复旦大学王鹏飞半浮栅晶体管Science论文补充材料

合集下载

复旦大学成功制备新型二维晶体黑磷

复旦大学成功制备新型二维晶体黑磷
最 慢 两 年 要 在 甘 肃 、福 建 和 河北 等 地 共 建 设 6 个 生产基地 ,
在未来 的集成 电路里” 。
面 向未 来的 二维 黑磷 材料
黑磷 的半导体 能隙是个直接 能隙 , 这个特性让 它的光学
年产值 总计超过 2 0 亿 元 ,目前河北 占地 5 0亩 的基 地 已经基 和 光 电性能 同其 他材料 ( 包括硅 和硫化钼 )相 比有 巨大 的优 本完成 ,设备正在陆续调 试中。
能吸光饱和 ,然后夜间再慢慢释放 出来,也不需要人工控制 , 模较大的投资推进 落实速度 又 比较慢 。 还能够最低限度地消耗能量 ,使 用寿命在 8 年 ~l 0年 。”纳 明新材料董事长张 明德介绍 ,纳 米稀 土产 品不仅更节能环保 , 还要 比传统照 明方式更省钱 。 凭着纳米稀土 发光材料这项技术 ,纳明新材料获 评国家 高新技术企业称号 ,目前 已经拥有 7 8 项 专利 ,其 中发 明专利 就达 1 6项 。但是 ,该产品并非完美 无瑕 ,据 了解 ,其在傍 晚 和清晨 ,当天色介于亮与不 亮之间时 ,它还在 吸收光源可 能
体。与石墨烯 最大的不 同是 ,黑磷有一个 半导体 能隙。 “ 两
年 前 中 国 科 技 大 学 的 陈 仙 辉 教 授 告 诉 我 他 们 可 以 生 长 黑 磷
产业扩张遇资金瓶颈
时, 当时直觉就告 诉我 , 这有 可能是一 个很 好的半导 体材料 , ”
“ 去年 接 到了 1 . 2亿 元 的订 单 ,但 是 只完 成 了 3 0 0 0万 张远波教 授说 , “ 果然 ,现在我们把 黑磷 做成纳米厚 度的二 元。”张明德表示 ,由于资金的缺乏 ,设 备等无法 满足生产 维 晶体 后 ,发现它有 非常好的半导体 性质 ,这样就 有可能用 的现实需求 ,而市场 的需求正在呈持续增长趋势 。 根据纳 明新 材料的规划 ,其将继 续 以松 山湖作 为总部 ,

复旦专家研究有机薄膜晶体管稳定性获突破

复旦专家研究有机薄膜晶体管稳定性获突破

科研团队已成功将有机薄膜迁移率从 1 0 — 4 c m i / V s 提高到 1 0 c m 2 / V s 左右 ,增加 了四个数量级 .接近多
晶硅 的水平 ,达到了可实用的量级 。如果能大幅提 升有机薄膜晶体管的稳定性 ,未来 的应用前景将
会 非 常广 阔 。
来 源 : 的难题。实验测试显示 ,这种石榴结构的阳极具有优 良的性能 .用它
制造 的电池在完成 1 0 0 0次充放电后还有 9 7 %的 电量 ,完全能够满足实用要求 。相关研究发表在 2月
1 7日出版的 《 自 然? 纳米技术》杂志上 。
崔毅说 ,为了使这项技术能够更快商业化 ,还需要解决两个问题 :一是简化制造流程 ,二是找到 更便宜的硅纳米颗粒来源。 目 前他们发现稻壳或许会成为一个很好的原材料。这种农作物副产 品来源
“ 石 榴” 电极 可让 电池 容量 增 1 0倍
受石榴启发 ,美 国斯坦福大学和美 国能源部下属 S L A C国家加速器 实验室副教授崔毅 ( 音译 )和
他 的研 究 小组 利 用硅 纳米 线 和纳 米颗 粒 开发 出一 种 硅纳 米颗 粒 和碳 制 成 的新 型 电极 ,成 功破 解 了此前
第1 1 卷第 1 期
2 0 1 4年 2月
显微 与测量
Mi c r o t e c h n i q u e& Me a s u r e me n t
V0 1 . 1 1 No . 1 F e b r u a r y 2 01 4
复旦专家研究有机薄膜晶体管稳定性获突破
最新一期的国际权威性学术期刊 《 自然一 通讯》 ( N a t u r e C o m m u n i c a t i o n s )发表了复旦大学信息科 学与工程学院仇志军与刘冉科研团队的一项成果 ,他们在揭示有机薄膜晶体管 ( O T F T )性能稳定性机

梯度Si_1_c_Ge_c混晶太阳电池长波光谱响应及其短路电流的理论计算与分析

梯度Si_1_c_Ge_c混晶太阳电池长波光谱响应及其短路电流的理论计算与分析

第21卷第11期半 导 体 学 报V o l.21,N o.11 2000年11月CH I N ESE JOU RNAL O F SE M I CONDU CTOR S N ov.,2000梯度Si1-c Ge c混晶太阳电池长波光谱响应及其短路电流的理论计算与分析李成虎 赵玉文 黎雪梅 王福永 王玉亭 阎晓彬 陆 炜(北京市太阳能研究所,北京 100083)摘要:提出了一种新型Si Si1-c Ge c太阳电池结构.对长波波段的响应区域—Si基区Si1-c Ge c梯度区光生少子分布进行了求解.根据电流连续性原理计算了光谱响应SR,讨论了电池结构和梯度区最大锗浓度c对光生电流J I的影响.结果表明:在一定条件下梯度区的引入使得SR明显提高,但c和梯度区厚度同时增加,窄能隙的复合效应显现出来.电池结构对器件性能的影响更为显著.当梯度区厚度L2远小于基区厚度L1时,复合及电场的抽运使得SR低于相同厚度Si基区之值.关键词:太阳电池;Si1-c Ge c;光谱响应;短路电流;计算分析PACC:7361;8460;0260;8530中图分类号:TM914;TN366 文献标识码:A 文章编号:025324177(2000)1121122207Theoretica l Com puta tion and Ana lysis of Spectra Respon se of L ong W ave and Photogenera ted Curren t i n Grad ien t M ixedCrysta l Si1-c Ge c Solar CellsL I Cheng2hu,ZHAO Yu2w en,L I Xue2m ei,W AN G Fu2yong,W AN G Yu2ting,YAN X iao2b in and LU W ei(B eij ing S olar E nergy R esearch Institu te,B eij ing 100083,Ch ina)R eceived21A ugust1999,revised m anuscri p t received27O ctober1999Abstract:A new structure of Si Si1-c Ge c so lar cells is p resented and the distributi on of pho togenerated carries in Si2based regi on and Si1-c Ge c gradient regi on fo r the long w ave radiati on is so lved.A cco rding to the current continuity p rinci p le,the Spectra R esponse(SR)w as calculated.T he influence of so lar cell L I Cheng2hu(李成虎)w as bo rn on N ovem ber7,1959,received the Ph.D.degreee from Shanghai Institute of T ech2 no logy Physics of Ch inese A cadem y of Sciences,now is in the research on so lar cells.ZHAO Yu2w en(赵玉文)w as bo rn in1939,graduated from the T ianjin U niversity,P rofesso r,the directo r of N ati onal Engineering R esearch Center fo r R enew able Energy.1999208221收到,1999210227定稿structure and m ax concentrati on in gradient regi on on the pho togenerated current has also been discussed .T he results show that under som e conditi ons ,the SR can be enhanced evidently by introducing the Si 1-c Ge c gradient regi on ,but due to the si m ultaneous increases of c and the th ickness of gradient regi on ,the recom 2binati on th rough t the narrow band begins develop ing .T he influence of the cell structure on device perfo r 2m ance becom es mo re consp icuous .W hen the th ickness of the gradient regi on is far s m aller than that of the base regi on ,the SR is low er than that of the sam e th ickness in the base regi on of Si so lar cells by the re 2com binati on and the electric field excluding .Key words :so lar cells ;Si 1-c Ge c ;spectrum responce ;sho rt circuit current ;computati on and analysis PACC :7361;8460;0260;8530Article I D :025324177(2000)11210002051 引言在硅太阳电池的发展趋势中,提高效率减少成本是研究工作的主流,其主要途径为:提高电池光电转换效率;减少昂贵电子级硅材料用量或提高其利用率;发展成本低廉的制造工艺等.为此提出了各种新型的工艺和器件结构,如钝化发射结,激光刻槽,表面织构及背场技术等.但要想进一步提高效率成本比,电池结构上的创新成为主要途径.锗硅混晶太阳电池[1,2]即是在器件结构上的创新之一,它本质上是用能带工程来提高电池效率.目前为止已有各种不同的电池结构模型[3]及各种实现其结构的工艺法如L PE 和CVD 等[4,5].通过锗硅混晶的引入,增加光子的产生率和拓宽吸收谱可有效地提高电池效率.但锗硅混晶异质结构电池的界面复合限制了效率的提高.为提高效率而增加锗的浓度又导致复合加剧.对单晶硅太阳电池来说,目前工艺上实现的以锗浓度为c =10%左右最好[2],并且此文作者的数值模拟结果预言,在此条件下和硅器件比较,其效率可达到15%.为消除Si 1-c Ge c Si 异质结界面的复合影响及,本文设计了一种在工艺上可以实现的新型太阳电池结构,即优化能带结构的空间设计,又尽量减小器件内的复合过程.且由电流连续性原理求得光生电流的严格解,并分析器件结构对光生电流的影响.2 锗硅梯度浓度太阳电池结构及光生电流的理论求解目前提出的锗硅异质结太阳能电池的典型器件结构如图1所示[1,2],相应的能带结构见图2.图1 Si 1-c Ge c Si 异质结太阳电池的结构F IG .1 C ro ss 2Secti on of Si 1-c Ge c SiH etero juncti on So larCell 图2 Si 1-c Ge c 梯度异质结太阳电池能带结构示意图F IG .2 D iagram of Energy Band Structure fo r Si 1-cGe Si H etero juncti on So lar Cell 尽管锗硅混晶的俄歇复合比硅严重,但此结构的优点是可以提高电池效率而不必具体确定电池的详细结构(掺杂浓度,顶层厚度等).此类锗硅异质结太阳电池目前的最高效率为11.4%[2](锗硅混晶厚度为321111期李成虎等: 梯度Si 1-c Ge c 混晶太阳电池长波光谱响应及其短路电流的理论计算与分析15Λm).为提高电池效率,电池的设计仍可优化.为消除晶格失配造成的界面复合及锗浓度增加导致的俄歇复合加强,可以对锗硅混晶区域采用梯度锗浓度的结构.从p2Si p2Si1-c Ge c界面开始,锗浓度由0开始线性递增直到p2Si1-c Ge c p+2Si界面为止,最大浓度为c.在这个器件结构模型下,求出光生少子分布的严格解.进而求出电流方程.根据实际情况,我们只考虑纵向分布.由于Ge的浓度梯度,此区域内平移对称性破坏,但我们假设在平移对称存在的条件下导出的物理定律仍然有效,即作为近似,我们忽略平移对称破缺的影响.顶层和空间电荷区是硅吸收的短波部分,已有严格解[6],本文不再论述.基区p2Si和梯度浓度锗硅混晶区对光生电流的贡献由于边值条件和文献[6]不同,需另考虑求解.Peop le[7]等人指出存在应变情况下的Si Si1-c Ge c异质结界面处导带底E c变化与锗浓度c无关.对本文的器件结构来讲,不存在浓度的突变和应变.我们近似地认为能带的不连续性与浓度有关.物理图象是这样的,由于Ge原子的引入,每一格点原子的价电子相互作用增强,使得价带抬高;而激发到导带上的电子受到有效原子实的作用增强而导致电子能量下降.容易看到价带电子的相互作用强于原子实对导带电子的作用.为此,假设导带的变化为能隙E g变化的1 4.首先写出锗浓度变化导致能隙变化而造成的梯度区内的电场.B raunstein等人最早发表了Si1-c Ge c的禁带宽度随成分c的变化[8].他们在假设单声子辅助直接电子跃迁的条件下测量了锗硅混晶的禁带宽度随成分的变化.但最近R uiz等在假定双声子辅助间接跃迁的条件下,分析吸收边光谱数据得出描写Si1-c Ge c能隙对组份c依赖关系的公式[9](c<0.87):E g=1124(1-c)+830c+130c(1-c)(1)式(1)的单位是mV.在我们所讨论的浓度范围内(0≤c≤28%)能带结构是类硅的.梯度区的电场为:E=-9E c9c=-0.259E g9c(2)忽略组份c二次项,电场的空间分布是一常数.则Si1-c Ge c梯度异质结太阳电池能带图如图2所示.电场对光谱响应的影响将在后面讨论.设波长为Κ处太阳辐射光谱强度为5(Κ),电池的吸收系数为Α(Κ),每吸收一光子产生一对电子空穴,则单位时间内电子产生率为G(Κ)=Α(Κ)5(Κ)exp(-Α(Κ)x).由最一般的电子连续方程:9(∃n)9t=G-∃nΣ+D n∃ ∃(∃n)+Λn n∃ Eψ+Λn Eψ ∃(∃n)(3)和电池内的实际物理过程,可以写出基区和梯度区的连续方程及其边值条件.基区的连续方程和文献[6]相同,但是由于梯度区的存在,边值条件要由衔接条件确定.即在x=L1+W处,基区和梯度区的光生电子密度∃n相等,相应光生电流密度J L相等.故有:d2(∃n) d x2-∃nL e=-(1-Θ)Α5e-Α(d+x)D n(4)对梯度区,由电中性条件,∃ E=0,连续性方程为:d2(∃n g) d x2-EkTd(∃n g)d x-∃n gL eg=-(1-Θ)Α5e-Α(d+x)D eg(5)其中 ∃n、∃n g、L e、L eg、D n、D ng分别为基区和梯度区的光生电子的浓度、扩散长度、扩散系数.而光生电子在基区和梯度区的迁移率分别由Λe、Λeg表示.Θ为电池对入射辐射的反射率,计算时取Θ=0.(4)和(5)的边值条件为:x=W: ∃n(W)=n p0(e q V kT-1)(6.1)x=W+L1: ∃n(W+L1)=∃n g(W+L1)(6.2)x=W+L1: D g d(∃n)d x=D ngd(∃n g)d x-Λeg∃(n cg)E(6.3)x=W+L1+L2:D ng d(∃n g)d x-Λeg∃n g E=-S n∃n g(6.4)4211半 导 体 学 报21卷其中边值条件(6.4)的物理意义是背表面的复合.背表面复合速率S n 一般为1000—2000c m s .由于(4)和(5)两式是标准的二阶常系数非齐次微分方程,两式有解析解:∃n (x )=C 1e x l e +C 2e -x l e -(1-Θ)Α5D n (Α2-l -2e )e -Α(d +x )(7)∃n g (x )=C 3e x E 2kT +E 2kT +l -2eg +C 4e x (E 2kT -(E 2kT +l -2eg )-(1-Θ)Α5D ng Α2+E kT Α-l -2eg e -Α(d +x )(8)边值条件导出下列方程组:C 1e W l e +C 2e -W l e =(1-Θ)Α5D n (Α2-l -2e )e -Α(d +x )+n p0(e q V kT -1)=U 1=U 1A +U 1B (9.1)C 1e (W +L 1) l e +C 2e -(W +L 1) l e -C 3e(W +L 1) (E 2kT +E 2kT )-C 4e (W +L 1) (E 2kT -E 2kT )=(1-Θ)Αe-Α(d +W +L 1)1D n (Α2-l -2e )-1D ng (Α2+E kT Α-l -2eg )=U 2(9.2)C 1D n l e e (W +L 1) l e -C 2D n l ee -(W +L 1) L e -C 3Λng E -D ng E 2kT+E 2kT 2+l -2eg e (W +L 1) (E 2kT +E 2kT )+C 4Λng E -D ngE 2kT -E 2kT 2+l -2eg e (W +L 1) (E 2kT -E 2kT )=(1-Θ)5Αe -Α(d +W +L 1)1D n (Α2-l -2e )-1D ng (Α2+E kT Α-l -2eg )=U 3(9.3)C 3D ng E 2kT+E 2kT 2+l -2eg -Λeg E +S n e (W +L 1+L 2) (E 2kT +E 2kT )+C 4D ng E 2kT +E 2kT 2-l -2eg -Λge E +S n e(W +L 1+L 2) (E 2kT -E 2kT )(9.4)=(1-Θ)5Αe -Α(d +W +L 1)1D ng (Α2+E kT Α-l -2eg )S n D ng Α-E kT Α=U 4代数方程组(9)中C i 系数行列式非0,我们注意到(9.1)的常数项U 1分解为两项U 1A 和U 1B ,前者是光生伏特项,后者是由于光电压V 的正偏作用所形成的p 2n 结正向电流抵消项.由于讨论光电流谱,设电池是p 2n 结短路的,所以我们只考虑光生伏特项U 1A .由电流连续性原理,计算光电流谱时只需求出x =W 处的光生电流即可.从(10)式我们看到,梯度区对光生电流的影响通过代数方程组(9)式的诸常数项U i 和C i 的系数体现出来.因为C 1和C 2是诸常数项U i 的函数. x =W 处的波长为Κ光生电流:J L (Κ)=q D n C 11l e e W l e -C 21l e e -W l e +(1-Θ)5Αe -Α(d +W )1D n (Α2-l -2e )(10)而光谱响应SR 为:SR (Κ)=J L (Κ)q 5(Κ)(1-Θ)(11)所有波长贡献的光生电流如下:J L =∫∞0J L(Κ)d Κ(12)计算时对各种不同最大锗浓度为c 的梯度锗硅混晶区,和输运过程有关各参数取相应锗浓度为c 的均匀混晶区之值.光生电子扩散长度l eg 值按文献[5]取实验值.c <10◊,l eg =220Λm ;c >10◊,l eg =100Λm .迁移率Λeg 按文献[10]取值,而扩散系数D eg 则按爱因斯坦关系D eg =kT Λeg q 得到.521111期李成虎等: 梯度Si 1-c Ge c 混晶太阳电池长波光谱响应及其短路电流的理论计算与分析3 计算结果与分析由方程组(10)看出C1、C2和电池结构及材料性质有关.吸收系数Α(Κ)近似地用文献[9]提供的实验数据拟合而得到.取c=0、4◊、7◊、13.5◊、19.5◊、28◊,计算了相应c值的光谱响应.为方便计在图3只给出了c=0、4◊、28◊的光谱响应曲线.从图3可分析电池结构和梯度区最大锗浓度c对SR的影响.图3(a)表明c从4◊增至28◊时,和单晶Si相同长度基区比较,SR明显提高,并且响应波段也有所拓宽.而且可以看出在保持锗硅混晶电池基区长度不变时(2Λm),基区长度增加20Λm时,长波处SR得到更加显著的改善.但图3(a)也表明在L=L1+L2=2Λm+28Λm的条件下,当c提高时在短波处SR低于Si.这正是由于能隙的减小导致复合增强所至.图3(a)的梯度区长度为8Λm,在少子寿命范围内,基本都能扩散到图3 梯度Si Si1-c Ge c长波光谱响应 (a)表明梯度区Si1-c Ge c提高了光谱响应SR,c增加使响应波段拓宽;且梯度区长度L2增加导致SR增加,但是对大c值复合作用使得SR开始从短波处降低;(b)揭示了薄梯度区电场的影响,SR下降;在薄梯度区长基区的情况下,复合与电场抽运的综合效果使得SR低于相应Si之值.F IG.3 Op tical R esponse in L ong W ave R ange fo r Gradient Si Si1-c Ge c6211半 导 体 学 报21卷空间电荷区而形成光电流.但在梯度区长度增加时,窄能隙的强复合作用就表现出来了.图3(b )揭示了梯度区厚度对SR 的影响.图3(b )与图3(a )相对应,只是基区为8Λm 而梯度区是2Λm .梯度区长度的减小导致其内电场强度E 增强,E 的作用是向与光电流方向相反的方向抽运电子,其结果是在基区和梯度区界面处造成指向基区的梯度,从而也加速了基区光生少子向梯度区的扩散.光谱响应的降低是电子复合和电场抽运的综合结果.在基区不是很长的情况下SR 稍高于Si .在基区长度增加时所有不同c 值的SR 均低于Si .尽管响应波段有所拓宽,此区域内的量子效率只有2%左右.图3(b )的物理图象这样的:吸收系数随波长增加而减小,但吸收长度增加了.基区长度增加使得长波吸收加强.且在我们所讨论的波段内,波长愈长,量子效率愈低.在梯度区电场恒定的条件下,基区长度不是大时界面处电场抽运造成的光生少子梯度尚不明显.但基区长度增加时,界面处量子效率比前者低,而E 保持不变故使得电子向光生电流反方向的扩散得到加强,而使得所有c 值的光谱响应均低于Si 相应值.由(13)式计算得到全部波长贡献的总电流J L 与电池结构和c 的关系.图4给出在本文指定的c 条件下,J L 与L 的关系.实心正方是硅太阳电池基区长度与其光生电流的关系.在基区厚度大于50Λm 后,光生电流增长趋于缓慢.这是和通常认为50Λm 厚Si 层能吸收入射光子总数的90%是一致的.对锗硅梯度结构,在保持基区长度等于2Λm 时,其光生电流远大于同长度的Si 基区光生电流.c =28%时,光生电流在50Λm 左右接近饱和.由于x 值增加导致复合得增强,当L =100Λm 处,c =28%贡献的电流小于c =1915%、13.5%贡献之值.当L 保持相同长度而L 2=2Λm 时,梯度区内电场增强,尽管随着L 增加,J L 也增加,但均在Si 基区光生电流之下,且随着浓度的增加,J L 依次由上至下排列.其原因我们已在分析光谱响应时讨论过.图4 基区 梯度区光生电流J L 与电池结构关系 可以看出L =80Λm ,c =28%,J L 开始下降F IG .4 L igh t 2Induced Current V ersus Cell Structurec 值对光电流的影响可以从图5得出.我们首先注意到基区和梯度区长度相等时光生电流随c 增加到一极大值时开始下降.其原因在分析光电流谱时讨论过.增加梯度区的长度,可以使得J L 极大值向高浓度发展,但由于复合增强最终还是随c 增加而减小.至于梯度区很短的情形,J L 随c 值增加而减小正是图3(a )和图3(b )的反映.721111期李成虎等: 梯度Si 1-c Ge c 混晶太阳电池长波光谱响应及其短路电流的理论计算与分析图5 梯度区最大锗浓度c 对J L 的影响(注意此图也反映了结构对J L 的影响)F IG .5 Influence of M axi m um Ge Concentrati on c on L igh t 2Induced Current J L4 结论本文提出一种新型锗硅混晶太阳电池结构,并对长波长光电转换起主要作用的基区和梯度区光生少子连续方程进行了严格求解.计算表明,当梯度区长度L 2远大于基区长度L 1时,长波处光谱响应得到明显提高,可以改善以往锗硅混晶异质结太阳电池的界面复合,窄能隙复合对光生电流的限制:但当c 和L 2同时增加,由窄能隙导致的复合使光谱响应得以下降;计算结果还显示当L 2µL 1且c 和L 2在一定范围内,光生电流明显提高饱和长度亦随之缩短.参考文献[1] S .A .H ealy and M .A .Green ,So lar Energy M aterials and So lar Cells ,1992,28:273.[2] K .Said ,J .Poo rts m ans et a l .,2nd W o rld Conference and Exh ibiti on on Pho tovo taic So lar Energy Conversi on ,V i 2enna ,A ustria ,1998,36.[3] L inda M Ko sch ier ,Stuart R ,W enham andM artin A .Green ,2ndW o rld Conference and Exh ibiti on on Pho tovo taicSo lar Energy Conversi on ,V ienna ,A ustria ,1998,292.[4] P .O .H ansson ,M .Konum a et a l .,1st W o rld Conference and Exh ibiti on on Pho tovo taic So lar Energy Conversi on ,H aw aii,U.S .,1994,1254.[5] K .Said et a l .,14th European Pho tovo ltaic So lar Energy Conference Barcelona,Spain,1997,986.[6] H aro ld J .Hovel ,So lar Cells ,Sem iconducto rs and Sem i m etals ,V o l .45,A cadem ic P ress .[7] R .Peop le ,Physics and A pp licati on of Si Si 1-x Ge x Strainted 2L ayer H etero structures ,IEEE J .Q uantum E lectron 2ics ,1986,22(9):1696.[8] R .B raunstein ,A .R .M oo re and F .H er m an ,Phys .R ev .,1958,109:695.[9] E .Ch ritoffel et a l .14th E ruopean Pho tovo ltaic So lar Energy Conference Barcelona ,1997,2097.[10] O tfried M adelung ,Sem iconducto rs Basic D ata ,2nd R evised Editi on ,Sp ringer 2V erlag Berlin 1996.8211半 导 体 学 报21卷。

复旦大学张卫教授领衔科研团队成功研制世界首个半浮栅晶体管

复旦大学张卫教授领衔科研团队成功研制世界首个半浮栅晶体管

复旦大学张卫教授领衔科研团队成功研制世界首个半浮栅晶体管作者:暂无来源:《师资建设》 2013年第10期我国科学家实现微电子领域重大原始创新8月9日,由复旦大学电子学院张卫教授领衔团队研发的世界第一个半浮栅晶体管(SFGT)研究论文刊登于((科学》杂志,这是我国科学家首次在该权威杂志发表微电子器件领域的研究成果,标志着我国在全球尖端集成电路技术创新链中获得重大突破。

.该成果的研制将有助于我国掌握集成电路的核心技术,从而在国际芯片设计与制造领域内逐渐获得更多的话语权。

多年来,我国的集成电路制造工艺长期处于全球产业链末端。

尽管我国在自主知识产权集成电路技术上取得了长足进步,但集成电路的核心技术基本上由国外公司拥有,我国集成电路产业主要依靠引进和吸收国外成熟的技术,在微电子核心器件及集成工艺上缺乏核心技术。

作为微电子领域的重大原始创新,半浮栅晶体管(SFGT)的技术突破将有助于我国掌握集成电路的关键技术,在芯片设计与制造上获得更大的核心竞争力。

目前,DRAM、SRAM和图像传感器技术的核心专利基本上都是被美光、三星、Intel、索尼等国外公司控制;在这些领域,中国大陆具有自主知识产权且可应用的产品几乎没有。

半浮栅晶体管作为一种基础电子器件,它在存储和图像传感等领域的潜在应用市场规模达到三百亿美元以上。

它的成功研制有助于我国掌握集成电路的核心器件技术,是我国在新型微电子器件技术研发上的一个里程碑。

张卫教授表示,下一步将本着三方合作的理念,即“设计公司出产品、制造企业生产、复旦大学提供技术支持”,加强产学研紧密合作,努力推进该技术的产业化推广。

潜在应用市场规模超过300亿美元分析人士表示,这是我国科学家首次在该杂志上发表微电子器件领域的论文,标志着我国在全球尖端集成电路技术创新链中获得了重大突破。

金属一氧化物一半导体场效应晶体管(MOSFET)是目前集成电路中最基本的器件,而常用的U盘等闪存器件多采用另一种被称为浮栅晶体管的器件。

免费下载-中子辐照半绝缘SiC单晶的光学性质_王鹏飞

免费下载-中子辐照半绝缘SiC单晶的光学性质_王鹏飞

2 结果与讨论
荧光光谱分析 辐照产生的缺陷可在禁带中引入施主或者受主 能级,从而形成发光中心,在光激发情况下很容易 检测到相应的发射峰,这对研究辐照缺陷微观构型 具有重要意义。图 1 为辐照前后和辐照后退火的样 品的荧光光谱。对于未辐照样品,荧光光谱出现两 2.1
个发射峰,一个是 424 nm 的弱峰;另一个是位于 606 nm 的强峰。前者属于本征发光,由于 SiC 单晶 属于间接带隙半导体,带间发射需声子参与,造成 发光效率很低,发射峰强度很弱。后者应为来自于 非故意掺杂的 Ti 杂质发光,SiC 单晶中的 Ti 杂质 的含量虽然很小, 但它具有很强的发光峰[11]。 值得 注意的是中子辐照发光特点, 可以看出荧光光谱未 出现任何发射峰,根据文献[4,12–13],电子和离子 辐照的 SiC 单晶存在较多发射峰, 这表明中子辐照 SiC 单晶光发射具有一定的独特性。这种情况一直 持续到 400、800 和 1 000 ℃退火,当退火温度增 加到 1 200 ℃时,荧光光谱出现了 3 个明显的发射 峰,即 510,540 和 575 nm 峰,并且在 1 400 ℃退 火后, 峰强度显著增大, 而 1 600 ℃退火后, 575 nm 发光峰消失,同时杂质 Ti 的 606 nm 峰再次出现。 这 3 个绿光峰的微观机制比较复杂, 一般认为是许 多能级接近的高能级缺陷态向低能级态跃迁的发 光[14]。 2.2 紫外–可见–近红外透射谱分析 为探究中子辐照后和低于 1 000 ℃退火的 SiC 单晶无任何光发射信号的原因, 进行了 UV–Vis–Nir 透射谱的测试, 结果如图 2 所示。 从图 2 可以看出, 未辐照 SiC 单晶的截止波长 λTW (截止波长: 透过率 为零的区域和不为零的区域的分界线;TW: threshold wavelength)为 393 nm, 且在 393~2 500 nm 波段的透过率约为 70%,说明 SiC 单晶在这一波段 具有较好的光透过率。辐照后样品内部存在大量缺 陷,将增加对光的吸收,导致光透过率降低。最值 得关注的是 λTW 由原来的 393 nm 增大到 1 726 nm,

技术产业化

技术产业化

技术产业化作者:来源:《纺织服装周刊》2013年第30期央视新闻联播报道了复旦大学研制出世界上第一个半浮栅晶体管的消息,确实激动人心。

这种技术如果用于电脑CPU,可让缓存面积缩小到目前的十分之一,用于内存的话,可让速度提高一至二倍。

对此,全球顶级学术权威期刊《科学》杂志给出了这样的评价:要突破目前微电子元器件的性能瓶颈就必须有更先进的设计,复旦大学的这一研究做到了这一点。

它使器件能够更加高速、低功耗地工作,预计在超高速存储、光传感和图像拍摄等领域可以得到成功应用。

一项核心技术诞生让人振奋,但只有将技术实现产业化,将技术成果从实验室搬到生产线上,才算是拥有了真正的竞争力。

正如央视报道的那样,核心技术一旦发表,技术的窗户纸就被捅破。

接下来,谁最先实现技术转化,最先推出新产品,谁就掌握了市场的主动。

否则,极有可能起了个大早,赶了个晚集,挤破了自己脑袋,装满了人家钱袋。

技术产业化是每个行业都必须解决好的问题,一项技术实现产业化的难度也大不相同。

有些技术研发难度大,推广实施相对容易;而另一些,比如节能环保领域,产业化的难度就很大。

日前,国务院印发了《关于加快发展节能环保产业的意见》,到2015年,节能环保产业总产值将达到4.5万亿元,成为国民经济新的支柱产业。

与节能环保相关的技术产业化,不仅与技术装备本身有关,更与技术成果的推广方式和政策环境密切相关。

就拿节能环保中的循环利用来说,要想做到真正高效率的回收再生利用,先期的技术研发投入固然重要,但是,真正的难点在于技术的推广应用,在于如何将技术成果变成经济效益和社会效益。

早在2008年纺织印染业就被纳入国家循环经济试点的重点行业,唐山三友、青岛凤凰印染、宜宾丝丽雅、福建凤竹等纺织印染企业成为行业首批试点企业。

多年来,中国纺织工业联合会一直将节能减排、低碳环保作为行业可持续发展的一项重要内容,坚持不懈地推动行业节能环保技术的产业化进程。

纺织行业的节能环保技术涉及面很广,即有电、光、热能源的节约使用,也有水、汽、料的循环利用。

聚偏氟乙烯压电薄膜的制备及结构

聚偏氟乙烯压电薄膜的制备及结构

第19卷第3期2005年6月材料研究学报CHINESEJOURNAL0FMATERIALSRESEARCHV01.19No.3June2005聚偏氟乙烯压电薄膜的制备及结构傅万里杜丕一翁文剑韩高荣(浙江大学硅材料国家重点实验室)摘要采用溶液浸渍提拉法制备了p型PVDF压电聚合物薄膜,采用x衍射、扫描电镜等手段测定了PVDF薄膜的物相结构,结晶形态及薄膜厚度等.结果表明,基板的诱导作用使PVDF在较低温度下结晶形成极性的正交相p晶,在较高温度下结晶形成非极性的单斜相Q晶.在低温下得到的PVDF薄膜中p晶呈典型的球晶结构,晶粒粒径随着膜厚的增加而增大,晶体颗粒的堆积疏松,薄膜致密性差;在较高温度下形成的Q晶相,以二维生长为主,晶粒大小随着膜厚的增加基本上保持不变,Q晶颗粒间接触紧密,薄膜致密性好.关键词有机高分子材料,卢晶型PVDF薄膜,浸渍提拉法,结晶形态分类号0631文章编号1005-3093(2005)03—0243-06PreparationandstructureofPVDFpiezoelectric6lmFUWanliDUPiyr+WENGWbll.jianHANGaorong(&oteKeⅣL口6o,&2icD佗讹tenozs,Zhe靠on口Uh伽er3记弘且帆口加Du3JDD2∞木SupportedbyNationalNaturalScienceFbundationofChinaNo.50372057andNo.50332030,andDoctoralDe盯eePro昏amFbundationofMinistryofEducationNo.20020335017.ManuscriptreceivedMay31,2004;inrevi8edform0ctober8,2004.木事’I、owhomcorrespondenceshouldbeaddressed,Tbl:(0571)87952324,E广_mail:dupy@zju.edu.cnABSTRACTPiezoelectricandferroeleCtricpoIymerf.1mofPVDFwassuccessfuIlypreparedbydip—coatingmethod.Thephasestatus.crystaImDrphoJogyandthjcknessofthefjhllwerecharacterizedbvXRDandSEM.Thinf.ImwithpoIarorthorhombiccryStaI.pphase.isformedwhenisothermaIcryStaI|izationat60℃:whilenon—poIarmonocIiniccrystaI,nphase.isformedat150℃.pcrystaIinthefiImexhibitsast、,picaIsphericaIstructure,anditsdiameterincreaseswithincreasingthefiIm’sthickneSs.DuetoitsIoosestacking.IargeinterSpaceandIowcompactnessareformedinthef_Im,whichresuItsinathickerfiIm.Hawever.thesizeofthencryStaI,mainIyfbrmedbytvv口_dimensiongrowthathighertenlperature.shawsbasicaIIyunchangedwhenthethicknessoffifmincreases.AthinnerfiImwithsmaIIerinterspaceandbettercompaCtnessisf_nalIyformedduetotheclosecontactingofacrystaI.KEYWoRDSo唱anicpolymermaterials,伊PVDF.dip_C0atingmethod,crystaImorphoIogyp型聚偏氟乙烯(伊PVDF)是一种典型的压电功能聚合物,在水声探测、红外报警、压电传感和超声换能等电子、通讯、军事和医学领域中得到广泛的应用[1~引.近年来,电子信息工业的4国家自然科学基金50372057和50332030和教育部高等学校博士点基金20020335017资助项目2004年5月31日收到初稿;2004年10月8日收到修改稿.本文联系人:杜丕一,教授,杭州市310027,浙江大学材料系无机非金属材料研究所244材料研究学报19卷飞速发展对器件的小型化和薄膜化提出了迫切的要求,因而对压电铁电功能薄膜、超薄膜(纳米薄膜)材料的研究引起了人们的广泛关注.薄膜材料的晶相结构、析晶能力、相变行为以及各种理化性能与块体材料有明显的不同【6 ̄10j.目前,制备伊PvDF的方法主要有高压结晶法【11]、电场极化法【12】和单轴热拉伸法【13】.前两种方法对实验条件和设备的要求比较苛刻,而用后一种方法容易产生缺陷.溶液结晶法简单易行,非常适合于压电聚合物薄膜材料以及聚合物基多相功能复合薄膜材料的制备【14】.本文采用溶液浸渍提拉法制备卢一PVDF压电聚合物薄膜,研究其结晶行为、晶相组成以及伊PVDF压电薄膜的形成条件和机制.1实验方法实验用PVDF为白色粉末状树脂,数均分子量40万.溶剂N,N一二甲基甲酰胺(DMF)为分析纯.先量取一定体积的DMF溶剂,加入5%一20%(质量分数,下同)的PVDF树脂粉末,逐步加热搅拌使其完全溶解后冷却至室温,得到无色透明的溶液.然后以玻片为基板,利用浸渍提拉法在玻璃基板上镀膜.制备时溶液中PVDF树脂的质量分数分别为5%、lo%、15%和20%,提拉速度为1cm/min.再将制备出的薄膜在温度为60一150℃的烘箱中,恒温热处理12—48h,得到厚度均匀的薄膜.为了对比,用流延法,将PVDF溶液滴在表面皿上在35℃热处理96h,得到厚度较大的PVDF膜.用x衍射仪(Rigakud/MaX—C)分析PVDF薄膜的相结构,在扫描电镜(xL30EsEM,PHILIPs)下观察PVDF薄膜的微观形貌.2结果与讨论2.1热处理温度对PVDF薄膜的结晶形态的影响及其机理探讨PVDF是一种链状半晶态多晶形热塑性聚合物,在不同条件下结晶成o、p、7和6相四种晶相,其中只有强极性的p相具有压电铁电性能.Q晶型和卢晶型PVDF的x衍射曲线分别列于图1和图2.图1a~PVDF的xRD曲线Fig.1XRDcurveofQ—PVDF28,(。

一种U型结构的半浮栅器件及其制造方法[发明专利]

一种U型结构的半浮栅器件及其制造方法[发明专利]

专利名称:一种U型结构的半浮栅器件及其制造方法专利类型:发明专利
发明人:王鹏飞,林曦,孙清清,张卫
申请号:CN201310548612.X
申请日:20131106
公开号:CN103579126A
公开日:
20140212
专利内容由知识产权出版社提供
摘要:本发明属于半导体器件技术领域,具体涉及一种U型结构的半浮栅器件及其制造方法。

本发明在U形凹槽形成后,保留原先的硬掩膜层;先通过淀积第一层多晶硅并回刻来定义出器件浮栅开口区域的位置,然后淀积第二层多晶硅;在对多晶硅进行刻蚀后,剩余的第二层多晶硅和第一层多晶硅形成器件的浮栅,之后再去除掉硬掩膜层;同时,在源漏接触区形成之后把控制栅牺牲层去除,再淀积金属栅极,使得U型结构的半浮栅器件可以集成金属栅极和高介电常数材料栅介质。

本发明采用自对准工艺,过程简单且稳定,可控性强,降低生产成本,而且可以精确控制浮栅的宽度,降低器件尺寸。

申请人:复旦大学
地址:200433 上海市杨浦区邯郸路220号
国籍:CN
代理机构:上海正旦专利代理有限公司
更多信息请下载全文后查看。

  1. 1、下载文档前请自行甄别文档内容的完整性,平台不提供额外的编辑、内容补充、找答案等附加服务。
  2. 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
  3. 3、如文档侵犯您的权益,请联系客服反馈,我们会尽快为您处理(人工客服工作时间:9:00-18:30)。

/cgi/content/full/341/6146/640/DC1Supplementary Materials forA Semi-Floating Gate Transistor for Low-Voltage Ultrafast Memoryand Sensing OperationPeng-Fei Wang,* Xi Lin, Lei Liu, Qing-Qing Sun, Peng Zhou, Xiao-Yong Liu, Wei Liu, YiGong, David Wei Zhang*Corresponding author. E-mail: pfw@Published 9 August 2013, Science341, 640 (2013)DOI: 10.1126/science.1240961This PDF file includes:Materials and MethodsSupplementary TextFigs. S1 to S8ReferenceMaterials and MethodsThe Semi-Floating-Gate (SFG) transistors were fabricated using the 0.18-μm manufacturing technologies with specifically designed ion-implant processes and photolithography masks. The process flow and device structures are summarized in Fig. S1. First, a phosphorous-doped drain extension region was formed by ion-implantation and a 6-nm gate oxide layer was grown (Fig. S1A). Then a contact window inside the gate oxide was opened by lithography and buffered-HF wet etching. Boron was implanted into this contact window using the photo-resist as a masking layer (Fig. S1B). After that, the first boron-doped poly-Si layer was deposited and contacted the drain extension region via the aforementioned contact window. The Semi-FG pattern was then formed by lithography and reactive ion etching process which stops on the underlying thin SiO2 layer. This plasma-damaged SiO2 layer was then etched away and a 6 nm SiO2 dielectric was grown and nitrided as the inter-poly-dielectric. An n+-doped poly-Si was then deposited and the control-gate structure shown in Fig. S1C was formed. After the spacer processes, arsenic ions were implanted to form the heavily n-doped source and drain regions. The transmission electron microscope (TEM) cross-sections of the device are shown in Fig. S1D. In Fig. S1E, there is a SiO2 layer between the p+ doped poly-Si and the silicon substrate. However, at the position indicated in Fig. S1F, that SiO2 layer was etched away. With the p+ poly contacts the n-doped drain extension region, a floating pn-junction is formed and became the floating-gate of the device. The processed structures show very good process compatibility with the commercial dual-poly-gate manufacturing technology. Only minor modifications on the standard manufacturing process flow are required and high device yield on 8-inch silicon wafer is obtained. To further reduce the cell size, self-aligned techniques can be used to minimize the size of contact window between the p+ poly and the n-extension region.Supplementary TextThe DC characteristics of the SFG transistors were measured using a parameter analyzer (model Agilent B1500) and a RF probe-station. Figs. S2A to S2D compare the device symbols, the typical transfer I-V curves for MOSFET, FG-MOSFET, and two types of SFG transistors. For comparison, a MOSFET has a gate over the semiconductor channel (Fig. S2A) and a floating-gate MOSFET has a floating-gate between the control-gate and the semiconductor channel (Fig. S2B). The floating-gate MOSFET stores logic “0” with high V th or “1” with low V th by changing the amount of charge inside the floating-gate. By adding a diode or a gated diode between the floating-gate and the drain electrode of FG-MOSFET, a SFG transistor is realized. When the SFG transistor is exposed under light, the photo-carriers generated from the pn-junction diode will be partially collected by the floating-gate and in turn the I D-V CG curves will shift accordingly when changing the light intensity and exposure time. The change in threshold voltage of SFG transistor ( ) is expressed as , where is the change of the amount of charges inside the floating-gate (), is the floating-gate capacitance, and is the coupling ratio of control-gate (S1). During the light exposure process, is increased due to the photocurrent flowing into the floating-gate. The device symbol andthe measured transfer I D-V CG curves for SFG transistor for light sensing are shown in Fig. S2C. It can be seen that V th decreases when the light intensity increases. The device symbol and the measured I D-V CG curves of a SFG transistor optimized for memory function are shown in Fig. S2D. Compared to the SFG transistor shown in Fig. S2C, this SFG cell extends the control-gate over the pn-diode. A gate controlled diode or Tunneling FET (TFET) is formed and connects the semi-floating-gate to the drain. As a result, the TFET can now be used to charge or discharge the floating-gate. The current of TFET depends on the band-to-band tunneling generation rate, i.e. . G BTBT is the band-to-band tunneling generation rate, and it is expressed as. A BTBT and B BTBT are constants, E g is the band-gap energy, and E is the magnitude of the electric field. With a higher V D, the electric field increases and the charge current increases exponentially. As shown in Fig. S2D, the threshold voltage decreases when the drain voltage increases. Fast switching behavior is also observed in the transfer I D-V CG curves. The subthreshold swing even reaches 30 mV/dec when V D equals 3.0V.Because SFG transistor has an entirely new device structure, process and device simulations were carried out using Silvaco TCAD tools before device fabrication. First, the process simulations were performed. The resulted structure was then imported to the device simulator for transient behavior simulation. The ion-implantation conditions were designed for the optimized device performance from the device simulation. Asymmetric doping profiles for the source and drain of the embedded P-TFET are specifically designed to obtain the favorable device electrical behavior. Since the processing parameters were obtained from the process simulation results, the simulated structure is almost identical to the experimental device structure shown in Fig. S1D. The models for simulating a conventional MOSFET were included in the SFG transistor simulation. In addition, Kane’s local band-to-band tunneling model was included for simulating the tunneling current during the device operation. Fig. S3A - S3D display the simulated device structures with current and potential contours. The writing-1 operation is with V CG = -2.0 V and V D = 2.0 V. During the writing-1 operation, the drain is positively biased and the semi-floating-gate has a negative potential, resulting in current flows from the drain to the semi-floating-gate (Fig. S3A). The writing-0 operation is with V CG = 2.0 V and V D = 0 V. During the writing-0 operation, the n+ drain is biased with 0 V and the semi-floating-gate potential is elevated by the capacitive coupling of increased V CG, resulting in current flows from the semi-floating-gate to the drain (Fig. S3B). After writing-1 and writing-0 operations, the potential contours for the “1” and “0” devices at the standby status are plotted in Fig. S3C and S3D. It can be seen that the semi-floating-gate potential of the “1” state is 0.75 V, which is 1.64 V higher than that of the “0” state with the same bias conditions.The transient electrical behavior of the simulated device is shown in Fig. S3E. With the band-to-band tunneling model turned on, the high-speed memory function is realized (black line in the read-out current subfigure of Fig. S3E). It is found from the simulation that the gate-controlled band-to-band tunneling current is the key to high speed writingoperation, especially at the nano-second level. The green line in the read-out current subfigure of Fig. S3E is simulated with the band-to-band tunneling model turned off. It can be seen that writing-1 operation does not work with the same voltage setting. For comparison, the measured transient operation characteristic of the SFG transistor is shown in Fig. S3F. The transient behavior of the measured device is similar to the simulated device with band-to-band tunneling model.The transient behaviors of SFG transistors were measured using the circuit configuration shown in Fig. S4A. The measurement environment includes a RF probe-station, a pulse generator, a transimpedance amplifier circuit, and a high-speed oscilloscope. Two channel signal pulses were generated by the pulse generator (Agilent 81160A) and coupled to the drain and the control-gate electrodes of SFG transistor. A transimpedance amplifier circuit (Fig. S4B) was designed to convert the read-out current of SFG transistor to voltage signal for oscilloscope tracing. In Fig. S4C, the examples of traces captured by the oscilloscope during transient operation are shown. The writing-1 operation was performed by pulses of V D = 2V and V CG = -2V for 3 ns. Then, during the read operation, the voltage drop on the feedback resistor R f was measured by the oscilloscope and then converted to the I drain signal shown in Fig. S4D. It can be seen the readout current is about 2 μA after a 3-ns writing “1” operation. Unlike the conventional DRAM cell, the read operation of the SFG memory device is non-destructive because of its gain cell nature, which means no write-back operation is needed after the reading operation.In order to evaluate the operation speed of SFG transistor for memory application, short writing time with the target pulse width of 0.6 ns was executed (Fig. S5A). Due to the parasitic capacitance of the measurement cables, the rising / falling edges were 0.9 ns and the actual peak width at half height (PWHH) was increased to 1.3 ns. However, the writing-1 and writing-0 operations were successfully demonstrated in Fig. S5B. Meanwhile, it was found that the off-chip read-out circuit oscillates at the beginning of reading operation. Figs. S5B to S5D show the transient operation of the devices with different reading operation time. The reading operation duration was 9 μs in Fig. S5B. First, a writing-1 operation was performed with V D= 2V and V CG= -1.8V. A read-out current of about 1.5 μA was measured for the “1” cell. Then, a writing-0 operation was performed with V D= 0V and V CG= 2V. A very small current representing “0” was measured for the “0” cell. It can be seen that the read-out current oscillates at the beginning of the reading operation. With longer reading time such as 22.4 μs and 45 μs shown in Figs. S5C and S5D, the read-out current became stable. The oscillation was caused by the amplification circuit board (Fig. S4B) which can be improved by using the on-chip read-out circuits. Meanwhile, higher operation speed is expected by reducing the rising / falling edges of the pulses when using the integrated supporting circuits.In Figure S6A, 40 test devices were tested. The operation sequence was writing-0, reading, writing-1, and then reading. A “0” could be written into the transistor with V D of -1V and V CG of 2 V for 3 ms. During the subsequent reading operation, a very small drain current of 30 nA was measured with V D = 2 V and V CG = 2 V. After a writing-1 operation at V D =2 V and V CG = -2 V for 3 ms, the threshold voltage was lowered and a large draincur rent of about 2 μA was measured with V D = 2 V and V CG = 2 V. The wafer map of the reading-1 drain current of 40 test devices showed high uniformity across the whole wafer. The retention time and operation endurance properties were evaluated. In Fig. S6B, the retention time for “0” cell reaches 1 second at room temperature, while the “1” cell does not have retention issue because the semi-floating-gate voltage will be pinned to the drain voltage in the standby state. The retention performance of “0” is good because of the low leakage current of reversely biased p-n structure in the standby status. In Fig. S6C, the device operation endurance was investigated by repeating the full operation sequences. Almost no degradation was observed for 1012cycles of operation. The extrapolated endurance reaches 1015, which is far beyond the endurance of 106 for the conventional floating-gate memory cell.As a memory device, the immunity to various disturb mechanisms is also very important. Figs. S7A to S7F show the disturb properties of SFG memory cell measured at room temperature. Six types of disturb mechanisms on “1” and “0” cells are characterized. Most of the disturb endurances exceed 100 ms except for the disturbance of the writing-0 control-gate voltage on the “1” cell when V D at standby status is set to 1.0 V. During this writing-0 V CG disturb with V CG = 2.0 V and V D = 1.0 V, the pn junction will be weakly forward-biased if “1” is already stored inside the cell. The disturb endurances can be further improved by optimizing the operation voltages and the operation sequences. Using the image sensing SFG transistor, an imaging array can be configured. In Fig. S8, the schematic view of the SFG image sensor array is shown. The source lines and bit lines are arranged above the shallow trench isolation (STI) structure to maximize the fill factor. Using SFG cell as an APS cell, the pixel density of image sensing chip can be increased and the reading operation becomes non-destructive.Fig. S1.Brief summary of the process flow for fabricating the SFG transistor (A to C) and the TEM pictures of SFG transistor along the channel direction (D to F). Inset E is taken from the MOSFET channel where poly-Si on SiO2 can be seen. Inset F is taken from the contact interface between semi-floating-gate and n-doped drain extension region whereno oxide interface exists.Fig. S2Device symbols and transfer characteristics for MOSFET (A), FG-MOSFET (B), and SFG transistors (C, D). The semi-floating-gate is realized by connecting the floating-gate of FG-MOSFET to the drain via a pn junction diode. When the diode works as a photo-diode, photo-sensing function can be realized. When extending the control-gate over the diode, an embedded TFET is formed and dramatically accelerates the writing-1operation.Fig. S3(A, B) Simulated current contours of writing-1 operation and writing-0 operation. During the writing-1 operation, the drain is positively biased and the semi-floating-gate has a negative potential. Current will flow from the drain to the semi-floating-gate. During the writing-0 operation, the n+ drain is biased with 0 V and the semi-floating-gate potential is elevated by the capacitive coupling of increased V CG. Current will flow from the semi-floating-gate to the drain. (C, D) Simulated potential contours for the “1” and “0” devices at the standby state. The semi-floating-gate potential of the “1” state is higher than that of the “0” state with the same external voltages. (E) Simulated transient operation of the SFG transistor. V CG, V D, and the read out I D are displayed separately. In the read out I D figure the simulation results with or without the band-to-band tunneling model are compared. Device does not work when the band-to-band tunneling model is turned off in the simulation. (F) Experimental transient operations are shown for comparison with thesimulation results of Fig. S3E.Fig. S4(A) Transient measurement circuit. The measurement environment includes a RF probe-station, a pulse generator, a transimpedance amplifier circuit, and a high-speed oscilloscope. (B) Specific transimpedance amplifier circuit for converting the current signal to voltage signal for oscilloscope tracing. (C) Examples of measured voltage signals shown on the screen of oscilloscope. (D) Control-gate voltage, drain voltage, and the read-out current of a SFG transistor using 3-ns writing-1 pulse. The voltages are measured directly and the read-out current is converted from the voltage-drop on thefeedback resistor.Fig. S5(A) Measured drain voltage and control-gate voltage pulses for writing “1” and writing “0”, where the peak-width-at-half-height (PWHH) is measured as 1.3 ns. Transient measurement sequence of writing “1” - standby - reading - writing “0” - standby -reading with various reading pulse widths of 9 μs (B), 22.5 μs (C), and 45μs (D).Fig. S6(A) Measured reading-1 drain current wafer map and two transient operation cycles of 40 SFG transistors. The operation sequence is writing-0, reading, writing-1, and then reading. (B) Data retention time of “1” and “0” at room temperature. The read-out operation is performed with various standby periods after the writing operation. (C) Measured operation endurance property. Full writing-reading sequences are repeated with3 ns writing time.Fig. S7Disturb performances of SFG memory cell measured at room temperature. (A) Writing-1 V D disturb on the “1” and “0” cells. (B) Writing-1 V CG disturb on the “1” and “0” cells.(C) Writing-0 V D disturb on the “1” and “0” cells. (D) Writing-0 V CG disturb on the “1” and “0” cells. (E) Reading-operation V D disturb on the “1” and “0” cells. (F) Reading-operation V CG disturb on the “1” and “0” cells.Fig. S8Schematic view of the SFG image sensor array. The source lines and bit lines arearranged above the shallow trench isolation (STI) structure to maximize the fill factor.ReferencesS1. P. Pavan, L. Larcher, and A. Marmiroli, Floating gate devices: operation and compact modeling (Kluwer Acdemic Publisher, Dordrecht, 2004), pp. 37-40.。

相关文档
最新文档