Enzyme-catalyzed quantitative chain-end functionalization of poly(ethylene glycol)s under solventless conditions

Abstract

This is the first report of quantitative vinyl chain-end functionalization of poly(ethylene glycol)s (PEGs) with reduced environmental effects and renewable catalysis. Divinyl adipate (DVA) was transesterified using Candida antarctica lipase B (CALB) supported on an acrylic resin at 50 °C under dry nitrogen and solventless conditions. After the reactions CALB was removed by filtration and excess DVA was recovered by hexane extraction followed by distillation for reuse. ¹H and ¹³C NMR spectroscopy and MALDI-ToF mass spectrometry was used to analyze the structure and purity of the products. The effects of DVA excess and PEG chain length were investigated. Model experiments with tetra(ethylene glycol) (TEG) led to polycondensation. At DVA/TEG 20/1 molar ratio ∼82% of the product was Vinyl–TEG–Vinyl, together with vinyl-telechelic dimers and trimers. When reacting monomethoxy MeO–PEG–OH of M_n = 1100 g mol⁻¹ under the same conditions, pure MeO–PEG–Vinyl was obtained with no coupling. MeO–PEG–OH with M_n = 2000 g mol⁻¹ gave pure MeO–PEG–Vinyl at 5 molar excess of DVA. With HO–PEG–OH₁₀₀₀ of M_n = 1000 g mol⁻¹ at DVA/PEG 20/1 molar ratio no polycondensation and only 2% coupled product was found, while HO–PEG–OH with M_n = 2000 g mol⁻¹ under the same conditions gave pure telechelic Vinyl–PEG–Vinyl.

---

Introduction

Enzymes, nature's catalysts have widely been used in biotransformations but are still not in the mainstream of catalysis.^1,2 Global environmental issues with the regulatory limitations on chemical and pharmaceutical industries have increased the interest in ‘Green Processes’ using enzyme catalysis, reagent recycling and other green chemistry principles.^3–5 Enzyme catalysis was found to be very effective in organic reactions involving small molecules. For example, the efficiency of Candida antarctica lipase B (CALB)-catalyzed transesterification of vinyl-acetate with 2-phenyl-propane-1-ol was demonstrated by comparison with tin octoate; this latter gave 95% conversion when reacting vinyl-acetate with 2-phenylpropane-1-ol in 12 hours, while CALB yielded 100% conversion of the vinyl-acetate in 2 hours.⁶ Enzymes have also been used to initiate polymerizations.^7,8 However, before the work of our group no quantitative chain-end functionalization of polymers had been reported. Earlier reports of enzyme-catalyzed polymer functionalization have been less than optimal. Telechelic carboxylic acid functionalized polydimethylsiloxanes (PDMSs) were reacted with α,β-ethylglucoside at 70 °C under vacuum for 34 hours in the presence of CALB, but the product was a mixture of mono- and difunctional esters.⁹ Recently the synthesis of telechelic methacrylate-functionalized oligoesters by the CALB-catalyzed polycondensation of ethylene glycol, hydroxyethyl methacrylate and divinyl adipate (DVA) was reported.¹⁰ However, 5–9% of the product was not difunctional methacrylate as shown by MALDI-ToF analysis. Together with difunctional methacrylate oligomers, chains containing a monofunctional methacrylate group on one end and a free acid end group on the other, oligomer chains containing a monofunctional methacrylate group on one end and a sodium carboxylate salt formed from the sodium ion during analysis and chains with unreacted HO-end groups and cyclic chains were identified. The authors stated that the free acid ends were the result of traces of water in the system that hydrolyzed some oligomers. Our group has shown that solid-supported CALB catalyzed quantitative methacrylate end-functionalization of poly(ethylene glycol)s (PEGs), polyisobutylenes and PDMSs without the use of solvents.^11–15 Specifically, low molecular weight PEGs were liquefied at 50 °C and were quantitatively end-functionalized by transesterification of vinyl methacrylate within 4 hours under a dry nitrogen atmosphere.¹⁵ No hydrolyzation of the ester functions were observed and pure telechelic PEG dimethacrylate was obtained.

Vinyl ester functionalized telechelic PEG building blocks are very attractive for further transformations. Transesterification of vinyl ester–PEGs with functionalized alcohols would lead to new functional groups with high efficiency: the vinyl alcohol product immediately tautomerizes into acetaldehyde, rendering the reaction irreversible as shown in Scheme 1.

Scheme 1 Transesterification of vinyl ester–PEG with functionalized alcohols. The red ball represents functional groups derived from the alcohol.

We set out to investigate the transesterification of DVA with PEGs without the use of solvents. The polycondensation of DVA with various small diols via enzyme catalysis in various solvents to produce polyesters has extensively been investigated and reviewed.^7,8,16–18 Based on experimental data and a mathematical model it was concluded that the polymerizations occurred by a step-condensation mechanism and the rate of polymerization decreased with increasing diol molecular weight.¹⁷ The transesterification of DVA with 1,4-butanediol in bulk catalyzed by CALB yielded the corresponding polyester with M_n = 8273 g mol⁻¹ and M_w/M_n = 2.78.¹⁸ Hydrolysis of the DVA invariably occurred so the product had a mixture of vinyl- and HO-end groups. The cause of the hydrolysis was believed to be the release of water attached to the enzyme. The presence of a solvent such as tetrahydrofuran was found to promote this process, leading to more hydrolysis.

This paper is the first report about the quantitative enzyme-catalyzed vinyl end-functionalization of PEGs using green chemistry principles.

Experimental

Materials

Lipase B from Candida antarctica immobilized on microporous acrylic resin (Novozym 435, 20 wt% CALB, recombinant, expressed in Aspergillus niger, 10 000 propyl laurate unit per g activity, Sigma), tetraethylene glycol (TEG, 99%, Aldrich), poly(ethylene glycol) monomethyl ethers (M_n = 1100 g mol⁻¹, M_w/M_n = 1.09, Polymer Source, Inc. MeO–PEG–OH₁₁₀₀ and M_n = 2000 g mol⁻¹, Aldrich MeO–PEG–OH₂₀₀₀ – with M_w/M_n = 1.46 measured in our lab), poly(ethylene glycol)s (M_n = 1000 g mol⁻¹, M_w/M_n = 1.08, Polymer Source, Inc. HO–PEG–OH₁₀₀₀, M_n = 2000 g mol⁻¹, M_w/M_n = 1.20, Alfa Aesar, HO–PEG–OH₂₀₀₀), divinyl adipate (DVA, >99%, TCI America), Tetrahydrofuran (THF), hexane, deuterated dimethyl sulfoxide (DMSO-d₆, 99.8%, Chemical Isotope Laboratories) and diethyl ether (Fisher Scientific) were used as received.

Procedures

Functionalizations. All reactions were carried out under dry nitrogen.

Model experiments. TEG (0.2 g, 1.03 mmol, 0.18 mL) was reacted with 3, 6, 10 or 20 equivalent of DVA in the presence of CALB (17 mg, 1.0 × 10⁻⁴ mmol) for 1 hour at 50 °C; [DVA] = [3.0 eq.: 0.61 g, 3.09 mmol, 0.58 mL, 3.60 mol L⁻¹; 6.0 eq.: 1.22 g, 6.18 mmol, 1.16 mL, 4.61 mol L⁻¹; 10.0 eq.: 2.04 g, 10.29 mmol, 1.94 mL, 4.85 mol L⁻¹; 20.0 eq.: 4.08 g, 20.59 mmol, 3.98 mL, 4.95 mol L⁻¹]. After the reaction, 5 mL of dried THF was added to the mixture and the solid CALB was filtered by a 0.45 μm PTFE syringe filter. The excess DVA was removed by silica gel chromatography using hexane–THF (5/1; v/v) as the eluent, and recovered by distilling off the THF and hexane.

Reactions with MeO–PEG–OH. MeO–PEG–OH₁₁₀₀ (1.0 g, 0.91 mmol) was reacted with 1.5, 3, 5, 10 or 20 molar equivalent DVA in the presence of CALB (100 mg, 6.0 × 10⁻⁴ mmol) for 4 hours at 50 °C; [DVA] = [1.5 eq.: 0.27 g, 1.37 mmol, 0.26 mL, 5.33 mol L⁻¹; 3.0 eq.: 0.54 g, 2.73 mmol, 0.52 mL, 5.31 mol L⁻¹; 5.0 eq.: 0.90 g, 4.55 mmol, 0.86 mL, 5.30 mol L⁻¹; 10 eq.: 1.80 g, 9.1 mmol, 1.71 mL, 5.31 mol L⁻¹; 20 eq.: 3.60 g, 18.2 mmol, 3.42 mL, 5.31 mol L⁻¹]. After 4 hours, 5 mL of dried THF was added to the mixtures and the solid CALB was removed by a syringe fitted with a 0.45 μm PTFE filter. The mixtures were cooled to room temperature, yielding solid polymers with excess liquid DVA. The excess DVA was removed by washing with hexane twice and recovered by distilling off the hexane using a rotary evaporator. The polymers were dried in a vacuum oven at room temperature. The same procedures were repeated with MeO–PEG–OH₂₀₀₀ up to 5 molar excess of DVA.

Reactions with OH–PEG–OH. HO–PEG–OH₁₀₀₀ (1.0 g, 1.00 mmol) was reacted with DVA [20.0 eq. (3.96 g, 20.00 mmol, 3.77 mL, 5.31 mol L⁻¹)] in the presence of CALB (100 mg, 6.0 × 10⁻⁴ mmol) for 4 hours at 50 °C. A sample was taken for MALDI-ToF analysis at 2 hours reaction time. After 4 hours, 5 mL of dried THF was added to the mixture and the solid CALB was removed by a syringe fitted with a 0.45 μm PTFE filter. The mixture was cooled to room temperature, yielding a solid polymer with excess liquid DVA. The excess DVA was removed by washing with hexane twice and recovered by distilling off the hexane using a rotary evaporator. The polymer was dried in a vacuum oven at room temperature. The same procedure was repeated with HO–PEG–OH₂₀₀₀.

Characterization

¹H (128 scans) and ¹³C NMR (5000 scans) spectra were recorded on a Varian Mercury-500 NMR spectrometer in DMSO-d₆ using 1 s relaxation time. The resonances at δ = 2.50 ppm (¹H NMR) and δ = 39.51 ppm (¹³C NMR) were used as internal references.

MALDI-ToF mass spectra were acquired on a Bruker Ultraflex-III ToF/ToF mass spectrometer (Bruker Daltonics, Inc., Billerica, MA) equipped with a Nd:YAG laser (355 nm). All spectra were measured in positive reflector mode. The instrument was calibrated prior to each measurement with an external poly(methyl methacrylate) (PMMA) standard. Individual solutions in anhydrous THF (99.5%, Aldrich) of polymer (10 mg mL⁻¹), 1,8,9-trihydroxyanthracene (dithranol, 20 mg mL⁻¹, 97%, Alfa Aesar), and sodium trifluoroacetate (NaTFA, 10 mg mL⁻¹, 98%, Aldrich) were mixed in the ratio of polymer : matrix : cationizing salt (1 : 10 : 2), and 0.5 μL of the resulting mixture were deposited on microtiter plate wells (MTP 384-well ground steel plate). After evaporation of the solvent, the plate was inserted into the MALDI source. The spectra were obtained an acceleration voltage of 20 kV. The attenuation of the Nd:YAG laser was adjusted to minimize unwanted polymer fragmentation and to maximize the sensitivity. The composition was quantified by using the intensity of each peak in the spectrum, directly proportional to the mass fraction of each component. The lower detection limit of the system is below 10 ppm.

Results and discussion

Model experiments

The effect of DVA concentration. TEG was reacted with 3, 6, 10 or 20 molar equivalent of DVA. Table 1 summarizes the product composition data for the various conditions after 1 h reaction time. Interestingly, the amount of oligomers decreased with increasing DVA concentration. At DVA/TEG 3/1 and 6/1, about 50% oligomers were observed. At DVA/TEG 20/1, 81.6% of the product mixture was V–TEG–V, with only 16.7% dimers and 1.7% trimers. The number average molecular weights M_n from MALDI-ToF were calculated based on eqn (1).where N_i is the signal intensity of the polymer with a mass of M_i.¹⁹

Table 1 Effect of DVA/TEG molar ratio on product composition^a

		Composition (%) by MALDI-ToF
		DVA/TEG molar ratio
		3.0	6.0	10.0	20.0
a [CALB] = 1.3 × 10^−4 mol L^−1 (3.0 eq.), 7.6 × 10^−5 mol L^−1 (6.0 eq.), 4.8 × 10^−5 mol L^−1 (10.0 eq.), 2.5 × 10^−5 mol L^−1 (20.0 eq.).
VV Vinyl–[TEG–DVA]n–TEG–Vinyl	V–TEG–V, n = 0	42.5	45.8	73.4	81.6
	Dimer, n = 1	35.5	34.8	23.3	16.7
	Trimer, n = 2	17.7	14.7	3.2	1.7
	Tetramer, n = 3	3.8	4.3	0.1	—
	Pentamer, n = 4	0.5	0.4	—	—
Total		100	100	100	100
M n (g mol^−1)		760	742	594	563

---

The MALDI-ToF of the product of the reaction at DVA/TEG = 6/1 (ESI Fig. S1†) shows that the major product is V–TEG–V, together with polycondensation products (up to pentamer). The signals are separated by m/z = 304.2, corresponding to the C₁₄H₂₄O₇ (TEG-adipate) repeat unit. It is important to recognize that each of these oligomers is symmetric telechelic. Significantly, no hydrolysis of the ester groups was observed.

Fig. S2 in the ESI† section displays the ¹H NMR spectra of the TEG and the product at DVA/TEG = 6/1. The resonance at δ = 4.55 ppm corresponding to the –OH (a) protons of HO–TEG–OH disappeared and the peak of the methylene protons adjacent to the hydroxyl groups shifted downfield from δ = 3.50 to δ = 4.22 ppm (b′). New resonances attributed to the vinylidene [δ = 4.87 (e), δ = 4.65 (e′)], and methine [δ = 7.24 (f)] protons appeared. In the starting HO–TEG–OH, (b), (c) and (d) overlap and (b + c + d) : (a) was found to be 8.00 : 0.98. In the product (b′) : (c′ + d) : (e) : (e′) : (f) = 2.00 : 6.44 : 0.95 : 0.97 : 0.94. The deviation from the theoretical 2 : 6 : 1 : 1 : 1 ratio is due to the mixture of oligomers.

Fig. S3† displays the ¹³C NMR. The carbons connected to the hydroxyl group at δ = 60.13 ppm in the HO–TEG–OH shifted downfield to δ = 63.09 ppm (B) and new carbon resonances of the vinyl groups [δ = 141.17 ppm (F) and δ = 97.76 ppm (E)] and carbonyl carbons resonances of adipic ester groups [δ = 172.60 ppm (K) and δ = 170.15 ppm (G)] appeared at the expected positions.

Synthesis of vinyl-functionalized PEGs

Effect of DVA concentration. As mentioned before, in the case of polycondensation of diols with diesters the rate of reaction was found to increase with increasing diol molecular weight.¹⁷ We investigated the effect of DVA concentration using MeO–PEG–OH₁₁₀₀. In this case no polycondensation can occur, but two MeO–PEG–OH₁₁₀₀ can be connected by a DVA molecule (coupling, Scheme 2).

Scheme 2 Reaction of DVA with MeO–PEG–OH₁₁₀₀ in the presence of CALB under solventless conditions.

Quantitative analysis was carried out by MALDI-ToF. The spectra of the products showed two distributions: that of the MeO–PEG–V₁₁₀₀ and the coupled MeO–PEG₁₁₀₀–V–PEG₁₁₀₀–MeO. The amount of the coupled product decreased with increasing DVA concentration. At DVA/MeO–PEG–OH₁₁₀₀ 1.5/1 and 3/1 30.5% and 15% coupled product was observed, in contrast with only 1 and 0.4% at DVA/MeO–PEG–OH₁₁₀₀ 5/1 and 10/1, respectively. No coupled product was observed at 20/1 ratio (Fig. 1).

Fig. 1 The effect of DVA/MeO–PEG–OH₁₁₀₀ ratio on coupling.

The sensitivity of the MALDI-ToF to identify side products is demonstrated in Fig. S4.† In the low molecular weight region (m/z 800–2000) there is one major distribution, corresponding to the sodium complex of the MeO–PEG–V₁₁₀₀. There is also a trace distribution corresponding to the potassium complex of the product (see also the expanded spectrum), it is known that K⁺ contamination from the glassware occurs during sample preparation.²⁰ Within the distributions, the peaks are separated by 44 Da, corresponding to the ethylene glycol repeat units. In the expanded spectrum, the peak at m/z 1309.79 corresponds to the sodium complex of the vinyl-functionalized 25-mer MeO–PEG–V₁₁₀₀. The calculated monoisotopic mass for this peak is 1309.75 Da [25 × 44.03 (C₂H₄O repeat unit) + 186.01 (C₈H₁₀O₃ end groups) + 22.99 (Na⁺)]. In the high molecular weight region (m/z 2000–2800), 1.0% MeO–PEG₁₁₀₀–V–PEG₁₁₀₀–OMe coupled product was observed.

The NMR spectra of the product at DVA/MeO–PEG–OH₁₁₀₀ 5/1 ratio is shown in Fig. S5,† verifying the structure of the product. However, NMR is not sensitive enough to show the presence of less than a few percent side products. The MALDI-ToF of the product at 20/1 DVA/MeO–PEG–OH₁₁₀₀ did not show the presence of any side products. These reactions showed that large excess of DVA is needed to obtain pure functionalized PEGs. However, the excess DVA can easily be recovered and reused (Fig. 3).

The effect of chain length

Based on the experiments with MeO–PEG–OH, HO–PEG–OH₁₀₀₀ was reacted with 20 molar equivalent of DVA in the presence of CALB under solventless conditions. The MALDI-ToF spectrum of the sample taken at t = 2 hours (Fig. S6†) shows the presence of a mixture of monosubstituted HO–PEG–V₁₀₀₀, disubstituted V–PEG–V₁₀₀₀ and coupled V–PEG₁₀₀₀–V–PEG₁₀₀₀–V. At t = 4 hours the signal of the monosubstituted HO–PEG–V₁₀₀₀ disappeared and only two sets of peaks were observed: V–PEG–V₁₀₀₀ (98.0%) and V–PEG₁₀₀₀–V–PEG₁₀₀₀–V (2.0%).

The same reaction was repeated with HO–PEG–OH₂₀₀₀. In the MALDI-ToF mass spectrum of the product (Fig. 2), the representative peak at m/z 2242.44 corresponds to the sodium complex of the 43-mer of telechelic V–PEG–V₁₀₀₀. The calculated monoisotopic mass for this peak [m/z = 43 × 44.03 (C₂H₄O repeat unit) + 326.14 (C₁₆H₂₂O₇ end groups) + 22.99 (Na⁺)] is 2242.42 Da. Traces of the potassium complex can also be seen in the spectra. No coupled product was observed in the MALDI.

Fig. 2 MALDI-ToF mass spectrum of the product of the reaction of DVA with HO–PEG–OH₂₀₀₀ at 4 hours. [DVA] = 5.29 mol L⁻¹, [HO–PEG–OH₂₀₀₀] = 0.26 mol L⁻¹; [CALB] = 1.6 × 10⁻⁴ mol L⁻¹.

Fig. 3 NMR spectra of the product of the reaction of DVA with HO–PEG–OH₂₀₀₀: ¹H NMR (1 s with 128 scans, top) and ¹³C NMR (1 s with 5000 scans, bottom) (solvent: DMSO-d₆). [DVA] = 5.29 mol L⁻¹, [HO–PEG–OH₂₀₀₀] = 0.26 mol L⁻¹; [CALB] = 1.6 × 10⁻⁴ mol L⁻¹.

In Fig. 3, the ¹H NMR spectrum shows the vinyl protons of the end groups [δ = 4.87 (e) and δ = 4.65 (e′) ppm, and δ = 7.26 ppm (f)], the adipic ester groups [δ = 2.32 (i) and δ = 2.43 (g) ppm, and δ = 1.56 ppm (h)], and the HO–PEG–OH₂₀₀₀ repeat units [δ = 4.11 ppm (b), δ = 3.59 ppm (c), and δ = 3.50 ppm (d)]. The degree of polymerization n = 43 was calculated from the protons of the –CH₂–CH₂–O– repeat units (d) and the proton next to the ester group of the end group (b). The NMR M_n = 2306 g mol⁻¹ (43 × 44 = 1892 from the repeat units + 2 × 207 from the end groups) is in good agreement with the M_n = 2228 g mol⁻¹ calculated from MALDI based on eqn (1). The ¹³C NMR spectrum shows the vinyl groups [δ = 141.17 ppm (F) and δ = 97.97 ppm (E)], the carbonyl carbons resonances of adipic ester groups [δ = 172.59 ppm (K) and δ = 170.17 ppm (G)] and the HO–PEG–OH₂₀₀₀ units [δ = 63.05 ppm (B), δ = 68.26 ppm (C), and δ = 69.98 ppm (D)]. The resonance of the carbons connected to the hydroxyl groups in the starting HO–PEG–OH₂₀₀₀ shifted downfield to δ = 63.05 ppm (B) and new carbon resonances of the vinyl groups (F and E) and carbonyl groups (G and K) appeared at the expected positions, confirming the structure of the product.

Based on the experiments with HO–PEG–OH, the effect of chain length was also investigated with MeO–PEG–OH. We found that pure MeO–PEG–V₂₀₀₀ could be obtained at 5 molar excess of DVA.

In summary, our studies have shown that both DVA excess and PEG chain length have important effects on polymer chain end functionalization.

Conclusions

The data presented in this work demonstrate that enzyme catalysis is very effective in end-functionalizing PEGs in the absence of solvents and under dry nitrogen, producing very pure products as demonstrated by quantitative MALDI-ToF analysis. The excess DVA was recovered for reuse, according to green chemistry concepts as discussed in the 12 principles of green chemistry.⁵ The vinyl groups enable further effective end-functionalization of PEGs and/or the attachment of drugs and proteins.

Acknowledgements

This material is based upon work supported by the National Science Foundation under DMR-0804878. We wish to thank The Ohio Board of Regents and The National Science Foundation for funds used to purchase the NMR (CHE-0341701 and DMR-0414599) and MS (CHE-1012636 and DMR-0821313) instruments used in this work.

References

1. J. R. Hanson, An Introduction to Biotransformations in Organic Chemistry, Oxford University Express, Inc., New York, 1997.
2. K. Faber, Biotransformations in Organic Chemistry, Springer-Verlag, New York, 5th edn, 2004.
3. A. S. Matlack, Introduction to Green Chemistry, Marcel Dekker, Inc., New York, 2001.
4. A. Liese, K. Seelback and C. Wandrey, Industrial Biotransformations, Wiley-VCH, Weinheim, 2006.
5. P. T. Anastas and J. C. Warner, Green Chemistry: Theory and Practice, Oxford University Express, Inc., New York, 1998.
6. J. R. Kasper, M. Y. Sen and J. E. Puskas, Polym. Prepr., 2008, 49, 1161–1162.
7. R. A. Gross, A. Kumar and B. Kalra, Chem. Rev., 2001, 101, 2097–2124.
8. S. Kobayashi, H. Uyama and S. Kimura, Chem. Rev., 2001, 101, 3793–3818.
9. B. Sahoo, K. F. Brandstadt, T. H. Lane and R. Gross, Org. Lett., 2005, 7, 3857–3860.
10. M. Eriksson, K. Hult, E. Malmström, M. Johansson, S. M. Trey and M. Martinelle, Polym. Chem., 2011, 2, 714–719.
11. M. Y. Sen, Ph.D. Dissertation, The University of Akron, Akron, OH, 2009.
12. J. E. Puskas and M. Y. Sen, US Pat., Appl. US20100240837A1, September 23, 2010.
13. J. E. Puskas, K. S. Seo and M. Y. Sen, Eur. Polym. J., 2011, 47, 524–534.
14. J. E. Puskas, K. S. Seo, M. Castano, M. Casiano and C. Wesdemiotis, Green Polymer Chemistry: Functionalization of Poly(ethylene glycol)s under Solventless Conditions, in Green Polymer 2, ed. H. N. Cheng and R. A. Gross, ACS Symposium Series, 2012, in press.
15. J. E. Puskas and M. Y. Sen, Green Polymer Chemistry: Biocatalysis and Biomaterials, ed. H. N. Cheng and R. A. Gross, ACS Symposium Series, 2010, pp. 417–424.
16. A. K. Chaudhary, E. J. Beckman and A. J. Russell, AIChE J., 1998, 44, 753–764.
17. R. L. Rodney, B. T. Allinson, E. J. Beckman and A. J. Russell, Biotechnol. Bioeng., 1999, 65, 485–489.
18. A. K. Chaudhary, J. Lopez, E. J. Beckman and A. J. Russell, Biotechnol. Prog., 1997, 13, 318–325.
19. M. J. Vergne and D. M. Hercules, J. Chem. Educ., 2007, 84, 81–90.
20. A. M. Belu, J. M. DeSimone, R. W. Linton, G. W. Lange and R. M. Friedman, J. Am. Soc. Mass Spectrom., 1996, 7, 11–24.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3ra46070c

---