Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009459
M. P. Patkar, T. Chin, J. Woodall, M. Lundstrom, M. Melloch
Ohmic contacts with low specific contact resistivity are essential for contemporary electronic devices such as HBTs, lasers, LEDs, and MESFETs. Alloyed Au-Ge-Ni contacts for nGaAs and alloyed Au-Zn contacts for p-GaAs have been widely used in GaAs semiconductor devices, but alloyed contacts produce a rough interface with the semiconductor and hence, are not good optical reflectors. Devices like light emitting diodes, vertical cavity surface-emitting lasers, and solar cells, need ohrmc contacts that are also good optical reflectors. Optimizing the reflectivity and the resistivity of the back contact to such devices poses a considerable challenge. We have developed a new technology for forming non-alloyed ohmic contacts to nand p-GaAs using Asrich layers. As-rich GaAs epi layers can be produced by molecular beam epitaxy (MBE) using lower than normal growth temperatures. Due to the amphoteric nature of Si, the typical dopant for n-GaAs, achieving high n-type doping concentrations is difficult. Simple reaction kinetics show that N+/ Np x n?/n2.
具有低比接触电阻率的欧姆触点对于当代电子器件(如hbt,激光器,led和mesfet)至关重要。nGaAs合金Au-Ge-Ni触点和p-GaAs合金Au-Zn触点已广泛应用于GaAs半导体器件中,但合金触点与半导体产生粗糙的界面,因此不是良好的光学反射器。像发光二极管、垂直腔面发射激光器和太阳能电池这样的设备需要欧姆触点,这些触点也是良好的光学反射器。优化这类器件的背触点的反射率和电阻率是一个相当大的挑战。我们开发了一种利用富砷层形成非合金欧姆接触的新技术。利用分子束外延(MBE)可以在低于正常生长温度的条件下制备富砷化镓外延层。由于n-GaAs的典型掺杂物Si的两性性质,实现高n型掺杂浓度是困难的。简单反应动力学表明N+/ Np x N ?/n2。
{"title":"Very low resistance non-alloyed ohmic contacts using as-rich GaAs","authors":"M. P. Patkar, T. Chin, J. Woodall, M. Lundstrom, M. Melloch","doi":"10.1109/DRC.1994.1009459","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009459","url":null,"abstract":"Ohmic contacts with low specific contact resistivity are essential for contemporary electronic devices such as HBTs, lasers, LEDs, and MESFETs. Alloyed Au-Ge-Ni contacts for nGaAs and alloyed Au-Zn contacts for p-GaAs have been widely used in GaAs semiconductor devices, but alloyed contacts produce a rough interface with the semiconductor and hence, are not good optical reflectors. Devices like light emitting diodes, vertical cavity surface-emitting lasers, and solar cells, need ohrmc contacts that are also good optical reflectors. Optimizing the reflectivity and the resistivity of the back contact to such devices poses a considerable challenge. We have developed a new technology for forming non-alloyed ohmic contacts to nand p-GaAs using Asrich layers. As-rich GaAs epi layers can be produced by molecular beam epitaxy (MBE) using lower than normal growth temperatures. Due to the amphoteric nature of Si, the typical dopant for n-GaAs, achieving high n-type doping concentrations is difficult. Simple reaction kinetics show that N+/ Np x n?/n2.","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122138545","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009431
J. Song, K. Chough, C. Palmstrøm, B. P. Van der Gaag, W. Hong
{"title":"Carbon-doped base InP/InGaAs HBTs with f/sub T/ = 200 GHz","authors":"J. Song, K. Chough, C. Palmstrøm, B. P. Van der Gaag, W. Hong","doi":"10.1109/DRC.1994.1009431","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009431","url":null,"abstract":"","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129025264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009436
T. Moise, Y. Kao
The resonant-tunneling diode (RTD) has been studied for many years because of its potential utility in high speed switching applications and compressed functionality circuits. 172 As a result of these investigations, the essential electronic characteristics of the RTD are now understood. However, the RTD's optoelectronic properties and its potential use in optical communication systems have not received much attention. With few exceptions,3 the optical investigations performed on the RTD have been directed towards understanding the charge build up characteristics within the quantum well (QW).4 In contrast, we have fabricated and characterized a series of InPand GaAs-based double barrier heterostructures (DB) that contain thick, undoped photon-absorption layers to maximize optical interactions. This new device, termed an optically-switched resonant-tunneling diode (ORTD), operates at room temperature and exhibits a 500 mV output voltage swing in response to a 0.2 mWinput optical signal at 880 nm.
{"title":"Optically-switched resonant tunneling diodes","authors":"T. Moise, Y. Kao","doi":"10.1109/DRC.1994.1009436","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009436","url":null,"abstract":"The resonant-tunneling diode (RTD) has been studied for many years because of its potential utility in high speed switching applications and compressed functionality circuits. 172 As a result of these investigations, the essential electronic characteristics of the RTD are now understood. However, the RTD's optoelectronic properties and its potential use in optical communication systems have not received much attention. With few exceptions,3 the optical investigations performed on the RTD have been directed towards understanding the charge build up characteristics within the quantum well (QW).4 In contrast, we have fabricated and characterized a series of InPand GaAs-based double barrier heterostructures (DB) that contain thick, undoped photon-absorption layers to maximize optical interactions. This new device, termed an optically-switched resonant-tunneling diode (ORTD), operates at room temperature and exhibits a 500 mV output voltage swing in response to a 0.2 mWinput optical signal at 880 nm.","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"212 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133463371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009416
S. Allen, U. Bhattacharya, M. Rodwell
W e have measured 0.68 p s electrical step functions, the fastest reported to date, generated by nonlinear transmission lines (NLTLs) integrated with sampling circuits which have a 3dB bandwidth of at least 725GHz. From the measured waveforms, the average velocity of the depletion edge of the varactor diodes on the NLTL is calculated to be 2.1. lo7 cm / sec. Because the other time constants in the circuit are much smaller than the measured 0.68 p s , this velocity saturation is believed to be the limiting phenomenon of the circuit performance. An NLTL is an electrical step function generator consisting of a high-impedance transmission line periodically loaded with varactor diodes [ I ] . The NLTL per-diode propagation delay q,Tllr,ov = [ L,,,,y (C,,,,, + Cdrode ( V))]1'2 is a function of the diodes' capacitance, and both decrease with increasing reverse bias voltage. The falling edge of the waveform becomes steeper during propagation since the delay for the waveform peak is greater than for the trough. A shock wave is formed whose transition time is limited by the diodes' cutoff frequency f , = (27~i?~C~/) ' and the Bragg frequency f,, = ( 7 ~ . Tlelrrv)-' that arises from the NLTL's periodicity. To make f,, high the diode separation must be decreased, and this spacing becomes limited by the diodes' physical size. As the line dimensions are reduced, the diodes' pad parasitics contribute a large fraction of C,,,,, and lower f,, . To address both these problems, coplanar waveguide (CPW) transmission lines were fabricated with the center conductor raised off' the substrate. Others have used a similar technique to reduce loss at the line input [2], but the impact is much greater at the high frequency output end. By elevating the center conductor 3 p m above the substrate, the wave velocity is doubled and, hence, for the same physical separation between diodes, f ' , is doubled. The process flow consists of forming ohmic contacts, ion implanting to provide isolation, depositing metal for the Schottky contacts and CPW ground planes, and then applying a layer of polyimide. The polyimide is subsequently etched in an 0 2 R.I.E. system until = 0 . 2 p m of the Schottky metal is exposed. The posts for the air bridge lines are formed on top of the polyimide and provide the contacts to the tops of the diodes. After electroplating the air lines, the polyimide is removed, leaving the contacts between the CPW and the diodes in air, substantially reducing the parasitic capacitance. By contacting the diodes this way, a small Schottky contact can be placed in the middle of a larger active region. The regions outside the active areas are H + implanted to render them semi-insulating, and the lateral straggle of the ions damages the active regions near the mask edge. By contacting the diodes from the top, l p m x l p m sampling diodes were fabricated that suffered no performance degradation from the Hf lateral straggle. The NLTLs are designed to have f,, = 1500GHz and
我们已经测量了0.68 p s的电阶函数,这是迄今为止报道的最快的电阶函数,它是由非线性传输线(nltl)与至少725GHz的3dB带宽的采样电路集成产生的。从测量的波形中,计算出变容二极管在NLTL上耗尽边缘的平均速度为2.1。由于电路中的其他时间常数远小于测量的0.68 p / s,因此这种速度饱和被认为是电路性能的限制现象。NLTL是一种电阶跃函数发生器,由周期性加载变容二极管的高阻抗传输线组成[1]。每个二极管的NLTL传播延迟q,Tllr,ov = [L,,,,y (C,,,,, + Cdrode (V))]1'2是二极管电容的函数,两者都随着反向偏置电压的增加而减小。波形的下降沿在传播过程中变得更陡峭,因为波形峰值的延迟大于波谷的延迟。形成激波,其过渡时间受二极管截止频率f (27~i?~C~/)的限制。’,布拉格频率f,, =(7 ~)。Tlelrrv)-'这是由NLTL的周期性引起的。为了使f高,必须减小二极管的间距,而这个间距受到二极管的物理尺寸的限制。随着线尺寸的减小,二极管的焊盘寄生占C,,,,,的很大一部分,占f,,的比例更低。为了解决这两个问题,共面波导(CPW)传输线的中心导体被制造出来。其他人使用类似的技术来减少线路输入[2]的损耗,但在高频输出端影响要大得多。通过将中心导体提高到基片以上3微米,波速翻倍,因此,对于二极管之间相同的物理距离,f '翻倍。工艺流程包括形成欧姆触点,离子注入以提供隔离,为肖特基触点和CPW接地面沉积金属,然后应用一层聚酰亚胺。聚酰亚胺随后在0.2 R.I.E.系统中蚀刻直到= 0。2 p m的肖特基金属暴露在外。用于空气桥线的柱子在聚酰亚胺的顶部形成,并提供到二极管顶部的接触。在空气线电镀后,去除聚酰亚胺,使CPW和二极管之间的接触处于空气中,大大降低了寄生电容。通过这种方式接触二极管,一个小的肖特基触点可以放置在一个较大的有源区域的中间。活性区域外的区域注入H +使其半绝缘,离子的横向分散破坏了掩膜边缘附近的活性区域。通过从顶部接触二极管,制备了1 p m × 1 p m的采样二极管,其性能不受Hf横向散射的影响。nltl设计为f, = 1500GHz,而具有N, = 1,1017 cm-'有源层的GaAs上的Ipin二极管具有f, = 4THz[3]。nltl提供单独的频闪和测试信号与采样电路集成,以提供片上测量。采样电路的带宽由频闪脉冲的孔径时间和采样二极管的RC时间常数决定,两者均< 0。采样波形具有3.7 V步进,衰减时间为0.68 p / s,衰减时间为10% / 90%。在以前的工作中,采样电路和nltl的时间常数是可比较的,使用简单的平方和反褶积来确定采样器带宽。使用该方法,这里给出了采样器带宽的保守估计为725 GHz。在NLTL的变容二极管中,耗尽边在0.68 p s内移动145nm,平均速度为2.1.107 cm / sec。由于电路中所有其他时间常数都远低于此,因此速度饱和(其他人已经分析过)似乎是限制现象。类似的电路将在N = 3的材料上制作。Cm -'和速度的提高是预期的,因为耗竭边缘必须移动的距离减少了1。
{"title":"725 GHz sampling circuits integrated with nonlinear transmission lines","authors":"S. Allen, U. Bhattacharya, M. Rodwell","doi":"10.1109/DRC.1994.1009416","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009416","url":null,"abstract":"W e have measured 0.68 p s electrical step functions, the fastest reported to date, generated by nonlinear transmission lines (NLTLs) integrated with sampling circuits which have a 3dB bandwidth of at least 725GHz. From the measured waveforms, the average velocity of the depletion edge of the varactor diodes on the NLTL is calculated to be 2.1. lo7 cm / sec. Because the other time constants in the circuit are much smaller than the measured 0.68 p s , this velocity saturation is believed to be the limiting phenomenon of the circuit performance. An NLTL is an electrical step function generator consisting of a high-impedance transmission line periodically loaded with varactor diodes [ I ] . The NLTL per-diode propagation delay q,Tllr,ov = [ L,,,,y (C,,,,, + Cdrode ( V))]1'2 is a function of the diodes' capacitance, and both decrease with increasing reverse bias voltage. The falling edge of the waveform becomes steeper during propagation since the delay for the waveform peak is greater than for the trough. A shock wave is formed whose transition time is limited by the diodes' cutoff frequency f , = (27~i?~C~/) ' and the Bragg frequency f,, = ( 7 ~ . Tlelrrv)-' that arises from the NLTL's periodicity. To make f,, high the diode separation must be decreased, and this spacing becomes limited by the diodes' physical size. As the line dimensions are reduced, the diodes' pad parasitics contribute a large fraction of C,,,,, and lower f,, . To address both these problems, coplanar waveguide (CPW) transmission lines were fabricated with the center conductor raised off' the substrate. Others have used a similar technique to reduce loss at the line input [2], but the impact is much greater at the high frequency output end. By elevating the center conductor 3 p m above the substrate, the wave velocity is doubled and, hence, for the same physical separation between diodes, f ' , is doubled. The process flow consists of forming ohmic contacts, ion implanting to provide isolation, depositing metal for the Schottky contacts and CPW ground planes, and then applying a layer of polyimide. The polyimide is subsequently etched in an 0 2 R.I.E. system until = 0 . 2 p m of the Schottky metal is exposed. The posts for the air bridge lines are formed on top of the polyimide and provide the contacts to the tops of the diodes. After electroplating the air lines, the polyimide is removed, leaving the contacts between the CPW and the diodes in air, substantially reducing the parasitic capacitance. By contacting the diodes this way, a small Schottky contact can be placed in the middle of a larger active region. The regions outside the active areas are H + implanted to render them semi-insulating, and the lateral straggle of the ions damages the active regions near the mask edge. By contacting the diodes from the top, l p m x l p m sampling diodes were fabricated that suffered no performance degradation from the Hf lateral straggle. The NLTLs are designed to have f,, = 1500GHz and","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133734143","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009441
S. R. in 't Hout, S. Middelhoek
A novel silicon photoconductor is presented which is sensitive to light only for one current direction. When the current is reversed its sensitivity to light is drastically reduced. This unique unipolar behavior allows for autocalibration, that is, the dark resistance of the photoconductor can be measured on line simply by reversing the bias current. Regular photoconductors do not show such a unipolar sensitivity and therefore require light chopping for the dark resistance to be measured. This new type of photoconductor can thus be used for applications similar to those for which regular photoconductors are used, but also for applications which require autocalibration. For example, the unipolar principle can be used in photoconductive HgCdTe infrared detectors which require measurement of the dark resistance because the material conducts relatively well even in the absence of infrared radiation. The unipolar photoconductor is a bulk type device consisting of a highly doped p-type silicon point contact on a low-doped p-type silicon substrate. A large-area ohmic contact is made at the bottom of the substrate and serves as a second contact to the device. The principle of operation is based on the effect of minority carrier accumulation2" at high-low junctions (p'p or n ' n junctions). High-low junctions present a small potential step which is a significant barrier for minority carriers but does not restrict the flow of majority carriers. Minority carriers flowing towards the highly doped region meet the potential barrier and thus accumulate near the junction lowering the local resistivity of the material. This effect is exploited in the photoconductor as follows. When the bias current is directed such that the holes (majority carriers) flow away from the point contact, which forms a high-low junction with the substrate, and the electrons (minority carriers) flow towards it, photogenerated excess electrons near the point contact will flow towards the point contact where they accumulate and lower the resistance of the device. When the current is reversed the photogenerated electrons flow away from the point contact into the bulk where they have little influence on the device resistance. The device was realized having a 6x6 pm square point contact with a doping level of approximately 1019 cm-3 on a 525 pm thick substrate with a doping level of about loi5 ~ m ~ , The dark resistance was about 10 kQ for both current directions. Sensitivity measurements were performed in a dark room using a 550 nm wavelength monochromatic light source varying the illumination power between 0.01 and 30 W/m2. Since the accumulation effect is strongly current dependent, the sensitivity was measured with currents varying from l FA up to 2 mA for both current directions. Optimal performance was observed for a bias current of 200 pA at which the resistance at 10 W/m2 illumination power decreased by about 75% for the sensitive current direction and only about 5% for the insensit
{"title":"Silicon unipolar photoconductor","authors":"S. R. in 't Hout, S. Middelhoek","doi":"10.1109/DRC.1994.1009441","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009441","url":null,"abstract":"A novel silicon photoconductor is presented which is sensitive to light only for one current direction. When the current is reversed its sensitivity to light is drastically reduced. This unique unipolar behavior allows for autocalibration, that is, the dark resistance of the photoconductor can be measured on line simply by reversing the bias current. Regular photoconductors do not show such a unipolar sensitivity and therefore require light chopping for the dark resistance to be measured. This new type of photoconductor can thus be used for applications similar to those for which regular photoconductors are used, but also for applications which require autocalibration. For example, the unipolar principle can be used in photoconductive HgCdTe infrared detectors which require measurement of the dark resistance because the material conducts relatively well even in the absence of infrared radiation. The unipolar photoconductor is a bulk type device consisting of a highly doped p-type silicon point contact on a low-doped p-type silicon substrate. A large-area ohmic contact is made at the bottom of the substrate and serves as a second contact to the device. The principle of operation is based on the effect of minority carrier accumulation2\" at high-low junctions (p'p or n ' n junctions). High-low junctions present a small potential step which is a significant barrier for minority carriers but does not restrict the flow of majority carriers. Minority carriers flowing towards the highly doped region meet the potential barrier and thus accumulate near the junction lowering the local resistivity of the material. This effect is exploited in the photoconductor as follows. When the bias current is directed such that the holes (majority carriers) flow away from the point contact, which forms a high-low junction with the substrate, and the electrons (minority carriers) flow towards it, photogenerated excess electrons near the point contact will flow towards the point contact where they accumulate and lower the resistance of the device. When the current is reversed the photogenerated electrons flow away from the point contact into the bulk where they have little influence on the device resistance. The device was realized having a 6x6 pm square point contact with a doping level of approximately 1019 cm-3 on a 525 pm thick substrate with a doping level of about loi5 ~ m ~ , The dark resistance was about 10 kQ for both current directions. Sensitivity measurements were performed in a dark room using a 550 nm wavelength monochromatic light source varying the illumination power between 0.01 and 30 W/m2. Since the accumulation effect is strongly current dependent, the sensitivity was measured with currents varying from l FA up to 2 mA for both current directions. Optimal performance was observed for a bias current of 200 pA at which the resistance at 10 W/m2 illumination power decreased by about 75% for the sensitive current direction and only about 5% for the insensit","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124833340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009392
I. Akasaki, H. Amano
Column III nitrides are one of the most promising materials for the applications to short wavelength light emitters as well as high temperature electronics. The wurtzite polytypes of AlN, GaN and InN form a continuous alloy system whose direct bandgaps range from 1.9 eV for InN, to 3.4 eV for GaN and to 6.2 eV for A1N at room temperature. These nitrides are physically harder and more stable than wide-bandgap II -VI compounds. Furthermore, their unique physical properties will make them suitable for piezo-electric, acousto-optic and opto-electric devices.
{"title":"Crystal growth and properties of column III nitrides for short wavelength light emitters","authors":"I. Akasaki, H. Amano","doi":"10.1109/DRC.1994.1009392","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009392","url":null,"abstract":"Column III nitrides are one of the most promising materials for the applications to short wavelength light emitters as well as high temperature electronics. The wurtzite polytypes of AlN, GaN and InN form a continuous alloy system whose direct bandgaps range from 1.9 eV for InN, to 3.4 eV for GaN and to 6.2 eV for A1N at room temperature. These nitrides are physically harder and more stable than wide-bandgap II -VI compounds. Furthermore, their unique physical properties will make them suitable for piezo-electric, acousto-optic and opto-electric devices.","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132546636","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009410
T. King, M. Hack
The variation of drain dopant profiles in two dimensions is shown to be superior to tradi- tional drain-engineering techniques for achieving low leakage in poly-Si thin-film transistors. Polycrystalline silicon (poly-Si) thin-film transistors (TFTs) are used in applications such as active- matrix liquid-crystal displays and image sensors, which have stringent TFT-leakage requirements. Experimental studies have shown that one of the main factors which affects leakage current is the electric-field strength in the channel region near the drain'. Lightly doped drain (LDD) structures have previously been used to reduce the electric field at the drain junction and thereby reduce TFT leakage current2. In these structures, the LDD region is placed between the channel and the heavily doped drain and spans the entire thickness of the channel, so that the drain dopant concentration varies only in one dimension (parallel to the gate). LDD structures are difficult to implement in poly-Si due to the poor controllability of dopant activation at concentrations below -lo1* ~m-~: underdoping of the LDD region results in decreased drive current in offset-drain structures, while over-doping results in a loss of the benefit of reduced leakage current. In this work, the two-dimensional (2D) engineering of drain doping profiles is demonstrated to provide significant advantages over conventional (1D) LDD engineering approaches for reducing TFT leakage current. Top-gate n-channel poly-Si TFTs were fabricated on quartz wafers using a conventional high- temperature (950°C) process3. After the formation of the heavily doped (nt) source and drain regions, some wafers were given additional implants at high tilt angles, to form lightly doped (n-) 1D or 2D LDD regions underneath the gate, extending -0.3 pm in from the gate edge. The 1D LDD regions were formed by deep implantation of phosphorus to a dose of 2 x 1013 The 2D LDD regions were formed using a combination of high-angle implants: a deep phosphorus implant (identical to the one used to form the 1D LDD structures), and a shallow counterdoping boron implant to confine the n- LDD region to the bottom region of the channel layer. A 1 hour anneal at 55OOC was used to activate the LDD implants. TFT measurements show that the 1D LDD structure provides a noticeable reduction in leakage com- pared with the standard drain structure, as expected: the median value is reduced from 0.94 pA/pm to 0.61 pA/pm, and the 20%-80% distribution spread is decreased from -16x to xllx. However, substantial further reduction is provided by the 2D LDD structure: the median value is reduced to 0.34 pA/pm and to 0.17 pA/pm with counterdoping boron implant doses of 4x 1OI2 cm-2 and 1 x 1013 cm-2, respectively. (The 20%-80% distribution spreads for the-2D LDD structures are -11 x.) Leakage was confirmed to scale linearly with TFT channel width for the devices fabricated in this work, in order to rule out edge contributions. Numerical device simulation
{"title":"Two-dimensional drain engineering for leakage reduction in thin-film transistors","authors":"T. King, M. Hack","doi":"10.1109/DRC.1994.1009410","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009410","url":null,"abstract":"The variation of drain dopant profiles in two dimensions is shown to be superior to tradi- tional drain-engineering techniques for achieving low leakage in poly-Si thin-film transistors. Polycrystalline silicon (poly-Si) thin-film transistors (TFTs) are used in applications such as active- matrix liquid-crystal displays and image sensors, which have stringent TFT-leakage requirements. Experimental studies have shown that one of the main factors which affects leakage current is the electric-field strength in the channel region near the drain'. Lightly doped drain (LDD) structures have previously been used to reduce the electric field at the drain junction and thereby reduce TFT leakage current2. In these structures, the LDD region is placed between the channel and the heavily doped drain and spans the entire thickness of the channel, so that the drain dopant concentration varies only in one dimension (parallel to the gate). LDD structures are difficult to implement in poly-Si due to the poor controllability of dopant activation at concentrations below -lo1* ~m-~: underdoping of the LDD region results in decreased drive current in offset-drain structures, while over-doping results in a loss of the benefit of reduced leakage current. In this work, the two-dimensional (2D) engineering of drain doping profiles is demonstrated to provide significant advantages over conventional (1D) LDD engineering approaches for reducing TFT leakage current. Top-gate n-channel poly-Si TFTs were fabricated on quartz wafers using a conventional high- temperature (950°C) process3. After the formation of the heavily doped (nt) source and drain regions, some wafers were given additional implants at high tilt angles, to form lightly doped (n-) 1D or 2D LDD regions underneath the gate, extending -0.3 pm in from the gate edge. The 1D LDD regions were formed by deep implantation of phosphorus to a dose of 2 x 1013 The 2D LDD regions were formed using a combination of high-angle implants: a deep phosphorus implant (identical to the one used to form the 1D LDD structures), and a shallow counterdoping boron implant to confine the n- LDD region to the bottom region of the channel layer. A 1 hour anneal at 55OOC was used to activate the LDD implants. TFT measurements show that the 1D LDD structure provides a noticeable reduction in leakage com- pared with the standard drain structure, as expected: the median value is reduced from 0.94 pA/pm to 0.61 pA/pm, and the 20%-80% distribution spread is decreased from -16x to xllx. However, substantial further reduction is provided by the 2D LDD structure: the median value is reduced to 0.34 pA/pm and to 0.17 pA/pm with counterdoping boron implant doses of 4x 1OI2 cm-2 and 1 x 1013 cm-2, respectively. (The 20%-80% distribution spreads for the-2D LDD structures are -11 x.) Leakage was confirmed to scale linearly with TFT channel width for the devices fabricated in this work, in order to rule out edge contributions. Numerical device simulation","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115752949","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009415
P. Bunyk, V. Semenov, A. Oliva, M. Bhushan, K. Likharev, J. Lukens, M. Ketchen, A. Kleinsasser, W. H. Mallison
Josephson junction integrated circuits using Rapid Single-nux-Quantum (RSFQ) logic are capable of performing logic operations at ultrahigh clock frequencies in excess of 100 GHz [ 11. In order to reach such a speed, Josephson junctions with high critical current density (jc=10kAkm2) and small area (S=lpm2) must be used. In this report, we will describe the first simple RSFQ circuits fabricated using two versions (A and B) of the PARTS technology [2]. The most important new feature of this technology is circuit planarization which allows superconducting contact between the counter electrode of the Nb trilayer and the wiring layer without opening a via inside the Josephson junction area S . Thus the area can be kept close to the minimum allowed by the particular patterning technology.
{"title":"Ultra-high-speed rapid single-flux-quantum digital circuits. using a planarized-niobium-trilaver josephson junction technology","authors":"P. Bunyk, V. Semenov, A. Oliva, M. Bhushan, K. Likharev, J. Lukens, M. Ketchen, A. Kleinsasser, W. H. Mallison","doi":"10.1109/DRC.1994.1009415","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009415","url":null,"abstract":"Josephson junction integrated circuits using Rapid Single-nux-Quantum (RSFQ) logic are capable of performing logic operations at ultrahigh clock frequencies in excess of 100 GHz [ 11. In order to reach such a speed, Josephson junctions with high critical current density (jc=10kAkm2) and small area (S=lpm2) must be used. In this report, we will describe the first simple RSFQ circuits fabricated using two versions (A and B) of the PARTS technology [2]. The most important new feature of this technology is circuit planarization which allows superconducting contact between the counter electrode of the Nb trilayer and the wiring layer without opening a via inside the Josephson junction area S . Thus the area can be kept close to the minimum allowed by the particular patterning technology.","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114191039","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009396
K. Domen, H. Ishikawa, M. Sugawara, M. Kondo, A. Furuya, T. Tanahashi
Using a 6 x 6 Luttinger-Kohn Hamiltonian1 considering the SO band effects, we calculated valence bands in (Alo~7Ga0~3~0~51n0~5P/GaxInl-xP QWs with Ga compositions (x) of 0.46 to 0.57, corresponding to strains ranging from 0.40% compressive to 0.43% tensile. We varied the well width until the bandgap energy equaled 1.96 eV, or 633 nm. For comparison, we also calculated the valence bands using a 4 x 4 Hamiltonian without SO band effects. Under tensile strain, the topmost band is light-hole (LH) like, but is coup!ed with the SO band. We found that this coupling widens the k-vector dispersion of the topmost band, and increeases the in-plane effective mass. Under compressive strain, the effective mass is smaller, due to less mixing of the topmost heavy hole band with both the LH band and the SO band. To obtain the relationship between modal gain Gln and current density J, we calculated the optical gain and. radiative lifetime from the valence band described above, assuming a parabolic conduction band. We found that structures under tensile strain have a larger transparent current density Ju and a larger differential gain than those under compressive strain. Those properties observed in structures under tensile strain result from the small spinorbit splitting in GaInP, because a large effective mass causes a large Jt, and a large differential gain. Therefore, the Gm curve for compressive strain crosses that for tensile. Comparing 0.40% compressive with 0.43% tensile strain, Gm for compressive is larger than that for tensile below Gm of 25 cm-l, while the opposite is true above 25 cm-l. As a result, we found that, for lasers with low threshold gain such as those with high-reflectivity optical coatings, compressive strain is effective in reducing Jth. In contrast, tensile strain is more effective in lasers with high threshold gain such as those with a saturatable absorber. These SO band effects have not been seen in GaInAsP long-wavelength strained QW lasers with a large spin-orbit splitting2.
{"title":"Calculation of threshold current density in 633-nm AlGaInP/GaInP strained quantum well lasers","authors":"K. Domen, H. Ishikawa, M. Sugawara, M. Kondo, A. Furuya, T. Tanahashi","doi":"10.1109/DRC.1994.1009396","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009396","url":null,"abstract":"Using a 6 x 6 Luttinger-Kohn Hamiltonian1 considering the SO band effects, we calculated valence bands in (Alo~7Ga0~3~0~51n0~5P/GaxInl-xP QWs with Ga compositions (x) of 0.46 to 0.57, corresponding to strains ranging from 0.40% compressive to 0.43% tensile. We varied the well width until the bandgap energy equaled 1.96 eV, or 633 nm. For comparison, we also calculated the valence bands using a 4 x 4 Hamiltonian without SO band effects. Under tensile strain, the topmost band is light-hole (LH) like, but is coup!ed with the SO band. We found that this coupling widens the k-vector dispersion of the topmost band, and increeases the in-plane effective mass. Under compressive strain, the effective mass is smaller, due to less mixing of the topmost heavy hole band with both the LH band and the SO band. To obtain the relationship between modal gain Gln and current density J, we calculated the optical gain and. radiative lifetime from the valence band described above, assuming a parabolic conduction band. We found that structures under tensile strain have a larger transparent current density Ju and a larger differential gain than those under compressive strain. Those properties observed in structures under tensile strain result from the small spinorbit splitting in GaInP, because a large effective mass causes a large Jt, and a large differential gain. Therefore, the Gm curve for compressive strain crosses that for tensile. Comparing 0.40% compressive with 0.43% tensile strain, Gm for compressive is larger than that for tensile below Gm of 25 cm-l, while the opposite is true above 25 cm-l. As a result, we found that, for lasers with low threshold gain such as those with high-reflectivity optical coatings, compressive strain is effective in reducing Jth. In contrast, tensile strain is more effective in lasers with high threshold gain such as those with a saturatable absorber. These SO band effects have not been seen in GaInAsP long-wavelength strained QW lasers with a large spin-orbit splitting2.","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129396159","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1994-06-20DOI: 10.1109/DRC.1994.1009451
M.A. Khan, J. N. Kuznia, D. Olson, W. Schaff, J. Burm, M. Shur
{"title":"Deep submicron AlGaN/GaN heterostructure field effect transistors for nficrowave and high temperature applications","authors":"M.A. Khan, J. N. Kuznia, D. Olson, W. Schaff, J. Burm, M. Shur","doi":"10.1109/DRC.1994.1009451","DOIUrl":"https://doi.org/10.1109/DRC.1994.1009451","url":null,"abstract":"","PeriodicalId":244069,"journal":{"name":"52nd Annual Device Research Conference","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1994-06-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130378981","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}