Laser removal of particles from solid surfaces

Transcription

1 RIKEN Review No. 32 (January, 2001): Focused on Laser Precision Microfabrication (LPM2000) Laser removal of particles from solid surfaces Y. F. Lu, W. D. Song, Y. Zhang, M. H. Hong, and T. C. Chong Laser Microprocessing Laboratory, Department of Electrical Engineering and Data Storage Institute, National University of Singapore, Singapore Laser removal of particles from solid surfaces was investigated both theoretically and experimentally. A cleaning model was established for laser-induced removal of particles from solid surfaces by taking Van der Waals force, capillary force and cleaning force into account. Laser cleaning forces are induced by fast thermal expansion of particles and/or solid surfaces irradiated by laser for dry laser cleaning and evaporating liquid film heated by laser irradiation for steam laser cleaning, respectively. It was found that cleaning efficiency depends on laser fluence, pulse number, wavelength, incident direction and liquid properties. Cleaning tesholds can be obtained by comparing cleaning force and adhesion force. The experimental results are good consistent with theoretical analysis. 1. Introduction As semiconductor and microelectronics device fabrication technology advances toward higher densities and smaller circuit dimensions, contamination control becomes one of the most critical problems in semiconductor and microelectronics industry. Similarly, in the disk drive industry, the distance between read-write head and disk surface has been continually reduced to increase the recording density. This implies that even submicron particles on slider or disk surfaces can damage both the slider and disk surfaces and hence lead to failure of the disk drive system. Thus, the cleanliness of solid surfaces becomes very critical. Consequently, there have been significant efforts to develop effective techniques to remove surface contaminants. These include high-pressure jet of a gas or a liquid or a gas doped with bits of ice or dry ice; mechanical wiping and scrubbing; etching in a gas, plasma or liquid; ultrasonic and megasonic cleaning; and laser cleaning. Some of them such as ultrasonic and megasonic cleaning require the immersion of a sample into a liquid bath, which has a number of serious drawbacks. First, it is widely known that wet techniques could add contaminants due to insufficient cleaning and filtering of the liquid at the submicron level. Second, the bulk usage of hazardous chemicals and solvents such as trichloroethylene and CFC s (carbon fluorochloride) becomes undesirable for environmental and industrial reasons such as causing cancers in humans and depleting ozone layer. Other problems associated with the wet techniques are rinsing/drying difficulties and incompatibility with other processes. Hence, dry cleaning techniques such as laser cleaning have emerged in order to overcome these drawbacks. Recently, two types of laser cleaning technique have been reported in the literature, 1 18) relying on pulsed laser heating of the surface without or with the presence of a thin liquid coating. We shall refer to these two types as dry laser cleaning and steam laser cleaning, respectively. For dry laser cleaning, particles can be ejected from particulatecontaminated surfaces by short-pulse laser irradiation. The proposed mechanism of ejection is fast thermal expansion of address: eleluyf@nus.edu.sg the particle and/or surface. For steam laser cleaning, the proposed mechanism is supposed to be the momentum transfer from the laser-heated and suddenly evaporating liquid film to the particles on the substrate. In this paper, we presented a cleaning model for laser-induced removal of particles from solid surfaces without or with the presence of a thin liquid coating by taking Van der Waals force, capillary force and cleaning force into account. This model cannot only explain the influence of incident direction, wavelength, laser fluence and liquid properties on cleaning efficiency, but also predict cleaning tesholds. 7,14 18) 2. Laser cleaning model 2.1 Adhesion forces Several important forces exist and can cause strong adhesion of a tiny particle on a solid surface. 1) Van der Waals force is typically the predominant adhesion force for particles less than a few microns in size. 1) It includes forces between molecules possessing dipoles and quadrapoles caused by the polarizations of the atoms and molecules in the material. This can include both natural as well as induced instantaneous dipoles and quadrapoles. The attractive Van der Waals force between an ideal spherical particle and a flat solid surface is given by 19,20) F 1 = 8πz, (1) 2 where r, h, and z is the particle radius, the material-dependent Liftshitz-Van der Waals constant and the atomic separation between the particle and surface. Typical values of h range from a low value of 0.6 ev for polymer/polymer interaction to a large value of 9.0 ev for gold/gold interaction. For Van der Waals-bonded crystals, the atomic separation z is approximately 4 angstroms. 19,20) This Van der Waals force can cause most of particles and/or surfaces to be deformed as a result. The amount of deformation will depend on the hardness of the particles and surfaces. The additional Van der Waals force due to deformation, F 2, is a function of the increased contact area caused by the deformation and is given as follow 19) 64

2 F 2 = hδ2 (2) 8πz 3 where δ is the radius of the adhesion surface area. Therefore, the total Van der Waals force between an actual spherical particle and a flat surface, F ν, is the sum of the Van der Waals force between an ideal spherical particle and a flat surface and the additional Van der Waals force due to deformation, that is given by F ν = F 1 + F 2 = 8πz 2 + hδ2 8πz 3. (3) For large deformation, such as δ/r 10%, the Van der Waals force between the ideal spherical particle and a flat surface is much less than the additional Van der Waals force. Hence, we approximate the Van der Waals adhesion force per unit area as 1) P = Fν πδ h 2 8π 2 z. (4) 3 For a plate-type particle on a flat solid surface, the Van der Waals force per unit area is given by 19) P = h 8π 2 z 3. (5) Due to high humidity or to an adhered particle/surface system having been immersed and then withdrawn from a liquid, a layer of liquid will be formed on the surface by capillary condensation or capillary action. This resulting capillary force is a function of the particle radius and liquid surface tension as shown by the following formula: F c =4πγr, (6) where γ is the liquid surface tension and r is the radius of the particle. 19) For a dry system, a capillary force will not be taken into the consideration while calculating the adhesion force of particles on substrate because there is no liquid layer between the particle and substrate (assuming negligible atmospheric moisture). In a dry system, Van der Waals force predominates for tiny particles with a particle size less than a few microns. 19) In semiconductor, disk drive and other microelectronic industries, particles are usually much smaller than 50 µm, Van der Waals force is dominant adhesion force for these tiny particles on dry solid surfaces. Therefore, the adhesion force between a tiny particle and a solid surface F is F = F ν = 8πz 2 + hδ2 8πz 3. (7) For a wet system, a capillary force and Van der Waals force will act to hold the particles. The adhesion force between a tiny particle and a solid surface F is F = F ν + F c = 8πz + hδ2 +4πγr. (8) 2 8πz3 2.2 Temperature distribution Since cleaning forces induced by laser irradiation, discussed in the following, depend on temperature distribution in a substrate or a particle, it is needed to know the temperature distribution in order to calculate the cleaning forces. When a pulsed laser irradiates a substrate surface, the temperature distribution in the substrate can be described by the onedimensional heat equation. The temperature at any point inside the substrate T (x, t) is a function of depth below the substrate surface x and time t, and is governed by the onedimensional heat equation 21,22) T(x, t) ρc = k 2 T (x, t) +(1 R)αI t x 2 0 exp( αx), (9) where ρ, c, K, R, α, and I 0 are density, specific heat, thermal conductivity, reflectivity, absorption coefficient of the substrate material and laser intensity on the substrate surface, respectively. 2.3 Dry cleaning force Due to the short pulse laser irradiation, there is a rapid increased in temperature of the particles and/or substrate. This increased in temperature will indirectly generate a cleaning force opposed the adhesion force. And it is this cleaning force that causes cleaning phenomena, ejecting of particles from the substrate surface. For transparent particles on an absorbing substrate, majority of the energy of the laser beam is absorbed only by the substrate, which results in temperature rise in the substrate. When the substrate surface under the particle experiences a fast expansion due to its temperature rise induced by pulsed laser irradiation, the expansion will be restrained by the particle. Therefore, an action force per unit area to the particle due to the thermal expansion of the substrate surface is produced and given by γe T (0,t), which is defined to be the cleaning force per unit area f 1. That is f 1 = γe T (0,t), (10) where γ and E are the linear thermal expansion coefficient and the elastic modulus of the substrate, respectively. The T (0,t) is the temperature rise at the substrate surface and time t, which is given by T (0,t)=T (0,t) T 0, where T 0 is the initial temperature at substrate surface. If the cleaning force (per unit area) exceeds the adhesion force (per unit area), the particle may be detached from the substrate surface. Therefore, the cleaning condition will be f 1 >P and the cleaning teshold fluence can be obtained by the following equation f 1 = P. For a transparent substrate, only particles are absorbing the energy from laser beam. Similarly, the temperature rise in the particles will also result in thermal expansion restrained by the substrate below it. Therefore, an action force to the substrate surface due to the thermal expansion of the particles is produced. Meanwhile, the substrate surface impose a reacting force to the particles given by γe T (d, t), where γ and E are the linear thermal expansion coefficient and the elastic modulus of the particles, respectively. The T (d, t) is the temperature rise at the interface between the particle and substrate. This force can be also derived by the following discussion. 65

3 When a pulsed laser irradiates a particle, the particle absorbs laser energy resulting in a non-uniform temperature rise in the particle. The change in temperature causes the particle to expand, which produces a thermal stress in the particle. If the particle is to be detached from the substrate surface, the particle surface at the interface between the particle and the substrate must experience a real displacement. According to the relationship between stresses and strains, 23) we can obtain that the cleaning condition is σ(d, t)/e + γ T (d, t) =ε(d, t) > 0, (11) where σ(d, t) and ε(d, t) are the thermal stress and displacement strain of the particle surface at the particle-substrate interface and time t, respectively. The T (d, t) is the temperature rise of the particle at the interface and is given by T (d, t) =T (d, t) T 0, (12) where T 0 is the initial temperature distribution of the particle at the interface. According to thermal stress analysis in the related literature, 23) the thermal stress is zero for a bar if it is unrestrained. γe T is for a bar completely restrained. The negative sign indicates a compressive stress resisting expansion of the bar. For a spring restraint, the magnitude of thermal stress times the cross section is equal to the magnitude of elastic force. Since the particle is restrained by the Van der Waals force being similar with the elastic force, the thermal stress σ(d, t) is given by 23) σ(d, t) = P, (13) where the negative sign indicates a compressive stress resisting expansion of the particle. Substitute the Eq. (13) into Eq. (11), we can get γe T (d, t) >P. (14) We define the cleaning force per unit area f 2 as f 2 = γe T (d, t). (15) Therefore, the cleaning condition will be f 2 > P and the cleaning teshold fluence can also be obtained by the following equation f 2 = P. 2.4 Steam cleaning force For laser with enough intensity irradiating on a solid surface coated by non-absorption liquid film, a sheet of liquid near the liquid/substrate interface can be superheated tough thermal diffusion. The growth of vapor bubbles on the interface can be generally observed over a range of temperature. 24,25) During the process of bubble growth, the pressure and temperature inside the bubble will lie in the ranges: 25) P P v P sat(t ) T sat(p ) T v T, (16) where P, P v, P sat, T sat, T v, T are the ambient liquid pressure, vapor pressure inside the bubble, saturation vapor pressure, saturation temperature, temperature inside the bubble and ambient liquid temperature, respectively. The process of bubble growth is divided into two regimes: inertial-controlled growth and heat-transfer-controlled growth. In cases of quickly imposed heat flux by laser irradiation and highly wetting liquid, inertial-controlled growth is more likely to occur. In this regime T v = T, and with the constraint of local thermodynamic equilibrium, P v = P sat(t v) = P sat(t ). The upper limit for velocity of bubble growth in inertialcontrolled process has been theoretically described as below: 25,26) ν(t )= 2 3 1/2 P ν(t ) P, (17) ρ l (T ) where ν(t ) and ρ l (T ) are the velocity of bubble growth, liquid density at temperature of T. P v and P are defined as before. The size of bubbles cannot be larger than the thickness of the superheated liquid. It has been known that coating of a thin water layer on metal surface can enhance the laser-generated pressure of stress wave by up to 20 db. 27) The enhancement of pressure has been attributed to plasma formation, rapid evaporation and bubble expansion. The plasma formation occurs only for high-intensity laser irradiation due to optical breakdown. For laser intensity below the breakdown teshold, the pressure of stress wave is dominated by the rapid evaporation and bubble expansion. Based on the results of other studies, 28) we can deduce the pressure of stress wave by assuming that: (1) in the region near liquid/substrate interface, the vapor layer created by evaporation of liquid acts as a plane piston, compressing its adjacent liquid layer and generating stress wave, although the value of volume fraction of vapor inside the superheated liquid layer is less than 1; 29) (2) due to the non-uniform temperature distribution in the liquid film, the vapor pressure inside bubbles is approximately considered to be the average saturation vapor pressure of the superheated layer; (3) the expansion velocity of vapor layer is equal to the growth velocity of the bubbles which is shown in Eq. (17). Then the average energy on unit area vapor/liquid interface obtained tough the expansion of vapor layer is R (P ν P )νfdt, where ν and f are expansion velocity and volume fraction of vapor. For the generated stress wave, its energy per unit area is calculated by: 30) Z P 2 E = dt, (18) 2ρc where ρ, c, and P are the liquid density, transmit speed and pressure of stress wave. According to energy conservation, the pressure of stress wave at the vapor/liquid interface is: P =[2ρc(P v P )νf] 1/2. (19) The cleaning force caused by this pressure on adhesion particles with radius of r can be obtained by Eq. (19). The teshold of laser cleaning is defined as the laser fluence at which the cleaning force is equal to the adhesion force. 66

4 Fig. 2. The relationship between the cleaning force induced by laser irradiation at 100 mj/cm2 and diameters of Al particles. Fig. 1. The optical micrographs of quartz surfaces with Al particles before and after laser cleaning both from the front-side (a) and reverseside (b) at a laser fluence of 100 mj/cm2, a pulse number of 100 and a repetition rate of 10 Hz. 3. Key factors in laser cleaning 3.1 Incident direction Since quartz is transparent to excimer laser radiation, a quartz substrate with Al particles can be cleaned by excimer laser irradiation both from the front-side or reverse-side of the sample. Therefore, we can investigate the influence of laser incident direction on cleaning efficiency using quartz as a substrate. Figure 1 shows the optical micrographs of the quartz surfaces with Al particles before and after laser cleaning both from the front-side and reverse-side at a laser fluence of 100 mj/cm2, a pulse number of 100 and a repetition rate of 10 Hz. The cleaning efficiency is only about 24% for laser irradiation from the front-side as shown in Fig. 1 (a). However, it becomes 100% for laser irradiation from the reverse-side as shown in Fig. 1 (b). It is apparent that laser cleaning from the reverse-side is more effective to remove the Al particles than that from the front-side. Based on above model, the relationship between the cleaning force induced by laser irradiation at 100 mj/cm2 and diameters of Al particles is shown in Fig. 2. Comparing the cleaning force induced by both the reverse-side and front-side irradiation, it is clear that the cleaning force induced by reverse-side irradiation is larger than that induced by front-side irradiation under the same fluence, which leads to higher cleaning efficiency for reverse-side irradiation. 3.2 Laser wavelength Comparison of laser cleaning efficiency for tee different wavelengths at 500 pulses and 10 Hz for removal of copper particles from quartz surfaces is shown in Fig. 3. From Fig. 3 (a), it was observed that the cleaning efficiency increases with increasing laser fluence for all tee different wavelengths and is higher for the shorter wavelength at a particular laser fluence for removing copper particles from quartz surfaces. The cleaning tesholds for removal of copper particles from quartz surfaces are 80 mj/cm2 at 1,064 nm, 40 mj/cm2 at 532 nm, and 20 mj/cm2 at 266 nm, respectively. These show Fig. 3. Laser cleaning efficiency for tee different wavelengths at 500 pulses and 10 Hz for removal of copper particles from quartz surfaces (a) and the peak cleaning force per unit area of copper particles on a quartz substrate as a function of laser fluence at different wavelengths of 1,064 nm, 532 nm, and 266 nm (b). 67

5 that the cleaning tesholds become smaller for the shorter wavelength. Above wavelength effects can be explained by following theoretical analysis. The peak cleaning force per unit area of copper particles on a quartz substrate as a function of laser fluence at different wavelengths of 1,064 nm, 532 nm, and 266 nm is shown in Fig. 3 (b). From Fig. 3 (b), it was found that for all tee different laser wavelengths, as laser fluences increases, the peak cleaning force per unit area increases, too. As the peak cleaning force increases, it is easier to overcome the adhesion force between the particles and substrate. Therefore, it is expected that the cleaning efficiency will increase with increasing laser fluence. Due to the fact that shorter laser wavelength can generate greater change in temperature which result in larger cleaning force, hence leads to higher cleaning efficiency. By plotting the peak cleaning force per unit area and the adhesion force per unit area as a function of laser fluence, the laser fluence at the intersection of these two curves will be the point where the cleaning force is equal to the adhesion force. This will be the theoretical cleaning teshold. The results show that the cleaning tesholds for removal of copper particles from quartz substrates decrease with decreasing wavelength. This is because wavelength decreases, it is easier to bring a large temperature rise in the particle, which results in a generation of a greater cleaning force. The theoretical cleaning tesholds for removal of copper particles from quartz surfaces are 65 mj/cm 2 at 1,064 nm, 27 mj/cm 2 at 532 nm, and 23 mj/cm 2 at 266 nm, respectively. Comparing theoretical and experimental tesholds, the same trends for both theoretical and experimental cleaning tesholds were observed. The theoretical cleaning tesholds are quite close to those of the experimental cleaning tesholds. Thus it can be seen that above the theoretical analysis successfully predicts and explains the experimental results. 3.3 Laser fluence Figure 4 shows the peaking cleaning force and cleaning efficiency at different laser fluences for dry laser cleaning. It indicates that peaking cleaning force increases with an increase of laser fluence, which results in easier removal of particles from substrates or higher cleaning efficiency at higher fluence. The theoretical cleaning teshold predicted by the cleaning models is 10 mj/cm 2. The experimental teshold as shown in Fig. 4 (b)is about 16 mj/cm 2, which is close to the theoretical teshold. The cleaning efficiency increases with an increase of laser fluence. Therefore, the experimental results can be explained and predicted by theoretical analyses. Figure 5 (a) shows the cleaning force at different laser fluences together with the adhesion force for steam laser cleaning. It indicates that the cleaning force increases with an increase of laser fluence and theoretical cleaning teshold is 27.5 mj/cm 2 for removal of 1 µm Al particles from IPA coated NiP surface. Figure 5 (b) shows laser fluence dependence of cleaning efficiency for removal of 1 µm Al particles from NiP surface with deposition of the IPA film. It is found that the Fig. 4. The peaking cleaning force (a) and cleaning efficiency (b) at different laser fluences for dry laser cleaning. Fig. 5. The cleaning forces (a) and cleaning efficiency (b) as a function of laser fluence for steam laser cleaning. 68

6 Fig. 6. The optical micrographs of slider surfaces with particles before cleaning (a) and after cleaning with liquid thin films at (b) 20 pulses, (c) 40 pulses, and (d) 60 pulses. Fig. 7. The dependence of cleaning efficiency on pulse number at 100 mj/cm2 and 10 Hz for Al particles, and at 50 mj/cm2 and 10 Hz for Sn particles on slider surfaces. cleaning teshold is about 30 mj/cm2 and the cleaning efficiency increases rapidly with laser fluence, which are close to theoretical analyses as shown in Fig. 5 (a). 3.4 Pulse number For steam laser cleaning, it was found that cleaning efficiency depends on pulse number. Figure 6 shows optical micrographs of slider surfaces with particles before and after laser cleaning at a laser fluence of 130 mj/cm2 and different pulses. It was observed that cleaning efficiency increases with an increase of pulse number and particles on slider surfaces have been removed by laser irradiation after 60 pulses. Similarly, it was also found that laser cleaning efficiency increases with increasing pulse number for dry laser cleaning. The dependence of laser cleaning efficiency on pulse number at 100 mj/cm2 and 10 Hz for Al particles and at 50 mj/cm2 and 10 Hz for Sn particles on slider surfaces are shown in Fig. 7. From Fig. 7, it was found that high cleaning efficiency could be obtained by increasing pulse number at lower laser fluence to avoid causing damage to surfaces. For example, the laser cleaning efficiency for Sn particles at 200 mj/cm2 and 100 pulses is about 100%, but this condition causes dam- Fig. 8. Differences between cleaning force and adhesion force on alumina particles with a size of 1 µm (A) and laser fluence dependence of cleaning efficiency for removal of 1 µm alumina particles from silicon surface with an acetone and ethanol film (B). age to slider surfaces. However, we can obtain the same cleaning efficiency at 100 mj/cm2 and 2,000 pulses without any damage. 3.5 Liquid films Laser removal of alumina particles from silicon surfaces with deposition of an acetone or ethanol film is performed. The cleaning behaviors are theoretically described in comparison with the experimental results. Figure 8 shows differences between cleaning force and adhesion force on alumina particles with a size of 1 µm and laser fluence dependence of cleaning efficiency for removal of 1 µm alumina particles from silicon surface with an acetone and ethanol film. It was found from Fig. 8 (A) that the cleaning tesholds of laser fluence are about 80 mj/cm2 with an acetone film and 90 mj/cm2 with an ethanol film. When the liquid film is acetone, laser cleaning has smaller cleaning teshold for its lower boiling point and volume heat capacity. With deposition of ethanol, the cleaning teshold is large, but its difference between cleaning force and adhesion force increases more quickly along with the laser fluence as shown in Fig. 8 (A). This will lead to better cleaning efficiency at higher laser fluence. This phenomenon has been proved by the experiment as shown in Fig. 8 (B). 69

7 4. Conclusion In summary, dry and steam laser cleaning was demonstrated to be an efficient cleaning tool for removing particles from solid surfaces both theoretically and experimentally. A cleaning model was established for laser-induced removal of particles from solid surfaces by taking adhesion forces and cleaning force into account. Laser cleaning forces are induced by fast thermal expansion of particles and/or solid surfaces irradiated by laser for dry laser cleaning and evaporating liquid film heated by laser irradiation for steam laser cleaning, respectively. With increasing laser fluence or decreasing laser wavelength, the difference between the cleaning force and the adhesion force increases, resulting in easier removal of particles from substrate surfaces or a higher cleaning efficiency. Comparing theoretical and experimental tesholds, the same trends for both theoretical and experimental cleaning tesholds were observed. The theoretical cleaning tesholds are quite close to those of the experimental cleaning tesholds, which means that the theoretical analysis can successfully predict and explain the experimental results. The authors would like to thank K. D. Ye, C. K. Tee, D. M. Liu and Y. W. Goh of the Data Storage Institute of Singapore for their valuable participation and contributions to this work. References 1) W. Zapka, W. Ziemlich, and A. C. Tam: Appl. Phys. Lett. 58, 2217 (1991). 2) K. Imen, S. J. Lee, and S. D. Allen: Appl. Phys. Lett. 58, 203 (1991). 3) A. C. Tam, W. P. Leung, W. Zapka, and W. Ziemlich: J. Appl. Phys. 71, 3515 (1992). 4) J. D. Kelley and F. E. Hovis: Microelectron. Eng. 20, 159 (1993). 5) Y. F. Lu, M. Takai, S. Komuro, T. Shiokawa, and Y. Aoyagi: Appl. Phys. A 59, 281 (1994). 6) H. K. Park, C. P. Grigoropoulos, W. P. Leung, and A. C. Tam: IEEE Trans. Compon. Packag. Manu. Technol. Part A 17, 631 (1994). 7) Y. F. Lu, W. D. Song, M. H. Hong, B. S. Teo, T. C. Chong, and T. S. Low: J. Appl. Phys. 80, 499 (1996). 8) K. Mann, B. Wolff-Rottke, and F. Muller: Appl. Surf. Sci. 98, 463 (1996). 9) R. Oltra, O. Yavas, F. Cruz, J. P. Boquillon, and C. Sartori: Appl. Surf. Sci. 98, 484 (1996). 10) D. A. Wesner, M. Mertin, F. Lupp, and E. W. Kreutz: Appl. Surf. Sci. 98, 479 (1996). 11) M. Afif, J. P. Girardeau-Montaut, C. Tomas, M. Romamd, M. Charbonnier, N. S. Prakash, A. Perez, G. Marest, and J. M. Frigerio: Appl. Surf. Sci. 98, 469 (1996). 12) I. Gobernado-Mitre, J. Medina, B. Calvo, A. C. Prieto, L. A. Leal, B. Perez, F. Marcos, and A. M. de Frutos: Appl. Surf. Sci. 98, 474 (1996). 13) J. B. Heroux, S. Boughaba, I. Ressejac, E. Sacher, and M. Meunier: J. Appl. Phys. 79, 2857 (1996). 14) W. D. Song, Y. F. Lu, K. D. Ye, C. K. Tee, M. H. Hong, D. M. Liu, and T. S. Low: Proc. SPIE 3184, 158 (1997). 15) Y. F. Lu, W. D. Song, B. W. Ang, M. H. Hong, D. S. H. Chan, and T. S. Low: Appl. Phys. A 65, 9 (1997). 16) Y. F. Lu, W. D. Song, K. D. Ye, Y. P. Lee, D. S. H. Chan, and T. S. Low: Jpn. J. Appl. Phys. 36, L1304 (1997). 17) Y. F. Lu, W. D. Song, K. D. Ye, M. H. Hong, D. M. Liu, D. S. H.Chan, and T. S. Low: Appl. Surf. Sci. 120, 317 (1997). 18) Y. F. Lu, W. D. Song, C. K. Tee, D. S. H. Chan, and T. S. Low: Jpn. J. Appl. Phys. 37, 840 (1998). 19) K. L. Mittal: in Particles on Surfaces, Vol. 1 (Plenum Press, New York, 1988) p ) K. L. Mittal: in Particles on Surfaces (Marcel Dekker, New York, 1995) p. 1; ) D. Bhattacharya, R. K. Singh, and P. H. Holloway: J. Appl. Phys. 70, 5433 (1991). 22) J. R. Ho, C. P. Grigoropoulos, and J. A. C. Humpey: J. Appl. Phys. 78, 4696 (1995). 23) D. Burgreen: in Elements of Thermal Stress Analysis (C. P. Press, Jamaica, 1971) p ) W. Zapka, W, Ziemlich, W. P. leung, and A. C. Tam: Microelectron. Eng. 20, 171 (1993). 25) V. P. Carey: Liquid-Vapor Phase-Change Phenomena (Hemisphere Pub., Washington, D.C., 1992). 26) A. Prosperetti and M. S. Plesset: J. Fluid. Mech. 85, 349 (1978). 27) D. A. Hutchins, R. J. Dewhurst, and S. B. Palmer: Ultrasonics 19, 103 (1981). 28) C. J. Knight: AIAA J. 17, 519 (1979). 29) O. Yavas, A. Schilling, J. Bischof, J. Boneberg, and P. Leiderer: Appl. Phys. A 64, 331 (1997). 30) P. D. Edmonds and F. Dunn: Methods of Experimental Physics-Ultrasonics, Vol. 19 (Academic Press, New York, 1981). 70