R. P. Donovan Research Triangle Institute P.O. Box Research Triangle Park, NC

Transcription

1 SUBSTITUTING FOR CFCs USED AS CLEANING AGENTS: (1) PARTICLE REMOVAL R. P. Donovan Research Triangle Institute P.O. Box Research Triangle Park, NC Abstract Freon-TF is no better a cleaning fluid than deionized water for removing either organic (polystyrene latex spheres) or inorganic (glass beads) particles from silicon wafer surfaces, as shown by comparative measurements of cleaning efficiencies in both an ultrasonic bath and a hydrodynamic cleaning apparatus. A primary reason for this lackluster performance is the higher van der Waals forces of adhesion for particles on silicon when immersed in Freon-TF. In the ultrasonic bath this difference in van der Waals adhesion plus the lower cavitation intensity of Freon-TF causes particle removal efficiency to be lower in Freon-TF than in deionized water. In the hydrodynamic cleaning apparatus, the higher drag forces exerted on the particles by the Freon-TF compensate for the higher van der Waals adhesion so that the particle cleaning efficiencies in Freon-TF and deionized water are about the same. Particle removal efficiency is only one minor criterion by which chlorofluorocarbon (CFC) replacements will be selected. Safety and nonparticulate contaminant removal will often be more important. A first task of the Surface Cleaning Technology Consortium (SCTC), a multiclient cooperative research program now starting up at the Research Triangle Institute, is to provide guidelines for selecting suitable CFC substitutes based on the full spectrum of performance criteria.

2 b ' Introduction Chlorofluorocarbons (CFCs) are widely used as cleaning agents in electronics and other industries. Particle removal from surfaces, such as silicon wafers or other electronic parts, is a common cleaning step. CFCs have been used in this role because of their low toxicity and lack of flammability rather than any spectacular success in particle removal, as will be shown in this paper. CFCs also prove useful in removing organic films, flux residues, and other nonparticulate contaminants. This cleaning role will not be evaluated by the data presented here but will be part of the criteria used to select suitable CFC substitutes as the first task of the Surface Cleaning Technology Consortium (SCTC) now starting up at the Research Triangle Institute. Comparing Freon-TF (and Freon-TMS) with other cleaning solutions, chiefly deionized water, with respect to particle removal from silicon wafers, is the sole subject of the data presented in this paper. Advantages of Freon-TF Among the desirable properties of Freon-TF, over and above its safety advantages, are its high density and low viscosity when compared to water. In a hydrodynamic spray, for example, higher denstty means a higher drag force for removing particles (Musselman and Yarbrough, 1987): Fd N p?a Fd = drag force on a particle in a fluid stream where p = fluid density V = fluid veloc'tty at the particle A = projected frontal area of the particle. Lower viscosity means a reduced boundary layer thickness, 6, which in turn implies a higher fluid veloctty near the surface being cleaned and hence a higher fluid velocity at the particle (I/ in Eq. 1): 6 = boundary layer thickness 1 R = Reynolds number^ - e U where u = fluid kinematic viscosity a = a constant, o<a< 1 Thus, when compared to deionized water, Freon-TF possesses several physical properties that make it superior as a cleaning fluid for dislodging particles from surfaces. Disadvantages of Freon-TF Other factors need to be considered, however. For example, the force holding a particle to a surface is an equally important term in the cleaning equation, and on this score Freon-TF does not fare as well as a cleaning medium as deionized water. One basic force, applicable to all particlesurface systems, is the van der Waals dispersive interaction. At separation distances comparable to the dimensions of a particle, charge induction effects between neighboring particles (or a particle and a surface) create an attractive force between the two. For a sphere and a plane surface, this force of adhesion, Fad, can be calculated by (Ranade, 1987): (2)

3 ' - *132 dp 12 z2 0 Fad - - where A132 is the Hamaker constant of the system composed of a sphere (material 1) on a plane surface (material 2) in a medium (material 3) (Figure 1) d is the sphere diameter P Zo is the separation between the sphere and the plane. The composite Hamaker constant, A132, can be calculated from the Hamaker constants of the constituent materials as follows: (3) A132 = *12 -k A33 - A13 - A23 where A.. = (A..A..) 112 and 1 '1 u A.. = the Hamaker constant of material i. II Figure 1. van der Waals' dispersive interaction between a sphere (1) and a planar surface (2) through a medlum (3). Using Equations (3) to (5) and published values of the Hamaker constant for the materials making up various particle, surface, and fluid systems (Table 1) leads to the calculated values of AI *, given in Table 2, and to the adhegion forces listed in Table 3 which also includes corresponding measured values (Menon et al., 1988). For the calculations, particle diameter, d P' Table 1. Hamaker Constants of Selected Materials AI (picoergs) Silicon 2.48 Glass beads 0.34 Polystyrene Latex 0.65 Deionized Water 0.44 Ethanol-acetone (50% v/v) 0.33 Freon-TMS (TF + 6% methanol) 0.18 Freon-TF 0.17 Source: Menon et al., 1988.

4 Table 2. A for the Particle/Plane Systems Studied (picoergs) Deionized water Ethanol-acetone Freon-TMS Freon-TF Glass beads on silicon wafer with native oxide pm PSL spheres on silicon wafer with native oxide Table 3. Calculated and Experimental Values of Sphere Adhesion in Various Cleaning Fluids Deionized water Force of adhesion (millldynes) Ethanolacetone Freon-TMS Freon-TF - Meas. (Xi, 2 in Fig. 1) van der Waals' (ultrasonic adhesion apparatus) Calc. Meas. Calc. Meas. Calc. Meas. +. Glass beads on sX lo x lo silicon wafer with native oxide 1 pm polystyrene W latex spheres on silicon wafer with native oxide t Zo = 1.5 nm. Source: Menon et al has been set equal to 1 km and the separation distance, Zo, 1.5 nm. The experimental removal force was measured,ultrasonically, using appropriate calibration curves to relate power setting to particle acceleration. The removal force was taken to be that force required to remove 50% of the particles. Figure 2 shows that particle removal occurs over a wide range of forces, even though a specific value at 50% removal is identified in Table 3 as the measured removal force. The Table 3 data show that for either glass beads or polystyrene latex spheres on silicon both the calculated and the measured forces of adhesion are higher in Freon-TF than in water. On the basis of adhesion force, then, Freon-TF is not a preferred cleaning fluid for these particles. The calculated higher adhesion forces in Freon-TF follow from the higher values of A13*, as given in Table 2.

5 I I 1 u Removal Force (Relative Units) Figure 2. Typical variation of cleaning efficiency with ultrasonic force. Measured removal force is the value corresponding to 50% cleaning efficiency. Discussion While the absolute measured values of adhesion force in Table 3 do not match the calculated values at all well, the trends and relative orderings among the systems do agremhen the calculated values increase, so do the measured values. That the measured values of adhesion force are so much lower than the calculated values suggests that forces other than ultrasonic acceleration are also contributing to particle removal. Cavitation-the formation of voids (bubbles) within the sonicated liquid--may contribute to particle removal. As the voids pass the surface, they cause particles to be explosively injected into the void by liquid trapped between the particle and the surface. Cavitation intensity describes the ease of yoid formation and on this measure also Freon-TF performs less effectively than water as listed in Table 4. Thus, one reason for the high measured values of particle adhesion forces in Freon-TF may be the absence of significant cavitation effects in Freon-TF. Cavitation effects are not accounted for in the Table 4. Cavitation Intensities Relative to Water Liquid Relative cavitation intensity Water 100 calculated values presented in Ethanol-acetone 45 Table 3 so that measuring detach- Freon-TMS 17 ment forces under conditions that Freon-TF 15 allow significant cavitation cont r i- Source: Niemczewski, butions would be expected to produce significant departures from. the calculated van der Waals forces.

6 Implications Ultrasonics One conclusion of the data presented is that Freon-TF is not as effective a cleaning solution as deionized water for the ultrasonic removal of particles from a silicon wafer. The two major reasons for this poorer performance are: (1) the larger van der Waals forces characteristic of particles resting on silicon surfaces when immersed in Freon-TF as opposed to being immersed in deionized water; (2) the reduced cavitation intensity of Freon-TF compared to deionized water. The measured values of particle adhesion force presented in Table 3 establish this conclusion. Hydrodynamic Cleaning Ultrasonics does not take advantage of the superior drag properties of Freon-TF previously noted. In a hydrodynamic cleaning system the higher van der Waals adhesion forces of a Freon-TF system are compensated, at least partially, by the increased drag force of a Freon-TF stream compared to deionized water. In Figure 3 the drag force on a 1 pm polystyrene latex sphere in a Freon-TF flow has been set equal to twice that of a deionized water flow at the same nozzle pressure. Also indicated is the calculated van der Waals force of a 1 pm polystyrene latex sphere resting on a silicon wafer immersed in: (1) Freon-TF and (2) deionized water. The intersectian of the drag force and the van der Waals force curves indicates the threshold nozzle pressure required for particle removal for each cleaning fluid. For the values plotted the water system requires less nozzle pressure than the Freon-TF, but for different materials this conclusion could change and, in general, the operating.levels of the two cleaning fluids in a hydrodynamic spray are expected to be comparable. Figure 4 shows that the removal of 1 pm polystyrene latex spheres from a silicon wafer (native oxide only) has a similar nozzle pressure dependence for both Freon-TF and deionized water. + 1 / O-l / I- / /DI Water 0 I I I I Nozzle Pressure, pslg Figure 3. Drag forces in hydrodynamic cleaning. van der Waals adhesion force (Table 31 lndlcated for aaarooriate Iloulds.

7 Rgure 4. Cleaning efficiencies in a hydrodynamic apparatus using: (1) Freon TF; (2) DI water. Conclusions Replacing Freon-TF as a cleaning fluid for removing particles from surfaces is not a major problem as many alternative fluids, such as deionized water, already outperform or match Freon-TF in terms of efficiency of particle removal from a surface. Other considerations, such as nonparticulate film and residue removal, freedom from post cleaning rinsing/drying, and safety, are more important and more difficult properties to match and suitably replace. It is satisfying these other requirements that brings the serious challenges to identifying satisfactory CFC replacements. Epilogue To rapidly respond to the urgent needs of many industrial users to replace CFCs as cleaning fluids the first task of AT13 newly formed Surface Cleaning Technology Consortium (SCTC), a cooperative research, program funded by industrial sponsors, is the preparation of "Guidelines for Selecting Aqueous Surfactants or other CFC Substitutes." Under this task RTI will provide assistance in selecting appropriate aqueous surfactants or other cleaning solutions for removing specific types of contaminants from specific surface types. This task will begin with an exhaustive literature search, the results of which will be tabulated for easy reference and analyzed for broad general relationships. Gaps in the existing literature will be identified. An important use of this initial literature survey and analysis will be as a guide for prioritizing laboratory studies of the cleaning efficiency and compatibility of selected aqueous surfactants with respect to selected contaminated surfaces. This task is designed to produce early tangible returns to sponsors-the first quarterly report on this task should provide information of immediate value.

8 The first step in the evaluation will be to tabulate cleaning solutions and methods suitable for each of the surface types of interest to the sponsors. This initial assessment will be based on an exhaustive open literature search, seeking out published descriptions of the cleaning of each of the surface types. Conversely, a tabulation fo the types of surfaces cleaned by candidate CFC substitutes will also be prepared. From these two basic search strategies and from established surface cleaning principles, the goal is to deduce general guidelines for selecting suitable CFC-substitute cleaning solutions, matching surface physical/chemical properties to the cleaning solutions and cleaning techniques that either have produced or are expected to produce the highest quality surface cleaning. Perceiving from the published literature what types surfaces are best cleaned by what type cleaning solutions is a first step in judicious selection of a CFC-substitute. Where no direct match-up occurs and no clear prior record exists, similarities to other surfaces successfully cleaned will be sought. Ideally, broad classifications of surface types can be associated with specific CFC-substitutes so that initial choices of cleaning solutions are based on more than random selection. The literature search is just the first step in the preparation of a complete evaluation/selection matrix. The open literature will undoubtedly contain gaps wherein specific surfaces of interest have no documented cleaning history other than perhaps its use with a CFC. For selected surfaces of high interest to the sponsors RTI will experimentally measure the cleaning efficiency of various candidate CFC-substitute solutions, using such energizing techniques as ultrasonics, hydrodynamics and megasonics. The intent is to use commercially available equipment exclusively in these cleaning measurements so that results will be more easily reproduced by others. This task is part of a recently begun (October 1) broader program (the SCTC) whose longterm goal is to improve practical surface cleaning technology for its sponsors through better understanding of contaminant surface media interactions. Additional memberships are available and solicited. References Menon, V. B., L D. Michaels, L. A. Hollar, and R. P. Donovan Removal Efficiencies of Organic and Inorganic Particles from Silicon Wafers Proceedings of the 34th Annual Technical Mfg. of the Institute of Environmental Sciences, pp IES, 940 East Northwest Highway, Mount Prospect, IL Musselman, R. P. and T. W. Yarbrough Shear Stressm Cleaning for Surface Departiculation. J. fnviron. Sci., 30 (1):51-56, January/February. Niemczewski, A Comparison of Ultrasonic Cavitation Intensity in Liquids. Ultrasonics, May, pp Ranade, M Adhesion and Removal of Fine Particles on Surfaces. Aerosol Sci. Tech., 7(2):