Anal. Chem. 1990, 62, 1167-1172
1167
Laser Desorption/Ionization Fourier Transform Mass Spectrometry and Fast Atom Bombardment Spectra of Nonvolatile Polymer Additives Carolyn L. Johlman' and Charles L. Wilkins* Department of Chemistry, University of California, Riverside, Riverside, California 92521
Jeremiah D. Hogan, Tracy L. Donovan, and David A. Laude, Jr.* Department of Chemistry, The University of Texas at Austin, Austin, Texas 78712
M.-Joseph Youssefi2 Research and Development, Union Carbide Corporation, Bound Brook, New Jersey 08805
Laser desorption Fourier transform mass spectra (LPFTMS) of a variety of nonvolatile polymer additives are compared wlth fast atom bombardment spectra (FAB) of the same materlals. Both a pulsed carbon dioxide laser and a neodymlwn-YAG laser with outputs of 10.6 and 1.064 pm, respectively, were used to obtain LD-FTMS spectra of ail samples. Three sterically hindered phenols and other additives containing a variety of functlonalitles lnciudlng thloester, phosphite, phosphonite, and hindered amine groups were examined. I n general, FAB spectra show undesirably large amounts of fragmentation, while molecular ion species domInate LPFlMS spectra. It Is concluded that LD-FTMS spectra are superior to FAB spectra for analysis of these common polymer addltives.
Despite continuing improvements in instrumental methods for chemical analysis, the reliable analysis of organic additives in polymers remains a formidable challenge because of the complexity of commercial polymer formulations ( I , 2). Successful analytical methodologies must not only distinguish the number of mixture components but also provide characteristic structural information about each additive. As an alternative to traditional methods of polymer analysis, Lattimer described in a recent review the application of mass spectrometry as a detector for the compounding materials (3). Mass spectrometry offers sensitivity and dynamic range advantages,provides substantial information about structure, and is routinely employed as a detector for chromatography. However, it is primarily used as an auxiliary detector, probably due to cost and its relatively late development in comparison with alternative analytical methods. Nevertheless, the application of mass spectrometry for the direct analysis of volatile additives is well established (4,5). With the move toward higher mass additives with reduced volatility, direct analysis by thermal desorption is not possible and more elaborate extraction or pyrolysis procedures must precede mass spectral detection (1,6, 7). Recently, Lattimer advocated the use of mass spectrometry for direct analysis of nonvolatile compounding agents in polymer matrices as an alternative to extraction procedures (7,8).This is possible because of advances in ion source and mass analyzer technology, specifically the development of sources such as fast atom bombardment 'Current address: Phillips Petroleum Co., Bartlesville,OK 74004. *Currentaddress: Firmenich SA, CH-1211Geneva 8, Switzerland. 0003-2700/90/0362-1167$02.50/0
(FAB) (9), organic secondary ion mass spectrometry (SIMS) (IO),plasma desorption (PD) (II), and field desorption (FD) (12,13). These techniques deposit large amounts of energy at the sample surface with consequent generation of high mass organic ions characteristic of the sample. Additional benefits of some of these sources are surface and depth profiling capabilities and the potential for mixture analysis. In recent studies of rubber vulcanizates (7,8), Lattimer compared FD and FAB spectra obtained by direct analysis and from extraction of polymer additives with molecular weights extending to 393. Concurrent with development of the desorption techniques described above was development of laser desorption/ionization (LD) sources for pulsed mass analyzers such as timeof-flight (TOF) and Fourier transform mass spectrometers (FTMS) (14-18).LD-MS has been applied with considerable success to the detection of many classes of nonvolatile compounds including organic polymers, porphyrins, and biopolymers such as carbohydrates and peptides. More recently LD also has been applied to surface and depth profiig studies with laser microprobe instruments. LD spectra are distinguished by excellent sensitivity, extended high mass range, and predominance of molecular ion species. In the present work, the feasibility of coupling LD with FTMS for detection of large molecular weight, nonvolatile polymer additives both by direct means and in extracts is evaluated. These results supplement and extend the thorough prior study by Asamoto, Young, and Citerin (19). Selected for evaluation are polymer additives with masses between 500 and 1300 daltons that are commonly used as primary and secondary antioxidants. Mass spectra of isolated polymer additives are obtained by LD-FTMS using both COZ and Nd:YAG lasers. FAB spectra are also acquired to permit comparison of the relative merits and potential of each ionization technique for these compounds, complementing previous studies which compared FAB and field desorption with LD-FTMS for analysis of polymers (20,21). Additional experiments indicate that direct measurement of extracts of additives at the part-per-thousand level is possible by LDFTMS.
EXPERIMENTAL SECTION Instruments. COzLD-FTMS Instrument. COz spectra were obtained at the University of California, Riverside, by using a Nicolet Analytical Instruments FTMS-1000 spectrometer equipped with a 3.0-T magnet and an 80% transmissive stainless steel mesh, single section, 5 cm3cubic trapped ion cell (15).The FTMS was coupled to a Tachisto 215G pulsed TEA COBlaser that delivered 300-400 mJ per 40 ns pulse at 10.6 pm. A zinc 0 1990 American Chemical Society
1188
ANALYTICAL CHEMISTRY, VOL. 62, NO. 11, JUNE 1, 1990
selenide lens with a 7.5 cm focal length was used to focus the laser beam through the cell into a 1 mm spot size on a stainless steel probe tip. Power density at the probe tip was in the range of 108-109W/cm2. The measurement was initiated by a computer-controlledtrigger of the laser pulse and was followed by a 3-10 s delay prior to data acquisition to permit a return to base pressure. Nd:YAG LD-FTMS Instrument. The Nd:YAG LD spectra were obtained at the University of Texas at Austin with a spectrometer assembled from components that comprise the Nicolet FTMS-2000 spectrometer (22). The system included a 3.0-T superconductingmagnet and dual 5 cm3cubic stainless steel trapped ion cells that share a common conductance limit. A Spectra Physics Model DRA 11 Nd:YAG laser provided, in the Q-switchedmode, an unfiltered 30 mJ per 9 ns pulse at 1064 nm. With a measured spot size of 500 pm diameter,corresponding laser power densities were 107-109W/cm2, depending on initial laser energy. Spectra were acquired 2-5 s following the laser ionization event, which was triggered by the FTMS. For both instruments, data acquisition and processing were performed with standard Nicolet 1280 data systems and FTMS software. Both low- and high-resolution spectra were obtained: in low resolution mode 64K data points were acquired over either 1.0 or 2.6 MHz bandwidths (46 or 18 dalton low mass cutoff at 3.0 T) following swept excitation over a 2.6-MHz bandwidth at 1000 Hzlps; higher resolution data were obtained by acquiring 64K data points over 100-200kHz bandwidths (232 to 400 dalton low mass cutoff at 3.0 T). Each spectrum presented resulted from a single laser firing followed by one FTMS detection event. The data were processed by adding 64K zeroes, sine-bell apodization, and magnitude mode Fourier transformation. Typical resolution values for narrow band measurements were (5-10) X lo4 at m/z 1000, which is sufficient for base-line mass resolution. It should be emphasized that, although there are some differences in spectrometer design, the COz and Nd:YAG spectra were obtained by using similar trapped ion cells, magnetic field strengths, and base pressures, and with identical data acquisition and processing parameters. Hence, differences between spectra are attributable either to differences in the desorption process associated with the use of different lasers or to uncontrolled differences in sample preparation. Sector and Quadrupole FAB Instruments. Fast atom bombardment spectra were acquired with both double sector and triple quadrupolemass spectrometers. The sector instrument employed was a VG ZAB-HF operating at 8 kV accelerating voltage with mass resolution of 1000 over a mass range of 100-2000 daltons. The FAB source was operated at 8 kV with a 1.mA discharge current. Quadrupole FAB spectra were acquired by using a Finnigan “SQ-70 triple quadrupole mass analyzer with unit mass resolution over a mass range extending beyond 3000 daltons. The FAB source was operated at 8 kV with a 0.2-mA discharge current. For measurements with both instruments the sample matrix was 3-nitrobenzyl alcohol. Samples. Both pure samples and extracts were prepared for LD-FTMS in the same manner. First an alkali salt, typically KBr, was applied to the stainless steel probe tip either by dissolving it in methanol and depositing it dropwise or by burnishing the tip directly with the salt. The polymer additives (typically, approximately 1 pg) were dissolved in methylene chloride and added dropwise to the probe tip. Following evaporation of the solvent, a thin uniform coating of the sample was usually apparent. Figure 1contains structures of the samples from which LDFTMS spectra were acquired. With one exception, the samples are antioxidants. These included the sterically hindered phenolic materials (molecular mass indicated in parentheses, with the exception of Spinuvex, where the mass of the repeating unit is indicated) Irganox 1010 (1176.78 daltons) from Ciba-Geigy and Goodrite 3114 (783.52 daltons) from Goodrich, Inc. Secondary polymer additives analyzed were the thioesters dilauryl thiodipropionate (DLTDP, 514.41 daltons), distearyl thiodipropionate (DSTDP,682.59 daltons), and Seenox 4125 (1160.82 daltons), and a phosphonite, Sandostab PEPQ (1034.65 daltons). A polyhindered amine, Spinuvex A36 (repeating unit 420.42 daltons), used as a UV-protecting agent in polypropylene, also was analyzed. Other polymer additives from which spectra could not be obtained are Westin 618 (a phosphte antioxidant, 732.56 daltons), Tinuvin
,&A0 I
4 OH Coodrite 3114
Irganox 1010
H
H
Spinuvex
Westin 618
Figure 1. Chemical structures of the commercial additives analyzed by LD-FTMS.
622 (a polymeric hindered amine), and calcium and zinc stearate, which are lubricant and mold release agents. However, spectra for Tinuvin 622 and the stearates recently were obtained by laser desorption using a dual section cell FTMS-2000 instrument at Nicolet Instruments (23).
RESULTS AND DISCUSSION As a first step toward direct analysis of polymer additives, the pure components were analyzed both alone and in simple mixtures to assess the potential performance of each ionization technique. Secondary Antioxidants. Considered first are DLTDP and DSTDP, which are thioesters that contain long hydrocarbon chain ester functionalities. Comparison spectra for the two additives are shown in Figures 2 and 3, respectively. In the FAB spectrum of DLTDP in Figure 2a only a small peak resulting from the potassium-attached molecular species appears. Fragmentation is substantial and corresponds to cleavage of the ester linkage with m / z 329, which further fragments to yield prominent ions with m/z 133,143, and 161. DSTDP spectra exhibited analogous fragment ions, but with no trace of a molecular ion species. Laser desorption spectra acquired by both COz (Figure 2b) and Nd:YAG (Figure 2c) laser desorption contrast substantially with the FAB spectra. Abundant molecular ion species are observed in the laser desorption spectra. These ions are a combination of (M + H)+, (M + Na)+, and (M + K)+, depending upon the relative abundance of alkali metal salts present in the sample. Present in the COz spectra of both DLTDP in Figure 2b and DSTDP in Figure 3a are fragment ions with m / z 329 and m / z 413 that correspond to ester cleavages, as observed in the FAB spectra. In contrast, Nd:YAG spectra of DLTDP in Figure 2c and DSTDP in Figure 3b each primarily contain a strong ion signal corresponding to cation attachment to the intact molecules.
ANALYTICAL CHEMISTRY, VOL. 62, NO. 11, JUNE 1, 1990
a)
'1
s, k
I 200
300
MASS I N A
500
400
U U
1.
ko
100
600
3do
1
I
1
A
Figure 2. Comparison of positive ion spectra of DLTDP acquired by
(a) FAB,
(b)
1189
CO, LD-FTMS, and (c) Nd:YAG LD-FTMS.
To verify the facility of LD-FTMS for detecting secondary antioxidants with thioester functionalities, Seenox 4125, with a substantially higher molecular weight of 1156, was examined. As with the earlier examples, the FAB spectrum of Seenox shown in Figure 4a is dominated by low mass ions, here corresponding to fragmentation along the hydrocarbon chain to yield Cl2HzsSCH2+with m / z 215, and ClzHz5SCHzCHz+ with m / z 229. Cleavage of the ester yields the highest observed mass ion with m / z 888, in low relative abundance. In contrast, the COz LD spectrum shown in Figure 4b exhibits a base peak correspondingto the cationized molecular species, with only minor fragmentation. Similarly, the Nd:YAG spectrum in Figure 4c contains only a peak arising from the (M + K)+ ion. On the basis of the spectral evidence in Figures 2 through 4, FAB generates neither molecular species nor characteristic ion fragments of these thioesters and is an unsuitable candidate for highly specific sample analysis. The LD-FTMS spectra acquired with both lasers provide a good indication that even direct analysis in the presence of polymers should be possible. The simple spectra generated by Nd:YAG LD, dominated by a single molecular ion species, would be par-
4do
d o
660
760
L eo0
m/z
Figure 3. Comparison of positive ion spectra of DSTDP acquired by (a) C02 and (b) Nd:YAG LD-FTMS. ticularly useful for mixture analysis. LD Spectra of Hindered Phenols. An important class of antioxidants used as radical scavengers contains hindered phenol functional groups. Of specific interest here are the more nonvolatile compounds with molecular weights around 1000 that are poor candidates for more routine sample preparation and detection schemes for chemical analysis. The first to be analyzed was Goodrite 3114, with three substituted di-tert-butylphenol functional groups. As was the case for the secondary antioxidant compounds, the FAB spectrum in Figure 5a reveals almost no molecular species. Instead, cleavage of the C-N bond yields an abundant characteristic hindered phenol fragment ion with m / z 219. Again, the C02 LD spectrum in Figure 5b exhibits (M + K)+as the base peak, along with a number of low abundance peaks resulting from fragment ions. The Nd:YAG LD spectrum in Figure 5c is again dominated by the cation-attached molecular ion, with little fragmentation observed. Moving up in molecular weight within the same class of polymer additives, Irganox 1010, with a molecular weight of 1176 and four hindered phenol substituents, also was analyzed. Trends from earlier spectra are maintained, as positive ion FAB spectra are dominated by the fragment ion with m/z 219, and only a low abundance of the M+ ion is observed; however, in the only negative ion spectrum included here, FAB yielded a relatively abundant molecular anion, as shown in Figure 6a. Only for the Irganox 1010 sample, among all the additives analyzed, was a useful negative ion spectrum obtained by FAB. Both COz and NdYAG LD spectra in parts b and c of Figure 6, respectively, primarily reveal cationized molecular species. In the negative ion spectrum, an abundant M- dominates. To this point it can be argued that LD-FTMS is superior to FAB for evaluating the particular classes of antioxidant
1170
ANALYTICAL CHEMISTRY, VOL. 02, NO. 11, JUNE 1, 1990
I,!,,,
, , , ,
,
,
,
k
9 MASS
I
I
IN A . M . U .
L,
,
I " "
MASS
*
,
I
,
600
:
,
,
,
,
,
, 1 . .
,
600
. 1000 m/ z
IN
A U
U
800
I I ' 1200
" " " ' I '
4CO
Figure 4. Comparison of positive ion -a of Seenox 4125 acquked by (a) FAB, (b) C02 LD-FTMS, and (c) NdYAG LD-FTMS.
additives containing thioesters and hindered phenols, in terms of both sensitivity and compound identification. The NdYAG instrument appears to offer a modest advantage over the COz laser for mixture analysis because interfering fragment ions are almost nonexistent. T o demonstrate this, a five-component mixture of five of the additives was analyzed. Figure 7 is a single-shot LD-FTMS acquired by using the ND:YAG laser showing, as desired, only the (M + K)+ molecular species. Compound identification is simplified, and especially for FTMS in which a limited trapped ion cell dynamic range is available, the reduced contribution of fragment ions is essential to mixture analysis. LD Spectra of Phosphites. Several other classes of polymer additives were evaluated by LD-FTMS, but with less success. Specifically, antioxidants containing phosphite and phosphonite functional centers tended to fragment far more readily than the hindered phenol and thioester additives. No molecular species were obtained for Weston 618 using COP LD. Shndostab PEPQ, a phosphonite with substituted dibutylphenols, also underwent substantial fragmentation by both FAB and LD. However, with the COz laser was it possible to generate substantially abundant molecular ion
-
I 400
m/
600
z
800
F W e 5. Comparison of positive ion spectra of ooodrite 3114 acquired by (a) FAB, (b) COPLD-FTMS, and (c) Nd:YAG LD-FTMS.
species, although this was accomplished infrequently and with reduced sensitivity; a representative spectrum is presented in Figure 8. Note that the mechanism for ionization of Sandostab PEPQ must differ from that of the other polymer additives in that cation-attached molecular species are not observed. FAB spectra yield no molecular ion species but instead show extensive fragmentation, including a prominent m l z 205 ion corresponding to cleavage a t the ether linkage. The Nd:YAG LD spectrum of Sandostab PEPQ closely resembles the FAB spectrum. However, as ND:YAG laser power is increased, the average mass of fragment ions increases and a low abundance M+is observed. This is in agreement with the general trend observed with the ND:YAG interface that as laser power from the relatively unfocused beam is increased, fragmentation is reduced. However, in contrast with the five previous polymer additives analyzed, i t was not possible to achieve conditions under which fragmentationwas eliminated. LD Spectra of a Polyhindered Amine. One final poly-
ANALYTICAL CHEMISTRY, VOL. 62, NO. 11, JUNE 1, 1990
1171
DSTDP
DLTDP
I
II
I
-1
,
II
4-L
12 0
11,
-.I
l4dO
1040
200
400
60b
MASS I N A . M . U .
1odo
1200
Flgure 8. UlFTMS positive ion spectrum of Sandostab PEW acquked by using the COP laser. 1
?OO
400
Figure 6. Comparison of spectra of Irganox 1010 acauired by (a) negative ion FAB, (b) posithk ion COP LbFTMS, and (c) positive on Nd:YAG LD-FTMS.
mer additive, Spinuvex A36, a polyhindered amine with UVstabilization properties, was examined. This additive is quite different in structure from the antioxidants previously discussed in that it is itself a polymer with a distribution extending beyond 4000 daltons. Because LD-FTMS has been demonstrated to yield effectively, without mass discrimination, the envelope of ions representing the oligomeric distribution of many classes of polymers (21),it was decided to examine its potential for this type of compound as well. Figure 9 is a C02LD-FTMS spectrum of Spinuvex A36, which demonstrates the expected distribution of (M + K)+ ions centered around n = 5 in a nearly symmetric envelope ranging from n = 1 to 9. Again FAB data yielded significant fragmentation, although ions with m / z 1723 corresponding to an (M + K)+ ion for n = 4 were observed. The Nd:YAG spectra were also incapable of providing the correct distribution, because the spectra generated were very much a function of laser power density. As laser energy was increased to the maximum possible through the fiber optic interface, the distribution shifted from one with a maximum at n = 2 to one with a maximum at n = 3, although no oligomeric ions beyond n = 4 were observed.
Flgure 9. LD-FTMS positive ion spectrum of Spinuvex A36 acquired by using the COz laser.
Analysis of Irganox 1010. Irganox 1010 was selected as a model compound for comparison of extraction and direct analysis procedures because it reproducibly generates a spectrum dominated by a cation-attached molecular ion. LD-FTMS of Irganox samples extracted with cyclohexane from commercial resins were easily obtained with the C02 laser system. The spectra were as expected, exhibiting abundant cationized molecular ions. Irganox 1010 mixed intimately in a polyethylene 2000 at varying levels of concentrationwas then examined directly with the C02LD. As shown in Figure 10, at the part-per-hundred and part-per-thousand levels, LDFTMS generates the characteristic molecular ion of Irganox,
1172
ANALYTICAL CHEMISTRY, VOL. 62, NO. 11, JUNE 1, 1990
the sample will be increased sufficiently to detect cationized molecular ions for the polymer additives not yet successfully analyzed with this system. Extension of this technique to commercial samples has begun, with the goal of reliably detecting nonvolatile additives directly in polymer matrices. Registry No. DLTDP, 123-28-4;DSTDP, 693-36-7;Irganox 1010, 6683-19-8; Goodrite 3114, 27676-62-6; Seenox 4125, 29598-76-3; Sandostab PEPQ, 38613-77-3; Spinuvex A36, 88003-10-5.
LITERATURE CITED Grichter.. R.; Muller. H. PlasffcAWAYve Hasndbook; Hanser Publishers: New York. 1984. Flick, E. W. plastic AaWHves: an It?dust&l GI&?; Noyes Publications: Park Ridge, NJ, 1986. LaWl”lM, R. P.; R. E. MaSS SpeCtrOm. R8V. 1985, 4,369-390. Yoshikawa, T.; Ushimi, K.; Kimura, K.; Tamura, M. J . Appl. P o r n . Sci. 1971. 75. 2065-2072. R&wlcz,’P.; Munson,-B. Anal. Chem. 1988, 58, 358-381. Riley, T. L.; Prater, T. J.; Getlock, J. L.; deVrles. J. E.; Schuetzle, D. Anal. Chem. 1984. 56, 2145-2147. Lather. R. P.: Hanis. R. E.: Rhee, C. K.; Schulten, H.43. Anal. Chem 1988, 58, 3188-3195. Lattlmer, R. P.; Harris, R. E. Rubber Chem. Technol. 1989, 6 1 , 839-657. Barber, M.; Borddi, R. S.; Sedgwich, R. D.; Tayler. A. N. J . Chem. Soc., Chem. Commun. 1981, 325-327. Grade, H.; Winograd, N.; Cooks, R. 0. J . Am. Chem. SOC.1977, 99, 7725-7726. Cotter, R. J. Anal. Chem. 1988, 60, 781A-791A. Beckey, H. D. J . Anal. At. specdwn.Ionphys. 1989, 12. 500-503. Lattimer, R. P.; Schulten, H A . Anal. Chem. 1989, 61, 1201A1215A. McCrery, D. A.; Ledford, E. B., Jr.; Gross, M. L. Anal. Chem. 1982, 5 4 , 1435-1437. Wilklns, C. L.; Well, D. A.; Yang, C. L. C.; Ijames, C. F. Anal. Chem. 1985. 57, 520-524. Cotter, R. J. Anal. Chem. 1981, 53,719-720. cotter, R. J.; Honovich, J. P.; Olthoff, J. K.; Lattimer, R. D.Macromolecules 1986, 19, 2996-3001. Karas. M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. Asamoto, B.; Young, J. R.: Citerin, R. J. Anal. Chem. 1990. 6 2 , 61-70. Nuwaysir, L. M.; Wilkins, C. L. Anal. Chem. 1988, 60, 279-282. Brown, R. S.; Weil, D. A.; Wilkins. C. L. MacromoLscuLss 1988, 19, 1255-1260. Laude, D. A., Jr.; Beu, S. C. Anal. Chem. 1989, 61, 2422-2427. David A. Weil, personal communication.
.
Flgure 10. LD-FTMS positive ion spectra of Irganox 1010 acquired directly from a polyethylene 2000 matrix at the (a) 3% level and (b) 0.1 % level.
which is sufficient chemical information for identification.
CONCLUSIONS Demonstrated in this paper is the potential of LD-FTMS for analysis of polymer additives. For all additives that were analyzed, LD is superior to FAB. The important classes of antioxidants containing hindered phenols and thioester functionalities yield high sensitivity spectra with dominant, characteristic molecular species. Although both C02 and Nd:YAG spectra exhibit abundant cation-attached molecular ions, NdYAG spectra generally show less fragmentation, suggesting their possible superiority for mixture analysis. However, with this laser, the spectra are more dependent on a variety of factors including laser energy, laser power, and sample preparation. These aspects of LD-FTMS are still not well understood. Progress continues toward a refinement of the Nd:YAG LD-FTMS interface and it is expected that, with additional beam focusing, the laser power distribution over
RECEIVED for review December 11,1989. Accepted February 26,1990. C.L.W. gratefully acknowledgespartial support of this research at the University of California Riverside, by Grant CHE-89-11685 from the National Science Foundation. D.A.L. acknowledges support for research conducted at the University of Texas at Austin from grants by the Welch Foundation and the Texas Advanced Research Program.
