Vibrational Spectroscopic Analysis of Silicones: A Fourier Transform

School of Chemical Sciences, University of East Anglia, Norwich NR4 7TJ, U.K., ... Corning, B-7180 Seneffe, Belgium, and ISIS Facility, Rutherford App...
0 downloads 0 Views 125KB Size
Anal. Chem. 2003, 75, 742-746

Vibrational Spectroscopic Analysis of Silicones: A Fourier Transform-Raman and Inelastic Neutron Scattering Investigation Linda Jayes,† Andrew P. Hard,† Christophe Se´ne´,‡ Stewart F. Parker,§ and Upali A. Jayasooriya*,†

School of Chemical Sciences, University of East Anglia, Norwich NR4 7TJ, U.K., Business and Technology Centre, Dow Corning, B-7180 Seneffe, Belgium, and ISIS Facility, Rutherford Appleton Laboratory, Chilton, Didcot, Oxon OX11 1QX, U.K.

An inelastic neutron scattering spectrum of a poly(dimethylsiloxane) (PDMS) is reported, and a spectrum simulated using a monomer molecular unit as a model for comparison. FT-Raman spectra of a series of PDMS derivatives are reported and structure spectra correlations are shown to exist for the estimation of (a) PDMS average chain length, (b) ratio of the number of monofunctional units to quadrifunctional units in silicone resins, and (c) the percentage weight of PDMS in silicone emulsions. Silicone polymers have unique properties resulting in their wide range of uses in industries including healthcare, personal care, aerospace, automotive, and construction materials for products used in the electronics and semiconductors industry.1 Unlike most organic polymers, silicones retain rubber properties even at high molecular weight and low temperature (typical Tg is -120 °C) and are excellent filmogens. Typical silicones are known to have very good thermal stability, good biocompatibility, low toxicity, and high chemical and UV/light resistance. Unlike 29Si/H NMR, FT-IR, SEC, TGA, and GC/RI-MS, which are the main techniques for the characterization of silicones, Raman spectroscopy remains an uncommon analytical tool in R&D or process analysis. However, a number of studies of the IR and Raman spectra of organosilicon compounds are found in the literature,2 with that of Smith and Anderson3 being the most complete to date. Several Raman spectroscopic applications are to be found including a study of the effect of temperature variation on PDMS itself.4 An investigation that reports the possible use of Fourier selfdeconvolution to the FT-IR spectra for chain length determination in PDMS is of relevance to the aim of the present study.5 However, the use of IR spectroscopy as an on-line measurement technique is likely to be more complicated than that of Raman spectroscopy. * Corresponding author. E-mail: [email protected]. † University of East Anglia. ‡ Dow Corning. § Rutherford Appleton Laboratory. (1) The Analytical Chemistry of Silicones; Smith, A. L., Ed.; John Wiley & Sons: New York, 1991. (2) See, for example: Lippard, E. D.; Smith, A. L. Chem. Anal. 1991, 112, 305. Marad, E.; Smith, L. M. Appl. Spectrosc. 1995, 49, 513. Deuring, H. Macromolecules 1995, 28, 1952. Curdes, J.; Hilf, E. R. Fresenius J. Anal. Chem. 1992, 344, 140. (3) Smith, A. L.; Anderson, D. R. Appl. Spectrosc. 1984, 38, 822. (4) Soutzidou, M.; Panas, A.; Viras, K. J. Polym. Sci., Part B 1998, 36, 2805. (5) Lipp, E. D. Appl. Spectrosc. 1986, 40, 1009.

742 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

The advent of FT-Raman spectroscopy, with its inherent advantages for avoiding the fluorescence problem and also the possibility of efficient coupling to fiber optics, should make this a viable technique for in situ monitoring of the industrial processes that generate these useful polymers. What follows is an investigation, first using the technique of inelastic neutron scattering (INS) in combination with IR and Raman spectroscopies to obtain a better understanding of the vibrations of the parent polymer, poly(dimethylsiloxane) (PDMS), followed by an exploration into some of the possible applications of FT-Raman spectroscopy to these polymer systems. The INS spectrum provides a hydrogen amplitude weighted density of vibrational states of the sample, whose spectral intensities are easier to model than those of IR and Raman spectra.6 This therefore should provide useful additional insights into the vibrations of the material. The range of polymers investigated during this study have provided quantitative correlations that show the potential for use of FT-Raman spectroscopy for in situ monitoring of industrially important properties such as chain lengths and degree of crosslinking of the silicone polymers. EXPERIMENTAL METHODS INS spectra of pure PDMS samples, kept at ∼20 K in aluminum sample containers, were run using the spectrometer Tosca at the ISIS facility of the Rutherford Appleton Laboratory. Raman spectra were recorded using a Bruker IFS 66 FT-IR spectrometer equipped with a FRA106 Raman module and a liquid nitrogen-cooled germanium detector. A diode laser operating at 1064 nm was used as the excitation source, with ∼100 mW of laser power at the sample. The spectral resolution was 4 cm-1, and each spectrum was an average of 1000 scans in backscattering geometry. The samples were held at ambient temperature, and there was no evidence of degradation from laser heating. All silicone samples were provided by Dow Corning and were used without further purification. The oil-in-water silicone emulsions used for this analysis were made starting with an oil-in-water emulsion containing 40 wt % of PDMS. The diluted samples were found to be stable. Samples were contained in sealed NMR tubes. (6) Fillaux, F. In The Enzyme Catalysis Process, Energetics, Mechanism and Dynamics; Cooper, A., Hoube, J. L., Chien, L. C., Eds.; Plenum Press: New York, 1989; pp 79-122. 10.1021/ac026012f CCC: $25.00

© 2003 American Chemical Society Published on Web 01/17/2003

Figure 1. FT-Raman spectrum of PDMS. Table 1. Observed FT-Raman and INS Wavenumbers for 58 100 Molecular Weight PDMS Raman (cm-1) 2966 2906 1411 1263 861 489 190 2966 1411 165 2966 1411

Figure 2. (a) Calculated INS spectrum of (CH3)2SiO2 and (b) a comparison with experimental data.

The silicone resin samples, MxQ, were held in a solid sample cell at room temperature. Each spectrum was an average of 500 scans in backscattering geometry. RESULTS AND DISCUSSION Vibrational Assignments of PDMS. FT-Raman and INS spectra of the sample of PDMS with a mean molecular weight of 58 100 and having a narrow molecular weight distribution are shown in Figures 1 and 2, respectively. The INS spectrum is dominated by the proton dynamics. Since these are localized on the methyl groups, an attempt was made to simulate the INS spectrum by considering only the immediate environment of the methyl group; a single (CH3)2SiO2 unit that is assumed to have C2v symmetry (shown as an inset to Figure 2) was used as the model. The dominant mechanism of scattering of neutrons is by nuclei not by electrons; thus, the electronic contributions (dipole

190 2966 2906 1411 1263 788 680 190

INS (cm-1)

1410 1264 870 709 190 54 1410 742 179 175 1410 742 185 183 1410 1266 802 687 190

assignment

symmetry

C-H sym str C-H sym str C-H asym bend C-H sym bend CH3 rock C-Si-C sym str Si-O sym str C-Si-C scissors O-Si-O scissors C-H asym str C-H asym bend CH3 rock CH3 torsion C-Si-C twist C-H asym str C-H asym bend CH3 rock CH3 torsion C-Si-C rock C-H asym str C-H sym str C-H asym bend C-H sym bend C-Si-C asym str CH3 rock C-Si-C wag

A1 A1 A1 A1 A1 A1 A1 A1 A1 A2 A2 A2 A2 A2 B1 B1 B1 B1 B1 B2 B2 B2 B2 B2 B2 B2

moment and polarizability) that make the intensities of infrared and Raman spectra so complex to calculate are absent in an INS spectrum. This has two implications: First, there are no selection rules and all modes are allowed. Second, from a conventional Wilson-Decius-Cross7 normal coordinate analysis, it is possible to calculate both the energies (from the eigenvalues) and intensities (from the eigenvectors) for INS spectra of molecular species. The program CLIMAX8 has been developed to carry out the normal coordinate analysis using both energy and intensity information as constraints. INS spectra differ from infrared and Raman spectra in that overtones and combination bands are allowed transitions in the harmonic approximation. CLIMAX can optionally calculate the spectrum with the inclusion of first overtones and binary combinations. (7) Wilson, E. B., Jr.; Decius, J. C.; Cross, P. C. Molecular Vibrations; Dover: New York, 1955. (8) Kearley, G. J. Nucl. Instrum. Methods Phys. Res. A 1995, 53, 354.

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

743

Table 2. Refined Force Constants for the PDMS Model force constant

valuea

C-H stretch Si-O Si-C H-C-H bend H-C-Si bend O-Si-O C-Si-C O-Si-C torsion Si-O/Si-O H-C-H/H-C-H Si-C/Si-C H-C-Si/H-C-Si Si-O/Si-C

475.1 500.0 491.7 43.3 36.6 10.0 40.0 5 5.5 -2.889 × 1012 -5.9 × 1010 9.7 × 1011 3.8 × 1010 1.97 × 1011

a Principle force constants in Nm-1; interaction force constants in Nm-2.

Figure 4. Correlation of peak height ratios with n, the number of repeating siloxane units.

Figure 3. FT-Raman spectra of different MW PDMS samples: (A) 3.5 siloxane units (MW 414); (B) 26 units (2090); (C) 68 units (5230); (D) 398 units (29 700); (E) 922 units (68 500).

From the comparison of the fits with and without binary combinations and first overtones, shown in Figure 2, it is apparent that a large part of the intensity in the 800-1000-cm-1 region is due to combination and overtones involving the very low frequency modes. The resulting assignments and the final force constants are given in Tables 1 and 2, respectively. No detailed assignment of the υ(C-H) region has been attempted, because the INS spectrum is too broad in the 3000-cm-1 region to make any 744 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

assignment meaningful. The band at ∼2802 cm-1 in the FT-Raman spectrum is assigned to the overtone of the asymmetric CH3 deformation at 1411 cm-1. Further weak bands resolved in the FT-Raman spectrum at 3144 and 3052 cm-1 are assigned respectively to the asymmetric and symmetric stretches of the CH3 end groups. These are clearly in environments different from those in the middle of the polymer chain and are expected to be present at much lower concentration than the rest of the methyl groups. Note that these frequencies are relatively high for aliphatic C-H. These assignments are in reasonable agreement with those found in the literature, particularly with those of Smith and Anderson.3 A novel observation and an assignment made possible by INS are the methyl group torsion modes at ∼173 cm-1. These modes in general involve no dipole moment or polarizability changes and, hence, are inactive in both IR and Raman spectroscopies. The accepted assignments of the asymmetric and symmetric stretching modes of the Si-O bonds are respectively at 1100 and 500 cm-1.3 This unusually large difference in frequency between these two modes is explained by the present analysis. The asymmetric stretch at 500 cm-1 is found to be relatively pure while the 1100-cm-1 band is due to an admixture of Si-C and Si-O bond-stretching coordinates. The limitations of the model are shown by the very large interaction constants for the Si-O stretches. It is also clear that the methyl rocking modes and the Si-C stretching modes are also highly mixed. Analysis of PDMS Chain Lengths. Five PDMS samples with varying chain lengths, but with narrow distribution of molecular weights in each case, were analyzed using FT-Raman spectroscopy. The distribution of the molecular weights is given by the polydispersity of the polymer, with a value of unity indicating a

Figure 7. Change in intensity of the Raman bands at 2905 and 2965 cm-1 with increasing percent by weight PDMS in the oil-in-water silicone emulsions.

Figure 5. FT-Raman spectra of group 1 and 2 silicone resins.

Figure 6. Correlation of peak area ratio with M/Q, the number of monofunctional units per quadrifunctional unit.

monodispersed sample. The FT-Raman spectra of these samples are shown in Figure 3. There are clear differences to be seen between the spectra, with the smallest mean molecular weight sample with only ∼3.5 repeat siloxane units showing the largest number of bands. The multitude of weak bands seen in this spectrum disappear with increasing molecular weight of the sample, with the spectra for the samples with the highest mean molecular weights appearing to be almost identical to each other. The higher concentration of end groups in the smaller molecular weight samples and the uniformity of the groups to be found with larger

molecular weight samples is clearly the reason for the observed spectral changes. Therefore, it should be possible to find a correlation between the number of repeat units in the polymer and the intensity ratios of a peak due to an end group and one due to the siloxane backbone. After investigating the various possible combinations of bands, it was found that the intensity ratios between the Raman bands at 490 and 3050 cm-1 as well as the intensity ratios between the Raman bands at 709 and 998 cm-1 provided the desired correlations with the mean number of repeat units per polymer sample. The observed correlations are shown in Figure 4. These correlations thus provide a quick and easy way to analyze the average chain lengths of these polymers, a measurement normally done using gel permeation chromatography, which is a time-consuming technique. Further, the latter technique has to overcome additional problems with higher molecular weight materials due to their high viscosity. Analysis of Silicone Resins. The silicone resins analyzed were labeled MxQ, indicating that there are x monofunctional units, M, for every quadrifunctional unit, Q. Therefore, one would expect to see intense Si-O bands for resins that are rich in Q units. The Raman spectra of the silicone resins shown in Figure 5 may be divided into two groups. Spectra of the first group consisting of M4Q, M1.15Q, M0.6Q, and M0.4Q exhibit stretches due to vinyl groups (3057 cm-1) probably present as an impurity. The second group containing M3Q, M2Q, M1.3Q, M0.9Q, and M0.5Q do not show any vinyl impurity. The band at ∼616 cm-1, which appears to decrease in intensity with the increase of monofunctional groups, is assignable to a normal mode confined to this group. This band intensity, when calculated as the ratio of the C-H stretching mode at 2905 cm-1, is found to correlate well with the ratio of monofunctional units to quadrifunctional groups, and the resulting correlation is shown in Figure 6. This correlation encompasses both groups of resins, with and without the vinyl impurity, and is an illustration of the possibility of targeting a certain feature of a reaction process irrespective of the impurities present. This correlation should therefore provide an easy estimation of the value of x and is another example of the power of FT-Raman spectroscopy for this type of analysis. The vinyl functionality is an important functionality in silicone technology for the manufacturing of two-part systems or the Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

745

of different dilutions used for this study. The emulsions were well shaken before each measurement to guarantee good homogeneity. The simplest of the techniques, monitoring the absolute intensity of a Raman band, was used to analyze for the silicone (PDMS) content of these emulsions. The bands at 2965, 2905, 1410, 710, and 490 cm-1 were found to give good correlations with percent by weight of silicone. The change in intensity of the bands at 2905 and 2965 cm-1 is clearly shown in Figure 7. These two bands gave the best fits. These correlations are given in Figure 8.

Figure 8. Correlations for the peak intensities (2905 and 2965 cm-1) with percent by weight PDMS.

polymerization of silicones in manufacturing processes. In Raman, the vinyl band (3057 cm-1) is intense and well resolved and therefore Raman spectroscopy could potentially be used to measure the vinyl content of silicone resins or silicone fluids. Analysis of Silicone Emulsions. An oil-in-water silicone emulsion containing 40 wt % PDMS was used to make emulsions

746 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

CONCLUSION A comparison of the experimental INS spectrum of a sample of PDMS with a simulation using a monomer unit showed reasonable agreement. The differences are seen to be mainly due to the low-lying vibrations, those that are expected to be introduced by the lengthening of the chain. We have therefore initiated a more systematic approach to this problem by planning to study the INS spectra of oligomers, starting from the monomer. Several clear spectra-structure correlations are shown to exist when FT-Raman spectroscopy is used. The latter technique is one that is most suitable for in situ applications. The results of this investigation clearly show the possibility of using FT-Raman spectroscopy for the estimation of the following properties: (a) PDMS average chain length; (b) ratio of the number of monofunctional units to quadrifunctional units in silicone resins; and (c) the percentage weight of PDMS in silicone emulsions. Functional analysis (vinyl group) of silicone resins or fluids could also potentially be achieved. The latter therefore illustrates the potential of FT-Raman spectroscopy as a technique for on-line monitoring of industrial processes involving silicones. ACKNOWLEDGMENT We thank the EPSRC for a studentship for A.P.H. and ISIS, Rutherford Appleton Laboratory, for neutron facilities. Received for review July 31, 2002. Accepted November 14, 2002. AC026012F