半导体硅片清洗工艺发展方向

对半导体硅片传统的RCA清洗办法中各种清洗液的清洗原理、清洗特点、清洗局限以及清洗对硅片表面微观状态的影响进行了详细的论述,同时在此基础上,对新的清洗办法(改进的RCA清洗办法)进行了一定的说明,指出了硅片清洗工艺的发展方向。     

MMST硅片清洗

德克萨斯仪器公司（ＴＩ）、预研规划局（ＡＲＰＡ）和空军三家联合的微电子制造科学与技术（ＭＭＳＴ）工程中，片子清洗工艺的目标是开发平片的全干法或气相清洗工艺，旨在全部地取代湿法清洗工艺。ＨＦ气相清洗和等离子工艺已成功地用于氧化物、氮化物和金属化后腐蚀的残余物的去除。但是还未找到用于替代炉前清洗的工业标准（ＲＣＡ）清洗和水冲洗的有效干法工艺。全干法清洗很可能要再花五到十年才能走向成熟。在未来的几年内，     

干冰微粒喷射清洗技术

简要介绍了随着工艺节点的缩小,传统RCA清洗方法在硅片清洗工艺中的局限性和弊端,进而提出了以CO2为介质的新型干冰微粒喷射清洗方法。从CO2的物理特性出发,论述了CO2流经喷枪后形成干冰微粒的机理,并简要分析了干冰微粒喷射技术对颗粒污染物和有机污染物的清洗机理。在此基础上,介绍了自主研发的一台基于干冰微粒喷射技术的半导体清洗设备,对该设备的结构和各部分的作用作了简要介绍,论述了使用该设备对硅片进行清洗的工艺流程。通过对比实验发现,采用压强为8 MPa、纯度为5N的CO2作为气源,喷嘴前压强设置为11 MPa,使用该设备可以达到很好的清洗效果。     

硅片清洗研究进展

综述了清洗液的组成、特点、清洗机理、对硅片表面质量的影响以及清洗技术和理论的发展;着重指出了,改进的RCAI对颗粒度、微粗糙度和金属沾污作用的机理,讨论了它与清洗顺序的关系,极度稀释的RCA2能使金属沾污降至10∧10原子/cm∧2以下,且不易使颗粒重新沉淀;最后介绍了清洗工艺的最新进展。     

硅片清洗设备的日常维护

对硅片清洗设备的日常维护从清洗工艺的角度进行了分析,并对具体问题的处理方案进行了阐述,从而使清洗设备工作更加稳定。     

硅片清洗原理与方法综述

对硅片清洗的基本理论、常用工艺方法和技术进行了详细的论述 ,同时对一些常用的清洗方案进行了浅析 ,并对硅片清洗的重要性和发展前景作了简单论述。     

光伏太阳能硅片清洗工艺的探索

对两种不同清洗方法的工作原理、清洗效果和适用范围等特点进行了分析。不同的工艺应采用不同的清洗方法才能获得最佳效果。介绍了硅片清洗机的清洗工艺和腐蚀工艺。指出了硅片清洗工艺的发展趋势。     

硅片传输单元洁净系统的研制

在集成电路制造的硅片传输过程中,硅片传输机器人等机械设备的运动会产生尘埃粒子污染硅片.为解决这一问题,设计并研制了硅片传输单元的洁净系统,分析了该洁净系统的设计方案和结构设计,计算了它的送回风管直径,并用计算流体力学的方法,对洁净系统内部的硅片传输机器人区域进行了流场的模拟分析.该洁净系统已达到设计要求.     

300 mm圆片后抛光清洗--满足纳米微粒去除的挑战

随着半导体制造关键尺寸的继续缩小,硅片表面清洗要求变得更加严格。当前这一要求包括有效地去除硅片表面的纳米微粒(<100nm),并控制主要金属杂质不超过1E+10原子/cm~2。传统的擦片机和兆声湿式批量清洗工艺面临达到这些目标的挑战。单片清洗澡机在硅片表面产生更加均匀的声强分布,更有效地去除了纳米微粒。介绍了湿式批量洗洗机和单片清洗澡机的兆声清洗效果。湿式批量浸泡术和兆声能量单片清洗机结合可以有效地去80mm磨料微粒。     

Ion beam-assisted planarization of chemically vapor deposited diamond thin films using electron cyclotron resonance plasma

The high inherent surface roughness of as-deposited polycrystalline diamond films has made effective planarization processing of these films essential for most industrial applications. We have investigated the efficacy of ion beam sources for planarization in an electron cyclotron resonance plasma system using both direct substrate biasing and an accelerating grid system. Rough polycrystalline diamond films were synthesized using hot filament chemical vapor deposition. Both the etching rates and the resultant surface roughnesses were found to decrease as the angle of incidence (relative to the substrate surface normal) of the ion beam was increased. In the case of direct biasing of the sample, acicular features were observed following processing at higher incident angles. The use of double ion-extraction grids in conjunction with concomitant sample rotation was found to produce more uniform planarization of the diamond films. The rate of surface roughness reduction was found to be nonlinear and decreased with time. For both ion extraction methods investigated, the average film roughness (R_a) was significantly reduced from 0.2 to 0.05–0.06 μm.     

Electron cyclotron resonance plasma preparation of CdZnTe (211)B surfaces for HgCdTe molecular beam epitaxy

CdZnTe wafers were inserted into a multi-chamber processing facility without prior preparation, cleaned by exposure to an electron cyclotron resonance Ar/H₂ plasma, and used as substrates for molecular beam epitaxy of HgCdTe. Changes induced in the wafer near-surface region during the cleaning step were monitored using in situ spectroscopic ellipsometry. Ellipsometric data were subsequently modeled to provide the time evolution of the thickness of a native overlayer. Auger electron spectra were consistent with surfaces free of residual contamination and which had the stoichiometry of the underlying bulk. Surface roughness values of 0.4 nm were obtained ex situ using interferometric microscopy. Electron diffraction patterns of plasma prepared wafers heated to 185°C (the temperature required for HgCdTe molecular beam epitaxy) were streaked. Structural and electrical characteristics of epilayers grown on these substrates were found to be comparable to those deposited on wafers prepared using a conventional wet chemical process. This demonstrates an important step in an all-vacuum approach to HgCdTe detector fabrication.     

用ECRC-VD方法制备非晶氟化碳薄膜

以C4F8为源气体,Ar为稀释气体,用电子回旋共振化学汽相淀积的方法制作了非晶氟化碳薄膜;使用XPS和FTIR分析薄膜的化学组分和成键类型;研究了微波功率对于沉积速率和薄膜光学性质的影响。沉积速率随Ar在混合气体中比例的增大先增大后降低,随微波功率的增加而增加,并最终趋于饱和值。沉积的薄膜介电常数约为2.0,在可见光区薄膜具有良好的透光性。在较高的微波功率条件下,沉积薄膜的光学带隙减小。     

Comparison of ECR plasma chemistries for etching of InGaP and AlGaP

Etch rates for InGaP and AlGaP are examined under electron cyclotron resonance (ECR) conditions in Cl₂/Ar, BCl₃/Ar, BCl₃/N₂, ICl/Ar, and IBr/Ar discharges. All the plasmas except IBr/Ar provide rapid etching of InGaP at rates above 1 μm min⁻¹. ICl/Ar provides the highest etch rates. Unlike the Cl₂/Ar and BCl₃-based chemistries, the rates in ICl/Ar and IBr/Ar are almost independent of microwave power in the range 400–1000 W. Much lower rates were obtained for AlGaP in every discharge due to the greater difficulties in bond breaking that must precede formation and desorption of the etch products.     

Effects of hydrogen on majority carrier transport and minority carrier lifetimes in long-wavelength infrared HgCdTe on Si

We present extended results on the use of a hydrogen plasma to passivate the effects of defects in long-wave ir HgCdTe/Si. Annealed and as-grown epilayers, in situ doped with indium, were exposed to a hydrogen plasma generated in an electron cyclotron resonance (ECR) reactor. Secondary ion mass spectrometry was used to measure the extent of hydrogen incorporation into the epilayers. Hall and photoconductive lifetime measurements were used to assess the efficacy of passivation. The passivation of defects responsible for the scattering and recombination of electrical carriers was observed for most ECR conditions over a range of dislocation densities.     

Use of electron cyclotron resonance plasmas to prepare CdZnTe (211)B substrates for HgCdTe molecular beam epitaxy

Without any additional preparation, Cd_1−yZn_yTe (211)B (y∼3.5%) wafers were cleaned by exposure to an electron cyclotron resonance (ECR) Ar/H₂ plasma and used as substrates for HgCdTe molecular beam epitaxy. Auger electron spectra were taken from as-received wafers, conventionally prepared wafers (bromine: methanol etching, followed by heating to 330–340°C), and wafers prepared under a variety of ECR process conditions. Surfaces of as-received wafers contained ∼1.5 monolayers of contaminants (oxygen, carbon, and chlorine). Conventionally prepared wafers had ∼1/4 monolayer of carbon contamination, as well as excess tellurium and/or excess zinc depending on the heating process used. Auger spectra from plasma-treated CdZnTe wafers showed surfaces free from contamination, with the expected stoichiometry. Stoichiometry and surface cleanliness were insensitive to the duration of plasma exposure (2–20 s) and to changes in radio frequency input power (20–100 W). Reflection high energy electron diffraction patterns were streaked indicating microscopically smooth and ordered surfaces. The smoothness of plasma-etched CdZnTe wafers was further confirmed ex situ using interferometric microscopy. Surface roughness values of ∼0.4 nm were measured. Characteristics of HgCdTe epilayers deposited on wafers prepared with plasma and conventional etching were found to be comparable. For these epilayers, etch pit densities on the order of 10⁵ cm⁻² have been achieved. ECR Ar/H₂ plasma cleaning is now utilized at Night Vision and Electronic Sensors Directorate as the baseline CdZnTe surface preparation technique.     

Chemical contribution of oxygen to silicon carbide plasma etching kinetics in a distributed electron cyclotron resonance (DECR) reactor

This paper deals with the influence of the oxygen additive on the fluorinated plasma etch rate of silicon carbide. The assumption according to which the oxygen has a direct contribution to silicon carbide etching, by chemical reaction with carbon atoms, is generally reported in the literature. Our etching experiments are performed in a distributed electron cyclotron resonance reactor, on both 3C- and 6H-SiC. An SF₆/O₂ gas mixture (avoiding the presence of C species in the plasma), fluorine saturation conditions and constant ion bombardment energy and flux are used, allowing the study of O₂ contribution exclusively. In these conditions, our results demonstrate the neutrality of O₂ on SiC etching mechanisms. These results will be discussed reinfored both by several other experimental observations.     

Dry etching of Hg1−xCdxTe using CH4/H2/Ar/N2 electron cyclotron resonance plasmas

An approach is presented which eliminates the problems caused by hydrocarbon polymer deposition during etching Hg_1-x Cd_xTe with CH₄/H₂ based plasmas. We find that the addition of N₂ to the plasma inhibits polymer deposition in the chamber and on the sample. We speculate that atomic nitrogen formed from N₂ in the plasma has several beneficial effects: the elimination of polymer precursors, the reduction of the atomic hydrogen concentration, and a potential increase of methyl radical concentration. Evidence for the reaction between the nitrogen and the polymer precursors is presented. It is also demonstrated that the addition of N₂ to CH₄/H₂ based electron cyclotron resonance (ECR) plasmas used to etch HgCdTe eliminates the roughness normally formed during etching and results in a steadier etch rate.     

Electron cyclotron resonance plasma etching of HgTe-CdTe superlattices grown by photo-assisted molecular beam epitaxy

In order to form HgTe-CdTe superlattice diode arrays, a well-controlled etch process must be developed to form mesa structures on HgTe-CdTe superlattice layers. Wet etch processes result in nonuniform, isotropic etch profiles, making it difficult to control etch depth and diode size. In addition, surface films such as a Te-rich layer may result after wet etching, degrading diode performance. Recently, a dry etch process for HgTe-CdTe superlattice materials has been developed at Martin Marietta using an electron cyclotron resonance plasma reactor to form mesa diode structures. This process results in uniform, anisotropic etch characteristics, and therefore may be a better choice for etching superlattice materials than standard wet etch processes. In this paper, we will present a comparison of etch processes for HgTe-CdTe superlattice materials using electron microscopy, scanning tunneling microscopy, surface profilometry, and infrared photoluminescence spectroscopy to characterize both wet and dry etch processes.     

Macro-loading effects of electron-cyclotron resonance etched II–VI materials

It has been observed in semiconductor processing that the etch rates for materials subjected to an electron-cyclotron resonance (ECR) plasma change with the total sample area. This phenomenon is known as loading. Loading effects can result in pattern definition errors during micromachining. In argon/hydrogen plasmas, designed to etch II–VI materials, loading appears to primarily affect photoresist deterioration. Using an 80% argon-20% hydrogen gas chemistry optimized for HgCdTe, we observe a factor of 2 variation in photoresist etch rate. Loading may also affect semiconductor etch rates to a lesser extent. The observed trends suggest that radical changes in the plasma are the likely cause of this phenomenon.