Factor analysis of chemical ionization experiments: Numerical simulations and an experimental case study of the ozonolysis of α-pinene using a PTR-ToF-MS: Numerical simulations and an experimental case study of the ozonolysis of α-pinene using a PTR-ToF-MS

TY - JOUR

T1 - Factor analysis of chemical ionization experiments: Numerical simulations and an experimental case study of the ozonolysis of α-pinene using a PTR-ToF-MS

T2 - Numerical simulations and an experimental case study of the ozonolysis of α-pinene using a PTR-ToF-MS

AU - Rosati, Bernadette

AU - Teiwes, Ricky

AU - Kristensen, Kasper

AU - Bossi, Rossana

AU - Skov, Henrik

AU - Glasius, Marianne

AU - Pedersen, Henrik B.

AU - Bilde, Merete

PY - 2019/2/15

Y1 - 2019/2/15

N2 - In this study we examine the application of a factor analysis technique for datasets recorded by the chemical ionization instrument Proton Transfer Reaction - Time-of-Flight - Mass Spectrometer (PTR-ToF-MS). Numerical simulations were carried out to test and optimize the performance of the factorization method Multivariate Curve Resolution - Alternating Least Squares (MCR-ALS). The simulations demonstrated that the choice of initial estimates for the factor analysis are crucial for the outcome of the factorization. For this purpose we describe a new method based on the Pearson correlation coefficient which, compared to the other tested methods, yields the best factorization results. Overlapping contributions of different substances to a certain mass bin and large uncertainties in the data deteriorate the performance of the factor analysis. The deconvolution technique was also applied to data obtained with a PTR-ToF-MS employed at a smog chamber. The response from single substances (fragmentation patterns) such as α-pinene, α-pinene oxide and pinonaldehyde as well as a more complex experiment involving the ozonolysis of α-pinene are investigated. For the single substances, the factorization method points to the presence of impurities which could be differentiated from the masses affiliated with the actual substances. The α-pinene ozonolysis experiment was analysed with the factorization technique and yielded four factors, one representing the decay of the reactant and three different products. The matrix factorization technique appeared to be a valuable tool allowing for a detailed analysis of complex mass spectra without the need of any prior information.

AB - In this study we examine the application of a factor analysis technique for datasets recorded by the chemical ionization instrument Proton Transfer Reaction - Time-of-Flight - Mass Spectrometer (PTR-ToF-MS). Numerical simulations were carried out to test and optimize the performance of the factorization method Multivariate Curve Resolution - Alternating Least Squares (MCR-ALS). The simulations demonstrated that the choice of initial estimates for the factor analysis are crucial for the outcome of the factorization. For this purpose we describe a new method based on the Pearson correlation coefficient which, compared to the other tested methods, yields the best factorization results. Overlapping contributions of different substances to a certain mass bin and large uncertainties in the data deteriorate the performance of the factor analysis. The deconvolution technique was also applied to data obtained with a PTR-ToF-MS employed at a smog chamber. The response from single substances (fragmentation patterns) such as α-pinene, α-pinene oxide and pinonaldehyde as well as a more complex experiment involving the ozonolysis of α-pinene are investigated. For the single substances, the factorization method points to the presence of impurities which could be differentiated from the masses affiliated with the actual substances. The α-pinene ozonolysis experiment was analysed with the factorization technique and yielded four factors, one representing the decay of the reactant and three different products. The matrix factorization technique appeared to be a valuable tool allowing for a detailed analysis of complex mass spectra without the need of any prior information.

KW - Matrix factorization

KW - MCR-ALS

KW - Pinonaldehyde

KW - PTR-ToF-MS

KW - α-pinene

KW - α-pinene oxide

UR - http://www.scopus.com/inward/record.url?scp=85056703703&partnerID=8YFLogxK

U2 - 10.1016/j.atmosenv.2018.11.012

DO - 10.1016/j.atmosenv.2018.11.012

M3 - Review

AN - SCOPUS:85056703703

SN - 1352-2310

VL - 199

SP - 15

EP - 31

JO - Atmospheric Environment

JF - Atmospheric Environment

ER -