Analysis of Poly (oxyethylene) and Poly (oxypropylene) Triblock

Apr 8, 2005 - Laboratoire Analyse et Environnement, Université d'Evry Val d'Essonne, CNRS UMR 8587,. Bd François Mitterrand, 91025 Evry Cedex, France...
0 downloads 0 Views 1MB Size
Download PDF
Anal. Chem. 2005, 77, 3292-3300

Analysis of Poly(oxyethylene) and Poly(oxypropylene) Triblock Copolymers by MALDI-TOF Mass Spectrometry Peran Terrier, William Buchmann,* Ghislain Cheguillaume, Bernard Desmazie`res, and Jeanine Tortajada

Laboratoire Analyse et Environnement, Universite´ d’Evry Val d’Essonne, CNRS UMR 8587, Bd Franc¸ ois Mitterrand, 91025 Evry Cedex, France

Triblock copolymers of ethylene oxide (EO) and propylene oxide (PO) are widely used in the chemical industry as nonionic surfactants. Triblock copolymers can be arranged in a EO-PO-EO or PO-EO-PO sequence. This arrangement results in an amphiphilic copolymer, in which the block sequence and block length determine the properties of the copolymer. MALDI-TOF MS was used to analyze various triblock copolyethers: EO-PO-EO (M h n )2000 g‚mol-1), PO-EO-PO (M h n ) 2000 g‚mol-1), and a random copolymer EO/PO (M h n ) 2500 g‚mol-1). Data treatment was assisted by using a homemade software allowing a picture of monomer composition of oligomers from the mass spectra. MALDI-TOF mass spectra of EO/PO copolymers were shown to depend strongly on the number of laser shots, relative proportions of polymer/salt, and the nature of the matrix. An unsaturated byproduct was detected. Its presence was demonstrated by prefractionation of copolymers by SEC before MALDI-TOF analysis, and its content was estimated by 1H NMR. The formation of layers inside the MALDI deposit was evidenced by varying the number of laser shots. Lighter oligomers of the copolymer, unsaturated byproduct, or both would be in the core of the deposit, coated with heavier oligomer. The layer formation depends on the nature of the matrix and the quantity of added salt. DHB matrix with a relative high sodium salt content induces layer formation inside the deposit, whereas dithranol matrix or low salt content does not. Consequently, an optimization of experimental parameters in order to estimate the lighter oligomers or unsaturated byproduct content or to obtain the actual representation of the monomer contribution in the copolymers from the MS data only seems obviously critical. MALDI-TOF mass spectrometry is obviously a powerful technique to analyze copolymers, but a careful survey of the experimental parameters is required. The combination of MALDI-TOF MS with separations techniques and NMR brings precious complementary information. Mass spectrometry plays a major role in structural studies of biopolymers; however, several other applications of mass spec* Corresponding author. E-mail: [email protected].

3292 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

trometry concern synthetic polymers.1-6 The development of soft ionization methods such as electrospray by Yamashita and Fenn7 and matrix-assisted laser desorption/ionization (MALDI) introduced by Tanaka et al.,8 and by Karas and Hillenkamp,9 contributed to improve the characterization of synthetic polymers. These well-established ionization techniques can now be used to determine average molecular weights (M h n and M h w) and calculate polymerization degrees DPn and polydispersity indices Ip. The obtained values are known to be relevant only if the polymer is little dispersed. Mass spectrometry can help to check the nature of the repeat units and of the ends. Until recently, works in this field aimed to mainly analyze homopolymers such as poly(ethylene glycol), poly(dimethylsiloxane), poly(methyl methacrylate), and polystyrene. A few of them concerned copolymers, i.e., polymers constituted of more than only one type of repeat unit because of the difficulties related to the interpretation of mass spectra.1,2,10 In this paper, we describe the characterization of triblock copolymers of ethylene oxide and propylene oxide. Block copolymers of ethylene oxide (EO) and propylene oxide (PO) are widely used in the chemical industry as nonionic surfactants. In particular, they are now studied as novel nonviral vectors in therapeutics for drug and gene delivery.11,12 These polymers are extensively used in complex formulations since they exhibit the original capability to arrange themselves into micelles. The EO block (-CH2-CH2-O-)n of the copolymer is well known to be hydrophilic while the PO block (-CH(CH3)-CH2-O-)n is rather hydrophobic. The triblock copolymers can be arranged in a EOPO-EO or PO-EO-PO sequence. This arrangement results in an amphiphilic copolymer, in which the block sequence and block (1) Montaudo, G.; Lattimer, R. P. In Mass Spectrometry of Polymers; Montaudo, G., Lattimer, R. P., Eds.; CRC Press: Boca Raton, FA, 2002. (2) Pash, H.; Shrepp, W. In MALDI-TOF mass spectrometry of synthetic polymers; Barth, H. G., Pash, H., Eds.; Springer-Vermag: Berlin, Germany, 2003. (3) Scrivens, J. H.; Jackson, A. T. Int. J. Mass Spectrom. 2000, 200, 261-276. (4) Hanton, S. D. Chem. Rev. 2001, 101, 527-569. (5) Nielen, M. W. F. Mass Spectrom. Rev. 1999, 18, 309-344. (6) McEwenn, C. N.; Peacock, P. M. Anal. Chem. 2002, 74, 2743-2748. (7) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-4459. (8) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, T. Rapid Commun. Mass Spectrom. 1988, 2, 151-153. (9) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (10) Montaudo, M. S. Mass Spectrom. Rev. 2002, 21, 108-144. (11) Kabanov, A. V.; Lemieux, P.; Vinogradov, S.; Alakhov, V. Adv. Drug Delivery Rev. 2002, 54, 223-233. (12) Kabanov, A. V.; Batrakova, E. V.; Alakhov, V. J. Controlled Release 2002, 82, 189-212. 10.1021/ac048193m CCC: $30.25

© 2005 American Chemical Society Published on Web 04/08/2005

length determine the properties of the copolymer. Efficient analytical methods are thus required to check the composition of these compounds. Classical analytical techniques, such as IR/FT spectroscopy, NMR spectroscopy, and size exclusion chromatography can provide some structural information, including the determination of the macromolecular chain ends, the average number of EO and PO units, and the average molecular weight. However, these methods may not be sufficient because they are averaging methods. A polymer is a mixture, and the great advantage of mass spectrometry is to allow the examination of each individual oligomer. Thus, mass spectrometry can, for example, easily distinguish a copolymer from a homopolymer blend whereas most classical techniques are not able to differentiate easily. EO and PO copolymers have already been evoked in several papers. Some authors used EO/PO copolymers in order to evaluate and solve the technical difficulties related to the use of trapping systems, i.e., ion trap or Fourier transform ion cyclotron resonance (FT-ICR) mass analyzers. For instance, a series of amine-terminated EO/PO copolymers (Jeffamines from Texaco Chemical Co.) were analyzed by laser desorption (no matrix was added) FT-ICR MS by Simonsick et al.13 MALDI FTICR results were combined and compared with those of 1H and 13C NMR data by Van der Hage and co-workers.14 Van Rooij and Heeren concentrated their efforts on the correction of the distorted molecular weight distributions induced by the use of an external source MALDI on a FT-ICR instrument.15 The same authors studied the performance of an ion trap equipped with an external source MALDI with respect to mass dependencies in trapping efficiency and mass shifts due to space-charge effects measuring the molecular weight distributions of copolymers.16 Although the use of the FT-ICR technique becomes very useful in the case of complex polymeric systems due its high-mass resolving power and high-mass accuracy, this technique is still very expensive and MALDI-TOF mass spectrometry remains the most affordable and easy to use technique. MALDI-TOF MS is often the most open mass spectrometry technique to polymer specialists. The work of Schriemer and Li illustrated the analytical merits of time-lag focusing MALDI-TOF MS in the analysis of amine-terminated PO-EO-PO copolymers.17 Chen and Li18 used MALDI-TOF MS to characterize four similar EO/PO copolymers bearing one double-bond end. Unfortunately, these works were carried out on a linear instrument resulting in peak broadening and overlapping. The same authors showed that variations in experimental conditions in MALDI can have a significant effect on the mass spectral appearance of two EO/PO copolymers bearing a C16 or C18 alkyl chain end.19 These authors emphasized that understanding how the spectra can be affected by the experimental conditions is (13) Nuwaysir, L. M.; Wilkins, C. L.; Simonsick, W. J. J. Am. Soc Mass Spectrom. 1990, 1, 66-71. (14) Van der Hage, E. R. E.; Duursma, M. C.; Heeren, R. M. A.; Boon, J. J.; Nielen M. W. F.; Weber, A. J. M.; de Koster C. G.; de Vries, N. K. Macromolecules 1997, 30, 4302-4309. (15) Van Rooij, G. J.; Duursma, M. C.; de Koster, C. G.; Heeren, R. M. A. Boon, J. J.; Wijnand Schuyl, P. J. Anal. Chem. 1998, 70, 843-850. (16) Van Rooij, G. J.; Boon, J. J.; Duursma, M. C.; Heeren, R. M. A. Int. J. Mass Spectrom. 2002, 221, 191-207. (17) Schriemer, D. C.; Whittal, R. M.; Li, L. Macromolecules 1997, 30, 19551963. (18) Chen, R.; Tseng, A. M.; Uhing, M.; Li, L. J. Am. Soc. Mass Spectrom. 2001, 12, 55-60. (19) Chen, R.; Zhang, N.; Tseng, A. M.; Li, L. Rapid Commun. Mass Spectrom. 2000, 14, 2175-2181.

clearly the first step in developing a mass spectrometric approach for compositional quantitation. The explored parameters were the analyte concentration, laser power, type of matrix, and solvent used in MALDI preparation. Nevertheless, their interpretations were based directly on mass spectra without any computerassisted treatment of the MS data. More recently, Suen et al.20 studied two copolymers, one random EO/PO copolymers derived from glycerin and one triblock PO-EO-PO. 3-D bivariate deconvolutions of the mass spectra (one for each copolymer) were presented, bringing important information on drift and heterogeneity of copolymer composition. Finally, they suggested that computer assistance needed to be developed to accelerate the analysis. To our knowledge, only a very few reports of computerassisted treatment of the MS data were reported.15 In this work, MALDI-TOF MS was used to analyze two hydroxyl-terminated triblock copolyethers: EO-PO-EO (M hn )2000 g‚mol-1), PO-EO-PO (M h n ) 2000 g‚mol-1), and a random EO/PO copolyether (M h n ) 2500 g‚mol-1). The influence of various experimental parameters (number of laser shots, concentration in added salt, nature of the matrix) was studied, bringing further information to the earlier work reported by Li and co-workers.19 Under certain circumstances, the number of laser shots was unexpectedly found to be a very important parameter. A homemade software was systematically used in order to obtain a rapid appreciation of the drift of the apparent copolymer composition as a function of the experimental parameters. We hope that the obtained results will improve the existing understanding of the MALDI experiment applied to copolymers, showing with some examples, how and why MALDI-TOF MS alone can experience some difficulties to give accurate average molecular weights or the actual monomer proportions in a copolymer chain. EXPERIMENTAL SECTION MALDI-TOF MS. All experiments were performed using a Perseptive Biosystems Voyager-DE Pro STR MALDI-TOF mass spectrometer (Applied Biosystems/MDS Sciex, Foster City, CA). This instrument was equipped with a nitrogen laser (λ ) 337 nm). The mass spectrometer was operated in the positive ion reflectron mode with an accelerating potential of +20 kV. Mass spectra were recorded with the laser intensity set just above the ionization threshold (2000-3500 in arbitrary units, on our instrument) to avoid fragmentation and to maximize the resolution (pulse width 3 ns). Typically, three acquisitions corresponding to 50 shots each were summed for each deposit to obtain representative mass spectra. External calibration using a mixture of two poly(ethylene glycol) standards (M h n ) 1000 and M h n ) 2000) was performed with the same matrix as in the experiment. The instrument is routinely capable of performing mass spectra of 6000-8000 mass resolution on the 500-2500 Da mass range with 80 °C during several days). Although the virgin material before thermal degradation was mostly dihydroxyl terminated, small amounts of ethylenic end groups were observed by 1H NMR. The presence of low molecular weight products was also reported by Lattimer23 in the MALDI-TOF mass spectra of poly(ethylene glycol) by increasing pyrolysis temperature, e.g., by deliberately degrading the polymer before analysis. Vinyl ether end groups due to dehydratation of hydroxyl end groups of poly(21) Barton, Z.; Kemp, T. J.; Buzy, A.; Jennings, K. R. Polymer 1995, 36, 49274933. (22) Gallet, G.; Carroccio, S.; Rizzarelli, P.; Karlsson, S. Polymer 2002, 43, 10811094.

(ethylene glycol) were identified. Moreover, the same author noted some fragmentation at higher laser powers. To our knowledge, whereas several authors have reported the degradation of polyethers at high laser power, there has not been any publication discussing the modification of the MALDI mass spectrum when the number of laser shots is increased. It must be emphasized that MALDI TOF mass spectra of copolymer fractions where byproducts were removed by SEC do not show any modifications upon increasing the number of laser shots (results not shown). Under our conditions, the observed effect in mass spectra does not result from decomposition during the laser shots; therefore, the sample already contained a byproduct. As a lesson to be learned about MALDI, the presence of byproducts can be evidenced by modifying the number of laser shots. Data Processing from MALDI-TOF Mass Spectra. Taking into account the very large number of ions in the mass spectra, (23) Lattimer, R. P. J. Anal. Appl. Pyrolysis 2000, 56, 61-78.

Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

3297

Figure 6. 2D plots created from MALDI-TOF mass spectra of EO-PO-EO 2000 and PO-EO-PO 2000 copolymers with different proportions of added salt. P/S denotes a [polymer]/[salt] ratio.

the assignment of the various peaks was automatically determined using a homemade software (Figure 3). Figure 3a shows a typical MALDI-TOF mass spectrum obtained from a PO-EO-PO triblock copolymer. Ions are fully assigned in a zoom of the middle of the mass range (Figure 3b). For example, the ion at m/z 1720 corresponds to EO21PO13,Na+ (noted 21/13) in the mass spectrum. The ion at m/z 1764 contains one EO repeat unit more (22/13 at +44 mass unit). After a deisotoping process (Figure 3c), the software determines the contribution of each monomer for each ion coming from the copolymer and generates representations of the relative abundance of the ions as a function of the number of EO and PO units (Figure 3e). These plots give an immediate representation of the apparent composition of the copolymer. The user needs only to enter the values of the exact molar masses of the two repeat units, the two ends, and the mass of the cation involved in the formation of the ions observed in the mass spectrum (Figure 3d). The software associates abundance with the various copolymer compositions. The algorithm only searches for ions corresponding to the copolymer; thus, unsaturated byproducts should have not been considered. Nevertheless, due to the proximity of masses, problems of assignment can be encountered. The masses of the byproducts interfered with those 3298

Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

of the copolymer chains. In Figure 4, two tables giving the theoretical masses of the sodium adducts of the copolymer EOPO-EO chains and those of the unsaturated byproduct as a function of the number of EO and PO units are shown. Since it is not necessary to consider the ions in detail in the Figure 4, these tables were simplified, some lines and columns have been deleted for clarification. Several areas of ambiguity have been revealed on comparing these two tables. The ions present in the two light gray frames and those present in the dark gray spaces show almost the same masses. For example, diol copolymer EOxPOy and unsaturated byproduct EOx+7PO)y-6 differ by only 0.036 mass unit in the dark gray spaces (∼25 ppm of mass difference for an ion at m/z 1500) while diol copolymer EOxPOy and unsaturated byproduct EOx-22PO)y+16 differ by 0.13 mass unit (light gray frames). Fortunately, the EO-PO-EO 2000 copolymer has a known average composition: 4EO, 31PO. The ions belonging to the copolymer are expected in “A” area in the Figure 4. Unsaturated byproducts were identified with masses expected in “B” area, which can be divided in two parts. The white part is not subject to ambiguity, whereas the dark part could. Consequently, some unsaturated byproducts can be assigned to the copolymer. A small contribution (“C” in Figure 4) due to the byproduct can be

Figure 7. 2D plots created from MALDI-TOF mass spectra of EO-PO-EO 2000, PO-EO-PO 2000, and EO/PO random copolymers using dithranol matrix.

observed in the 2D plot of the composition of the copolymer after computation. Figure 5 shows the contour plots we obtained from EO-POEO 2000 after computer processing as a function of the number of laser shots (50 or 300). These contour plots evidence the laser shot effect discussed above. The extent of the low molecular weight contribution finally can appear less important than expected compared to the differences we observed in corresponding mass spectra (compare Figure 1a and c). It should be kept in mind that only the ions in the dark part of “B” area can be accounted for the copolymer, not all the ions in “B” area. Moreover, relative abundance of all ions over the entire mass range is not exactly as it appears at first sight. Isotopic contribution is in fact important with higher masses. Influence of the Salt Concentration and Matrix. Figure 6 shows the plots we obtained from EO-PO-EO 2000 and POEO-PO 2000 copolymers after computer processing depending on different proportions of added salt. P/S denotes a [polymer]/ [salt] ratio. All mass spectra were recorded with 50 laser shots in order to limit the emergence of the transfer reaction products in the 2D plots. When comparing the various 2D plots of EO-POEO 2000, it appears that the expected copolymer distribution (“A” in Figure 4) and the byproduct contribution, i.e., short diol or ethylenic chains (“C” contribution in Figure 4) are perfectly distinguishable when the salt is not in excess (P/S ) 1/0.5) even if the acquisition is made with only 50 laser shots. In the case of

an important excess (P/S ) 1/50) an increase of the sodium acetate concentration leads to the disappearance of the byproducts. Moreover, the copolymer distribution appears narrower. A similar conclusion can be made from the PO-EO-PO 2D plots. Only the expected copolymer distribution appears with an important excess of salt (P/S ) 1/50). One possible explanation could be that layer formation is favored in the presence of an important salt excess only (i.e., low P/S ratio). Several observations lead to this conclusion. EO-POEO 2000 and PO-EO-PO 2000 copolymers give at P/S ) 1/0.5 almost the same ions in the low-mass range as those reported with 300 laser shots and P/S ) 1/50. Moreover, whatever the number of laser shots at a P/S ) 1/0.5 ratio, mass spectra do not undergo strong modifications as mentioned at a P/S ) 1/50 ratio. Consequently, there is no layer formation if sodium salt is not in sufficient excess. In addition, layer formation as a funtion of P/S ratio was not clear in the case of the EO/PO random copolymer (2D plots not shown). Layer formation could be related to the capacity of EO-PO-EO or PO-EO-PO copolymers to arrange themselves into micelles. Figure 7 shows what happens when dithranol matrix was used instead of DHB. All mass spectra were recorded with 50 laser shots, and the ration P/S was kept 1/5. Whatever the copolymer sequence, EO-PO-EO, PO-EO-PO, or random EO/PO, the ions corresponding to the byproducts or PO-rich chains were strongly intensified using dithranol (compare to Figure 6). It Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

3299

appears clearly that, using a more hydrophobic matrix, desorption-ionization of the most hydrophobic chains is favored. It is worth noting that ions of higher intensity in the low-mass range were obtained with dithranol instead of DHB from the very first laser shots and that their abundance was kept constant whatever the number of laser shots. Dithranol matrix did not seem to induce the formation of layers inside the deposit. Since layer formation inside the MALDI deposit prevents the desorption-ionization of some chains for a given laser shot number, it can be interesting to work with low quantities of sodium salt in order to avoid this phenomenon. Nevertheless, without an important excess of sodium salt, others types af adducts can complicate the mass spectrum. Furthermore, even in the absence of layers inside the deposit, it is well known that preferential desorption-ionization can occur during MALDI process. Chains bearing different end groups can be detected with different signal intensities even if they are in similar amounts in the sample. Due to the high affinity of the sodium cation to the hydroxyl group, sodium salt should promote the preferential desorption-ionization of the diol copolymer chains since they exhibit two hydroxyl ends instead of only one as in the case of the unsaturated byproduct. Consequently, low amounts of sodium salt should promote preferential desorption-ionization of the diol copolymer before unsaturated byproducts whereas high amounts of sodium salt should give ions coming from the copolymer chains plus the unsaturated byproducts. It is surprising to observe the inverse phenomenon if we assume that the low-mass contribution in the contour plot comes from unsaturated byproducts mainly. POEO-PO 2000 gives the same result. One of the possible reasons could be that the sodium salt has the highest affinity to the byproduct due to the fact that it is rather EO-rich when compared to the diol copolymer. Indeed, in not shown MALDI-TOF experiments in which intensity of equimolar homopolymers EOn and POn were compared for different P/S ratios, the most intense ions for low P/S were EOn while they were POn at high P/S ratios. This result suggests a higher affinity of sodium cation for EOn than for POn. This is in agreement with the fact that the number of oxygen atoms coordinated to Na+ is higher for EOn than for POn.24 In the presence of a limiting concentration of salt, the unsaturated byproduct would be more intense because of its better affinity for sodium. (24) Gidden, J.; Wyttenbach, T.; Jackson, J. H.; Scrivens, J. H.; Bowers, M. T. J. Am. Chem. Soc. 2000, 122, 4692-4699.

3300

Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

CONCLUSION Characterization of various triblock copolyethers, EO-POEO, PO-EO-PO, and a random EO/OP copolymers (M h n ≈ 2000 g‚mol-1), can be achieved by MALDI-TOF mass spectrometry. The determination of the copolymer composition was assisted by using a homemade software in order to obtain a rapid qualitative analysis of the material. However, MALDI-TOF mass spectra of EO/PO triblock copolymers depend strongly on MALDI parameters such as the number of laser shots, relative proportions of polymer/salt, and nature of the matrix. The formation of layers inside the MALDI deposit was evidenced by varying the number of laser shots. The layer formation depends on the matrix, the quantity of added salt, and very likely the heterogeneity of the copolymer (presence of byproduct or bimodal distribution). The contribution of reaction transfer byproducts appeared when increasing the number of laser shots or decreasing the concentration of sodium acetate. When dithranol matrix is used instead of DHB, the ions corresponding to the most hydrophobic chains are intensified. Layer formation was involved with DHB matrix with a high sodium salt content, whereas dithranol matrix did not induce the formation of layers inside the deposit. Under these circumstances, an optimization of experimental parameters in order to obtain the actual representation of monomer unit contribution in the copolymers or that of the byproduct contribution from the MS data only seems obviously critical. Additional 1H NMR data were required to refine the knowledge of the nature of the unsaturated byproduct and estimate its content. Prefractionation of copolymers by SEC was very useful to demonstrate the presence of a byproduct. However, it is worth noting that, as we have seen, even within a SEC fraction, block size and block arrangement can be heterogeneous. In this case, consecutive 1H NMR analysis can be insufficient and more efficient separations are required. MALDI-TOF mass spectrometry is obviously a powerful technique to analyze copolymers but a careful survey of the experimental parameters is required. The combination of MALDI-TOF MS with separation techniques and NMR brings precious complementary information.

Received for review December 7, 2004. Accepted March 16, 2005. AC048193M