Radio Frequency Glow Discharge Mass ... - ACS Publications

A radio frequency (rf)-powered glow discharge (GD) atomization/ionization source is utilized to determine the applicability of the technique for direc...
1 downloads 0 Views 215KB Size
Anal. Chem. 1996, 68, 2113-2121

Radio Frequency Glow Discharge Mass Spectrometry for the Characterization of Bulk Polymers Charles R. Shick, Jr., Patrick A. DePalma, Jr., and R. Kenneth Marcus*

Department of Chemistry, Howard L. Hunter Chemical Laboratory, Clemson University, Clemson, South Carolina 29634-1905

A radio frequency (rf)-powered glow discharge (GD) atomization/ionization source is utilized to determine the applicability of the technique for direct polymer analysis. A series of PTFE-based polymers are studied to assess their fingerprint mass spectra and to distinguish each sample by its differing base peaks and relative peak intensities. A parametric study with respect to discharge gas pressure and rf power is conducted to evaluate their respective roles in the sputtering process as well as possible ionization mechanisms. The results of the GD sputtering processes are examined by scanning electron micrographs of a sputtered PTFE surface. Excellent discharge stabilization characteristics (10-9 A. This is the same level of ion production as seen for the analysis of metals and glasses by this same source. This indicates that the energy coupling into the plasma is the same for the polymers as for other solids. With the dual-detector system employed on this instrument, ion currents down to ∼10-18 A can be effectively quantified. Thus, it appears that the rf-GDMS approach to polymer analysis holds the promise of excellent sensitivity. Positive ion mass spectra from SIMS analysis of PTFE have been previously reported using different MS systems and sample preparations.12,15,17,36,37 The positive and negative mass spectra of (36) Wheeler, D. R.; Pepper, S. V. J. Vac. Sci. Technol. 1982, 20, 226-232.

Analytical Chemistry, Vol. 68, No. 13, July 1, 1996

2115

Figure 2. rf-GDMS (logarithmic scale) mass spectrum of a 3.0 mm thick PTFE-co-PFA sample: (a) m/z 4-80, (b) m/z 81-160, (c) m/z 161-240 (rf power, 20 W; argon pressure, 0.075 mbar).

compacted PTFE pellets were acquired by Muller with a quadrupole-based SIMS system utilizing a filament (electron) source for charge compensation.15 The author noted that the fragment intensities decreased with higher masses due to a combination of lower secondary ion yields at larger masses and the inherent mass discrimination of the quadrupole mass filter. Werner also used a quadrupole-based SIMS system to produce a positive ion spectrum with intense 31CF+, 69CF3+, and 131C3F5+ peaks from a thick PTFE target, again with signal intensity decreasing for higher mass fragments.17 Briggs and Wootton examined 0.25 mm thick PTFE samples as a function of electron current density from an electron flood source, illustrating electron-stimulated desorption (ESD) of secondary ions.12 The electron current density had a profound effect on the PTFE fragment ion intensities, demonstrating a PTFE positive ion spectrum with electron currents of as little as 8 nA cm-2. Other mass spectra of PTFE samples have been reported for characterizing the chemical modifications of (37) Ingemarsson, P. A.; Keane, M. P.; Gelius, U. J. Appl. Phys. 1989, 66, 35483553.

2116

Analytical Chemistry, Vol. 68, No. 13, July 1, 1996

polymer samples induced by X-rays36 and MeV ion beams.37 Comparison of these examples and the spectrum shown in Figure 1 reveals no appreciable differences, which is surprising given the wide variety of experimental procedures. The coupling of SIMS and time-of-flight mass analyzers (TOF-SIMS) has enhanced polymer analysis, permitting simultaneous detection of secondary ions, high transmission efficiency compared to quadrupole mass analyzers, and unlimited mass range (in theory). The enhanced power of TOF-SIMS systems has been used to analyze polymer samples toward the determination of structural information, molecular weight distribution, and direct identification of additives.38-42 (38) Bletsos, I. V.; Hercules, D. M.; Greifendorf, D.; Benninghoven, A. Anal. Chem. 1985, 57, 2384-2388. (39) Bletsos, I. V.; Hercules, D. M.; Magille, J. H.; Van Leyen, D.; Niehuis, E.; Benninghoven, A. Anal. Chem. 1988, 60, 938-944. (40) Lub, J.; Benninghoven, A. Org. Mass Spectrom. 1989, 24, 164-168. (41) Linton, R. W.; Mawn, M. P.; Belu, A. M.; DeSimone, J. M.; Hunt, M. O., Jr.; Menceloglu, Y. Z.; Cramer, H. G.; Benninghoven, A. Surf. Interface Anal. 1993, 20, 991-999.

Figure 3. rf-GDMS (logarithmic scale) mass spectrum of a 0.75 mm thick THV sample: (a) m/z 4-40, (b) m/z 41-80, (c) m/z 81-140 (rf power, 20 W; argon pressure, 0.075 mbar).

Figure 2 is the rf-GDMS mass spectrum (logarithmic scale) of the copolymer, PTFE-co-PFA. Here, the backbone fluorocarbon chain yields characteristic CxFy+ fragments akin to PTFE with slightly different intensities, indicating the influence of the second component, perfluoromethyl vinyl ether. In comparison to the data in Figure 1, this mass spectrum has an overall decrease in intensity, that can be attributed to a greater sample thicknesses, an important experimental aspect, which will be addressed in a subsequent section. Similar to Figure 1, this mass spectrum has an overall decrease in intensity with regard to fragment mass. The 31CF+, 50CF2+, 69CF3+, 100C2F4+, and 131C3F5+ fragments are the most intense fragments, as in the case for the PTFE sample; yet, the perfluoromethyl vinyl ether characteristic fragment, 85CF2OF+, is observed in this fingerprint spectrum. The combination of the 85CF2OF+ fragment and different relative fragment intensities clearly distinguishes this mass spectrum from that in Figure 1. (42) Reichlmaier, S.; Hammond, J. S.; Hearn, M. J.; Briggs, D Surf. Interface Anal. 1994, 21, 739-746.

Figure 3 is the rf-GDMS mass spectrum (logarithmic scale) of the terpolymer, THV. An overall decrease in CxFy+ fragments assignable to PTFE is observed on the Faraday cup detector with no observed fragments above 131 m/z; however, the high-gain electron multiplier reveals subsequent higher mass PTFE fragments, including 231C5F9+. PTFE and hexafluoropropylene have very similar SIMS fragmentation patterns; however, their spectral fragments have different relative intensities.43 Figures 1 and 3 show similar base peaks with widely differing intensities. The third component of THV, poly(vinylidene fluoride), yields characteristic peaks such as 47C2H4F+ and 113C3HF4+. Here, the overall decrease in fragment ion intensities can be attributed to structural differences between these fluorocarbons. Gardella and Hercules have also observed that, for a series of polymethacrylates, secondary ion peaks are directly related to structure and are assignable to bond-breaking events along either the backbone (43) Newman, J. G.; Carlson, B. A.; Michael, R. S.; Moulder, J. F.; Hohlt, T. A. Static SIMS Handbook of Polymer Analysis; Perkin-Elmer Corp.: Eder Prairie, MN, 1991.

Analytical Chemistry, Vol. 68, No. 13, July 1, 1996

2117

hydrocarbon chain or the side chain pendant group.10,11 Figures 1-3 allude to the fact that varying structures result in different base peaks, different relative intensities, and different fragmentation patterns; thus, similar polymers could be distinguished from one another by rf-GDMS in a manner similar to SIMS. rf-GD Parametric Influences and Ionization Processes. Previous reports have shown that plasma parameters influence the analyte intensity and spectral character from various dcpowered GDMS44-46 and rf-GDMS sources.32,33,35,47 This particular rf-GD ion source has been evaluated for elemental analysis of metals and glass samples.32 Analysis of organic species by GDMS sources has also been performed by other research groups. Chapman and Pratt obtained mass spectra using a sector-based mass spectrometer with a thermospray interface and GD accessory, inducing fragmentation of caffeine in a controlled manner to obtain structural information.48 A GDMS source designed by Carrazzato and Bertrand exhibited picogram range detection limits with a 6 orders of magnitude linear response range for organic compounds introduced via a liquid chromatography pump.49 That work illustrated that many ionization processes occur simultaneously in the GD plasma, leading to different fragmentation pathways. GD ionization can produce fragments which are similar to those obtained by electron impact (EI) and chemical ionization (CI) methods. DePalma and co-workers have characterized fragmentation patterns for GD, EI, and CI ionization sources for organometallic, inorganic, and organic compounds.50 While the previously cited applications were based on GC or LC sample introduction, Mason and Milton have, in fact, acquired dc-GD mass spectra of several organic compounds, including sucrose and tyrosine, deposited directly onto a copper probe tip.51 That work indicated that sputtering of organic samples was, indeed, possible and capable of producing mass spectra of intact molecular ions. The physicochemical processes occurring within the GD have been extensively studied.19,22 Neutral atoms and clusters are normally removed from the sample surface by a cathodic sputtering process and diffuse into the negative glow region for subsequent ionization. It is important to note that the molecular ions seen in these spectra are the result of the sputter-removal of neutral species and not the gas-phase association of isolated atoms/ions. The fragment ions observed in the spectra shown in Figures 1-3 indicate that various ionization processes are quite possible. The 31CF+, 69CF3+, 181C4F7+, and 231C5F7+ fragments could be produced by several different pathways. Among the possible CI reagents present in the discharge, 41ArH+ and 18H3O+ could produce the various protonated species like 32CFH+, 70CF3H+, and 101C F H+. In Figure 3, more protonated species are observed 2 4 since one of the pendant side chains is a hydrocarbon. Also, the amount of sample wetness will influence the formation of proto(44) Saito, M. Anal. Chim. Acta 1993, 274, 327-334. (45) Jakubowski, N.; Stuewer, D.; Toelg, G. Int. J. Mass Spectrom. Ion Phys. 1986, 71, 183-197. (46) King, F. L.; McCormack, A. L.; Harrison, W. W. J. Anal. At. Spectrom. 1988, 3, 883-886. (47) Myers, D. P.; Heintz, M. J.; Mahoney, P. P.; Hieftje, G. M. Appl. Spectrosc. 1994, 48, 1337-1346. (48) Chapman, J. R.; Pratt, J. A. E. J. Chromatogr. 1987, 394, 231-237. (49) Carazzato, D.; Bertrand, M. J. J. Am. Soc. Mass Spectrom. 1994, 5, 305315. (50) DePalma, P. A., Jr.; You, J. Z.; Marcus, R. K.; Willoughby, R. C. Particle Beam LC/MS using a Glow Discharge Ionization Source. Presented at the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, GA, May 22-26, 1995; Poster TPB 094. (51) Mason, R.; Milton, D. M. P. Int. J. Mass Spec. Ion Processes 1989, 91, 209225.

2118 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996

Figure 4. (a) Effect of discharge gas pressure on PTFE fragment intensities for 12C+, 31CF+, 69CF3+, 131C3F5+, and 181C4F7+ (rf power, 20 W). (b) Effect of rf power on PTFE fragment intensities for 12C+, 31CF+, 69CF +, 131C F +, and 181C F + (argon pressure, 0.075 mbar). 3 3 5 4 7

nated species. Charge exchange and EI processes could also be rationalized for the fragmentation patterns observed in Figures 1-3. Penning transfer in the negative glow of the GD is likely an important mechanism for the formation of organic ions and radicals. In a seemingly unrelated work, Hess and co-workers studied the effect of methane addition on the argon metastable populations in a GD.52 Their work indicated that argon metastables efficiently ionize methane, removing this mechanism as a means of ionizing atomic species present in that discharge. The extent of Penning ionization present in this rf-GD source cannot be determined within the discharge source design employed in these studies. Figure 4 illustrates the response of selected PTFE fragments generated from the sputtering of a 1.5 mm thick PTFE sample as a function of discharge gas pressure and rf power. By changing the discharge pressure from 0.05 to 0.20 mbar, insights about GD processes can be inferred. In Figure 4a, the 12C+ intensity is seen to increase with increasing pressures, while the CxFy+ fragment intensities decrease. This effect is more prominent as the size of the fragment increases. These data suggest that higher discharge pressures likely increase collisional dissociation of the sputtered (52) Smith, R. L.; Serxner, D.; Hess, K. R. Anal. Chem. 1989, 61, 1103-1108.

fragments, yielding more 12C+. An analogous response is seen in atomic GDMS, where collisional dissociation of metal dimers (M2+) has been implied for both rf- and dc-powered GDs with increases in discharge pressure.32,46 The mechanism of increased gas-phase dissociation as a function of discharge pressure is discounted when the role of discharge power is considered. For a given discharge pressure, subsequent rf power increases have been shown to enhance analyte signal intensities in elemental analysis applications32-35 and would be expected to do so as well for the polymer samples. This is seen to be the case in Figure 4b, wherein all of the selected fragment intensities become more intense across the 15-25 W rf power range at a constant 0.075 mbar discharge pressure. Similar rf power and discharge pressure results were observed for both the PTFE-co-PFA and THV samples. The dc bias voltage characteristics (on the sample surface) of rf-GDs have been extensively studied.53 As might be expected, increases in applied power result in proportional increases in dc bias values, while an inverse relationship exists between the discharge pressure and the dc bias. One would expect that higher dc bias values would produce more energetic argon ions arriving at the sample surface with more kinetic energy to transfer to the sample material. Within certain limitations, higher energy ions would be more effective at liberating larger clusters/fragments from the sample surface. This is, in fact, the case in Figure 4b, where the increases in rf power result in the 12C+, 31CF+, and 69CF + fragment intensities slightly increasing, while the 131C F + 3 3 5 and 181C4F7+ fragment intensities have more pronounced increases. This must be interpreted as larger fragments being sputtered from the sample. The data in Figure 4a support this sort of mechanism, where the 131C3F5+ and 181C4F7+ fragment intensities are higher at lower discharge pressures, i.e., where the dc bias values are highest. It can be concluded, therefore, that the sputtering process controls the extent of fragmentation in the rf-GDMS spectra as opposed to gas-phase collisions. PTFE-based samples were chosen for this work because PTFE mass spectrometric fragmentation has been well characterized and PTFE has a high melt viscosity and a maximum usage temperature of 260 °C.54 All of the samples analyzed here by rf-GDMS produced stable ion intensities during the course of these parametric studies, for periods of up to 5 h. This level of plasma stability suggests that sample degradation is not occurring. Scanning electron micrographs of a gold-coated 1.5 mm thick PTFE sample reveal the effect of sputtering on the crystalline structure of the PTFE sample. Figure 5a is a micrograph (×1000) which includes both sputtered and nonsputtered regions in its field of view. The nonsputtered region was not exposed to the GD, and so the micrograph reveals a quite smooth region indicative of the highly crystalline structure. The sputtered region has many conelike shapes and ribbon projections. Further magnification of ×7000 is shown in Figure 5b, enhancing these structural features. Juger and Blum have shown formation of these same sorts of cones from the sputtering of gold and brass samples.55 The choice of discharge gas and sample composition affect the type of surface fractures produced by the sputtering process. Clearly what is observed here is the result of argon ion sputtering of the sample surface, a kinetic process. There is no (53) Parker, M.; Marcus, R. K. Spectrochim. Acta 1995, 50B, 617-638. (54) Engineering Material Handbook, Engineering Plastics; ASM International: Materials Park, OH, 1995; Vol 2, pp 114-119. (55) Juger, H.; Blum, F. Spectrochim. Acta 1974, 29B, 73-77.

Figure 5. Scanning electron micrographs of a 1.5 mm thick PTFE sample: (a, top) ×1000 magnification of sputtered and nonsputtered region and (b, bottom) ×7000 magnification of sputtered region (rf power, 20 W; argon pressure, 0.075 mbar; sputtering time, ∼30 min).

visual indication of a thermal desorption process occurring at the sample surfaces. Discharge Stabilization. While scanning electron micrographs provide strong evidence for the sputtering of the PTFE sample, Mason and Milton observed that heating due to argon ion and atom bombardment caused rapid desorption and chemical ionization in the analysis of organic residues.51 Their studies revealed a 20-30 s delay before organic ions were detected, with these ions only remaining present in the mass spectra for ∼60 s longer. Further investigation of the apparent thermal desorption process was performed by monitoring molecular ion intensities for several organic samples with respect to discharge burn time and operating conditions. Their time profiles supported a process whereby some sample sputtering occurs early in the discharge lifetime and, after some induction period, the sample sublimes. For any sequential instrument such a quadrupole mass filter, excellent ion source stability is necessary for confident measurement. Temporal response (break-in) curves (raw counts per second) for the sputtering of a fresh 1.5 mm thick PTFE sample are shown in Figure 6. A steady state production of both small and large fragment ions is reached within 3 min after discharge ignition. Over the 15 min following discharge stabilization, the ion intensities of the monitored fragments varied over the limited Analytical Chemistry, Vol. 68, No. 13, July 1, 1996

2119

Table 2. Relative Ratios between Selected Fragment Intensities for 3 mm Thick PTFE and PTFE-co-PFA Samples

31CF+/12C+ 31CF+/50CF + 2 31CF+/62C F + 2 2 31CF+/69CF + 3 31CF+/93C F + 3 3 31CF+/119C F + 2 5 31CF+/131C F + 3 5 31CF+/181C F + 4 7

Figure 6. Temporal response curves for 1.5 mm thick PTFE sample: 69CF3+, 50CF2+, 100C2F4+, 181C4F7+, and 59ArF+. Table 1. Maximum Intensity and Stability for Selected PTFE Fragment Intensities as a Function of Sample Thickness 1 mm thick

1.5 mm thick

% % fragment intensity (A) RSD intensity (A) RSD 12C+ 31CF+ 50CF + 2 62C F + 2 2 69CF + 3 93C F + 3 3 119C F + 2 5 131C F + 3 5 181C F + 4 7

1.84 × 10-12 5.93 × 10-11 1.75 × 10-11 6.02 × 10-13 2.11 × 10-10 3.61 × 10-12 5.83 × 10-12 1.70 × 10-11 7.11 × 10-12

0.9 1.6 0.2 0.2 0.3 1.4 0.7 0.6 0.8

2.61 × 10-12 6.47 × 10-11 1.48 × 10-11 5.85 × 10-13 1.67 × 10-10 2.76 × 10-12 1.46 × 10-12 6.38 × 10-12 1.54 × 10-12

0.6 0.7 3.2 2.6 0.7 1.4 2.5 5.5 2.4

3.0 mm thick intensity(A)

% RSD

4.14 × 10-12 6.95 × 10-11 9.18 × 10-12 1.89 × 10-13 1.09 × 10-10 2.73 × 10-13 2.18 × 10-13 2.04 × 10-13 1.61 × 10-13

3.5 0.8 0.2 1.2 0.2 2.3 1.4 0.9 4.1

range of 1-5% RSD. This overall stability, coupled with the electron micrographs, substantiates the direct sputtering of PTFE without sublimation or decomposition, which would produce a highly unstable discharge. One should note that the total analysis time for this study was under 30 min (3 min sample loading time and 20 min sample analysis). This time frame is much shorter than that required for most SIMS analyses case, where the deleterious effects of residual gases are more prominent and sputtering rates are much lower than in this rf-GDMS source. Effect of Sample Thickness. Previous studies in this laboratory have shown that the thickness of nonconductive samples affects the amount of rf power losses, wherein thicker samples have larger power losses than the thinner ones. Work with rfGD-AAS and rf-GDMS have shown reductions in sputtering rates for nonconducting samples of increasing sample thickness.32,53 These reductions in sputtering rates are a direct result of lower dc bias voltages on the sample surface. To assess such power loses for polymer samples, a set of 1.0, 1.5, and 3.0 mm thick PTFE samples were sputtered to examine the resultant ion fragment intensities. The maximum intensity and internal stability for the selected PTFE fragments are reported in Table 1. The overall average stability values are 0.8, 2.2, and 1.6% RSD for the 1.0, 1.5, and 3.0 mm thick PTFE samples, respectively. These stability values are similar to the data shown in Figure 6. As expected, the 1 mm thick PTFE sample had higher overall ion intensities. The identity of the base peaks provides several insights about the sputtering process. Both 12C+ and 31CF+ 2120 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996

PTFE

PTFE-co-PFA

17 7 370 0.6 256 323 357 435

71 5 45 0.4 9 9 4 15

intensities increase with the sample thickness, yet the larger CxFy+ intensities all decrease with sample thickness. Thus, the power loss through the nonconductive sample affects the relative fragment intensities. Basically, the 3 mm thick PTFE sample is sputtered by lower energy argon ions than those for the thinner PTFE samples; therefore, there is less energy available to liberate large fragments in the sputtering process. Because there is a range of ion kinetic energies impacting the surface, there is still production of high-mass fragments, albeit at greatly reduced relative intensity. This can also be seen in the comparison of Figures 1 and 2. The 3 mm thick PTFE-co-PFA sample (Figure 2) has a similar low-mass peak pattern when compared to the 1.5 mm thick PTFE (Figure 1), which should be expected since both polymers have the same polymer backbone, with lower ion intensities. A more realistic test of the production of unique spectra between similar polymers would entail comparison of spectra for samples of the same thickness. A comparison of relative fragment intensities between 3 mm thick samples of PTFE and PTFE-co-PFA is given in Table 2. The ratios of the intensities of the 31CF3+ fragment relative to those of selected higher mass fragments are clearly different. Specifically, the simple PTFE spectra produce greater fractions of high-mass species in comparison to the copolymer, which is not be unexpected. Thus, the differentiation between the polymer samples of similar thickness and of the same polymer backbone with differing pendant side chains can be performed by rf-GDMS. Applications in Depth-Resolved Analyses. The level of stability for temporal response and internal repeatability observed here bodes very well for sequential analysis by a quadrupole mass filter. This is essential for depth-resolved rf-GDMS analysis with this system. Payling and co-workers utilized rf-GD-AES to perform depth profiles for both commercial prepainted metallic-coated steel samples and in-house-prepared pigmented polymer-coated steel samples.56,57 That work employed a multielement polychromator system to study polymer layers which contained C, H, O, Ti, and Fe. The respective emission intensities were monitored simultaneously as the paint layers were sputtered away, after which the substrate (steel) component emission intensities became prominent. Those studies clearly illustrated that rf-GD-AES was capable of measuring the elemental compositions of polymer coatings. While painted (polymer) coatings on metal surfaces are important in many industrial applications, the complementary case (56) Payling, R.; Jones, D. G.; Gower, S. A. Surf. Interface Anal. 1993, 20, 959966. (57) Jones, D. G.; Payling, R.; Gower, S. A.; Boge, E. M. J. Anal. At. Spectrom. 1994, 9, 369-373.

Figure 7. rf-GDMS depth profile analysis of an ∼1 µm copper layer on a 1.5 mm thick PTFE substrate (rf power, 20 W; argon pressure, 0.075 mbar).

of metal coatings on polymer bases is also relevant. For example, the adhesion between metal and polymer has been a common concern for the use of polymers in electronic packaging and device fabrications. Figure 7 illustrates this sort of application in the depth-resolved analysis (intensity versus time) of an ∼1 µm thick copper layer on a 1.5 mm thick PTFE base. This sample was prepared in-house by sputter deposition. After ∼5 min, the 63Cu+ intensity decreases as the metal/polymer interface is sputtered. The exposure of the PTFE substrate was monitored by the 69CF3+ fragment intensity. Particularly attractive in this case is the rapid stabilization (