 27·如請求項16之系統,苴 ,、r 4 g理電路會施加一隔離電壓 至位於δ亥被選擇的儲左 子7L件近知的第一儲存元件且施加 I23781.doc 200832411 至位於該被選擇的儲存元件遠端的第二儲存元件。 h求員27之系統,其中該管理電路會施加 電壓至正好位於該第1存元件旁邊的第三儲存元件I 施加至JL好位於該筮-心—_ 、 一错存疋件旁邊的第四儲存元件。 29.如請求項27之车鲚,甘士》— ^ 糸‘其中該管理電路會施加一閘極保護 私£至正好位於該第一儲存元件之相反側上旁邊的第三 與第四儲存元件且施加至正好位於該第二儲存元件之相 反側上旁邊的第五與第六儲存元件。 3〇·如明求項27之系統,其中該管理電路會施加該第—昇壓 電壓至位於和該等第一與第二通道區相對應處的儲存元 件’以及施加該第二昇壓電壓至位於和該第三通道區相 對應處的儲存元件。 ^200832411 X. Patent application scope: 1' - #心程m non-volatile storage element isolation and such non-volatile storage (4): Zone 'where the first channel zone and the second channel zone = at least three channel components, and , s, dad sin near - the selected storage while the second channel area is far from the selected potential to enhance each channel area, Α中 / storage components; system utilization is lower than the first and second: second The potential of the zone is increased. The boost voltage of the potential of the & 2 is as requested by the method of the requester, which uses the flute to "use the potential of the two-channel area with the 箓-曰广忒 and the second-channel area of the younger brother. The voltage is boosted. 3. According to the method of claim 2, the voltage is boosted by the voltage in the middle. The middle 4 brother - the external voltage is slightly larger than the second 4. If the request voltage is large, the voltage is > The first boost voltage is approximately one to two volts greater than the second boost package I. 5 • As claimed in claim 1, < 10,000, wherein the first one is selected on the opposite side of the storage element on which the money is selected. 6 · If the request item $ 俜 俜 uses the method 'where the potential of the first- and second channel regions is boosted by the second boosting ink, and the potential of the third channel region is tied to the voltage To boost, and wherein the first boost voltage is slightly greater than the 3 boost voltage. 7. If the request item 5 is A, it further comprises: 4: using the first use of the boost voltage to boost the potential of the first channel region, external voltage to increase the potential of the second channel region, and 123781 .doc 200832411 uses a second boost voltage to boost the potential of the third channel region, wherein the third boost voltage is greater than the first boost voltage, and the second boost voltage is greater than the second boost voltage. 8. In the method of claim 7, the third boost voltage is slightly larger than the third boost voltage. 9. The method of claim 8, wherein the third boost voltage is about one to two volts greater than the first boost voltage. 10. The method of claim 7, further comprising: isolating - a fourth channel region, wherein the fourth channel region is remote from the selected storage element and is located adjacent to the third channel region, and wherein the second channel region The system is located beside the first channel region; the first boost voltage is used to boost the potential of the fourth channel region. 11. The method of claim 1, further comprising: ??? increasing a potential of the first channel region with a third boost voltage, wherein the third boost voltage is greater than the first boost voltage. 12. The method of claim 1, wherein the isolating step comprises: ^ isolating a voltage to a first storage 7C member located at a proximal end of the selected storage element and applying to a first storage located at a distal end of the selected storage element element. 13. The method of claim 12, wherein the step of isolating comprises: a gate protection voltage to a third storage element located just adjacent to the first storage element and applied to the side of the second storage element Fourth storage element. 14. The method of claim 1, wherein the isolating step further comprises: 123781.doc 200832411 applying a gate protection voltage to the second and fourth stores located just above the opposite side of the first storage element The components are applied to the fifth and sixth storage elements just above the opposite side of the second storage element. The method of claim 12, wherein the step of boosting comprises: applying a first boost voltage to a storage element located at a location corresponding to the first and second channel regions, and applying the second boost The voltage is to a storage element located at a location corresponding to the second channel region. 16. A system for programming a set of non-volatile storage elements, comprising: a plurality of non-volatile storage elements; a logic circuit that is in communication with the storage elements, wherein the management circuit isolates and At least three channel regions associated with the non-volatile storage element, wherein the first channel region and the second channel region are adjacent to a selected storage piece, and the third channel region is remote from the selected storage element, and S enhances each The potential of the channel region, wherein the potential of the third channel region is boosted by a boost voltage that is lower than the potential of the first and second channel regions. 17. The system of claim 16, wherein the potentials of the first and second channel regions are boosted by the management circuit using a first boost voltage, and the potential of the third channel region is utilized The second boost voltage is boosted. 1 8 as claimed in claim 1 wherein the first boost voltage is slightly greater than the second boost voltage. 19. The system of claim 18, wherein the first boost voltage is about one to two volts greater than the second boost voltage. 2 0 · If the request item & i i μ t , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , The system of 20: wherein the potentials of the first and second channel regions are boosted by the management circuit using a first boost voltage, and the potential of the third region is utilized by the management circuit The second boost voltage is boosted, and wherein the first boost voltage is slightly greater than the second boost voltage. "The system of the priest 20, in which the management circuit will use the third boosting power: to: raise the potential of the first channel region, and use the first boosting voltage to raise the second channel region of the ό ό ,, -曰^ 1 扪 using a voltage of the brother to increase the potential of the third channel region, 1 φ 筮 曰 曰 where the second boost voltage is greater than the first boost voltage and the first boost voltage is greater than the second Boost voltage 23. As in the system of claim 22, the voltage of the second voltage of the 曰中44 brother is slightly larger than the first boosting voltage. 24·The system of claim 23, 1 曰, Τ 弟二二The voltage of the voltage is about one to two volts greater than the first voltage of the same voltage. ^ 25. The system of claim 20, 1 of which, the mouth of the mouth is further isolated from the fourth: the road area, and used a second rising electric dust to raise the potential of the fourth channel region, wherein the fourth channel region is away from the selected storage element and is located in the third channel - ^ ^ and wherein the third channel region is located Beside the first passage area. 26. As in the system of claim 25, Gans-. The official circuit further uses the third boosting voltage to boost the potential of the skirt, wherein the third boost voltage is greater than the first boost voltage. 27. The system of claim 16,苴,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, The second storage element. The system of claim 27, wherein the management circuit applies a voltage to the third storage element I located just next to the first storage element, and is applied to the JL, which is located in the 筮-heart__, a defective component Next to the fourth storage element. 29. The rut of claim 27, Gans-- 糸' wherein the management circuit applies a gate protection to the side of the opposite side of the first storage element. Third and fourth storage elements and applied to the fifth and sixth storage elements located on the opposite side of the second storage element. The system of claim 27, wherein the management circuit applies the - boost And a voltage to a storage element located at the first and second channel regions and applying the second boost voltage to a storage element located at a location corresponding to the third channel region. 123781.doc123781.doc
TW096131530A 2006-09-06 2007-08-24 System and method for programming non-volatile memory with improved boosting TWI349286B (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US51715406A 2006-09-06 2006-09-06
US11/516,976 US7440326B2 (en) 2006-09-06 2006-09-06 Programming non-volatile memory with improved boosting

Publications (2)

Publication Number Publication Date
TW200832411A true TW200832411A (en) 2008-08-01
TWI349286B TWI349286B (en) 2011-09-21

Family

ID=38982715

Family Applications (1)

Application Number Title Priority Date Filing Date
TW096131530A TWI349286B (en) 2006-09-06 2007-08-24 System and method for programming non-volatile memory with improved boosting

Country Status (2)

Country Link
TW (1) TWI349286B (en)
WO (1) WO2008030775A2 (en)

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5715194A (en) * 1996-07-24 1998-02-03 Advanced Micro Devices, Inc. Bias scheme of program inhibit for random programming in a nand flash memory
JP3957985B2 (en) * 2001-03-06 2007-08-15 株式会社東芝 Nonvolatile semiconductor memory device
JP3863485B2 (en) * 2002-11-29 2006-12-27 株式会社東芝 Nonvolatile semiconductor memory device
US7161833B2 (en) * 2004-02-06 2007-01-09 Sandisk Corporation Self-boosting system for flash memory cells

Also Published As

Publication number Publication date
WO2008030775A2 (en) 2008-03-13
WO2008030775A3 (en) 2008-05-02
TWI349286B (en) 2011-09-21

Similar Documents

Publication Publication Date Title
TWI326487B (en) Non-volatile storage system and method for programming non-volatile storage
TWI402856B (en) Enhanced bit-line pre-charge scheme for increasing channel boosting in non-volatile storage
EP1759393B1 (en) Concurrent programming of non-volatile memory
JP4431139B2 (en) Self-boost technology for non-volatile memory
JP5334954B2 (en) Selective use of non-volatile memory programming prohibition scheme
EP2446443B1 (en) Forecasting program disturb in memory by detecting natural threshold voltage distribution
JP5439488B2 (en) Improved data retention of the last word line in non-volatile memory arrays
JP4490977B2 (en) Nonvolatile memory programming method
US7355888B2 (en) Apparatus for programming non-volatile memory with reduced program disturb using modified pass voltages
US20070171719A1 (en) Method for programming non-volatile memory with reduced program disturb using modified pass voltages
TW201324513A (en) Substrate bias during program of non-volatile storage
KR100984563B1 (en) Last-first mode and method for programming of non-volatile memory of nand type with reduced program disturb
TW200837764A (en) Non-volatile storage with early source-side boosting for reducing program disturb
TWI386944B (en) Non-volatile storage with boosting using channel isolation switching and method thereof
EP2304733A1 (en) Nand flash memory cell array and method with adaptive memory state partitioning
TWI378456B (en) Method and system for reducing program disturb in non-volatile storage
TW200832411A (en) System and method for programming non-volatile memory with improved boosting
JP4950299B2 (en) Reducing program disturb in non-volatile memory using multiple boost modes

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees