Surface Chemical Analysis Using Scanning Probe Microscopy

STR/03/067/ST Surface Chemical Analysis Using Scanning Probe Microscopy A. L. K. Tan, Y. C. Liu, S. K. Tung and J. Wei Abstract - Since its introduction in 1986 as a tool for imaging and creating three-dimensional micrographs with resolution down to the nanometer and angstrom scales, the scanning probe microscope (SPM) has increasingly been acclaimed as a quantitative probe of surface forces such as adhesion. The SPM is able to study these important parameters using a technique that measures forces on the probe as it approaches and retracts from a surface. The technique of measuring the adhesion force between the probe and the different organic contaminants and fluxes is established. To ensure accuracy of the adhesion force measured, the spring constant of each cantilever used in the force measurement was accurately determined by measuring the exact dimensions of the cantilever by means of the scanning electron microscope and the modulus determined using the nano-indentation system. Keywords: Force measurement, SPM, Solvents, Fluxes 1 BACKGROUND Contaminants introduced during the manufacturing of many high-tech components, such as in the electronics packaging sector, may lower the quality, performance and productivity of the manufacturing process. Major contamination on device surfaces during the manufacturing processes can arise in processes such as soldering, cleaning and etc. For example, the detrimental effects of residues from solder fluxes that remain on the assemblies, due to the use of inefficient cleaning agents and methods, can be major problems. Cleaning media and application techniques, if not carefully chosen, used and controlled, may be the cause and origin of many different types of contamination. It is not unusual that cleaning agents and processes, while accomplishing the removal of some types of contaminants, can introduce other types at the same time [1]. Since its introduction in 1986 [2] as a tool for imaging and creating three-dimensional micrographs with resolution down to the nanometer and angstrom scales, the scanning probe microscope (SPM) has increasingly been acclaimed [3-4] as a quantitative probe of surface forces such as van der waals, capillary, electrostatic, capacitive, double layer, magnetic and adhesive forces. Microscopic adhesion affects a huge variety of events, from the behaviour of paints and glues, ceramics and composite materials, to deoxyribonucleic acid (DNA) replication and the action of drugs in the human body. The SPM is able to study these important parameters on the micron to nanometer scale using a technique that measures forces on the AFM probe as it approaches and retracts from a surface [4]. A better understanding of these forces has pushed the SPM from a simple high resolution profilometer to a tool for measuring and imaging a variety of sample properties [5]. SPM is thus a powerful tool for the examination of surfaces with atomic resolution. They therefore possess the capability to analyze monolayers of contaminant molecules or atoms which Electron Dispersive X-ray Spectroscopy (EDX), Fourier Transform Infrared Spectroscopy (FTIR) and the Raman imaging microscope fail to achieve. This type of analysis is presently performed by sophisticated and expensive instruments like Auger Electron Spectroscopy (AES), X-Ray Photoelectron Spectrometry (XPS) and Secondary Ion Mass Spectrometry (SIMS). It is therefore worthwhile to explore the capability of the SPM to do surface analysis. 2 OBJECTIVE The objective of the project was to explore and establish the capability of SPM in the analysis of surface chemical species to complement EDX, FTIR and Raman imaging microscopy for applications in Micro-mechanical Micro-Electro- Mechanical Systems (MEMS), Microsystems Technology (MST) and electronics packaging components. 3 METHODOLOGY 3.1 Deposition of contaminants The organic solvents deposited on the silicon substrate include electronic grade methanol, isopropanol, dichloromethane and trichloroethylene. Si (100) substrates were dipped into the solvent and were allowed to dry naturally. 1

The residue left behind by the solvent was analysed by the SPM. The aqueous-based solder paste (SP800) was obtained from Kester Solder. The paste was applied to the Si (100) substrate and heated up to ~220 o C on a hot plate for 20 to 30 s. The cooled paste was then removed by immersing in hot water at 70 o C. The residue left behind was analysed by the SPM. The organic-based soldering flux paste (TSF6522 No-clean tacky paste flux) was also obtained from Kester Solder. The content of the paste includes modified rosin, diethylene glycol dibutyl ether, petroleum distillate and organic thickener [6]. The paste was applied to the Si (100) substrate and heated up to ~220 o C on a hot plate for 20 to 30 s. The cooled paste was then removed by immersing and shaking in a Kester solvent (5252M). The residue left behind was analyzed by the SPM. 3.2 Measurement of spring constant of cantilever The dimensions of each cantilever (length, width and thickness) were measured by means of the scanning electron microscope (SEM). The modulus measurements were performed using a depth-sensing nano-indention system (Nanotest 550, Micro Materials Limited, Inc.) with load and depth resolutions of 0.05 µn and 0.04 nm respectively. A Berkovich indenter was used in all the experiments. The indenter was first loaded to a penetration depth of 150 nm at a rate of 0.2 mn/s and held there for 10 s before unloading. The Oliver and Pharr model [7] was used for the analysis and the modulus was obtained from the average of three measurements. Thus, the spring constant, k, for the thin beam cantilever is given by [8]: Ew t = 4 l 3 k (1) where E: Modulus of cantilever w: Width of cantilever t: Thickness of cantilever l: Length of cantilever 3.2 Measurement of adhesion force The SPM used was a Dimension 3000 scanning probe microscope of Digital Instruments, Inc., equipped with a G-type scanner. Data was collected in the force volume-imaging mode. In this mode, force curves were taken while the tip was raster scanned across the two-dimensional field of view and thus a two-dimensional array of force curve was acquired. The scan size of the sample was 5 x 5 µm 2 and 16 by 16 force curves were recorded. The number of data points per force curve was set at 512. The data was recorded in the trigger mode, which ensures that the tip is only pressed with a pre-set maximum force against the sample. The trigger threshold controlling this feature was typically set to 50 nm. Each result is obtained from the average of at least 50 force curves using at least three different tips. 4 RESULTS 4.1 Imaging using tapping mode Figs. 1(a) and (c) show the tapping mode height images of residues left behind by methanol deposited on Si wafer. Figs. 1(b) and (d) show the corresponding phase images respectively. The morphology of the methanol residues appears to be different when the deposition was conducted on two different occasions using the same method of preparation. Thus, the height and phase images of the methanol residues obtained by tapping mode can vary with the different occasions whereupon the deposition was conducted despite using the same method. Therefore, force measurements were conducted in an attempt to differentiate/characterise the residues of methanol, isopropanol, dichloromethane and trichloroethylene. 4.2 Spring constant of cantilever Since its invention, the AFM has become an increasingly important tool for studying surfaces at the atomic level. Beyond the informative images obtained for the surfaces, the AFM has the potential to give quantitative information about local forces and interactions. However, the accuracy of such measurements de- pends upon the knowledge of the physical properties of the spring and tip that probe these forces. In particular, the determination of the spring constant of the cantilever is essential in order to obtain accurate quantitative results. However, an accurate determination of the spring constant will 2

(a) (b) (c) (d) Fig. 1. Tapping mode height images (a,c) and phase images (b,d) of methanol deposited on Si substrate (5 µm scan). require an accurate determination of the dimensions and modulus of the cantilevers. Despite the need for calibration of individual tips and cantilevers, measurements of these properties have either been based on theoretical estimates or have been made only on supposedly representative samples. In the former case, a nominal figure is used to represent an average for a batch of micro-fabricated tips while in the latter, measurements are difficult and are likely to destroy the tip. Since the spring constant depends on the cantilever thickness as t 3 (equation (1)), even small variations in the thickness will result in large variations among the spring constants of nominally identical cantilevers. Similarly, the modulus of the cantilevers is rarely measured and they are often inferred from theoretical estimates based on the elastic properties of the cantilevers. Table 1 tabulates the dimensions of the cantilevers measured by SEM. The two types of tips include the force modulation etched silicon tips (F6, FM2 to FM10) and the magnetic etched silicon tips (M1 to M5). Table 1 shows that the dimensions of the cantilevers vary greatly among the cantilevers from the same batch (FM2 to FM10) and are different from the dimensions reported by the manufacturer. For example, the length and width of the cantilevers (FM2 to FM10) was reported to be 228 µm and 27.0 µm respectively but SEM measurements indicate that the length and width of the cantilevers can range from 218 to 228 µm and 30.0 to 31.7 µm respectively. Similarly, the thickness of 3

the tip was reported to be 2.80 µm for FM2 to FM7 and 2.90 µm for FM8 to FM10 but SEM measurements indicate that the thickness can range from 2.79 to 3.16 µm. Thus, an accurate determination of the dimensions of the cantilevers is essential and data reported by the manufacturer should only be served as a guide and should not be used in precise force measurements. Table 2 tabulates the modulus of cantilevers M1, F6, FM8 and FM9 measured using the nano-indentation system at different depths of indentation. It can be observed that the modulus of cantilever decreased when the depth of indentation increased. However, when the depth of indentation was increased to 300 nm for FM8, cracks can be detected. Thus, modulus measurements for all the cantilevers were conducted at a standardised depth of indentation of 150 nm whereby no cracks were detected for all the cantilevers. In addition, the modulus of cantilevers FM8 and FM9 (which belong to the same batch of tips) were found to be similar (161 ± 6 GPa for FM8 and 162 ± 5 GPa for FM9). No significant difference in value can be found for the two cantilevers from the same batch and thus, a modulus of 161 ± 6 GPa was used for all the cantilevers (FM2 to FM10) in subsequent calculations. Similarly, a modulus of 170 ± 2 GPa was used for cantilevers M1 to M5, based on the value obtained for M1 as M1 to M5 belong to the same batch. Table 1. Dimensions of cantilevers measured by SEM. Length (µm) Width (µm) Thickness (µm) F6 226 32.8 3.45 M1 219 35.5 3.33 M2 226 31.0 3.25 M3 217 32.6 3.78 M4 221 33.8 3.58 M5 222 32.0 3.02 FM2 226 (228) 31.7 (27) 2.88 (2.80) FM3 224 (228) 31.3 (27) 2.79 (2.80) FM4 223 (228) 30.9 (27) 2.85 (2.80) FM5 221 (228) 31.2 (27) 2.84 (2.80) FM6 223 (228) 31.1 (27) 2.93 (2.80) FM7 221 (228) 30.7 (27) 2.88 (2.80) FM8 218 (228) 30.0 (27) 3.16 (2.90) FM9 228 (228) 30.5 (27) 2.98 (2.90) FM10 224 (228) 31.0 (27) 3.12 (2.90) *Value in brackets represents manufacturer s reported data enclosed with tips Table 2. Modulus of cantilevers M1, F6, FM8 and FM9 measured using the nano-indentation system at different depths of indentation. Depth of indentation Modulus (GPa) (nm) M1 F6 FM8 FM9 150 170 ± 2 132 ± 5 161 ± 6 162 ± 5 300 146 ± 2-157 ± 3-600 125 ± 4 115 ± 9 - - 1200-86 ± 7 - - 4

With an accurate determination of dimensions and modulus of cantilevers, the spring constant of the cantilever can be determined accurately using equation (1). Table 3 lists the determined values as compared with the specifications supplied by the manufacturer. Large deviation from the reported specifications can be observed. Thus, an accurate determination of the dimensions and modulus of the cantilevers is essential to obtain an accurate spring constant of the cantilever. 4.3 Obtaining a good force curve (a) (b) In order to calculate the interaction force between the tip and sample, it is important to obtain a good force curve, which shows the typical features as illustrated in Fig. 2(a). The hysteresis in the curves corresponds to the adhesive interaction between the tip and sample when the deflection is converted to the force with the cantilever spring constant. However, depending on the surface conditions, variations to the force curve can be obtained as illustrated in Figs. 2(b) and (c). The basic approach to obtaining a good force curve entails adjusting the Z motion of the piezo relative to the sample (with the Z scan start and Z scan size parameters) and shifting the graph (with the setpoint parameter) so the pull-off point for the tip can be seen on the graph. (c) Fig. 2. Force curves showing typical features. Table 3. Comparison of calculated spring constant of cantilever against manufacturer s reported values. Spring Constant (N/m) Calculated Manufacturer s specifications value F6 3.8 - M1 5.3 - M2 3.9 - M3 7.3 - M4 6.1 - M5 3.4 - FM2 2.7 2.1 FM3 2.5 2.0 FM4 2.6 2.0 FM5 2.7 2.1 FM6 2.9 2.0 FM7 2.8 2.1 FM8 3.7 2.5 FM9 2.8 2.5 FM10 3.4 2.4 5

4.4 Trigger threshold As the trigger threshold is increased from 10 nm to 150 nm, the tip deflection, which is proportional to the adhesive force, remains almost the same (576 to 612 nn) as illustrated in Fig. 3 and tabulated in Table 4. The trigger threshold was fixed at 50 nm for subsequent force measurements. SP800 and TSF6522 tends to be rounder and the tip deflection is more gradual Figs. 4(e) and (f). (a) Force (nn) 800 700 600 500 400 (b) 300 200 0 50 100 150 Trigger threshold (nm) Fig. 3. Variation of adhesion force between the Si tip and the Si substrate with different trigger threshold. Table 4. Adhesion force between the Si tip and the Si substrate at different trigger thresholds. (c) Trigger Threshold (nm) Adhesion Force (nn) 10 612 ± 29 30 584 ± 87 50 596 ± 35 100 580 ± 28 150 576 ± 45 (d) 4.5 Contaminants organic solvents and fluxes The tip deflection versus cantilever displacement () curves recorded on surfaces contaminated with residues from methanol, isopropanol, dichloromethane, trichloroethylene, SP800 and TSF6522 are shown in Figs. 4 (a), (b), (c), (d), (e) and (f) respectively. It can be observed that the hysteresis in the curves for the organic solvents (methanol, isopropanol, dichloromethane and trichloroethylene) is sharper as compared to that of SP800 and TSF6522 as illustrated in Fig. 4. This shows that the cantilever returns to its non-deflected, nocontact position faster for the less viscous organic solvents whereas the more viscous fluxes (SP800 and TSF6522) may have delayed the cantilever on its return to its non-deflected position. Thus, the hysteresis in the curves for (e) 6

(f) However, there seems to be no distinct differences in the adhesion force measured for hydrophobic (dichloromethane and trichloroethylene) and hydrophilic (methanol and isopropanol) solvents as illustrated by the adhesion forces measured and tabulated in Table 4. A study was conducted by Frisbie et. al. [9] who used chemically modified tips to map and measure the adhesion force of some organic monolayers terminating in a lithographically defined pattern of distinct functional groups. The functional groups studied include CH 3 and COOH. It was found in the study that the interaction between hydrophilic groups (COOH) is stronger than that between hydrophobic groups (CH 3 ). Similarly, Ducker et. al. [10] have used hydrophobic AFM colloidal probes to study the nature of the hydrophobic force. Nakagawa et. al. [3] also found that the AFM could discriminate between monolayers of four different alkyltrichlorosilanes (CH 3 (CH 2 ) n SiCl 3 with n = 1, 8, 13, 17) deposited on silicon. One possible reason no distinct differences in adhesion force can be obtained could be due to the tips being used for the experiments. The tips used in this study were unmodified Si tips. Thus, modifying the tips chemically may help in differentiating the different contaminants more effectively. Possible functional groups to be used are OH and CH 3 groups. Fig. 4. Typical force curve for residues of (a) methanol, (b) isopropanol, (c) dichloromethane, (d) trichloroethylene, (e) SP800 and (f) TSF-6522. Hysteresis in the curves corresponds to the adhesive interaction between the tip and sample when the deflection is converted to the force with the cantilever spring constant. The adhesion force between the tip and the residue of each organic solvent (methanol, isopropanol, dichloromethane and trichloroethylene) were measured to be 467 ± 89, 358 ± 123, 321 ± 177 and 412 ± 113 nn respectively. By comparison, the force curve on the uncontaminated Si surface reveals a larger hysteresis due to a stronger adhesion between the tip and the surface. The adhesion force between the tip and the Si surface was measured to be 596 ± 134 nn. On the other hand, the adhesion force between the tip and the fluxes seems to be lower (262 ± 70 and 251 ± 83 nn for SP800 and TSF6522 respectively). 5 CONCLUSION Force measurements using the scanning probe microscopy were conducted to characterise the residues of commonly used organic solvents and fluxes on Si substrates. The followings have been achieved from the present study: The spring constant of each cantilever used in the force measurement was determined accurately by measuring the exact dimensions of the cantilever by means of the scanning electron microscope and the modulus determined using the nanoindentation system. The determined spring constant of the cantilevers was compared against the specifications provided by the manufacturer and discrepancies were found between the two. Thus, an accurate determination of the spring constant of the cantilever is essential for precise force measurements. The technique of measuring the adhesion force between the tip and the different organic contaminants and fluxes has been established. Thus, the applications of scanning probe microscopy have been expanded beyond the commonly used imaging applications to include quantitative information. 6 INDUSTRIAL SIGNIFICANCE The applications of SPM beyond the commonly used imaging applications developed can be used to characterise the commonly found contaminants on Si substrates, which is a common problem in the electronics industry. REFERENCES [1] C.J. Tautscher, Contamination effects on electronic products, Marcel Dekker, Inc., New York, (1991). [2] G. Binnig, C.F. Quate and C. Gerber, Phys. Rev. Latt., Vol. 56, pp. 930, (1986). [3] Application notes from Digital Instrument Probing nano-scale forces the atomic force with microscope. 7

[4] P.K. Hansma, V.B. Elings, O. Marti and C.E. Becker, Science 242, pp. 209, (1988). [5] M. Radmacher, J.P. Cleveland, M. Fritz, H.G. Hansma and P.K. Hansma, Biophysical Journal, Vol. 66, pp. 2159, (1994). [6] Material safety data sheet (MSDS number: TSF-6522, 31 Aug 2001), Kester Solder. [7] W.C. Oliver, G.M. Pharr, J. Mater Res., Vol. 7(6), pp. 1564, (1992). [8] T.J. Senden, W.A. Ducker, Langmuir, Vol. 10, pp. 1003, (1994). [9] C.D. Frisbie, L.F. Rozsnyai, A. Noy, M.S. Wrighton and C.M. Lieber, Science 265, pp. 2071, (1994). [10] W.A. Ducker, Langmuir, Vol. 8(7), pp. 1831, (1992). 8