Analysis of poly(bisphenol A carbonate) by size exclusion

RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
Analysis of Poly(bisphenol A carbonate) by Size
Exclusion Chromatography/Matrix-assisted
Laser Desorption/Ionization. 1. End Group and
Molar Mass Determination
Concetto Puglisi1, Filippo Samperi1, Sabrina Carroccio2 and Giorgio Montaudo2*
1
Istituto per la Chimica e la Tecnologia dei Materiali Polimerici, Consiglio Nazionale delle Ricerche, Viale A. Doria, 6 - 95125
Catania, Italy
2
Dipartimento di Scienze Chimiche, Università di Catania, Viale A. Doria, 6 - 95125 Catania, Italy
The determination of molar mass (MM) data for polydisperse polymers by SEC/MALDI involves the
fractionation of samples through analytical size exclusion chromatography (SEC). Selected SEC fractions
are then analyzed by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) and the mass
spectra of these nearly monodisperse samples allow the determination of the average molar masses. The
SEC/MALDI procedure has now been applied to two polycarbonate samples, PC1 and PC2. The results
show that the MALDI spectra of the SEC fractions allow not only the detection of linear and cyclic
oligomers contained in these samples, but also the simultaneous determination of their average molar
masses. Two slightly differing SEC calibration plots were obtained, due to the smaller hydrodynamic
volume of the polycarbonate cyclic chains with respect to the linear ones. In agreement with theory, the ratio
(Mcycle/Mlinear)Ve at a fixed elution volume was found to be 1.22, independent of the molar mass values.
Copyright # 1999 John Wiley & Sons, Ltd.
Received 17 August 1999; Revised 24 September 1999; Accepted 26 September 1999
Matrix-assisted laser desorption/ionization time-of-flight
Mass Spectrometry (MALDI-TOFMS)1,2 allows the desorption and ionization of very large molecules even if present
in complex mixtures, and is widely used with increasing
success for the characterization of synthetic polymers,
particularly for end group analysis.
Although MALDI spectra of synthetic polymers can
reach masses up to 106 Da and beyond, there is a much
lower limit to the mass spectral resolution of individual
oligomers, and this depends both on the resolution of the
MALDI instruments and on the polymer repeat unit. The
mass limit at which peaks corresponding to individual
oligomers are no longer mass-resolved reaches a value of
about 15 kDa with conventional MALDI-TOF (continuous
extraction), whereas with the development of delayed
extraction (time-lag focusing) it can be extended to about
50 kDa.
For the analysis of polymers it has been found that molar
mass (MM) estimates provided by MALDI agree with the
values obtained by conventional techniques only in the case
of samples with narrow molar mass distributions (MMD),
whereas with polydisperse polymers MALDI fails to yield
reliable MM values.2 This implies that the relative
intensities of the MALDI signals as a function of their m/z
values is far from that expected from the actual MMD of the
polymer sample. Lighter molecules (which outnumber the
larger ones in a most probable distribution) tend to saturate
*Correspondence to: G. Montaudo, Dipartimento di Scienze Chimiche
Università di Catania, Viale A. Doria, 6 - 95125 Catania, Italy.
Contract/grant sponsor: Italian Ministry for University and for
Scientific and Technological Research.
Contract/grant sponsor: Progetto Finalizzato Materiali per Tecnologie
Avanzate II.
Contract/grant sponsor: National Council of Research, Italy.
CCC 0951–4198/99/222260–08 $17.50
the MALDI detector, suppressing the detection of the laterarriving larger molecules. Furthermore, the small molecules
reach the detector with a higher kinetic energy and are thus
preferentially revealed in the MALDI process. Even when
the detectors currently employed in MALDI instruments are
equipped with a post-accelerator, the detection of high
masses may still be partially discriminated against.3
Synthetic polymers may show a wide range of molar mass
distributions according to the synthetic method used in their
preparation, and therefore the direct application of the
MALDI technique for the MM measurements appears
limited to narrow distributions, although several attempts
are currently being performed to extend it to polymers with
wider distributions.3–5
A widely accepted method for MM determination in
polydisperse polymer samples by MALDI is6–12 the
analytical size exclusion chromatography (SEC) fractionation of polydisperse samples. We have proposed this
approach earlier,6,7 and the results appear encouraging.
Furthering our studies on the off-line SEC/MALDI method,
we have now applied it to the characterization of end groups
and molar masses in poly(bisphenol A carbonate) (PC).
The characterization of poly(bisphenol A carbonate) by
mass spectrometry has received continued attention, due to
the fact this polymer is an important engineering thermoplastic material.8,13–17 End group analysis and molar mass
characterization are among the major concerns in the
structural characterization of PC. It is well known that the
production of this polymer is accompanied by the formation
of sizeable amounts of cyclic oligomers, as is the case for
other polymers produced by condensation and ring-opening
reactions, and it is of great interest to determine their size
and relative abundance.18 The presence of a mixture of
linear and cyclic chains in low mass polycarbonates causes
Copyright # 1999 John Wiley & Sons, Ltd.
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
2261
Figure 1. MALDI-TOF spectrum of unfractionated PC1 sample.
difficulties in establishing an appropriate SEC calibration
curve for such samples.8
It is well known that, for a given polymer, there is a small
but well-defined difference between the hydrodynamic
volumes of linear and cyclic oligomers of the same molar
mass,19 and this difference is reflected in the elution
volumes (Ve) of the linear and cyclic species of the same
molar mass.19 Thus, it has been shown by SEC-light
scattering experiments,19 that cyclic oligomers of polydimethyl siloxane (PDMS) are eluted at slightly higher
volumes with respect to linear oligomers of the same MM,
owing to the smaller hydrodynamic volume. The ratio
(Mcycle/Mlinear)Ve at a fixed elution volume (Ve) was found to
be 1.24, independent of MM values. This result was recently
confirmed by performing the SEC/MALDI calibration of
two PDMS samples (linear and cyclic, respectively), and the
ratio (Mcycle/Mlinear)Ve was found to be 1.22.20
In the case of PC a small but well-defined difference was
found in an early study21 on linear versus cyclic PC samples,
by measurements of molar masses and intrisic viscosities of
Figure 2. MALDI-TOF spectrum of unfractionated PC2 sample.
Copyright # 1999 John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
2262
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
Table 1. Structural assignments of the peaks appearing in the inset sections of the MALDI spectra of polydisperse samples PC1 and PC2
reported in Figs 1 and 2
M
n
Na‡
A
11
12
13
3034
3288
3542
A'
13
14
15
3654
3908
4162
B
12
13
14
3168
3422
3676
B'
14
15
16
3732
3986
4240
11
12
13
14
15
16
3048
3302
3556
3810
4064
4318
12
13
14
15
16
17
3074
3328
3582
3836
4090
4344
Mass
series
Oligomer structure
C
D
a series of monodisperse BPA-polycarbonate samples. Two
Mark-Houwink relationships could be discerned for the
linear and cyclic PC samples, but no quantitative estimates
of the ratio (Mcycle/Mlinear) were made.21
In a recent investigation,8 different slopes for the SEC/
MALDI calibration lines of linear and cyclic PC chains
were reported. This result implies that the ratio (Mcycle/
Mlinear)Ve changes with the MM value, and that the
theoretical19 constant value of 1.24 is not valid for all the
polymers. However, the conditions used in that SEC
experiment8 are likely to promote sizeable self-association
of the PC chains,22 and this might affect the SEC/MALDI
calibration line.
We have applied the SEC/MALDI procedure to two
polycarbonate samples, PC1 and PC2, which contain both
linear and cyclic oligomers. The results show that the
MALDI spectra of the SEC fractions allow the individual
detection of linear and cyclic oligomers contained in these
samples, and also the simultaneous determination of the
average molar masses of the linear and cyclic oligomers. As
expected, two parallel SEC calibration lines were obtained
for linear and cyclic PC chains, and the ratio (Mcycle/Mlinear)
was found to be constant (about 1.22).
In earlier works,19–21 in order to investigate the molecular
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
K‡
3670
3924
4178
3748
4002
4256
3090
3344
3598
3852
406
4350
dimensions of a set of linear and cyclic polymers, samples
of pure linear and pure cyclic polymers were analyzed. This
effort was needed in order to check the value of the (Mcycle/
Mlinear) ratio over an extended range of MM. Of course, pure
linear and pure cyclic samples are not always available for
every polymer, whereas the most common case is that of an
essentially linear polymer sample which contains cyclic
oligomers. The advantage of MALDI mass spectra is that
they yield separate and simultaneous information on the
molar masses of both cyclic and linear oligomers present in
each SEC fraction.
EXPERIMENTAL
Materials
Poly(bisphenol A carbonate) sample PC1 was supplied by
Aldrich Chemical Co.(Italy), whereas PC2 was a L220 GE
polycarbonate sample. Both samples were dried at 70 °C,
under vacuum, for one week before use. 2-(4-Hydroxyphenylazo)benzoic acid (HABA) and trans-3-indoleacrylic
acid (IAA) were supplied by Aldrich Chemical Co. (Italy)
and used as supplied.
Copyright # 1999 John Wiley & Sons, Ltd.
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
2263
Figure 3. SEC traces of samples PC1 and PC2 together with the calibration line obtained by the SEC/MALDI
method.
SEC analysis and fractionation
Size exclusion chromatography (SEC) analyses were
performed using a Waters 600 A apparatus, equipped with
five Ultrastyragel columns (7.8 300 mm; in the order 105,
103, 500, 104 and 100 Å pore size) connected in series, and a
Waters R401 differential refractometer. 90 mL of sample
solutions (2.5 mg/mL) were injected and eluted at a flow
rate of 1 mL/min using CHCl3 as the mobile phase. The
fractionation was performed by collecting about 100 equal
volume fractions (12 drops each) corresponding to about
0.165 mL.
MALDI-TOF analysis
Matrix-assisted laser desorption ionization time-of-flight
(MALDI-TOF) mass spectra were obtained using VoyagerDE2 STR and the Voyager-DE2 (PE Biosystems) instruments, both equipped with a nitrogen laser emitting at
337 nm with a 3 ns pulse width and working in positive ion
mode. The accelerating voltage was 20–25 kV, the grid
voltage and delay time (delayed extraction, time-lag) were
optimized for each sample to obtain the higher molar mass
values. The laser irradiance was maintained slightly above
threshold. The MALDI-TOF mass spectra of the unfractionated PC1 and PC2 samples, and those of fractions with
molar mass lower than 5000 Da, were acquired in reflectron
mode with the Voyager-DE2 STR instrument, which gave a
mass resolution greater than 3000, measured from full peak
width measured at half-height.
MALDI sample preparation
HABA (0.1 mol Lÿ1 in tetrahydrofuran (THF)) was used as
the matrix for the two unfractionated PC samples and for
SEC fractions containing low molar mass polycarbonate
Copyright # 1999 John Wiley & Sons, Ltd.
(mass range 1000–12000), whereas IAA (0.1 mol Lÿ1 in
THF) was used as the matrix for SEC fractions with higher
molar mass.
Solutions of the unfractionated PC1 and PC2 polycarbonate samples were prepared in THF at a concentration of
2mg/mL, whereas the PC1 and PC2 SEC fractions were
dissolved in 80 mL of THF after complete evaporation of the
eluent. Equal volumes of polycarbonate sample solutions
and matrix solution were mixed in order to obtain a 1:1 or
1:3 ratio (v/v, sample/matrix). One microliter of each
sample/matrix solution was spotted on the MALDI sample
holder and allowed to dry slowly in order to improve matrix
crystallization.
Molar mass calculations
The molar mass of the two polydisperse PC samples were
calculated from the SEC curves by the Polymer Lab Caliber
software using the absolute calibration curves obtained by
plotting the log Mw of each SEC selected fraction
(calculated from the MALDI-TOF spectra) as a function
of the corresponding elution volume.
The average molar masses (Mn and Mw) of all SEC
fractions were calculated from MALDI-TOF analysis by the
appropriate Perseptive-Voyager-Grams software, using
Eqns ((1)) and ((2)):
Mn ˆ Ni Mi =Ni
…1†
Mw ˆ Ni Mi2 =Ni Mi
…2†
where Ni is the relative abundance of the polymer chain with
molecular mass Mi. The Ni values were measured as the
relative intensity of the molecular ion peak with a molecular
mass Mi in the MALDI spectrum.
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
2264
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
Figure 4. MALDI-TOF spectra of the SEC fractions from sample PC1 collected at: 26.65 mL (a), 27.8 mL (b) and
28.54 mL (c).
Viscometry
Inherent viscosity [inh = (ln r)/C;C = 0.5 g/dL] measurements were performed at 30 0.1 °C in CHCl3 as solvent
using a Desreux-Bishoff suspended-level viscometer, yielding the values of 0.42 and 0.44 for PC1 and PC2, respectively.
RESULTS AND DISCUSSION
End group determination
The MALDI spectra of the two unfractionated polycarbonate samples (PC1 and PC2) are reported in Figs 1 and 2,
respectively, showing well-resolved peaks up to 16 000 Da.
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
These spectra were recorded in reflectron mode with a mass
resolution of about 3000, allowing the accurate structural
characterization of the two polymer samples.
The inset expansions in Figs 1 and 2 show the presence of
peaks belonging to six mass series, assigned in Table 1,
which are desorbed as [M ‡ Na]‡ adducts accompanied by
less intense potassiated adducts. Cyclic oligomers (species
of type D, Table 1) appear as the most abundant species at
low masses in both samples and their peak intensity
decreases rapidly with size. Different end groups are
attached to the linear chains of the two polycarbonates.
Oligomers of type A, bearing phenylcarbonate groups at
both chain ends (Table 1), predominate in sample PC1 at
masses higher than 2500 Da (Fig. 1).
Copyright # 1999 John Wiley & Sons, Ltd.
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
2265
Figure 5. MALDI-TOF spectra of the SEC fractions from sample PC1 collected at: 30.2 mL (a), 35.3 mL (b) and
40.0 mL (c).
In the MALDI spectrum of sample PC2 the peaks
corresponding to the sodiated oligomers of type A' (Fig. 2),
containing p-tert-butylphenylcarbonate at both chain ends
(Table 1), become the most intense starting from 4000 Da and
are accompanied by satellite peaks (plus 16 mass units) due to
the potassiated species. Oligomers of type C (Fig. 1, Table 1)
can be assigned to polycarbonate chains containing phenol
groups at both ends. In the inset expansion in Fig. 1, species C
appearat 14 Da higherthan the most intense peak series (due to
species A, Table 1). Oligomers of type C can also be identified
in the inset expansion of the MALDI spectrum in Fig. 2,
Copyright # 1999 John Wiley & Sons, Ltd.
although they appear with low intensity. In Figs 1 and 2 the
oligomer series B and B' (Table 1), containing phenol/
phenylcarbonate and phenol/p-tert-butylphenylcarbonate
chain ends, respectively, can also be detected.
Our results are in agreement with recent reports on end
group analysis of some industrial BPA-polycarbonates.8,16,17 A minor ion fragmentation process due to the
use of HABA as a matrix (see experimental section) has
been claimed to occur in the MALDI spectra of PC
samples,8 but the latter claim was not confirmed in the
present work.
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
2266
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
Figure 6. SEC calibration lines of linear (&) and cyclic (*) PC chains, calculated by separate summation of molar
masses of mass spectral peaks corresponding to linear and cyclic oligomers, respectively, observed in the MALDI-TOF
spectra of the SEC fractions of sample PC1.
SEC/MALDI measurements
The average molar masses (Mw and Mn) of the PC1 and PC2
samples were determined by the off-line SEC/MALDI
method6,7 (see experimental section), which consists of the
analytical SEC fractionation of polydisperse polymers to
collect numerous fractions (about 100) containing nearly
monodisperse samples. The amount of sample present in
each fraction (about 2.5 mg on average) exceeds many times
the quantity needed for a MALDI spectrum. The weight
average molar mass values (Mw) of these fractions are
calculated from the MALDI spectra and are plotted against
the corresponding elution volumes (Ve), producing an
absolute MM calibration line which enables the computation of the average MM and MMD of the unfractionated
sample directly from its SEC trace.
Figure 3 reports the SEC traces of samples PC1 and PC2,
together with the calibration line constructed after the
fractionation of these two samples and the MALDI analysis
of the narrow distribution fractions collected. The mass
range covered by the MALDI spectra of the PC fractions
covered the mass range 1000 to 100 000 Da (see below). The
average MM values computed by using the calibration line
in Fig. 3 are as follows: PC1: Mw = 22 200, Mn = 10 300;
PC2: Mw = 28 000, Mn = 13 200. These values are lower by a
factor of two with respect to those calculated by using the
calibration line obtained from polystyrene standards (PC1:
Mw = 55 800, Mn = 23 600; PC2: Mw = 59 300, Mn = 27
300). (The universal calibration curve was obtained using
ten PS, ten PMMA, and five PEG samples. These samples
were primary SEC standards obtained by anionic polymerization, and their MMs were all accurately measured by
conventional techniques.20)
The strong discrepancy between the two sets of data is
most likely ascribed to the higher hydrodynamic volume of
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)
polycarbonate with respect to the random coiled polystyrene
standards. These results are in agreement with data reported
in the literature, where the SEC was performed both with
polystyrene SEC standards and with monodisperse polycarbonate samples.23
Figures 4(a)–(c) show the MALDI spectra of the SEC
fractions collected from sample PC1 at the lower elution
volumes (26.7, 27.8, 28.5mL, respectively). These spectra
display well-resolved oligomer peaks up to 70 000Da (Fig.
4(b)), allowing the identification of the polymer structure. In
fact, the MALDI spectra in Figs 4(b) and 4(c) are mainly
constituted of linear chains terminated by phenylcarbonate
groups at both ends (oligomers of type A, Table 1).
Figures 5(a)–(c) display instead the MALDI spectra of
the SEC fractions collected from sample PC1 at higher
elution volumes (30.2, 35.3, 40.0mL, respectively). The
MALDI spectra in Figs 5(a) and 5(b) still show oligomers of
type A as major components, but in the inset of Fig. 5(b)
other, less abundant, species (oligomers B, C and D, Table
1) are identified. In the MALDI spectrum in Fig. 5(c),
obtained from the PC1 fraction taken at 40.0 mL, the most
abundant peaks are those corresponding to cyclic oligomers
(species D, Table 1). Similar results were obtained in the
MALDI analysis of PC2 SEC fractions.
It should be remarked that even those fractions collected
(Fig. 3) at high elution volumes (>37 mL), where the SEC
traces are flat suggesting that the PC had been totally eluted,
in fact produced excellent MALDI spectra. This is due to the
high sensitivity of MALDI, and also to the fact that,
although the weight fraction of these low-mass eluted
polycarbonate oligomers is quite low, the number fraction is
still sizeable.
Owing to the presence of both linear and cyclic chains,
the SEC calibration line reported in Fig. 3 shows a deviation
from linearity at elution volumes higher than 34.0 mL.
Copyright # 1999 John Wiley & Sons, Ltd.
ANALYSIS OF POLYCARBONATES BY SEC/MALDI 1
Although such deviations are not unusual and might appear
to not merit further consideration, in the present case it has a
very specific origin. It depends on the fact that the molar
masses of the SEC fractions were calculated from the
MALDI spectra as summations of all the peaks appearing in
the spectra, irrespective of whether they belonged to linear
or cyclic oligomers.
It is well known19–21 that cyclic species are eluted at later
times with respect to linear chains having the same
backbone length. As a consequence, the calibration lines
for cyclic and linear PDMS chains are different19–21 and,
according to theory, the ratio (Mcycle/Mlinear)Ve should be
1.24.19 However, the MALDI spectra of the low mass
fractions (Fig. 5) allow the detection of linear and cyclic
oligomers contained in these PC samples, and separate
average molar masses for linear and for cyclic oligomers
can be calculated. When this procedure was applied, two
slightly differing SEC calibration plots were obtained (Fig.
6), due to the smaller hydrodynamic volume of the
polycarbonate cyclic chains with respect to the linear ones.
In agreement with the theory, the ratio (Mcycle/Mlinear)Ve was
found to be 1.22.
CONCLUSIONS
Because of its unprecedented sensitivity, MALDI-TOF
constitutes an excellent SEC detector and the combination
of SEC with MALDI brings new possibilities to polymer
analysis. Our results show that the MALDI spectra of the
SEC fractions allow not only the detection of linear and
cyclic oligomers contained in these samples, but also the
simultaneous determination of the average molar masses of
linear and cyclic oligomers. Two parallel SEC calibration
plots were obtained, and the ratio (Mcycle/Mlinear) was found
to be in agreement with theory.
Acknowledgements
Partial financial support from the Italian Ministry for University and
for Scientific and Technological Research (MURST), from Progetto
Finalizzato Materiali per Tecnologie Avanzate II (CNR, Rome), and
Copyright # 1999 John Wiley & Sons, Ltd.
2267
from the National Council of Research (CNR, Rome) is gratefully
acknowledged.
REFERENCES
1. Bahr U, Deppe A, Karas M, Hillenkamp F, Giessman U. Anal.
Chem. 1992; 64:2866.
2. Montaudo G, Montaudo MS, Puglisi C, Samperi F. Rapid
Commun. Mass Spectrom. 1995; 9:453.
3. Larsen BS, Michelle Byrd HC, McEwen CN. Symposium on
Synthetic Polymer Analysis by Mass Spectrometry, Pittcon, March
12, 1999, Orlando, FL; 1365.
4. Scamporrino E, Maravigna P, Vitalini D, Mineo P. Rapid
Commun. Mass Spectrom. 1998; 12:646.
5. Servaty S, Kohler W, Meyer WH, Rosenauer C, Spickermann J,
Rader HJ, Wegner G. Macromolecules 1998; 31:2468.
6. Montaudo G, Garozzo D, Montaudo MS, Puglisi C, Samperi F.
Macromolecules 1995; 28:7983.
7. Montaudo G, Montaudo MS, Puglisi C, Samperi F. Rapid
Commun. Mass Spectrom 1995; 9:1158.
8. Nielen MWF, Malucha S. Rapid Commun. Mass Spectrom. 1997;
11:1194; Nielen MWF. Anal. Chem. 1998; 70:1563.
9. Dwyer J, Botten D. Internat. Lab. January, 1997; 13A.
10. Fei X, Murray KK. Anal. Chem. 1996; 68:3555.
11. Montaudo G. Polym. Prep. (A.C.S., Div. Polym. Chem.) 1996;
36:290.
12. Danis PO, Saucy DA, Huby FJ. Polym. Prep. (A.C.S., Div. Polym.
Chem.) 1996; 36:311.
13. Wiley RH. Macromolecules 1971; 4:264.
14. Foti S, Giuffrida M, Maravigna P, Montaudo G. J. Polym. Sci.,
Polym. Chem. Ed. 1983; 21:1567.
15. Puglisi C, Sturialc L, Montaudo G. Macromolecules 1999;
32:2194.
16. Montaudo G, Montaudo MS, Puglisi C, Samperi F. Anal. Chem.
1994; 66:4366.
17. Przybilla L, Rader HJ, Mullen K. Eur. Mass Spectrom. 1999;
5:133.
18. Maravigna P, Montaudo G. in Comprehensive Polymer Science
Vol. 5, Pergamon Press Oxford, 1989; 63; Montaudo G. Macromolecules 1991; 24:5289.
19. Wright PV, Beevers MS. In Cyclic Polymers, Semlyen JA (ed).
Elsevier, 1986; 109.
20. Montaudo G, Montaudo MS, Puglisi C, Samperi F. Rapid
Commun. Mass Spectrom. 1998; 12:519.
21. Horbach A, Vernaleken H, Weirauch K. Makromol. Chem. 1980;
181:111.
22. Puglisi C, Samperi F, Montaudo G, Carroccio S. Rapid Commun.
Mass Spectrom. 1999; 13:000.
23. Bruessau RJ. Macromol. Symp. 1996; 110:15.
Rapid Commun. Mass Spectrom. 13, 2260–2267 (1999)