Wafer Cleaning Research Articles

A novel detection system for defects and chemical contamination in semiconductors based upon the Scanning Kelvin Probe

A novel wafer scale detection system for defects and chemical contamination in semiconductors based upon the Scanning Kelvin Probe (SKP) is presented. It incorporates traditional work function ( φ) topography with high resolution (<0.1 meV), surface photovoltage (SPV) measurement and surface photovoltage–deep level transient spectroscopy (SPV–DLTS). This permits quasi-simultaneous determination of surface charge, surface potential and carrier lifetime, as well as the characterisation of deep level defects. The SKP system is introduced and the SPV and SPV–DLTS measurement modes are discussed. Surface charge measurements on clean and Fe-contaminated p-type Si(100) wafers show that iron induces negative charge in both native and thermally grown oxide layers. SPV–DLTS is demonstrated on defects in thermally oxidised Si. The system is both non-contact and non-destructive, requires no wafer preparation, and can be applied in both ambient and vacuum environments.

Alkaline cleaning of silicon wafers: additives for the prevention of metal contamination

The paper presents a new strategy for cleaning of silicon wafers. A novel class of chelating agents added to alkaline cleaning mixtures provides promising performance without negative effects such as metal redeposition due to residual metal contamination of the cleaning solution. The superior capability of the new cleaning process is confirmed by the results obtained from wafer surface metal analysis as well as from minority carrier lifetime and diffusion length measurements and gate oxide integrity tests. Particle densities and surface roughness are not influenced by the presence of the chelating agents in the cleaning solution. TOF–SIMS measurements do not indicate any deposition of chelating agent on the wafer surface. With this type of modified SC-1 cleaning procedure the acid SC-2 step used in conventional RCA cleans to remove the metals deposited in the preceding SC-1 step is unnecessary resulting in substantial cost savings with respect to chemicals, waste, equipment and space.

Cost-effective cleaning and high-quality thin gate oxides

Some recent findings in the area of wafer cleaning and thin oxide properties are presented in this paper. Results are shown for a practical implementation of a simplified cleaning concept that combines excellent performance in terms of metal and particle removal with low chemical and DI-water consumption. The effect of organic contamination on ultrathin gate-oxide integrity is illustrated, and the feasibility of using ozonated DI water as an organic removal step is discussed. Metal outplating from HF and HF/HCl solutions is investigated. Also, the final rinsing step is critically evaluated. It is demonstrated that Si surface roughness without the presence of metal contaminants does not degrade gate-oxide integrity. Finally, some critical remarks on the reliability measurements for ultrathin gate oxides are given; it is shown that erroneous conclusions can be drawn from constant-current charge-to-breakdown measurements.

The surface chemistry of 1,1,1-trifluoro-2,4-pentanedione on clean and oxygen pre-covered Cu(210): A comparison with 1,1,1,5,5,5-hexafluoro-2,4-pentanedione and 2,4-pentanedione surface chemistry on the same surfaces

In the recent literature there has been some interest in the use of diones as metal etchants during wafer cleaning and metal patterning. In previous work we found that trifluoro-2,4-pentanedione was a better etchant for nickel than either hexafluoro-2,4-pentanedione or 2,4-pentanedione. In this paper we examine the decomposition of trifluoro-2,4-pentanedione on copper and etching of copper with the same compound. We find that trifluoro-2,4-pentanedione is not a good etchant for copper, even though trifluoro-2,4-pentanedione is a good etchant for nickel. Trifluoro-2,4-pentanedione decomposes via a β-scission process beginning at about 300 K on copper. While we did observe a tiny amount of etching as evidenced by a tiny Cu(CF 3COCHCOCH 3) 2 peak at 380 K, the etching rate was more than an order of magnitude smaller than that observed with hexafluoro-2,4-pentanedione. These results show that copper is more reactive than nickel for trifluoro-2,4-pentanedione decomposition, even though nickel is usually more reactive than copper for carbon–carbon bond scission processes.

Small Signal AC-Surface Photovoltage Technique for Non-Contact Monitoring of Near Surface Doping and Recombination-Generation in the Depletion Layer

ABSTRACTSmall signal non-contact ac-SPV method for monitoring near surface doping (NSD) in silicon has recently been introduced in commercial diagnostic tools. High chopping frequency light with a submicron penetration depth is used to generate small SPV signal and this signal is in turn monitored using a transparent pickup electrode. This technique has the advantage of producing fast, non-destructive full wafer measurement. Under certain conditions, the magnitude of this ac-SPV signal is inversely proportional to the depletion layer capacitance. If a depletion layer barrier height is known this allows the calculation of the concentration of ionized donors or acceptors in the depletion layer. NSD measurements by ac-SPV method were typically done for doping concentrations up to about 1016 cm−3. Only recently this range has been extended to 1018 cm−3, making it a very attractive technique for monitoring low and medium dose implants and especially for wafer scale mapping of implant uniformity and activation efficiency. In addition, the frequency dependence of the SPV signal provides a mean for evaluating the minority carrier lifetime in the near surface region of bulk and epitaxial wafers.The influence of surface/interface traps upon small ac-SPV signal has never been fully understood. This paper quantifies the role of interface traps in the monitoring of NSD. The effect of typical surface treatments such as HF, SCI and SCI+SC2 wafer cleans are examined.

Post-CMP Megasonic Cleaning Using Dilute SCI Solution

Non contact cleaning or wet-cleaning processes, were the megasonic play a key role in the separation of the particles from the wafer is a commonly used technique in semiconductor manufacturing. CMP process can be very damaging to the production yield if not followed by an effective post clean process. McQueen identified the effect of the acoustic boundary layer and its role in the removal of small particles at high frequency. Busnaina et alt studied ultrasonic and megasonic particle removal and the effect of acoustic streaming. They showed that the cleaning efficiency increased with power until a certain range and then decrease slightly. Busnaina et al result indicted that SCI removes more particles than DI water particularly at lower megasonic powers specially in the case were the slurry particles are deposited onto the wafer surface by dipping experiments. But they also demonstrated that it was possible to achieve 100 % removal in DI water when using the optimum conditions. This paper presents the latest results of the post-CMP megasonic cleaning process, this study is focused on the cleaning of thermal oxide silicone wafer polished using silica based slurry and cleaned using diluted SC1 (H20/H202/NH4OH: 40/2/1).

Use of Dilute HF With Controlled Oxygen for Post Cu CMP Cleans

Wet cleaning of wafers during the semiconductor production process often requires uniform removal of a few nanometers of material. Ideally, a single cleaning chemistry can be found that etches all exposed features at a comparable rate. Etch rates near 1 nm/min are desired for batch process and near 10 nm/min for single-wafer processes. A mixture of 500:1 DHF (dilute HF) with dissolved oxygen controlled near parts-per-million (ppm) levels has been found to meet these requirements for post copper CMP (chemical-mechanical polishing) cleans with exposed SiO2 and Cu metal.

半導体プロセスを支えるクリーン化技術 ウェハ洗浄技術

半導体プロセスを支えるクリーン化技術 ウェハ洗浄技術

Silicon Cleaning Methods Compared at Metal Concentrations Below 1E10 atoms/cm2

AbstractAn alkaline aqueous silicon wafer cleaner has been developed which reduces trace metal contamination levels on “p-type” unpatterned silicon wafers to below 1E10 atoms/cm2 without acidic cleaning. This patented technology uses a specially formulated buffering system consisting of an oxidation resistant chelating agent and salt of a weak acid. The aqueous cleaner maintains a stable pH of about 9.5 over a wide range of dilutions, temperatures, hydrogen peroxide concentrations and bath aging times. A single-step bath, megasonic bath or spray clean with this formulation leaves the chemical oxide on the silicon wafer surface free of particles, atomically smooth, free of organics and lower in trace metal contamination levels than similar surfaces cleaned with conventional formulations.An analytical method was developed which allows the reliable detection of trace metals on silicon wafer surfaces down to 1E8 atoms/cm2 for aluminum, calcium, copper, iron, nickel, sodium and zinc. The procedure uses an ICP-MS with a concentric nebulizer and a desolvator. The acids used were ultrapure to keep the blanks to a very minimum and analyses were run in a class 10 clean room. The trace metals on the wafer were extracted using known amounts of ultrapure acids and were directly aspirated using a special arrangement with the concentric nebulizer. The J.T. Baker cleaner was compared to and outperformed the conventional RCA-1&amp;2 clean and dilute RCA-1&amp;2 chemistries using ultrapure ammonium hydroxide, hydrochloric acid and hydrogen peroxide.

The Effects of Post Chemical Mechanical Planaization Buffing on Defect Density of Tungsten and Oxide Wafers

Tungsten sheet wafers and tetraethylorthosilicate (TEOS) wafers were planarized on chemical mechanical planarization (CMP) tools and cleaned using a mechanical wafer cleaner. Post‐CMP buffing processes on the primary or secondary polisher platens were investigated. Using laser scattering wafer inspection systems and atomic force microscopy we demonstrated that the buffing process strongly affects the defect density on both the TEOS and tungsten CMP wafers and the roughness power spectrum density of the tungsten CMP wafers. A “pH shock” to the TEOS wafers during oxide CMP on the primary platen resulted in a high defect density. The changes in zeta potential of the slurry particles and the TEOS surfaces during the pH shock might have caused the variations in the defect density. TEOS wafers polished with a tungsten slurry were also cleaned on a wafer cleaner with diluted HF solutions. The HF solution cleaning further enabled a reduction of the defect density. Defect densities measured on the same TEOS wafers using SFS6220 and SFS6420 were compared. The results indicated that some of the defect counts may be attributed to the intrusion defects and background noise from the surface roughness.

Wafer-Bonding and Thinning Technologies

“Wafer bonding” refers to the phenomenon in which mirror-polished, flat, and clean wafers of almost any material—when brought into contact at room temperature—are locally attracted to each other by van der Waals forces and adhere or “bond” to each other. Wafer bonding is alternatively also known as “direct bonding” or “fusion bonding,” or more colloquially as “gluing without glue.” Although this is by no means required, in most cases, the wafers involved in actual applications are typical semiconductor wafers consisting of single-crystalline material used in microelectronics or optoelectronics such as silicon or gallium arsenide.

Influence of the Dissolved Gas in Cleaning Solution on Silicon Wafer Cleaning Efficiency

Influence of the Dissolved Gas in Cleaning Solution on Silicon Wafer Cleaning Efficiency

Characterisation of HF-last cleaning of ion-implanted Si surfaces

Issues of HF-last cleaning of implanted silicon are studied. Contact angle measurement and ellipsometry are employed and the results are discussed based on SiO 2 and Si surface chemistry. The cleaning of BF 2-implanted wafers is not straightforward because of low etch rate of surface oxide and slow removal of surface hydroxyl groups. This problem is not significant in the case of As-implanted silicon, however the surface is found to oxidise faster than that of BF 2 or non-implanted silicon. Oxygen chemisorption starts at the silicon surface immediately after HF etching. Surface charge measurement can be a good alternative to contact angle measurement for the characterisation of silicon surface cleaning.

SEMICONDUCTOR WAFER BONDING

▪ Abstract When mirror-polished, flat, and clean wafers of almost any material are brought into contact at room temperature, they are locally attracted to each other by van der Waals forces and adhere or bond. This phenomenon is referred to as wafer bonding. The most prominent applications of wafer bonding are silicon-on-insulator (SOI) devices, silicon-based sensors and actuators, as well as optical devices. The basics of wafer-bonding technology are described, including microcleanroom approaches, prevention of interface bubbles, bonding of III-V compounds, low-temperature bonding, ultra-high vacuum bonding, thinning methods such as smart-cut procedures, and twist wafer bonding for compliant substrates. Wafer bonding allows a new degree of freedom in design and fabrication of material combinations that previously would have been excluded because these material combinations cannot be realized by the conventional approach of epitaxial growth.

Detection of Metallic Contaminants on Silicon by Surface Sensitive Minority Carrier Lifetime Measurements

We present the use of in situ carrier lifetime measurement by radio‐frequency photoconductance decay (RFPCD) to detect metallic contamination during aqueous HF cleaning of silicon wafers. The effect of metal ion deposition from dilute HF solutions on carrier recombination at the Si/HF junction is examined. Ions of the more electropositive metals, such as and , do not undergo spontaneous reduction from dilute HF. When present in HF, they do not measurably increase surface recombination and hence cannot be detected by RFPCD. Reduction of ions at the Si/HF interface occurs by a slow (conduction band) process which involves nucleation of nanometer sized precipitates, whose areal density can be measured by RFPCD. Using RFPCD in conjunction with atomic force microscopy, we have found that the presence of 2 ppm of chloride ions in dilute HF increases considerably both nucleation and growth of the copper clusters. Air exposure prior to RFPCD measurement in HF increases the recombination activity of deposited copper on the surface drastically. The reason for this effect remains poorly understood at present. In contrast with copper, gold ions deposit from HF on Si by a much faster valence band process. The initial deposition rate of gold ions is diffusion limited, as evidenced by the t−1/2 dependence of the surface minority carrier lifetime, τsurf (t is immersion time in gold‐dosed solution).

A study of platinum electrode patterning in a reactive ion etcher

This article addresses the problem of Pt electrode etching through the use of a batch load production reactive ion etch (RIE) tool to study etching characteristics and the cleanliness of patterned films with pressure, total gas flow, and percent of Cl2 in Ar as variables, and considers some of the environmental, health, and safety issues. The results show that Pt etching is primarily a sputter etch process in which the Cl2 percentage has little impact on the Pt removal rate, but does significantly affect etch uniformity across the wafer and the surface cleanliness as analyzed with Auger electron spectroscopy. The maximum Pt etch rate achieved was about 5 nm/min with good etch uniformity and surface cleanliness. X-ray photoelectron spectroscopy of the etch by-products shows the presence of PtCl2 and PtCl4 when the Ar–Cl2 etch chemistry was used. These results provide useful information to address material redeposition, wafer cleaning, and etch chamber cleaning safety issues, major concerns in the RIE of Pt.

Susceptibility of silicon wafers to acoustic microcavitation damage

Acoustic methods are being increasingly used in semiconductor processing. Given the increase in circuit densities being implemented on semiconductor chips the requirements for particulate contamination tolerance are becoming more stringent. The semiconductor industries road-mapped specifications for 1998, 0.25 μm feature size require control of all particulate contaminants larger than 0.08 μm. Fine cleaning of post CMP wafers is currently being done by using the megasonics process. The cleaning effectiveness depends on the strength of the acoustic fields used. In using high-intensity, high-frequency acoustic fields it is crucial to ensure that the surface of the silicon wafer does not suffer any damage due to cavitation. This paper assesses the damage potential of semiconductor wafers to acoustic microcavitation. Acoustic microcavitation is brought about by low megahertz acoustic fields giving rise to micron size bubbles that live a few microseconds. In exposing a surface to continuous waves one could obtain cavitation effects in an average, overall sense; the details of nucleation, evolution of inertial events, however, get glossed over. Both, pulsed and cw insonification were studied. Experimental measurements of the thresholds for cavitation damage of semiconductor wafers will be presented.

A model of wafer bonding by elastic accommodation

Two clean and flat wafers can adhere spontaneously. The technique has recently led to novel electronic and optoelectronic devices. The adhesion arises from short-range interatomic forces between wafer surfaces, which can be represented by a reduction in surface energy associated with the transformation of two surfaces into one interface. The wafers, however, are seldom perfectly flat; the misfit has to be accommodated by elastic distortion, plastic deformation, or mass transport. We model elastic accommodation in this paper. The distortion causes the wafers to gain elastic energy. If the surface reduction dominates over the elastic energy gain, the wafers will bond. We solve the three dimensional elastic field in the misfit wafers analytically. The conditions for bonding are established, and practical implications discussed.

Influence of initial wafer cleanliness on metal removal efficiency in immersion SC-1 cleaning: limitation of immersion-type wet cleaning

When contaminated silicon wafers are immersed in an ultra-pure cleaning solution of an NH/sub 4/OH/H/sub 2/O/sub 2//H/sub 2/O mixture known as the RCA Standard Clean 1 (SC-1), in which the impurity concentration is negligibly low, the level of wafer-surface metallic contamination after the cleaning treatment depends on the amount of metallic impurities brought into the solution by the to-be-cleaned wafers themselves. Even if the chemicals are disposed of after each wafer cleaning, the surface metallic contamination is still dominated by the amount of impurities brought into the fresh solution by the wafers themselves. In the past, purer chemicals have been sought to improve metal removal efficiency, but after reasonably purer chemicals are obtained the efficiency is not governed by the initial chemical purity but by the initial wafer cleanliness. Because of this, scrubbing of dirty wafers-both the backand front-surfaces-before immersion-type wet cleaning is recommended. However, to meet future stricter wafer cleanliness requirements, new cleaning methods in which fresh chemicals are continuously supplied, such as single-wafer spin cleaning, will have to be employed.

Usefulness of surfactants in reducing particle adhesion and their effectiveness in cleaning silicon wafers

The effectiveness of surfactants (sodium dodecyl sulfate and dodecyl amine hydrochloride) in the cleaning of silicon wafers has been investigated using atomic force microscopy and image analysis. Recognizing that the surface of SC-1-cleaned silicon wafers is essentially SiO2 atomic force microscopy was used to determine the interaction forces for the silica/silica and silica/alumina systems with and without surfactant present in solution. Experimental force vs. separation distance curves were found to be in good agreement with theoretical predictions based on electrostatic and van der Waals interactions. Interestingly, the pull-off forces obtained from atomic force microscopy measurements were very close to the adhesion forces calculated on the basis of van der Waals interactions at very close separations. The results from image analysis further confirmed the usefulness of surfactants in reducing particle adhesion and their effectiveness in cleaning silicon wafers. Finally, the results from this study suggest that a more complete understanding of particle interaction forces should be of considerable importance to the electronics industry.