Letter

pubs.acs.org/ac

Extension of the Kendrick Mass Defect Analysis of

Homopolymers to ​Low Resolution and High Mass Range

Mass Spectra Using Fractional Base Units Thierry

​ Fouquet​* ​and
Hiroaki Sato​*
National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki
305-8565, Japan
•​S ​Supporting Information
ABSTRACT: ​Beyond the high resolution/low mass range data traditionally used, a Kendrick mass defect analysis
(KMD) using the new concept of fractional base units has been successfully conducted on low resolution/low mass
range and high resolution/high mass range data for the ​fi​rst time. Relying on a mathematical framework to
rationalize the e​ff​ect of the ​fractional base units, the electrospray ionization single stage and
​ multistage mass spectra
of a poly(vinylpyrrolidone) recorded from a low resolution ion trap analyzer were turned into information-rich KMD
plots using vinylpyrrolidone/112 and pyrrolidone/86 as base units. The distributions detected in the matrix assisted
laser desorption ionization spiralTOF mass ​spectra of high molecular weight poly(ethylene oxide) and
poly(caprolactone) were conveniently discriminated in KMD plots using (ethylene oxide)/45 and caprolactone/113
as base units with an unprecedented resolution at such a mass range. The high resolution KMD analysis using
fractional base units opens new perspectives for the acquisition, visualization, and presentation of mass spectra of
polymers with less restrictions in terms of required resolution and molecular weights.

M​ analytical ​
ass spectrometry ​
technique ​
(MS) ​
for the ​
has ​
characterization ​
emerged as a major ​
of poly- mers​1

thanks to soft ionization sources such as electrospray ionization (ESI)​2​,​3 ​and matrix assisted laser desorption
ionization (MALDI)​4​,​5 ​and a variety of mass analyzers and experiments.​6​,​7 ​However, technical polymers are often
blended samples containing several homopolymers/copolymers and additives. Consequently, mass spectra of
polymers are notoriously complex, and the data mining is extremely time- consuming. To circumvent this issue, the
Kendrick mass defect analysis​8 ​(KMD) has been recently adapted from petroleomics​9 ​to the ​fi​eld of polymer
chemistry.​10 ​In lieu of the traditional CHin the ​2 base
​ Kendrick unit​11 ​scale), (at 14.0157 Sato et in al.the 10
​ ​IUPAC
have used scale, the ​© 2017 American Chemical Society ​2682 ​DOI: ​10.1021/acs.analchem.6b05136 ​Anal. Chem. ​2017, 89, 2682−2686 ​set at 14.0000
repeat unit of a polymer backbone as the new base unit for the calculation of the Kendrick masses (KM). Thus, all
the congeners of a given distribution possess the same mass defect and line up horizontally in the associated KMD
plot (KMD vs nominal Kendrick mass NKM) while other distributions with di​ff​erent end-groups or a di​ff​erent repeat
unit line up horizontally at di​ff​erent KMD values or obliquely, respectively. Since its ​fi​rst use for polymers, several
authors have re​fi​ned the KMD analysis such as by a comonomer using higher for order a copolymeric mass defects​12
​ ,​13​,​14 ​base or units the expelled neutral for tandem mass spectra (MS/MS).​15 ​The latest improvement
sampleor other 10​
consists of the use of a ​fraction o​ f the repeat unit ((repeat unit)/​X ​with ​X ​being a positive integer) as a base
unit chemically senseless but mathematically acceptable. Depending on the value of the divisor ​X​, the computed
KMD plots discriminate with an unprecedented resolution the isotopes of a given distribution (one horizontal line
per ​12​C/​13​C​x oligomers
​ isotope) and/or the components of a blend or co- without any modi​fi​cation of the raw data.​16
 If the main drawback of a regular KMD analysis remains its need for high resolution mass spectral data (i.e., low
mass range and high resolution analyzers such as Fourier transform ion cyclotron resonance, orbitrap, or
spiralTOF​17​), the introduction of the fractional base units and the associated ​high resolution KMD plots ​open new
perspectives. As a founding postulate, improving the resolution of the KMD analysis would allow a decrease of the
required resolution for the MS analysis with no loss of information. This Letter validates the postulate and reports on
the successful KMD analysis of data recorded with (a) a low resolution ion trap device in a low mass range (average
resolution at ​m​/​z ​1000: ​∼​4000) and (b) a high resolution spiralTOF analyzer in high mass range (average resolution
at ​m​/​z ​15 000: ​∼​20000) using a fractional base unit with a rationally chosen divisor.
Received: ​December 27, 2016 ​Accepted: ​February 9, 2017 ​Published: ​February 9, 2017
Analytical Chemistry ​Letter
Figure 1. (A) ESI-MS spectrum of PVP (inset: peak shapes recorded with ion trap and spiralTOF analyzers). (B) Full scale
VP-based KMD plot. (C) Zoom from (B) (KMD range: ​−​0.14 to +0.04). (D) Full scale VP/112-based KMD plot.
Experimental protocols are detailed in an extended experimental part in the ​Supporting Information​, and the
​ cheme S1​. Concisely, the ESI-MS and MS​n ​(​n ​= 3, 4) spectra of a
structures of the polymers are depicted in S
polyvinylpyrrolidone (PVP) have been recorded using an amaZon SL-STT2 ion trap (Bruker, Bremen, Germany).
The MALDI-MS spectra of a polyethylene glycol (PEO) and polycaprolactone (PCL) were recorded using a
JMS-S3000 SpiralTOF mass spectrometer (JEOL, Tokyo, Japan).​17 ​For every experiment, mMass 5.5.0.0​18 ​was used
for data processing and artworks. The mass-to-charge ratios of ions (​m/​ ​z)​ were converted to KM, NKM, and KMD
according to ​KM(ion) = ​m ​/ ​z (​ ion) · r​ ound( ​m
m
(baseunit) ​
(baseunit))
with ​“​round(​x)​ ​” ​being ​
the rounded value of ​x t​ o the nearest integer, NKM(ion) = round (KM(ion)), and KMD(ion)
= NKM(ion) ​− ​KM(ion). The KMD plot displays the KMD of the detected oligomeric adducts as a function of their
corrected ​NKM (CNKM) using a bubble chart (data triplet: CNKM, KMD, abundance), CNKM(ion) = NKM(ion) ​−
​ ​“​Full scale​”
ceiling(NKM(ion) ​− ​m/​ ​z(​ ​ion)​ ) with ​“​ceiling(​x​)​” ​being the smallest integer greater than or equal to ​x​.16
plot refers to KMD plots with KMD ranging from ​−​0.5 to 0.5. MSRepeatFinder (JEOL) was used to extract points
of interest from the KMD plots and highlight the corresponding peaks in the mass spectra (so-called grouping
mode).
As the missing link between the founding article​16 ​and the present Letter, the rigorous mathematical framework
account- ing for the expansion of the KMD dimension using a fractional base unit (repeat unit)/​X i​ s reported in
Figure S1 ​and the associated comment. In particular, it provides a rationalized method to choose the divisor ​X,​
avoiding a time-consuming screening of possible values. Brie​fl​y, the best values of ​X ​to improve the resolution of
the KMD plot in a controlled extent are close to round(​R​) and round(2​R)​ . Relying on this ​fi​nding, the divisors used
throughout the text have been voluntarily chosen near round(​R​) following a rapid trial-and-error step
2683

DOI: ​10.1021/acs.analchem.6b05136 ​Anal. Chem. ​2017, 89, 2682−2686

(eight values tested per data set, from round(​R​) ​− ​4 to round(​R​) + 4). The low resolution ESI-MS spectrum of PVP is
depicted in ​Figure 1​A and displays a main distribution of sodium adducts ([PVP + Na]​+​, red squares) and two minor
proton adducts ([PVP + H]​+​, black triangles) and sodiated dimers ([2PVP + Na]​+​, black circles). Two representative
peak shapes from the low resolution ESI-MS (ion trap) and high resolution MALDI-MS data (spiralTOF, ​Figure S2​)
are depicted in the inset for the sake of comparison of the resolving power (2-digit ​m​/​z v​ alue with the ion trap,
4-digit value with the spiralTOF). The data set recorded from the ion trap device is normally not suited to a KMD
analysis owing to the lack of accuracy in the measurement of ​m/​ ​z​. The VP-based full scale KMD plot is depicted in
Figure 1​B, and its magni​fi​cation (KMD: ​−​0.14 to +0.04) is depicted in ​Figure 1​C. The KMD analysis clearly fails as
points are roughly aligned along an oblique direction with no clear separation of either the three distributions or the
isotopes. On the contrary, a KMD analysis using VP/112 as a newly introduced fractional base unit produces a
su​ffi​ciently resolved KMD plot (​Figure 1​D) with three distinct point alignments readily assigned to the three series
of sodiated PVP (red dots), protonated PVP, and sodiated PVP dimers (​R ​= 111.0684, round(​R)​ = 111, optimal ​X = ​
112, cf. ​Supporting Information​). It should be pointed out that the average mass di​ff​erence between two consecutive
oligomer peaks in the ESI-MS spectrum is not 111.0684 (the exact mass of a VP unit) but 111.0860 (​δ ​∼ ​159 ppm)
owing to the low resolution. Consequently, data points do not line up along the horizontal axis but slightly obliquely
while the KMD plots computed with 111.0860 and 111.0860/112 as base units display horizontally aligned series
(​Figure S3​). Such deviation is nevertheless not detrimental to the interpretation of the KMD plot, and the exact
masses of the repeat units will be used for sake of simplicity. The selection of points from the KMD plot to highlight
the associated peaks in the MS spectrum is easily achievable (​Figure S4​) since the discrimination of the series is
clear enough.
Analytical Chemistry ​Letter

Figure 2. (A) ESI-MS​3 ​spectrum of a dehydrated PVP 7-mer at ​m​/​z ​820 (red squares: eliminations of pyrrolidone P). (B) Full
scale P-based KMD plot. (C) Zoom from (B) (KMD range: ​−​0.08 to +0.10). (D) Full scale P/86-based KMD plot.
Figure 3. (A) MALDI-spiralTOF-MS spectrum of a 10 000 g mol​−​1 ​PEO. (B) Full scale EO-based KMD plot. (C) Zoom from (B)
(KMD range: ​−​0.03 to +0.09). (D) Full scale EO/45-based KMD plot.
DOI: ​10.1021/acs.analchem.6b05136 ​Anal. Chem. ​2017, 89, 2682−2686

Using an ion trap for the mass analysis of polymers
​
is obviously within the scope of recording MS​n mass
spectra. The fragmentation pathways of polymer ions
is indeed rich of information in terms of end-groups
and architectures.​6 ​So far, the KMD analysis has
been successfully extended to the MS/ MS stage
using data recorded from a high resolution
orthogonal-acceleration TOF.​15 ​The previous results
suggest a KMD analysis would be satisfactorily
conducted on the MS​n ​data recorded with a low
resolution ion trap considering the high resolution of
the KMD analysis achieved with the fractional base
units. The ESI-MS​3 ​spectrum of a dehydrated
PVP 7-mer (formed upon the release of water from
the protonated precursor ion​19​) is depicted in ​Figure
2​A. The main product ion series arising from the
consecutive expulsions of the pyrrolidone pendant
group (abbreviated as P, 85.0528 Da) is marked with
red squares. Several minor product ion series are
slightly seen and formed upon the disruption of the
polymer backbone and the loss of the end-groups.
The full scale and magni​fi​ed P-based KMD plots are
depicted in ​Figure 2​B,C, respectively. As for the MS
stage, a rough point alignment is reached but the
KMD analysis fails at distinguishing the product ion
series.​15 ​On the contrary, a KMD plot computed

2684

Analytical Chemistry ​Letter


Figure 4. MALDI-spiralTOF-MS spectrum peaks from the
two PCL (H, OH) and PCL scale CL/113-based KMD plot.
of (H, a SEC-fractionated C​3​H​7​O) series adducted PCL.
(A) with ​M​p sodium ​∼ ​7000 and ​g mol​potassium). ​−​1 ​and
​ ​
​ ​MD)
(B) (C, ​ ​p ​∼ Full
​ ​15 000 scale
​ ​g CL-based
​ ​mol​−​1 ​(inset:
KMD. ​representative (E,
​ F) Full
with P/86 as the base unit (​Figure 2​D) displays ​fi​ve
well- resolved lines (​R =​ 85.0528, round(​R)​ = 85,
​ 86, cf. ​Supporting Information​) with the
optimal ​X =
main series plotted in red. It constitutes (a) an
alternative visualization of the MS​n ​data with
intuitive alignments and (b) an easy way to extract
data points using any appropriate software. Similar
results are found at the MS​4 ​step (​Figure S5​, using ​X
= 89) and con​fi​rm the possibility to conduct a KMD
analysis for low resolution mass spectral ​data at any
stage (MS or MS​n)​ with a well-chosen ​fractional
base unit.
The KMD analyses for polymer ions conducted with
high resolution mass analyzers have been reported
for samples of ​low molecular weights only owing to
the degradation of the resolution
​ with the increase of
the mass range. It is worth noting that low
resolution/low mass range (<4000 Da) and high
resolution/high mass range (>10000 Da) mass
spectra exhibit a similar peak shape. As such, the
high resolution KMD analysis proposed above for
low resolution data in the low mass range should be
extendable to the case of high resolution data in the
high mass range, as a ​fi​rst KMD analysis for high
molecular weight polymers. The MALDI-MS
spectrum of a 10000 g mol​−​1 ​PEO is depicted in
Figure 3​A. A high laser ​fl​uence has been voluntarily
used to form ions in the low mass range in addition
to the main distribution in the high mass range as
this situation is commonly faced when analyzing
high molecular weight samples. The mass spectrum
displays four
2685
DOI: ​10.1021/acs.analchem.6b05136 ​Anal. Chem. ​2017, 89, 2682−2686
main peak series: the main distribution centered
around 9000 Da ([PEO + Na]​+​), a doubly charged
series ([PEO + 2Na]​+​), and in source decay (ISD)
product ions [l​n ​(linear product ions, same elemental
composition as the intact chains) and c​n ​(cyclic
product ions, no end-groups)].​6 ​A EO-based KMD
analysis ​provides the KMD plots depicted in ​Figure
3​B (full scale) and Figure
​ 3​C (magni​fi​cation, KMD:
−​0.03 to +0.09). If the points are satisfactorily
aligned along the horizontal axis, no discrimination
is achieved between singly charged, doubly charged,
and product ions, but an unresolved cloud is plotted
instead. Using a fractional EO/45 as the base unit (​R
= 44.0262, round(​R​) = 44, optimal ​X =​ 45, cf.
Supporting ​Information​), the full scale KMD plot
 depicted in ​Figure 3​D o​​ ff​ers a strikingly enhanced
visualization of the data. The doubly charged
congeners (green dots), the cyclic product ion c​n
(blue dots), and the singly charged oligomers (+ the
l​n ​product ions, same composition/KMD but in low
​ now distinctly grouped at di​ff​erent
mass range) are
KMD values. Signals can then be readily extracted
from the KMD plots for a convenient ​interpretation
Analytical Chemistry ​Letter
2686
of the mass spectra (​Figure S6​). Last but not least,
​
the EO/45-based KMD plot is isotopically resolved
in the high mass range with several horizontal lines
assigned to the ​13​C​x isotopes (e.g., the 13​
​ C​5 series
​ ​
plotted in red, ​Figure 3​D). Exemplifying another
situation for which several distributions are found in
the high mass range, the MALDI mass spectra of an
8000 g mol​−​1 ​PCL sample fractionated by size
exclusion

DOI: ​10.1021/acs.analchem.6b05136 ​Anal. Chem. ​2017, 89, 2682−2686 ​chromatography (FY2015) 7000 g ​Representative (​​ Figure S7​) are

depicted in ​Figure 4​A mol​−​1​) ​peaks and

​ ​are Figure
​ ​depicted 4​​ B ​in (​​ M​the ​p ∼
​ ​15000 g ​insets to highlight mol(​
​ M−​​ 1​p ∼
​ ​).

the resolving
​ power of a spiralTOF in this mass range and shows four series of peaks (H, OH) and (H, adducted with
​ ​scale (​Scheme ​CL-based
sodium assigned C​3​and H​7​O) potassium. to as two end-groupsPCL The distributions full 10
carrying ​S1​)

KMD plots Figure

​ 4​D clouds from ​are (​​ M​which ​p ​depicted in ​Figure ∼
​ ​15000 Da) and ​4​C displayed
​ ​(​M​p ∼
​ ​7000 Da)

two unresolved and

no information can be extracted. With CL/ 113 as the base unit (​R ​= 114.0681, round(​R​) = 114, optimal ​X ​= 113, cf.
Supporting Information​), the full scale KMD plots of the two fractions now display four well-discriminated groups

readily assigned to the sodium and potassium adducts of the two

​ PCL (H, OH) (red and green dots, respectively) and
(H,

and a Grant-in-Aid ​“​JSPS KAKENHI​” ​(Grant number: ​

■ ​ REFERENCES ​JP 15F15344). ​(1) Montaudo,

G.; Lattimer, R. P. ​Mass spectrometry of polymers​; CRC Press, Taylor and Francis group: Oxford, 2001.
(2) Hakkarainen, M. ​Mass spectrometry of polymers ​− ​new techniques​; Springer-Verlag: Berlin, Heidelberg, New York, 2012.
(3) Gruendling, T.; Weidner, S.; Falkenhagen, J.; Barner-Kowollik, C. ​Polym. Chem. 2​ 010, ​1,​ 599​−​617.
(4) Pasch, H.; Schrep, W. ​MALDI-TOF mass spectrometry of synthetic polymers;​ Springer-Verlag: Berlin, Heidelberg, New
York, 2003.
(5) Charles, L. ​Mass Spectrom. Rev. ​2014, ​33​, 523​−​543. (6) Wesdemiotis, C.; Solak, N.; Polce, M. J.; Dabney, D. E.;
Chaicharoen, K.; Katzenmeyer, B. C. ​Mass Spectrom. Rev. ​2011, ​30​, 523​−​559. ​Cfractional ​3​H​7​O)-ended CL/113
distributions. as The two KMD plots with the the base units provide an indubitably enhanced visualization of the
mass spectral data as compared to the mass spectrum itself, and the regular CL-based KMD plot and grouping of
points would be readily done with no ​additional data treatment using the appropriate software. ​In conclusion, the
KMD analyses of (a) a low resolution mass spectrum in low mass range at the MS and MS​n ​stages and (b) a high
resolution mass spectrum in high mass range are thus successfully conducted. The use of a fractional base unit
(7) Wesdemiotis, C. ​Angew. Chem., Int. Ed. 2​ 017, ​56,​ 1452​−​1464. (8) Sleno, L. ​J. Mass Spectrom. ​2012, ​47​, 226​−​236. (9)
Marshall, A. G.; Rodgers, R. P. ​Acc. Chem. Res. 2​ 004, ​37,​ 53​−​59. (10) Sato, H.; Nakamura, S.; Teramoto, K.; Sato, T. ​J. Am.
Soc. Mass Spectrom. 2​ 014, ​25,​ 1346​−​1355.
(11) Kendrick, E. ​Anal. Chem. 1​ 963, ​35,​ 2146​−​2154. (12) Dimzon, I. K.; Trier, X.; Fromel, T.; Helmus, R.; Knepper, T. P.; de
Voogt, P. ​J. Am. Soc. Mass Spectrom. 2​ 016, ​27,​ 309​−​318.
(13) Fouquet, T.; Nakamura, S.; Sato, H. ​Rapid Commun. Mass Spectrom. 2​ 016, ​30,​ 973​−​981.
(14) Fouquet, T.; Aizawa, H.; Sato, H. ​Rapid Commun. Mass i​ mproves the resolution of the KMD analysis itself and

​ basis of these ​fi​ndings

Spectrom. 2​ 016, ​30,​ 1818​−​1822. ​counterbalances the lower resolution of the MS analysis. On the
and relying on a rational choice of the divisor, the concept of fractional base units is expected to widely extend the
scope of the KMD analysis in the ​fi​eld of polymer chemistry by allowing the use of more a​ff​ordable low resolution
mass analyzers and making high molecular weight ​samples suited to the mass defect analysis with all its advantages

(fast data mining, easy visualization, and presentation to wide audiences) ​

■ ​ •​S ​Supporting ​ASSOCIATED

as long ​Information
​ ONTENT ​the mass spectra are isotopically resolved. ​(15) Fouquet, T.; Sato, H. ​Rapid Commun. Mass Spectrom.
as C
2016, ​30​, 1361​−​1364.
(16) Fouquet, T.; Sato, H. ​Mass Spec. (Tokyo),​ submitted for publication.
(17) Satoh, T.; Tsuno, H.; Iwanaga, M.; Kammei, Y. ​J. Am. Soc. Mass Spectrom. ​2005, ​16​, 1969​−​1975.
(18) Strohalm, M.; Kavan, D.; Novak, P.; Volny, M.; Havlícek, V. ​Anal. Chem. 2​ 010, ​82,​ 4648​−​4651.
(19) Fouquet, T.; Torimura, M.; Sato, H. ​Mass Spectrom. 2​ 016, ​5,​ A0050.
The Supporting Information is available free of charge on the ​ACS Publications website ​at DOI: ​10.1021/acs.anal-
chem.6b05136​.
Extended experimental section; PVP, PEO, and PCL structures; discussion of the fractional base units;
MALDI-spiralTOF MS spectrum and KMD plot of PVP; KMD plots from the ESI-MS and ESI-MS​4 ​spectra of
PVP; mass spectrum and KMD plot of PEO; SEC

■ ​ AUTHOR ​chromatogram ​INFORMATION ​of PCL (​PDF​) ​Corresponding Authors ​*​E-mail:

thierry.fouquet@aist.go.jp​. ​*​E-mail: ​sato-hiroaki@aist.go.jp​. ​ORCID ​Thierry Fouquet: ​0000-0002-9473-9425

Notes ​The ​
■ ​ ACKNOWLEDGMENTS ​authors declare no competing ​fi​nancial interest. T.F. and H.S.

acknowledge the ongoing ​fi​nancial support by the Japan Society for the Promotion of Science (JSPS) under the
postdoctoral fellowship for overseas researchers program