Mass-Remainder Analysis (MARA): a New Data Mining Tool for

Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Mass-Remainder Analysis (MARA): a New Data Mining Tool for Copolymer Characterization Tibor Nagy, Á kos Kuki, Miklós Zsuga, and Sándor Kéki* Department of Applied Chemistry, University of Debrecen, H-4032 Debrecen, Egyetem tér 1., Hungary S Supporting Information *

ABSTRACT: A new data mining method is proposed for the determination of the copolymer composition from moderate/ low resolution complex mass spectra. The Mass-remainder analysis (MARA) does not require a “Kendrick-like” transformation to a new mass scale, it is simply based on the calculation of the remainder after dividing by the exact mass of one of the repeat units of the copolymer (e.g., B of an A/B copolymer). Plotting the remainder of this division (MR) versus m/z the homologous series differing only by a number of base units (e.g., B unit) can be visualized. The number of A units (nA) and subsequently nB is assigned to the m/z peaks using the bijective nA, MR mapping. Simultaneously, our algorithm removes the isotopes from the peak list. However, the intensities of the monoisotopes are increased to the value corresponding, approximately, to the total intensity of their isotope peaks. The correction of the mass spectral peak intensities enables the accurate calculation of the usual polymer and copolymer quantities: the molecular weight-average, the number-averaged molecular weight of A and B units, the composition drift, or the bivariate distribution, among others. Our Mass-remainder analysis method was demonstrated by the analysis of various ethylene oxide/propylene oxide copolymers.

M

ethylene oxide (nEO) and propylene oxide (nPO) units can be easily calculated for every m/z peak. This easy assignment of the mass spectrum peaks enables the calculation of the copolymer composition, for example, the molar fraction of the ethylene oxide unit in the copolymer. However, the isotope peaks have slightly different KMD values than the corresponding monoisotopic peaks that may distort the calculations. For example, nEO = 0.00 and nPO = 15.00 are calculated for the m/z 911.628 peak of the ethylene oxide/propylene oxide copolymer, but its first isotope peak m/z 912.631 is assigned as nEO = −0.49 and nPO = 15.39. Very recently a fraction of the repeat unit R/X, with X being a positive integer, was used as the base unit of the KMD analysis resulting resolution-enhanced KM values.7−9 The introduction of fractional base unit broadens the scope of the applications of the KMD analysis for polymers to low resolution and high mass range mass spectra and improves the separation of isotope peaks. Fouquet et al. developed a referenced Kendrick mass defect (RKMD) procedure (KMD referenced to the terminal group and adduct composition) with fractional base units for the calculation of copolymer composition.10 The information about the degrees of polymerization (DP) for the two comonomers (e.g., the number of EO and PO units in EO/PO copolymer) is basically contained in

ass spectrometry plays a very important role in the characterization of polymers. Combined with soft ionization techniques, such as matrix-assisted laser desorption/ionization (MALDI)1,2 and electrospray ionization (ESI),3 mass spectrometry yields highly sensitive, selective, and specific data within a relatively short time and provides information on the individual, intact molecules. However, the mass spectra of complex polymer formulations can contain a huge number of m/z peaks arising from different homopolymer/copolymer series. Therefore, the manual assignment of the individual peaks is particularly challenging and time-consuming. To solve this issue, complex mass spectra can be visualized in mass defect (i.e., exact mass subtracted from the nominal, integer mass) versus nominal mass plots.4 Kendrick suggested a new mass scale setting the mass of CH2 to an integer value of 14 Da, instead of the IUPAC mass of 14.01565 Da.5 In the Kendrick mass defect (KMD) plots the homologous series differing only by a number of CH2 base units can be easily identified. Kendrick mass analysis was historically first used for crude oil samples. Other mass scales have also been applied in the data mining procedure outlined by Kendrick. Sato et al. created KMD plots by using the repeat unit (R) of the polymer backbone (e.g., propylene oxide) as the base unit for the analysis of polymers and copolymers (e.g., ethylene oxide/ propylene oxide; EO/PO copolymer).6 Their KMD plot visually represents the structural distribution of the polymer components, and furthermore, the KMD value and nominal Kendrick mass (NKM) value, and subsequently the number of © XXXX American Chemical Society

Received: November 15, 2017 Accepted: February 14, 2018 Published: February 14, 2018 A

DOI: 10.1021/acs.analchem.7b04730 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

summed. The recorded mass spectra were evaluated with the DataAnalysis 3.4 software from Bruker. Samples for MALDI-TOF MS were prepared with DHB matrix dissolved in methanol at a concentration of 20 mg/mL, polymers were solved in methanol also at a concentration of 10 mg/mL. Ionizing agent was sodium trifluoroacetate at a concentration of 5 mg/mL, methanol was used as solvent. The mixing ratio was 10/5/1 (matrix/polymer/cationizing agent) and 0.5 μL of the solution was dropped onto the sample plate and allowed to air-dry.

the RKMD plots, but due to the strong aliasing (explained in detail in ref 10) an antialiasing algorithm was applied. In our opinion, the slight differences in the RKMD values of the isotope peaks may hinder this antialiasing filter. Furthermore, although the resolution-enhanced KMD analysis can be successfully applied for the lower resolution spectra recorded by a typical time-of-flight (TOF) analyzers, it is not able to handle the overlapped peaks of the ethylene oxide/propylene oxide copolymers (EOxPOy overlaps the second isotope peak of EOx−4POy+3). In this paper, we propose a new and robust data mining method for the determination of the copolymer composition from moderate/low resolution complex mass spectra. Our simple algorithm, the Mass-remainder analysis (MARA), can handle the issues related to the isotope peaks, and can be applied for the m/z peaks with lower mass accuracy in the higher mass ranges which can not be handled by the other methods. Our method is fundamentally different from those based on the calculation of mass defect, in particular Kendrick mass defect, in the following aspects: (i) MARA does not require any transformation to a new mass scale, it is based on the calculation of the remainder after dividing by the exact mass of one of the repeat units of the copolymer; (ii) The number of A units of an A/B copolymer (nA) and subsequently nB is assigned to the m/z peaks using the bijective nA − MassRemainder (MR) mapping; (iii) MARA removes the isotopes from the peak list; (iv) It handles the overlapped isotopes of the EO/PO copolymers; (v) The intensities of the monoisotopic peaks are increased approximately to the total intensity of their isotope peaks.

■

RESULTS AND DISCUSSION Block copolymers consist of ethylene oxide and propylene oxide blocks arranged in a triblock structure are widely used as nonionic surface-active agents in industrial, and domestic applications, cosmetics, pharmaceuticals, and so on. The diversity of applications justifies the importance of the structural characterization of these copolymers. Furthermore, the commercially available EOx−POy−EOx copolymers (e.g., Pluronics or Synperonics) may contain admixtures of the PO homopolymer and di- and triblock copolymers with lower degrees of polymerization than expected.12 It emphasizes the strong need for effective copolymer analysis methods, particularly if it is suitable for commonly used mass spectrometers. We show our new data mining method for copolymer mass spectra, the Mass-remainder analysis, by the characterization of EO/PO copolymers. Figure 1 shows the

■

EXPERIMENTAL SECTION Chemicals. Pluronic copolymers (PE3100, PE3500, PE4300, PE6100, and PE6120) were purchased from BASF Ludwigshafen, Germany. Methanol (HPLC-MS grade) was received from VWR International (Leuven, Belgium). Electrospray Quadrupole Time-of-Flight Mass Spectrometry (ESI-QTOF MS). The MS measurements were performed with a MicroTOF-Q type Qq-TOF MS instrument (Bruker Daltonik, Bremen, Germany) using an ESI source with positive ion mode. The sample solutions were introduced directly into the ESI source with a syringe pump (Cole-Parmer Ins. Co., Vernon Hills, IL, U.S.A.) at a flow rate of 3 μL/min. The spray voltage was set to 4 kV. The temperature of the drying gas (N2) was kept at 180 °C. The MS spectra were accumulated and recorded by means of a digitizer at a sampling rate of 2 GHz. The recorded mass spectra were evaluated with the DataAnalysis 3.4 software from Bruker. In order to suppress the multiple charged ions in the ESI source high csample/cionizing agent concentration ratio was applied as it was suggested by S. Kéki.11 Therefore, the concentrations were 5 × 10−4 M and 5 × 10−5 M for the polymer and ionizing agent, respectively. Matrix-Assisted Laser Desorption/Ionization Time-ofFlight Mass Spectrometry (MALDI-TOF MS). The MALDITOF MS measurements were performed with a Bruker BIFLEX III mass spectrometer equipped with a time-of-flight (TOF) mass analyzer. In all cases 19 kV acceleration voltage was used with pulsed ion extraction (PIE). The positive ions were detected in the reflectron mode (20 kV). A nitrogen laser (337 nm, 3 ns pulse width, 106−107 W/cm2) operating at 4 Hz was used to produce laser desorption and 1000 shots were

Figure 1. MALDI-TOF mass spectra of the EO x −PO y −EO x copolymer with approximately 50 wt % EO content.

mass spectra of the EOx−POy−EOx copolymer with approximately 50 wt % EO content recorded by typical time-of-flight analyzers equipped with a MALDI ion source. Additional spectra of triblock copolymers with various EO contents and number-average molecular weights (Mn) are given in the Supporting Information in Figures S1 and S2a−c. As seen in Figure 1 (and Figures S1 and S2), some EOxPOn series can be recognized differing in the number of EO units (x = 0, 1, 2, and 3 in Figure S1). As examples, some m/z peaks are identified in Figure 1, but the manual assignment of all of the observed peaks, including isotope peaks (over an intensity or relative intensity threshold), is time-consuming or almost impossible. An additional phenomenon that can distort the calculation of EO/PO copolymer composition that these mass spectra contain peaks with 2 Da intervals, because the m/z values of EO and PO equal 44 and 58, respectively, and 4 × 44 − 2 = 3 × 58. At m/z 2000 and 5000 resolutions of approximately 73000 and 182000 are required, respectively, to separate the monoisotopic peak of EOxPOy and the second B


Article


Figure 2. (a) Mass remainder (MR) vs m/z plot and (c) zoomed MR − m/z plot of the mass spectrum shown in Figure 1. (b) MR vs m/z plot after deisotoping and intensity correction. (d) The nEO − nPO plot after deisotoping and intensity correction.

S3a, S4a, S5a, and S6a. Each dot row of Figure 2a represents an EOxPOn series with constant x values, as indicated by the labels EO0 to EO39. Moreover, as seen in the zoomed MR plot in Figure 2c, the 13 Cx isotopes are clearly separated exactly by ΔMR = 1 Mass Remainder value difference. However, of course, the overlapped peaks EOxPOy(13C2) and EOx+4POy−3(13C0) (see Figure S1a, inset) are represented as unseparated dots, see, for example, EO5PO18(13C2) = EO9PO15(13C0) in Figure 2c. The 2 Da intervals between the EO/PO copolymer peaks result additional (even triple or multiple) overlaps, for example, EO5PO18(13C3) = EO9PO15(13C1) or EO5PO18(13C4) = EO9PO15(13C2) = EO13PO12(13C0), as seen in Figure 2c. In spite of all these issues the good resolution of the MR versus m/z plot enables the accurate determination of the copolymer composition. A simple algorithm was developed using a regular spreadsheet software and its built-in programming tools for the processing of the raw mass spectrum with the following main steps (see Scheme 1): 1. Calculation of the Mass-Remainder Values by Eq 1. All of the observed peaks, above an intensity threshold, were directly used without deisotoping. 2. nEO, nPO Assignment and Deisotoping. As seen in Figure 2a, the MR values contain the implicit information about the number of EO units in the copolymer. An auxiliary table was created in the spreadsheet program containing the nEO − MR mapping. For instance, the MR = 16.927 value of m/z = 10 × 44.02622 + 40.99979 was assigned to the value nEO = 10,

isotopic peak of EOx−4POy+3 (see, e.g., Figure S1a, inset, EO4PO13(13C0) and EO0PO16(13C2) at m/z 971). The latter resolution (and often the lower one) is out of the range of a conventional TOF analyzer. However, the differences between the measured and the calculated isotope ratios of the EO0PO16 peak reveal the presence of EO4PO13. As the inset of Figure S1b shows, the relative intensity of the EO0PO16(13C2) isotope is 18%, as calculated with the IsotopePattern software of Bruker Daltonics, against the measured intensity ratio of 28%. (The measured intensity ratio of the first isotope peak EO0PO16(13C1) agree well with the calculated one, 53% and 54%, respectively.) We will take into consideration these intensity ratio differences, as it will be detailed later. In order to identify the homologous series differing only by a number of base units (e.g., a number of PO units) and determine the number of ethylene oxide and propylene oxide units (nEO and nPO, respectively) in the copolymer, we propose a simple method, which does not require a “Kendrick-like” transformation to a new mass scale. Mass-Remainder values of the measured m/z peaks are calculated according to ⎛ m/z ⎞ ⎟ × R MR = m /z − int⎜ ⎝ R ⎠

(1)

where R is the exact mass of the PO repeat unit (i.e., C3H6O = 58.04187 Da) and int(x) rounds x down to the nearest integer. Figure 2a depicts the MR versus m/z plot of the mass spectrum shown in Figure 1. Additional MR plots of various triblock copolymers are given in the Supporting Information in Figures C


Article

Analytical Chemistry Scheme 1. Flowchart of the Mass Remainder Analysis (MARA) applied for EO/PO copolymers

nPO = y = round((m /z − x × 44.02622 − 40.99979) /58.04187)

(2)

where round(a) rounds a to the nearest integer. Using the found nEO = x; nPO = y values an m/z value can be calculated as m/zcalcd = x × 44.02622 + y × 58.04187 + 40.99979. An x; y assignment is accepted, only if two criteria are fulfilled: (i) abs(m/z − m/zcalcd) < ε = 0.05, and (ii) y ≥ 0. These criteria filter the bad assignments and the isotope m/z peaks as well, offering this way an alternative method for deisotoping, namely, after this step, the mass list contains only monoisotopic EOxPOy peaks. For example, nEO = 181 and nPO = −120 are assigned to the peak m/z = 1044.710 of the isotope EO3PO15 (13C1), but this assignment is rejected due to the negative nPO value (criterion (ii)). Besides, if the calculated nPO value is positive, only the monoisotopic peaks fulfill the criterion (i). The deisotoping and the subsequent intensity correction of the monoisotopic peaks (see steps 3 and 4) is one of the main advantages of our method over the KMD analysis. For example, the procedure for the nEO calculation developed by Fouquet et al.10 results nEO = 24 for the EO3PO15 (13C2) peak. 3. Correction of Overlapped Peaks. Another novel aspect and advantage of our method, that it does not require the separation of the overlapped isotopes, expanding thereby the scope of the KMD method toward lower resolution mass spectra and higher mass ranges. The measured intensities of the monoisotopic peaks are corrected by the overlapping isotope peaks based on the theoretical isotope distributions, as follows (see an example in the Supporting Information in Table S1). The mass spectrum is already filtered in step (2), it contains only monoisotopic EOxPOy peaks (see Table S1, Step 2 Deisotoping column). In the further operations, the intensity of the isotope peaks of an EOxPOy copolymer will be calculated (see Figure S1b inset, theoretical distribution). The dependence of the relative intensity of the second and fourth isotope peaks (13C2, 13C4, respectively) on the number of carbon atoms was determined for the EO/PO copolymers as a second order and fourth order polynomial function, respectively, using the IsotopePattern software from Bruker Daltonics. The intensity of a monoisotopic peak EOxPOy is decreased by the intensity of the second isotope peak of EOx−4POy+3 and the intensity of the forth isotope peak of EOx−8POy+6 calculated by the second order and fourth order polynomial function, respectively (see Table S1, Step 3 Overlapped peak correction column, for example, the intensity of EO8PO10 is decreased by the intensity of the second isotope peak of EO4PO13 and the intensity of the forth isotope peak of EO0PO16). Of course, if the peak EOx−4POy+3 and EOx−8POy+6 does not exist, no correction occurs. 4. Summing Isotope Peak Intensities. In the last step, the intensity of every monoisotopic peak is increased to the value corresponding, approximately, to the summarized intensity of its isotope peaks (see Table S1, Step 4 Summing isotope peaks column). To do this, a correction function was created, similarly to the ones in step 3. For example, 63% of the EO−PO copolymer of mass 800 Da is monoisotopic, whereas, for the copolymer at 1600 Da this will be the case for only 40%. Figure 2b shows the MR versus m/z plot after processing the raw spectrum shown in Figure 2a. The dot diameters in Figure 2b now indicate the relative concentration of each copolymer species. The 2-dimensional nEO − nPO plot can be easily

where C2H4O = 44.02622 Da and [H2O + Na]+ = 40.99979 Da are the exact masses of the EO repeat unit and the end group with the adduct ion, respectively. In the range nEO = 0−350 this assignment is bijective with a gap at least 0.16 between two adjacent MR values. It means that the MR values of the copolymers in this nEO range are all different and the step between them is 0.16 or greater. Our algorithm finds an nEO = x value if abs(MRmeasured − MRx) < ε = 0.05, where abs(a) returns the absolute value of a, MRmeasured, and MRx are the mass-remainder values calculated from the measured m/z (eq 1) and the table MR values, respectively. The ε = 0.05 criterion means, that the correct nEO assignment requires a 0.05 Da mass accuracy, which fulfilled by the common TOF analyzers with ease in the mass range to 20 kDa (which corresponds to the nEO = 0−350 range). If an nEO = x value was found, the number of propylene oxide units in the copolymer is calculated as follows: D


Article


quantity for a copolymer sample is the weight of copolymer chains at each composition and at each length, that is, the “Bivariate Distribution”.15 As seen in Figure 4, the asymmetrical bivariate distribution of the Pluronic PE 3500 sample also shows the composition drift reported in Figure 3.

constructed after running our mass-remainder algorithm (MARA), and it clearly visualizes the copolymer composition, the distribution of the EO and PO units (see Figure 2d, and additionally in the Supporting Information, Figures S3b, S4b, S5b, and S6b). The correction of the overlapped peaks and the summation of all the isotope peaks performed by our method allows the accurate calculation of the usual molecular weight averages Mn (number-average), Mw (weight-average), and the polydispersity index Mw/Mn. Moreover, additional quantities can be determined describing the chemical composition of the copolymer in terms of the monomers X (X = EO, PO), such as the average molar fraction of X unit in the copolymer, cX;13 the average weight fraction of X, wX; the number-average number of units and weight-average number of units for X, nnX and nwX, respectively;14 and the polydispersity index for monomer X, nwX/nnX.14 The values of these quantities are listed in Table 1. Table 1. Chemical Composition of Pluronic PE 3500 (EOx− POy−EOx Copolymer with Approximately 50 wt % EO Content) Calculated from the Mass Spectra in Figure 1 by the Mass-Remainder Analysis (MARA) Mn Mw Mw/Mn cEO wEO nnEO nwEO nwEO/nnEO nnPO nwPO nwPO/nnPO

Figure 4. Bivariate distribution of the EOx−POy−EOx copolymer with approximately 50 wt % EO content. The figure was created by TeraPlot 1.4 (Kylebank Software Ltd., U.K.) graphing software (trial version).

1540 1630 1.05 0.533 0.464 16.1 19.7 1.23 14.1 14.7 1.04

The distribution of the EO and PO units of the copolymer with approximately 10 wt % EO (see Figure S3b) tells us how the EOx−POy−EOx copolymers were synthesized. The molecular weight averages (Mn and Mw) can be calculated both for the PO homopolymers (nEO = 0) and for the POy segment of the triblocks. The Mn 951 and 964 Da was calculated for the PO homopolymers and POy segments, respectively, while the polydispersity indexes are 1.038 and 1.034 for the PO homopolymers and POy segments, respectively. The calculated PO oligomer length distributions are shown in the Supporting Information in Figure S6. The good agreement between the corresponding quantities suggests, that the copolymer was synthesized from a PO block using it as a macroinitiator for polymerization with EO to obtain EOx− POy−EOx triblock copolymers

As Table 1 shows the average weight fraction of the ethylene oxide unit is slightly lower than the value 50 wt % provided by the supplier. Once the number of EO and PO units is assigned to the individual mass spectra peaks by the mass-remainder analysis, not only the average polymer properties can be determined, but even more complex analysis can be performed. Figure 3 shows the variation of copolymer composition with the molar mass, usually called “composition drift”.13 As seen in Figure 3, the EO block length strongly depends on the degree of polymerization of the copolymer. An additional important

■

CONCLUSIONS A new method was proposed for the processing of the mass spectra of copolymers. Our Mass-remainder analysis does not require any transformation to a new mass scale, it is based on the calculation of the remainder after dividing by the exact mass of one of the repeat units of the copolymer. In our algorithm both EO and PO can be used as the divisor. The strengths of our method are (1) the true m/z peak assignment (i.e., the calculation of the number of ethylene oxide and propylene oxide units) does not require a high resolution/high mass accuracy instrument; (2) the monoisotopic peaks can easily be separated from the isotope peaks offering an alternative method for deisotoping; (3) it is able to handle the overlapped peaks; (4) the isotope peak intensities are summarized, which is necessary for the correct calculation of the copolymer composition. Our mass-remainder analysis (MARA) method can be applied for other types of complex copolymers. The key of the MR analysis of an A/B copolymer (using B as the divisor unit) is the one-to-one correspondence of the nA and MR values with a large enough gap between two adjacent MR

Figure 3. Composition drift of the EOx−POy−EOx copolymer with approximately 50 wt % EO content. E


Article


(10) Fouquet, T.; Cody, R. B.; Sato, H. J. Mass Spectrom. 2017, 52, 618−624. (11) Kéki, S. Rapid Commun. Mass Spectrom. 2006, 20, 3374−3378. (12) Kabanov, A. V.; Lemieux, P.; Vinogradov, S.; Alakhov, V. Adv. Drug Delivery Rev. 2002, 54, 223−233. (13) Montaudo, G.; Lattimer, R. P. Mass Spectrometry of Polymers; CRC Press, 2001. (14) van Rooij, G. J.; Duursma, M. C.; de Koster, C. G.; Heeren, R. M. A.; Boon, J. J.; Schuyl, P. J. W.; van der Hage, E. R. E. Anal. Chem. 1998, 70, 843−850. (15) Montaudo, M. S. Mass Spectrom. Rev. 2002, 21, 108−144.

values (e.g., 0.1 Da). For example, for isobutylene/stryrene (IB/S) copolymers, using IB as the divisor unit, the nS − MR mapping is bijective in the range nS = 0−300 with a gap at least 0.175 Da between two adjacent MR values.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.7b04730. Figure S1: (a) MALDI-TOF and (b) ESI-TOF mass spectra of the EOx−POy−EOx copolymer with approximately 10 wt % EO content. Figure S2: (a) PE4300 (30 wt % EO content), (b) PE6100 (10 wt % EO content), and (c) PE6120 (12 wt % EO content) Pluronics. Table S1: The intensity values of some mass peaks (EOx− POy−EOx copolymer with approximately 10 wt % EO content) calculated by the MARA method. Figure S3: (a) Mass remainder (MR) and (b) copolymer composition plots of PE3100 (10 wt % EO content) Pluronic. Figure S4: (a) Mass remainder (MR) and (b) copolymer composition plots of PE4300 (30 wt % EO content) Pluronic. Figure S5: (a) Mass remainder (MR) and (b) copolymer composition plots of PE6100 (10 wt % EO content) Pluronic. Figure S6: (a) Mass remainder (MR) and (b) copolymer composition plots of PE6120 (12 wt % EO content) Pluronic. Figure S7: PO oligomer length distributions of the EOx−POy−EOx copolymer with approximately 10 wt % EO content. Blue, PO homopolymers; red, POy segments (PDF).

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +36 52 518662. ORCID

Sándor Kéki: 0000-0002-5274-6117 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The work was supported by the GINOP-2.3.2-15-2016-00041 and GINOP-2.3.3-15-2016-00021 projects. These projects were cofinanced by the European Union and the European Regional Development Fund. Furthermore, this paper was also supported by the Grant K-116465 and supported through the New National Excellence Program of the Ministry of Human Capacities, Ú NKP-16-3 (T.N.).

■

REFERENCES

(1) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299−2301. (2) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T.; Matsuo, T. Rapid Commun. Mass Spectrom. 1988, 2, 151−153. (3) Wong, S. F.; Meng, C. K.; Fenn, J. B. J. Phys. Chem. 1988, 92, 546−550. (4) Sleno, L. J. Mass Spectrom. 2012, 47, 226−236. (5) Kendrick, E. Anal. Chem. 1963, 35, 2146−2154. (6) Sato, H.; Nakamura, S.; Teramoto, K.; Sato, T. J. Am. Soc. Mass Spectrom. 2014, 25, 1346−1355. (7) Fouquet, T.; Sato, H. Mass Spectrom. 2017, 6, A0055. (8) Fouquet, T.; Sato, H. Anal. Chem. 2017, 89, 2682−2686. (9) Fouquet, T.; Sato, H. Rapid Commun. Mass Spectrom. 2017, 31, 1067−1072. F


Mass-Remainder Analysis (MARA): a New Data Mining Tool for

Recommend Documents