ABSTRACTS

The following is a list of the abstracts for papers which will be presented in EIGHTH INTERNATIONAL SYMPOSIUM ON PARTICLES ON SURFACES: DETECTION, ADHESION AND REMOVAL. The listing is alphabetical by presenting author. This list is updated continually to add abstracts as they become available and make appropriate corrections. This list may be conveniently searched by using the editor provided with most popular browsers (e.g. Microsoft Explorer, Netscape, ... etc.)

Performance Criteria for Copper Damascene Post-CMP Wafer Cleaning

Copper damascene chemical mechanical polishing (CMP) is becoming a standard process in the semiconductor industry. Successful integration of this process into wafer-level fabication requires efficient and robust post-CMP cleaning methods. This cleaning step provides a unique set of engineering challenges: known defects such as slurry particles and dissolved metals must be removed from the polished wafer surface without damaging the substrate and all circuitry contained therein. The performance criteria for this process will be reviewed in detail, covering such topics as surface potentials, corrosion effects, etching, particle dispersion, and tool qualification as these topics relate to the damascene post-clean. Also, specific examples will be shown of cleaning-related defects and electrical performance metrics.

Overview about Particle Cleaning Technologies for Semiconductor Wafer Including Wet Cleaning as Well as Alternative Cleaning Methods like Laser Cleaning and Cryogenic Cleaning DROPPED OUT

Submicrometer particle as well as nanoparticles on wafer surfaces can have detrimental effects on quality and therefore on wafer yields. As feature sizes shrink, defects caused by particles during Si-wafer production will have also effects on device yield. Therefore cleaning processes are developed with excellent particle performances in order to make sure that particles have no influence on wafer performance.

In Si- wafer manufacturing cleaning is necessary for rough surfaces as well as for smooth surfaces.

Cleaning of rough surfaces is done after sawing, grinding or lapping. Slurries with particle sizes as big as 20m as well as Si "dust" together with organic chemicals, partly mixed with water have to be removed from surface and also from sub surface damage.

Unremoved particles on rough surfaces lead to:

a)scratches which are only eliminated with high removal of Si for example in polishing (high cost and detrimental for flatness)

b)dimples if a hard chuck or something similar is used for the removal step. A dimple is the consequence of a deepening at the position of the particle on the frontside.

For cleaning of rough surfaces, wet cleaning is still the most common cleaning method. Cleaning of such surfaces must remove particles, organics, metals, moisture and silicon dust along with subsurface damage. Therefore some etching (removal of Si) is needed to facilitate the removal of particles beneath the surface.

Alternative cleaning methods like Cryogenic -, CO2 snow - or laser cleaning are not capable of subsurface cleaning.

Cleaning of smooth surfaces is done for example after polishing or epitaxy to remove polishing slurry and /or particles from measurements, handling and environment.

Particles left on smooth surfaces lead to increasing yield losses if more stringent particle specifications are requested from the device manufacturer.

Wet cleaning is also still the most common cleaning technique for smooth surfaces, but other cleaning technologies are under investigation. It is still the method of choice for removal of high densities of particles after polishing because the wafer surface is still wet and particles are more easily removed when wet.

Smooth surfaces must not be etched during cleaning. Etching has a negative impact on roughness and COP`s are more likely to be opened up.

Wet wafers cannot be treated by low temperature cleaning like snow cleaning or cryogenic cleaning and also not by laser cleaning. The removal of the last few particles after a wet cleaning and drying step could be done with low temperature or laser cleaning but methods are still in evaluation status.

The potential of laser cleaning for this application is dependent on the surface condition. Damage of silicon with the so-called DLC (Dry laser cleaning) has been observed, and therefore SLC (Steam laser cleaning) looks better at the moment. Moreover, it seems to be very difficult to remove particles from the entire wafer especially when speaking about 300mm wafers. Particles smaller than 0.12 mm seem to be more difficult to remove than larger particles.

With low temperature cleaning methods using CO2 , it is difficult to find the right parameter set to remove particles while not damaging the surface. It appears that organic contamination cannot be removed with this method.

Cryogenic cleaning is possible without damaging the wafer surface. Organic removal is dependent on the composition of gases used. This method currently only cleans the frontside, and its ability to clean the backside without recontamination of the fronside appears to be a challenge.

The Adhesion of Rough, Nonuniform Particles to Rough Surfaces

A model has been developed that predicts the van der Waals interaction force for micron-scale, asymmetrical, rough particles interacting with rough substrates. It accounts for particle geometry, asperity deformation on the particle and substrate, and surface roughness on the particle and substrate. The predicted interaction forces for materials of interest in chemical mechanical polishing during integrated circuit manufacture are compared with AFM measurements for these materials. A prior adhesion model described asperities as hemispheres and simulated roughness on the particle and substrate by distributing hemispheres over approximate mathematical descriptions of these surfaces. This prior model was compared with new roughness descriptions, including: 1) a direct surface topographical map obtained from an AFM analysis of the interacting surfaces and 2) a surface generated using the Fourier transform of the surface height variations. While the hemisphere model required the least computational effort, it was not as accurate as the Fourier transform model in predicting interaction forces. The Fourier transform approach required less computational effort than the direct topographical approach, and provided a very accurate representation of the adhesion force. The accuracy of the Fourier transform approach was equal to that from the direct topographical map. The validation of these different roughness representations in the van der Waals adhesion model will be described.

Applied Materials, 3050 Bowers Avenue, MS 0104, Santa Clara, CA 95054

Particle Performance of Modified SC-1 Solutions

The RCA clean is widely used in the semiconductor industry for many wet-chemical cleaning processes. The RCA clean consists of a particle removal step, the Standard Clean 1 or SC-1 and metallic impurity removal step, the Standard Clean 2 or SC-2 step. In this work we have investigated the addition of chelating agents and surfactant in SC-1 solutions to provide an "all-in-one" cleaning solution. We also have studied the effect of surfactants in such solutions on sub-micron particle removal. This leads to the development of a very fast and efficient single step RCA replacement clean.

Characterization of Particles on Surfaces using X- Ray Fluorescence Methods with Grazing Geometries

Total Reflection X-ray Fluorescence spectrometry (TXRF) is a technique that has found wide usage in the semiconductor industry for the detection and characterization of contamination on silicon wafer surfaces. State of the art commercial instruments with conventional energy-dispersive detection systems have detection limits for transition metal impurities of the order of 10⁹ atoms/cm². Measurements of the angular dependence of the fluorescence intensity which exploit the X-ray standing wave fields developed above flat and smooth surfaces enable additional information about the vertical distribution of contamination (e.g. particle-like or film-like) or even the sizes of particles on silicon wafer surfaces to be obtained.

In order to meet the challenges set by the increasingly stringent demands on contamination control, a complementary method Grazing Emission X-ray Fluorescence spectrometry (GEXRF) is currently being developed. Detection of the fluorescence radiation emitted at grazing angles from flat surfaces following X-ray or electron excitation at normal incidence has the advantage that both energy-dispersive and wavelength-dispersive detection systems can be employed. The latter considerably extends the range of application by enabling the analysis of the light elements.

In this contribution, details of the GEXRF instrumentation and the results of modelling calculations and initial experimental measurements will be presented. These will be discussed in terms of the information that can be extracted from the angular dependencies of the fluorescence intensities in order to characterize particles on surfaces.

1) NSF Center for Microcontamination Control, Northeastern University, Boston, MA 02115-5000

2) IBM Microelectronics Division, 32A, IBM, Hopewell Junction, NY 12533

Removal Of Nano and Micro-Scale Particle from Silicon Oxide Substrates

There is a need to physically manipulate, control or remove nanoscale particles. The removal of nanoscale particles physically without substrate damage or alteration is needed in nanoscale manufacturing. It is needed to remove existing contaminants from a substrate. Even in the semiconductor industry, such a need is a projected requirement in 2011. However, this need is required toady in nanoscale fabrication. Physical non-contact removal using high frequency acoustic streaming had been used to remove submicron particles. However, the removal of 100 nm particles and smaller is becoming a serious challenge. Busnaina et al¹ studied megahertz streaming particle removal and evaluated the effect of acoustic streaming on the cleaning process.

High frequency acoustic streaming is a promising technique for nano-scale particle removal. Using DI water, the removal of nano-size particles down to 10 nm can be best accomplished using acoustic streaming with frequency above 1.3 MHz. Softer particles (such as PSL) are more difficult to remove than hard particles (such as silica), because of adhesion-induced deformation, needing a much higher frequency. The experimental results show that a complete removal of silica particles down to 100 nm is achievable. The removal of nanoscale particles from submicron trenches and cavities will also be presented and discussed. The mechanism of removal from deep cavities will also be explored.

REFERENCES

1. Busnaina, A. A. and Gale, G. W, Journal of Particulate Science and Technology, 17(3), 1999.

1) NSF Center for Microcontamination Control, Northeastern University, Boston, MA 02115-5000

2) IBM Microelectronics Division, 32A, IBM, Hopewell Junction, NY 12533

Physical Cleaning of Patterened Semiconductor Substrates

According to the international technology roadmap for semiconductors, the industry will face the challenge of cleaning 100 nanometer trenches with high aspect ratio (30-60) in the next five years. Cleaning high aspect ratio deep trenches is a challenging task because of the need to rinse or remove contaminants from the bottom of the trench. Megasonic cleaning is known as one of the most effective techniques in blanket wafer cleaning. Although megasonic cleaning is currently used in patterned wafer cleaning, the mechanism of megasonic cleaning process for patterned wafers is not well understood. Pulsating flow rinse shows a significant advantage in pattered wafer cleaning because the vortex oscillating mechanism enhances the mixing. In this paper, the removal of contaminants from high aspect-ratio submicron trenches using high frequency pulsating flow (megasonic rinse) is studied using physical modeling. Both parallel rinse and normal (with respect to the wafer surface) rinse are investigated.

The results show that modeling of pulsating flow passing a series of rectangular cavities has been verified and shows excellent agreement with the numerical and experimental results of Perkins. Compared to steady flow rinse, pulsating flow rinse shows a significant advantage in blanket and patterned wafer cleaning. When parallel oscillating flows are used to rinse deep and sallow trenches larger than 100 mm, maximum mixing occurs at the same Strouhal number (St = 0.133). However, for submicron trenches, higher frequency consistently gives higher cleaning efficiency. Using the same average rinsing frequency to rinse deep trench, normal flow shows orders of magnitudes better cleaning efficiency than the parallel flow for both steady flow rinse and oscillating flow rinse.

Nanoparticle Removal from Substrates with Pulsed-laser Induced Plasma

Removing of nanoparticles from substrates is a challenging task with various critical applications. A new removal method for nanoparticles is developed and tested. The technique, which is dry and non-contact, takes advantages of shock wavefronts initiated by plasma formation under a focused laser beam pulse and its interaction with the substrate. Our experimental results indicate that silica particles down to 500 nm on silicon wafers can be removed without substrate damage. In the current experiments, a Q-switched Nd:YAG laser with a 5-ns pulse width and 360-mJ pulse energy at 1064 nm wavelength is utilized as a plasma generation source. The traditional dry laser cleaning method based on the rapid thermal expansion under direct laser irradiation often results in surface damage due to light diffraction around nanoparticles and/or stress localization in the thermal skin. In the laser plasma method, a number of mechanisms are interactively responsible for the removal effect. The strong shock wave in air generates complex pressure wavefields resulting in both drag/lift on the particle and acceleration of the substrate. However, no shock wave is transmitted to the solid substrate due to a large difference between the relevant wave phase speeds in the two media. The effects of the number of shots and the distance between the surface and the plasma boundary on the removal efficiency are reported.

* To whom all correspondence should be addressed

1) Evans East, 104 Windsor Center, Suite 101, East Windsor, NJ 08520

2) Charles Evans & Associates, 810 Kifer Road, Sunnyvale, CA 94086

3) Evans PHI, 6509 Flying Cloud Drive, Eden Prairie, MN 55344

Surface and Micro-Analytical Methods for Particle Identification

The ultimate tool for particle analysis would provide elemental composition, chemical bonding, and exact compound identification of organic, inorganic, and metallic compounds of any size and shape. Since this hypothetical tool does not exist, scientists are forced to use a variety of tools that accomplish a subset of these requirements. Surface analysis methods such as Auger Electron Spectroscopy (AES), Time-of-Flight Secondary Ion Mass Spectrometry (TOF-SIMS) and X-ray Photoelectron Spectroscopy (XPS, also known as ESCA) fulfill many of the desired criteria while confining analysis to the outer surface of the particles. Micro-beam methods such as Energy Dispersive x-ray Spectroscopy (EDS), micro Fourier Transform Infrared Spectroscopy (m-FTIR), and Raman Spectroscopy fulfill many of the same criteria but examine larger analytical volumes. The fundamental principles of these six tools will be discussed in light of the specific requirements and restrictions needed for particle analysis. Typical case studies from a variety of industries will be used to highlight the tool selection process.

The Use of Surfactants to Enhance Particle Removal from Surfaces

Effective particle removal from surfaces is critical to many important technologies, and surfactants are important tools in enhancing particle removal from surfaces in solution environments. The behavior of surfactants in solution under a variety of conditions will be presented in the context of adsorption and aggregation. The effects of surfactant hydrocarbon chain length and solution ionic strength will be analyzed. Surface forces between particles and surfaces will be discussed, and the important role that surfactant molecules can play in altering such forces as it relates to enhancing the removal of particles from surfaces during cleaning will be presented.

Influence of Transient pH on Particle Redeposition During Rinse

In semiconductor device manufacturing the negative impact of particulate contamination (sub micron sized particles) on the wafer surface is well known. Various wet cleaning steps are therefore developed in order to remove these particles from the surface. The chemistry of the employed cleaning mixture is optimized to enhance particle removal or at least to prevent deposition during the course of the cleaning step. A typical way to obtain this is by modifying the pH in order to induce like charges on particles and substrates, hence causing electrostatic repulsion.

In a next step, the wafers are rinsed and dried. During the rinse process, the processing liquid is gradually being replaced with clean rinse liquid, usually of a different chemical composition. As the liquid composition changes, so do the interactions between the particles and the wafer surface. In this work we will investigate the deposition behavior of sub micron sized spherical silica particles during the course of a neutral rinse step that follows a low pH cleaning step.

The Use of the Rectangular Jets for Surface Processing

High speed waterjet constitutes the most effective and the most common cleaning tool. The use of the jets ranges from street cleaning to decontamination of food products. Practically all jets currently utilized for material decontamination are generated by cylindrical nozzles. The use of the round geometry is due to the simplicity of the fabrication of round orifices. This geometry, however, does not assure effective utilization of the compressed water. The round impact zone brings about uneven distribution of the energy transferred from the water to the workpiece. The energy exchange between the cylindrical jet and a substrate results in energy losses due to insufficient energy delivery to the periphery of an impact zone and the excessive energy delivery to the central part. Depending on the kind of an application the water stream should operate as a knife, scraper or brush. The cylindrical jet however cannot meet the diverse requirements of various cleaning processes. The use of a rectangular jet with a wide range precisely controlled aspect ratio enables us to alleviate the shortcomings above.

Several versions of the slot nozzle generating the rectangular jet were constructed and tested at the NJIT's Waterjet laboratory. The principal advantage of these nozzles was feasibility to change in a wide range the width and the thickness of the jet. This enabled us to increase the rate of the surface processing (cleaning, decoating) and to reduce water consumption. The fabrication and the restoration of the nozzles were simple and inexpensive. The nozzles were successfully used for metal depainting, derusting and graffiti removal. The special experiments were carried out in order to compare the performance of the developed and commercial nozzles. The tests showed significant superiority of the slot nozzle even at early stage of its engineering. Particularly, the specific water consumption by the nozzle was six times less than that of comparable commercial nozzles. The performed experiments evidently demonstrated that the rectangular jets constitute more effective surface processing tool than commonly used round jets.

Development of a Technology for Glass Cleaning in The Course of Demanucturing of Electronic Products

The recycling and reuse of post-consumer products constitutes one of the most challenging engineering problems. Although the environmental considerations make current practice of dumping after the end-of the-life products unacceptable, available technologies do not provide adequate means for commercially viable recovery of these products. A recent survey of the major glass manufacturers indicates that a significant market exists for post-consumer Cathodic Ray Tubes (CRT) glass cullet provided that it is acceptably clean, sorted and attractively priced. Consequently, an efficient and effective CRT glass cleaning process is critical in achieving high rates of recovery and recycle of computer monitors and televisions.

The primary objective of this exploratory study is to determine the technical feasibility of fluidized bed-based technology for cleaning CRT glass cullet. During the CRT production process, layers of various coatings are applied to the interior surface of the CRT. These coatings plus any soil or other contaminants must be removed before the glass can be used as recycled cullet for the glass industry Concerns have been raised regarding the viability of current cleaning technologies; consequently, this study has been undertaken to examine an innovative new approach to CRT glass cleaning which has the potential to achieve high levels of productivity and low cost. One advantage over current technology is that the fluidized bed approach can handle any size cullet whereas traditional cleaning systems are limited to larger sizes of cullet. It was shown in the course of our experiments that fluidized bed cleaning brings about generation of fine glass powder. Glass powder is a high value product and it was another objective of the study to identify its potential applications.

The technical feasibility of using fluidized beds to clean glass cullet has been demonstrated at a laboratory scale experimental setup. A range of process characteristics and bed formulations were examined with an objective of producing visually clean cullet without undue reduction in the cullet size. As the result preliminary process characteristics for the experimental bed were determined and conceptual process design was suggested.

Ice-air Blasting: Case Studies

Ice-airjet constitutes one of the most promising emerging cleaning tools. The ice-air blasting generates minimal off-streams, requires portable, comparative inexpensive facilities, the and the most of all does not involve any realistic possibilities of the substrate damage. The paper discusses the results of experimental study of ice cleaning performed by the Waterjet Lab as well as results of practical process applications. Different techniques for powder production are analyzed and the conceptual design of various ice blasting systems is discussed. Special attention is paid to the current and potential applications of the ice based cleaning. The case studies involve decontamination of electronic devises, precision mechanical parts, soft plastics, etc. Special attention is paid to the infrastructure maintenance (derusting, depainting, graffiti removal) and the biomedical applications. The performed analysis enables us to identify the potential applications of the use ice blasting in industry, infrastructure maintenance and medicine.

Adhesion of Nanoparticles

Adhesion of nano-particles is important in a wide range of applications, from contamination of silicon wafers to setting of Portland cement[1]. This paper examines the fundamental processes involved in such nano-particle adhesion, including the deformation of the particles, the formation of molecular bonds, the reversibility of the mechanism, the effects of drying and the structuring of the particulate assembly. Experiments using Atomic Force Microscopy are described to illustrate the theoretical arguments.

[1] K. Kendall, 'Molecular Adhesion and its Applications', Kluwer Academic, New York 2001, ch 13.

The Nature and Characterization of Small Particles

Nano-scale particles are of fundamental and practical interest. As feature sizes shrink, nano-particle contamination will become increasingly important and will present an ongoing challenge to achieve and maintain high product yields. In order to employ appropriate preventive and remediation strategies to control particle contamination, it is necessary to understand the nature of nano-scale particles and to characterize the contaminant particles. Particles in the range 1 nm to 100 nm present unique challenges and opportunities for their imaging and characterization. Critical information for this purpose is the number and size of the particles, their morphology, and their chemical structure. Following a review of the nature of small particles, many of the techniques available for counting, sizing and analyzing small particles will be discussed. These techniques include optical counting, measurement of mass and number concentration, nucleated condensation, electrical mobility analysis, particle impaction, and optical and electron microscopy, as well as chemical analytical methods. The advantages and limitations of these techniques will be discussed, and recent developments will also be described.

The Future of Industrial Cleaning and Related Public Policy-Making

In this paper, the author discusses some of the findings of her doctoral thesis, The Search for Safer and Greener Chemical Solvents in Surface Cleaning,^a as well as some of the new directions that the science of cleaning may take in the next five to ten years.

Behaviors of Microparticles on Solid Surfaces During Laser Surface Cleaning

Laser cleaning is a prospective cleaning method that can be widely used in microelectronics fabrication, archive restoration and optical apparatus cleaning. Removal of microparticles and nanoparticles from solid substrate is an important aspect of laser cleaning. Although many studies have been carried out on this subject, few of them are objected to the characterization of the ejected particles in laser cleaning. In this study, a method was developed to "capture" the particles ejecting from the substrate after laser irradiation. Detection of both angular distribution and ejecting energies was achieved with this method. It was found that the angular distribution of the ejected particles fitted to a Gaussian curve when the laser irradiated normally to the substrate. The distribution curve for the particles ejected from a rough surface has a wider FWHM than that from a smooth substrate. It was also found that the particle ejecting energy increased obviously with laser fluence, therefore the laser cleaning efficiency was promoted sharply as laser fluence increased. Laser cleaning efficiencies to remove particles have also been investigated with different incident angles ranging from 0 deg. to 60 deg. It is found that when the laser light irradiated normally to the subsrate surface, the particle could be removed most efficiently. In this direction, the cleaning efficiency was also most sensitive to the light intensity. A sharp drop of cleaning efficiency occurred with a small change of the incident angle. Theoretical calculations based on the Lorentz-Mie theory and an accurate solution of the boundary problem, indicate that the light intensity near the contacting point is sensitive to the incident angle even though the incident light is uniform.

Thermodynamic Theory of Adhesion of Particles on Surfaces

In his theory of the Brownian motion, Einstein [1905] introduced into thermodynamics a basic concept of central importance in Newtonian mechanics: the rate of change of momentum, associated with the thermal motion of particles suspended, or dissolved in a liquid. Einstein's objective was to study the particle transport in the liquid bulk, excluding the surface. This paper applies a generalization [Melehy,1997] of Einstein's theory to interfacial systems. A basic consequence results: If certain thermodynamic parameters vary across a surface, a membrane, or any other interface, the first and second laws require the existence of electric charges at such sites [Melehy, 1998]. This nearly universal result explains numerous interfacial phenomena. In the particular case of macroscopic, or microscopic particles on surfaces, significant electric dipole charges are formed that exert mutually attractive Coulomb forces. Such forces can, in principle, be calculated, in terms of the thermodynamic, physical, and geometric parameters.

_________________________

A. Einstein, On the theory of Brownian Motion, Ann. d. Phys., Vol. 17, pp. 549-560, 1905.

M. A. Melehy, Thermal momentum in thermodynamics, Part 1, Phys. Essays, Vol. 10, No. 2, pp. 287-303, 1997.

M. A. Melehy, Thermal momentum in thermodynamics, Part 2, Phys. Essays, Vol. 11, No. 3, pp. 430-443, 1998.

Laser Cleaning - a Complex Process

The removal of particle contamination from surfaces is one of the crucial prerequisites for a further increase in the integration density of ICs and for the progress in nanotechnology. Although the traditional wet cleaning methods have been continuously improved, with further shrinking of line widths there is a definite need to complement traditional methods by new cleaning technologies.

One of these new approaches is called laser cleaning. In Dry Laser Cleaning (DLC) the surface to be cleaned is irradiated by a short laser pulse. In Steam Laser Cleaning (SLC) prior to the application of the laser pulse a liquid, e.g. a water-alcohol mixture, is condensed onto the surface. More than one decade after laser cleaning has been suggested for wafer cleaning for the first time, there is still a lack of understanding of the underlying physical processes and in the determination of the optimum process parameters for efficient cleaning. We have performed studies on laser induced bubble nucleation in liquids and optical near field effects at the contaminating particles. A combination of the results with the knowledge gained from laser cleaning experiments performed in well-defined environments (high vacuum, precise control of liquid film in SLC), using lasers of various wavelength (UV - IR) and pulse length (femtosecond - nanosecond) and particles of a broad size range (60 nm - 2500 nm) and shape, enables us to draw conclusions on the particle removal process. Laser cleaning" is a complex interplay of several processes: thermal expansion of substrate and/or particle, local substrate ablation due to near-field effect and the explosive evaporation of atmospheric or by purpose applied liquid. The contribution of each of the above processes to particle removal is determined by parameters as laser wavelength, laser pulse duration, particle size, particle shape, optical properties of particle and substrate or amount and composition of the liquid. In SLC, the precise control of the amount of liquid is essential for successful cleaning.At first sight the complexity of laser cleaning seems to be unfavorable for its implementation as industrially applied cleaning technique. However we will show that, based on the knowledge of the underlying physics, it is possible to specify large process windows where efficient and safe particle removal are possible.

Particle Adhesion on Nanoscale Rough Surfaces

Formulae are developed for calculation of capillary adhesion between a smooth adhering particle and a surface with roughness on the nanometer scale. Adhesion force was measured using AFM for different materials with roughness in the range of 1-10 nm. The values of adhesion forces vs. relative humidity of air indicated that the RMS roughness of 1 nm decreases the adhesion by ten times in dry systems. It was observed that critical humidity for the onset of capillary forces increases as roughness increases, and that the smallest scale of roughness primarily controls the adhesion on surfaces. Experimental results and theoretical calculations indicate that the known criteria of critical humidity, based on thermodynamic equilibrium between liquid film, annulus and the gas medium, yield overestimated values of critical humidity for the onset of capillary forces between particles and surfaces with nanoscale roughness.

1) Phrasor Scientific Inc.,1536 Highland Ave., Duarte, CA 91010

2) Axcelis Technologies, Beverly, MA

Particle Removal by Collisions with Energetic Clusters

A novel method for precision dry cleaning is described that is capable of cleaning in-situ, within process chambers, enabling smooth operation between process steps. This method, named NanoClean, employs charged liquid clusters generated from an aqueous solution and electrically accelerated to remove submicron debris and thin films by momentum transfer. With anticipated low operating costs and non-hazardous chemical usage, environmental impact issues are virtually non-existent.

Using an electrified capillary nozzle, the NanoClean head produces an energetic beam of charged micro-clusters, directly from the liquid state, that are accelerated by the electric field. The clusters reach supersonic velocities and impart impulsive forces sufficient to physically remove surface films as well as particles. The match between the impacting cluster size, well below one micron, with that of the submicron particles to be removed, assures efficient momentum transfer even in trenches and vias. The cleaning action can be dynamically regulated to be ultra gentle and non-destructive to the surface, or forceful enough to remove films and oxide coatings. In addition low energy electrons can be injected into the beam when required to compensate for charge build-up on insulating surfaces. NanoClean projects a small footprint and can be integrated into a cluster tool or used as a supplement for removing submicrom contaminants following wet cleaning.

Tests reported in this presentation were performed in-situ in a Scanning Electron Microscope (SEM), operating at pressures in the E-5torr range. Pieces of a wafer, contaminated with particle debris produced by intentional scratch marks, were placed on a stage in the SEM, then examined and photographed. The wafer piece was transported to a position to be cleaned by the microcluster beam, then re-staged to the examination position, without breaking vacuum. SEM photos, taken of the same location before and after cleaning, clearly show the ability of energetic microclusters to completely remove particles ranging from about 5 microns down to 0.05 microns (the limits of the resolution of the microscope). There is no theoretical lower limit of the contamination size that can be removed by the NanoClean process.

This series of tests, in an electron microscope, provided verification of the following performance features:

• •In-situ operation
• •Submicron particle removal down to .05 micron
• •Observation of impingement angular effects
• •Ability to avoid or cause surface effects
• •Line-of-sight cleaning

Ideal Ultrasonic Parameters for Delicate Parts Cleaning

The various ultrasonic parameters, or degrees of freedom, available to the process engineer define what the ultimate limits are for the cleaning process. The traditional degrees of freedom available in an ultrasonic cleaning system have included modulation of a single center frequency (sweep), variable duty cycle, and amplitude control at a single frequency. All of these variables allow control of gross, or macroscopic, variables such as raw power into the fluid. The latest class of aqueous cleaning technology allows all of the fore mentioned parameters, but at multiple center frequencies in a single process tank. Multiple center frequencies allow precise microscopic tuning of the energy in the individual cavitation event. Understanding and optimizing the parameters yields maximal cleaning efficiency with minimal substrate damage. This paper intends the ambitious task of illuminating the various physical effects of these parameters in such a way as to shed as much light and as little heat as possible on this often less than intuitive subject. It is the authors' desire that this paper is not a final description, but the beginning of a dialog.

The primary physical phenomenon behind the technology of ultrasound is an event known as cavitation. Indeed, any proper treatment of ultrasonic cleaning must begin with a discussion of cavitation. Cavitation is the creation and subsequent collapse of microscopic bubbles within a liquid. These bubbles are formed when a large pressure gradient is introduced into a fluid. During the low-pressure part of a sound wave the fluid is put into tension. When the amplitude of this sound wave exceeds the local tensile strength of the fluid a void, or cavity, is created in the medium. This cavity grows for the rest of the half cycle of sound. As this bubble grows, both dissolved gasses and fluid vapor diffuse through the walls of the cavity and into the bubble via a process known as rectified diffusion. As the pressure associated with the sound wave begins to go positive one of two things can happen. The bubble, which grew to a certain size R0 during the half cycle, can collapse partially. In this case the entrained gasses act as a shock absorber and the bubble may undergo further stable oscillations. The second possibility is that the bubble can suffer complete implosion, an event labeled transient cavitation. Both of these processes re-radiate absorbed energy from the incident acoustic field. It is this re-radiated energy that impinges upon a substrate and does the bulk of the cleaning. Cavitation is one of nature's most efficient and dramatic amplifiers of energy density currently known. Each bubble collapse is accompanied by the local generation of temperatures on the order of thousands of degrees centigrade and pressures exceeding hundreds of atmospheres. Though recognized for almost a century , physicists have yet to construct a complete description of this final collapse. Although much of the final implosion event is shrouded in mystery, the bubble dynamics prior to this is quite well understood. , ,

Armed with this phenomenological explanation of cavitation we can begin to examine the effects of the various ultrasonic parameters. The relevant questions to be asked are, "How does the generator impart energy to the transducers, how do the transducers impart that energy to the cavitating cleaning solution and then how is the sonic energy exciting the part being cleaned?" The cleaning system is composed of both the cleaning solution and the part or parts being cleaned.

Advances in Particle Removal Technology

Progression of integrated circuit technology requires smaller particles to be removed from wafers surfaces. As linewidths decrease, particles specs tighten and cleaning of wafers becomes more complicated and more critical. This paper will review the upcoming particle removal challenges and give an overview of various new methods to remove small particles from the surface. Addressed will be some of the issues with these new technologies.

Particle removal methodology evolves around various techniques: removal based on undercutting the particle, particle release by changing the Zeta potential, physical bombardment to shock the particle off the substrate, and physical removal based on brush cleaning where the particle is wiped off the surface.

These methods must not affect the underlying material on which the particle resides. Oxidation and corrosion of the copper or other metals, roughening the silicon or oxide surface, and modification of the films due to the particle removal process must be eliminated or not allowed to occur. Cleaning high aspect ratio vias and contact are additional obstacles.

Some of the promising new dry cleaning technologies for particle removal are: cryogenic cleaning with argon or CO2 snow, laser cleaning, supercritical CO2 cleaning using surfactants, and water cluster bombardment. Wet cleaning techniques are evolving: advanced brush cleaning materials and chemistries, the use of non-damaging megasonics, and methods to reduce the surface tension of water such as pressure pulsations and foam cleaning, and the addition of surfactants. Particles removal is becoming increasingly important in integrated circuit manufacturing. Future requirements for particle removal are now being investigated, for example the need for particle removal for the next generation lithography (NGL) is not well understood.

1) Towson University, Towson, MD 21252

2) University of Rochester, Rochester, NY 14627-0132

3) NexPress Solutions LLC, Rochester, NY 14653-6402

Engineering Aspects of Particle Adhesion: Applications to Surface Cleaning

A fundamental knowledge of particle adhesion is necessary to address a broad range of technical problems, particularly those associated with manufacturing of miniature precision components. As mechanical and electrical components reduce in size, the control of micrometer and nanometer particles becomes vital. This includes the controlled placement of components, the minimization or elimination of particles (cleaning), or a combination of both. The following talk will provide a summary of particle adhesion from the perspective of engineering micro devices. Initial discussion will focus on basic mechanisms and theories of particles adhesion, with a short description of theories for deforming systems and irregular particles. A discussion of experimental investigations and characterization of particle adhesion will then be presented. Topics such as detachment force measurement techniques, effects of time, RH, and particle size will be addressed. Finally, a brief discussion of surface roughness effects, impact effects, and cohesive failure will be included.

Surface Contamination by Alumina Particles

Micron and submicron alumina particles are used for the mechanical polishing of the GaAs wafers processed in the microelectronic industry. A better understanding of the detachment mechanism of the alumina particles resting on surface is crucial for the optimisation of the industrial chemical cleaning. Indeed, the nature and the strength of the complex interactions between asymmetrical alumina particles and the surface remains unclear. To this end, a shear stress flow chamber was used to study the detachment of alumina particles in adhesive contact with a glass plate. A series of experiments was performed to measure the wall shear stress of the laminar flow necessary to detach individual micron alumina particles of 3 and 0.3µm with various chemical solutions (using diluted ammonia, surfactant and glycerol), pH and resting time. It was found that the longer the resting time, the more adherent the particles are. Moreover, the detachment was highly facilitated as the pH was basic, as suggested by the DLVO theory.

Removal of Ink Particles from Ceramic Anilox Rolls Used in Flexographic Printing

As the cleaning industry has moved from solvents to water-based cleaners, the printing industry has moved from solvent-based to water-based inks. Within the printing industry, this has led to the growth of Flexographic printing. This is used on labels, decals, flexible packaging, corrugated boxes, balloons, and so much more. The essential part of Flexographic printing is the laser engraved ceramic Anilox rolls which are used to pick up and deposit the dots of ink on the image plate before placement on the various substrates. These expensive surfaces become imbedded with ink residues over time that are very difficult to remove.

This paper will discuss the current cleaning procedures used in the printing industry as well as propose a better way to avoid having to clean these surfaces and, thereby, increase the life of these Anilox rolls. Many printers are still using oven cleaner and other corrosive chemistries to clean a highly specialized surface. Excessive use of Ultrasonics as well as blasting can also have detrimental effects, especially as the cells on the surface get smaller and smaller to achieve high definition printing.

Possible Post CMP Cleaning Processes for STI Ceria Slurries

CMP is an established semiconductor process step in the integrated production of logic and memory devices. DI water is a critical component in these processes. A typical fab consumes 240 million¹ gallons of water for both BEOL and FEOL processes (~1500 gals/200 mm wafer) and the CMP process accounts for about 5-7% of the total water. There is the critical need to eliminate particle and metal ion contamination² while attempting to reduce water usage levels, especially at the gate oxide structure.

The STI (Shallow Trench Insulator) structures involve polishing processes to planarize CVD silicon oxide films³ as part of the gate oxide structure. Both silica⁴ and ceria-type^5,6 slurries have been used for this process. Pourbaix diagrams for cerium oxide, between pH 2 to 7 indicate that Ce+3 and +4 can be in equilibrium with the solid abrasive. There has been some concern that ceria ions, besides other metal ions, will be absorbed onto the very sensitive STI structure. It is critical that an aggressive post CMP cleaning process be developed to remove trace metal and particulate contamination without damaging the STI structure. A sulfuric acid/H₂O₂ or peroxide-only solutions has been used with only partial success for removing particles. The use of NH₄OH, dilute SC1 or dilute HF has been recently reported⁷ with marginal success. The use of the HF cleaning chemistries pose problems of surface etching which can "decorate" the STI

structure and influence device yield.

VPD-ICP-MS data in figure 1 shows that Buffered Chelating solutions (BCS) at pH 4.2 for post CMP cleaning procedures on a single wafer spray tool⁸ can reduce the ceria ionic contamination levels (99.8%) from 9E10 to 2E8 atoms/cm². A DI water only rinse could only reduce the ceria ion level 95% (to 5E9 atoms/cm²). At pH 4.2 the best overall procedures were either the BCS or the BCS/peroxide using both the rollers (R) and megasonic (M) units. Though the peroxide-only chemistry (P/R&M) was effective for lowering ceria ions, it is not as effective for the other metals. The dilute SC1 results⁷ were only able to reduce the residual Ce+4 to ~5E11 atoms/cm².

Table 1 SP 1 Defectivity Data

Table 1 shows preliminary TEOS defectivity data (particles and scratches) for the CeO₂ slurries with different chemistries at pH values between <1 to 8.5.

This work will discuss results for eight different chemistries, ranging from pH<1 to ~10 with both oxidizing and reducing potentials removing metal ions and particles. The effect of megasonics and brush cleaning for metal ion and particle contamination will be examined.

References:

1. Corlett, G., "CMP Technology for ULSI Interconnection"; SEMI p. 81, 1998

2. Lester, M., "Future Cleaning Role of DI Water;" Semiconductor International p. 26, March, 2002.

3. Zhao, E., Xu, C. S., "Direct CMP for STI;" Semiconductor International p. 145, June 2001.

4. Lee, S. et al, "The Effects of Slurries with Pattern Size and Step Height in Shallow Trench isolation Chemical Mechical polishing," 2000 CMP-MIC Confer. p. 163, Santa Clara, CA.

5. Bonner, B. et al, "Development of a Direct Polish Process for Shallow Trench Isolation Modules," 2001 CMP-MIC Conf. p. 572, Santa Clara, CA.

6. Leduc, P. et al, "CMP: Aiming for Perfect Planarization," 2002 CMP-MIC Confer. p. 239, Santa Clara, CA.

7. Li, H et. al.; "Post-Cleaning of High Selectivity Slurry for STI CMP," 2002 CMP-MIC Confer. p. 359, Santa Clara, CA.

8. Small, R. Scott, B, "Post CMP Cleaning for STI Ceria Slurries," 21^th SPWCC, p.171, Santa Clara, CA 2002

Figure 1. Cesium Ionic Contamination

Mechanics of Nanoparticle Adhesion

The fundamentals of particle adhesion are presented as suitable combination of nanomechanics and continuum mechanics: The models for elastic, adhesive, dissipative, plastic, elastic-plastic and viscoplastic contact response of normal loaded, isotropic, smooth spheres are briefly discussed.

The decreasing contact stiffness with decreasing particle size is the major reason for adhesion effects in nanoscale. By means of the model "stiff particles with soft contacts", the combined influence of elastic-plastic and viscoplastic repulsion in particle contacts is shown. The attractive particle adhesion term is described by a sphere-sphere-model without any contact deformation plus a plate-plate-model for contact flattening. Now, the contact deformation path for loading, unloading, reloading or contact detachment is discussed. Thus, the adhesion level between particle and surface contacts depends directly on the consolidation history.

For colliding particles the correlation between particle impact velocity and contact deformation response is obtained by means of energy balance.

Finally, the combined elastic-plastic, dissipative, adhesive particle contact displacement with time dependent viscoelastic relaxation and strain rate dependent viscoplastic consolidation is explained.

Advanced Wet Cleaning of Sub-micron Sized Particles

(Abstract not yet available)

Particle Transport and Adhesion in an Ultra-Clean Ion-Beam Coating Process*

Extreme Ultraviolet Lithography (EUVL) is the expected lithographic technology for the IC industry for the 45nm critical-feature-size node and beyond. EUVL is analogous to current production optical lithography, but uses light at = 13.4nm (in the soft x-ray or extreme ultraviolet regime). All the optics, including the mask are reflective, and reflectivities near 70% are achieved with a 0.28 m thickness Mo/Si multilayer coating. A key remaining challenge is final development of the low-defect coating process for the mask. The coating for the mask is deposited by ion-beam sputtering at P ~ 1´10^-4torr. The multilayer coating is highly conformal, and defects in the final mask blank can result from particles arriving during multilayer coating or from replication of substrate defects. Estimates of the critical printable defect size in the multilayer are 30-50nm physical size, depending on proximity to features. The current mask coating process can achieve 0.04 defects/cm² (10 added defects 80nm or larger on a 200mm Si-wafer test substrate), but this must be reduced by about 5X to meet mask cost requirements for EUVL.

In order to understand the sources of particle defects on the masks, transport and adhesion of particles in ion-beam coating conditions have been studied, using test particles and particles native to the coating process. At process pressure, gas drag is negligible for particles above 100nm, so particles travel ballistically until they hit a surface. Bounce from chamber walls allows particles to reach all surfaces in the chamber for initial velocities above ~100m/s. The ion beam has sufficient momentum to entrain slower particles and accelerate them toward the sputter target, where some can bounce to the substrate. Results on dominant particle sources and transport paths will be presented, with an assessment of best methods of improving process cleanliness for ion-beam sputtering.

*(Work performed under the auspices of the U. S. Department of Energy by the Lawrence Livermore National Laboratory under Contract No. W-7405-ENG-48. Funding was provided by the Extreme Ultraviolet Limited Liability Corporation under a Cooperative Research and Development Agreement.)

P. Mertens¹, M. M. Heyns¹ and C. Vinckier²

1) IMEC vzw, Kapeldreef 75, B-3001 Leuven, BELGIUM

2) Chem. Dept. of KULeuven, Celestijnenlaan 200F, B-3001 Leuven, BELGIUM

Nano-sized Particle Deposition and the Correlation Between Haze and Particle-count on Wafers

Surface cleaning mechanisms have been successfully used for integrated circuit manufacturing for more than 30 years. As the critical dimensions of the devices are continuously scaled down, new targets for cleaning should be set. For the upcoming sub100nm technology node it is required to remove particles on the order of a few tens of nano-meters. By using state of the art light scattering equipment, the current limitation to measure particles amounts to sizes of 50nm for wafers with the extremely low surface roughness. In order to measure even smaller particles, haze measurements can be used if a high density of particles is present on the wafer surface. For instance, this is especially important because in most chemical mechanical polishing applications sub-micron sized particles are being used. This report describes the deposition of SiO₂ slurry particles with different sizes on two different substrates (nitride and silicon - O3 last) with individual zeta-potentials. The particle number deposited on substrates appears to be proportional to the concentration of particles in the contamination bath. Two models are developed to estimate the number of particles on substrates after contamination. The first model assumes that the size of the particles deposited on the wafer surfaces follows a Gaussian distribution function. By measuring the tail of the particle size distribution, the total amount of particles deposited can be calculated. In a second approach, the number of particles deposited is calculated using the particle diffusion ability in the liquid. The correlation between the added haze of high particle-density wafers and the number of particles deposited on the wafer will be demonstrated. Since the added haze was found to be proportional to the number of particles deposited on substrates, it can be used to estimate the particle-count instead of measuring the particles as individual light point defects. Finally, the size-reliability of particle counts determined by using the haze signal is investigated, which leads to the formulation of specifications on the size distribution of particles to be used with this method.