- Email: [email protected]
MIR and NIR group spectra of n-alkanes and 1-chloroalkanes
Accepted Manuscript MIR and NIR group spectra of n-alkanes and 1-chloroalkanes Michał Kwaśniewicz, Mirosław A. Czarnecki PII: DOI: Reference:
S1386-1425(15)00208-5 http://dx.doi.org/10.1016/j.saa.2015.01.134 SAA 13346
To appear in:
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy
Received Date: Revised Date: Accepted Date:
9 November 2014 19 January 2015 29 January 2015
Please cite this article as: M. Kwaśniewicz, M.A. Czarnecki, MIR and NIR group spectra of n-alkanes and 1chloroalkanes, Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy (2015), doi: http:// dx.doi.org/10.1016/j.saa.2015.01.134
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
MIR and NIR group spectra of n-alkanes and 1-chloroalkanes
Michał Kwaśniewicz, Mirosław A. Czarnecki* Faculty of Chemistry, University of Wrocław, F. Joliot-Curie 14, 50-383 Wrocław, POLAND
Address for correspondence: Prof. Mirosław Czarnecki Faculty of Chemistry, University of Wrocław F. Joliot-Curie 14, 50-383 Wrocław, POLAND E-mail address: [email protected]
Tel. 48-71-3757238 Fax 48-71-3282348
Abstract Numerous attempts were undertaken to resolve the absorption originating from different parts of alkanes. The separation of the contributions from the terminal and midchain methylene units was observed only in the spectra of solid alkanes at low temperatures. On the other hand, for liquid alkanes this effect was not reported as yet. In this study, ATR-IR, Raman and NIR spectra of eight n-alkanes and seven 1-chloroalkanes in the liquid phase were measured from 1000 to 12000 cm-1. The spectra were analyzed by using two-dimensional (2D) correlation approach and chemometrics methods. It was shown that in 2D asynchronous contour plots, constructed from the spectra of n-alkanes and 1-chloroalkanes, the methylene band was resolved into two components. These two components were assigned to the terminal and midchain methylene groups. For the first time, the contributions from these two molecular fragments were resolved in the spectra of liquid n-alkanes and 1-chloroalkanes. MCR-ALS resolved these spectra into two components that were assigned to the ethyl and midchain methylene groups. These components represent the group spectra that can be used for assignment, spectral analysis and prediction of unknown spectra. The spectral prediction based on the group spectra provides very good results for n-alkanes, especially in the first and second overtone regions.
Keywords: n-alkanes, 1-chloroalkanes, group MIR/NIR spectra, midchain methylene groups, spectral prediction, two-dimensional correlation analysis, chemometrics.
2
1. Introduction Conformation of hydrocarbon chains determines the structure and properties of biological systems and polymers. In solid state the chains usually take all-trans conformation, while in the liquid state there exists a conformational disorder [1,2]. Many efforts were undertaken to resolve the absorption from different parts of hydrocarbons. Particularly, separation of the contribution from the terminal and midchain methylene units is not easy task. Ricard-Lespade et al. reported that the MIR spectrum of polymethylene chains in solid state is almost insensitive to the flexibility of the chain [3]. However, Raman spectra were complicated by Fermi resonances between the CH2 symmetric stretchings and the overtones and combinations of CH2 bendings. The authors did not find any differences in peak positions for the midchain CH2 and those adjacent to the CH3. On the other hand, Hill and Levin observed in the Raman and MIR spectra of n-hexadecane at liquid nitrogen temperatures two bands at 2855 and 2846 cm-1 that were assigned to the symmetric CH2 stretching of the terminal and midchain units, respectively [4]. MacPhail et al. also resolved these two bands at 2853 and 2846 cm-1 in the MIR spectrum of crystalline n-CD3-(CH2)20-CD3 at low temperatures (<10 K) [5]. Snyder et al. evidenced that the methylene band of the crystalline polyethylene (at 2922 cm-1) was affected by the chain conformation, whereas the methyl band (at 2938 cm-1) was influenced by the solvent [6]. Another conclusion was obtained by Šebek et al. from the studies of the CH stretching band in selected organic molecules in the liquid and gas phase [7]. It was shown that the CH3 symmetric and asymmetric frequencies are almost independent on the surroundings and the corresponding bands always appear at the same position. In contrast, the location of both symmetric and asymmetric CH2 stretching bands strongly depends on the neighboring groups. Fang et al. evidenced the conformational sensitivity of the overtones of methyl C-H stretching vibrations resulting from the influence of the heteroatom (N, O, S) [8]. This effect was explained in terms of donation of electron
3
density form the lone pair into the trans oriented antibonding orbital of a C-H bond. Numerous works demonstrated that the overtones region is also sensitive on the chain conformation [9-14]. The better peaks separation in the NIR spectra made possible to resolve the absorption from the axial and equatorial CH bonds in NIR spectrum of cyclohexane [13]. The first overtones have both the local and normal nature. As a result, the corresponding spectra are complex due to possibility of various resonances. In contrast, the higher overtones spectra are simple due to the local character of the vibrations [9-14]. Iwamoto suggested that NIR spectrum is a useful tool for separation of the characteristic spectral features of heptene from the spectra of solvents like heptane or toluene [15]. This separation can be performed down to concentrations of 0.1%. Reflectance spectra of alkanes from the ultraviolet to MIR show spectral similarity within the alkane family. On the other hand, the spectra retain sufficient diversity to distinguish among family members from methane to decane [16]. Klavarioti et al. demonstrated that NIR spectra can be used for discrimination between different classes of hydrocarbons (alkanes, aromatics, chlorinated alkanes) [17]. Discrimination within the class of alkanes is more difficult but still possible through the comparison of the peak ratios. Another interesting idea was proposed by Parker et al. [18]. The authors demonstrated that the spectra of liquid aliphatic hydrocarbons measured from 3900 to 6000 cm-1 can be represented as the sum of a limited number of group spectra that were obtained from principal factor analysis (PFA). Using PFA to the set of the spectra of related structure (e.g. homologous series) one can determine the factors necessary to reconstruct the original spectra with a high precision. For n-alkanes spectra from n-hexane to n-hexadecane the authors considered only two structural units: CH3 and CH2. However, it was later shown that this was an oversimplified picture. The authors concluded that in the absence of deuterated species at either the terminal CH2 group, it is not possible to separate the contribution from the terminal and midchain CH2.
4
This study evidences that the separation of the vibrational absorption from the terminal and midchain CH2 is possible also for non-deuterated samples in the liquid phase. In this purpose we recorded ATR-IR, Raman and NIR spectra of eight n-alkanes and seven 1chloroalkanes in the liquid phase. The spectra were analyzed by using 2D correlation approach and chemometrics methods. The main aim of present study was (a) evaluation of the number of groups contributing to the total MIR and NIR absorption, (b) determination of the group spectra for homologous series of n-alkanes and 1-chloroalkanes, (c) performing spectral prediction based on the group spectra. Besides, we examined the effect of substitution by Cl on the chain properties. In contrast to previous studies [18], we used a different computational approach and modified spectral basis to obtain the group spectra. Besides, our study covers both MIR and NIR region up to the third overtones.
2. Experimental and Computational Methods 2.1 Materials and spectroscopic measurements. All samples of eight n-alkanes (from n-hexane to n-dodecane and n-tetradecane) and seven 1-chloroalkanes (from 1-chloropropane to 1-chlorooctane and 1-chlorododekane) were distilled and dried under freshly activated molecular sieves (4A). The water content was monitored by inspection of NIR spectra in the 5000-5300 cm-1 region were absorb the ν2+ν3 band of water. The spectra were recorded on Nicolet Magna 860 spectrometer with DTGS or InGa detectors. NIR and Raman spectra were measured at resolution of 4 cm-1 (512 scans), while ATR-IR spectra were collected at resolution of 2 cm-1 (256 scans). NIR spectra were recorded in a variable-temperature quartz cells (Hellma) of 1 mm (4500-6200 cm-1), 5 mm (6200-9000 cm-1) and 50 mm (9000-12500 cm-1) thickness at 30 oC. ATR-IR spectra were measured with multi-reflection ZnSe crystal (PIKE). 2.2 Computational procedures.
5
At first, the baseline fluctuations were minimized by an offset at 3200 cm-1 (MIR, Raman) or 6200/7500 cm-1 (NIR). Next, to eliminate the density variations from sample to sample and the other effects (like different scattering in Raman spectra), all spectra were normalized. The normalization was achieved by division of the given spectrum by its total integrated intensity. This way, the normalized spectra include mostly the information on the relative changes at particular wavenumbers due to the chain length variation. In addition, ATR-IR spectra were subjected to advanced ATR correction (OMNIC 5.12). The sample refractive indexes were taken from www.sigmaaldrich.com/catalog. The generalized 2D correlation spectra were calculated according to Noda's algorithm [19, 20], using MATLAB 7.0.4 (The Math Works Inc.) based software [21] and an average spectrum was used as a reference. To facilitate interpretation of the asynchronous contour plots, the asynchronous intensity was multiplied by corresponding synchronous intensity. However, this procedure may lead to sudden change in sign of the asynchronous peaks. The rules for interpretation of 2D correlation spectra were presented elsewhere [19, 21]. The first step in chemometric analysis is an estimation of the number of species present in the data set. These values were primarily obtained from principal component analysis (PCA) [22], and then verified by evolving factor analysis (EFA) [23]. Initial estimates of the concentration profiles were obtained from EFA, or from selective regions of the spectra (where the absorption originates mostly from a single component). The actual concentration profiles and the pure component spectra were resolved by multivariate curve resolution-alternating least squares (MCR-ALS) method with constraints (non-negativity on concentrations and spectra) [24]. The chemometric calculations were realized by PLSToolbox 6.2 (Eigenvector Research Inc.) for use with MATLAB.
3. Results and Discussion
6
3.1 ATR-IR, Raman and NIR spectra Figs. 1 and 2 show ATR-IR and NIR spectra of n-alkanes and 1-chloroalkanes, respectively, while the corresponding Raman spectra are displayed in Fig. 3. In ATR-IR spectra of n-alkanes (Fig. 1a) appear four peaks assigned to the symmetric stretching vibrations of CH2 at 2860 cm-1 and CH3 at 2873 cm-1, and the corresponding asymmetric vibrations of CH2 at 2922 cm-1 and CH3 at 2958 cm-1. The bands due to the methyl groups appear at the same positions, while the methylene bands are more or less shifted, confirming previous observations [6, 7]. This shift suggests that the CH2 band consists of at least two components located at slightly different positions, however in the raw spectra these components are not resolved. The spectrum of 1-chloropropane (Fig. 2, dashed line) is significantly distinct from the remaining spectra in the data set. Therefore, this spectrum was not used in the further analysis. On the other hand, this difference reveals sensitivity of the chain structure and properties on the chain length. This effect is more evident for short chains, confirming the observation by Rest et al. [25]. The authors proved that significant differences are observed in NIR spectra of methane and ethane, while the differences for the longer chain alkanes are smaller. In 1-chloropropane atom of Cl affects all parts of molecule, including the terminal CH3 group. As expected, an increase in the chain length decreases this effect. Comparing the band shift, it is clear that the methyl group is less affected by Cl as compared with the methylene groups. In going from 1-chlorobutane to 1-chlorooctane the position of the CH3 band is red-shifted by 6 cm-1, whereas the analogous shift for the CH2 band is around 12 cm-1 (Fig. 2a). An inspection of Fig. 3 confirms the conclusion by Ricard-Lespade et al. that the Raman spectra are complicated by possibility of Fermi resonances [3]. In contrast, the complex appearance of the first overtone spectra (Figs. 1b, 2b) results from the coexistence of the normal and local vibrations. The second and third overtones (Figs. 1c-d, 2c-d) are relatively simple due to the local character of vibrations [9-13].
7
Figure 1. Normalized ATR-IR spectra in the range of 2780-3020 cm-1 (a) and NIR spectra in the range of 5300-6100 cm-1 (b), 6500-9000 cm-1 (c), 9000-11700 cm-1 (d) of liquid n-alkanes at 30 oC.
8
Figure 2. Normalized ATR-IR spectra in the range of 2780-3020 cm-1 (a) and NIR spectra in the range of 5300-6100 cm-1 (b), 6500-9000 cm-1 (c), 9000-11700 cm-1 (d) of liquid 1chloroalkanes at 30 oC. The spectrum of 1-chloropropane was drawn by dashed line.
9
Figure 3. Normalized Raman spectra of liquid n-alkanes (a) and 1-chloroalkanes (b). The spectrum of 1-chloropropane was drawn by dashed line.
3.2 2D correlation analysis. A simple inspection of the raw spectra does not provide detailed information, hence, we used more advanced methods of data analysis. Fig. 4 displays 2D asynchronous contour plots constructed from the normalized ATR-IR spectra of n-alkanes and 1-chloroalkanes. It is of particular note that the methylene peaks are resolved into two components, whereas the methyl peaks mostly do not show this splitting. The asynchronous peaks of the symmetric and asymmetric vibrations of CH2 for n-alkanes (Fig. 4a) are located at 2851/2857 and 2920/2925 cm-1, respectively. The corresponding peaks for 1-chloroalkanes (Fig. 4b) appear at 2853/2862 and 2922/2933 cm-1. In accordance with previous results [3], the high frequency component peaks were assigned to the terminal CH2, whereas the low frequency components
10
Figure 4. Asynchronous 2D correlation contour plots constructed from ATR-IR spectra of nalkanes (a) and 1-chloroalkanes (b). In red and blue are drawn the positive and negative peaks, respectively.
11
Figure 5. Asynchronous 2D correlation contour plots constructed from Raman spectra of nalkanes (a) and chloroalkanes (b). In red and blue are drawn the positive and negative peaks, respectively.
12
Figure 6. Asynchronous 2D correlation contour plot constructed from NIR spectra of nalkanes in the range of 5600-6000 cm-1. In red and blue are drawn the positive and negative peaks, respectively.
were assigned to the midchain CH2. The separation between the resolved peaks is larger for 1chloroalkanes, suggesting that Cl has more significant effect on the adjacent methylene group as compared with the methyl group. In the asynchronous spectrum of 1-chloroalkanes (Fig. 4b) is observed the splitting of the asymmetric stretching band of CH3 at 2955/2966 cm-1. The reason of this splitting is not clear as yet. The asynchronous contour plot constructed from the Raman spectra (Fig. 5) develops the splitting only for the symmetric stretching band of CH2. Probably it results from weak intensity of the asymmetric stretching band of CH2 in the Raman spectra. The absorption from the terminal and midchain methylene units in n-alkanes is also resolved in the first overtone region (Fig. 6) at 5771/5790 cm-1. The value of this splitting (19 cm-1) is close to that estimated by Parker et al. (22 cm-1) [18].
3.3 Chemometrics analysis. 13
The main aim of chemometric analysis was evaluation of the group spectra from the basis spectra of n-alkanes and 1-chloroalkanes. These group spectra represent basic structural units present in the data sets. In contrast to work by Parker et al. [18] we used a different computational procedure (MCR-ALS) and a modified spectral basis. Our basis set included the spectra of n-alkanes from n-hexane to n-dodecane and the spectrum of n-tetradecane, whereas Parker et al. used the spectra from n-hexane to n-decane and the spectrum of nhexadecane. Besides, previous studies were limited to the first overtones and the first order combination bands in the 3900-6000 cm-1 region, while our study covers MIR region from 1000 to 4000 cm-1 and NIR region from 5000 to 12000 cm-1. In Figs. 7 and 8 are displayed the results obtained from MCR-ALS for n-alkanes and 1-chloroalkanes, respectively. The results obtained from Raman spectra (not shown) are not as clear as those from ATR-IR spectra. It results from different sensitivity of the Raman and MIR intensities to vibrations of particular molecular fragments. In addition, the Raman spectra of alkanes are complicated by Fermi resonances [3]. As a result, the relative peak intensities in the Raman and MIR spectra are very different and difficult to compare. Assignments of MIR and NIR peaks in the group spectra of n-alkanes and 1-chloroalkanes are collected in Table 1. From 2D correlation analysis it results the presence of three independent molecular fragments contributing to the total absorbance: methyl, midchain methylene and terminal methylene groups. However, the reasonable results from MCR-ALS were obtained by using two components only. As it can be seen from Fig. 7a, the spectrum of the methyl group is more complex as compared with the spectrum of the methylene group. The spectral profile attributed to CH3 has two additional peaks at 2861 and 2927 cm-1. These peaks appear close to the CH2 peaks at 2852 and 2921 cm-1 and definitely do not originate from vibrations of the methyl group. Hence, one can conclude that the spectral profile of methyl group includes also contribution from the terminal methylene group, as was suggested by Parker et al. [18].
14
Figure 7. Spectral profiles obtained from MCR-ALS using the spectra of n-alkanes in the MIR (a) and NIR (b-d) regions. Red - midchain CH2, blue - CH3 + terminal CH2 (C2H5). 15
Figure 8. Spectral profiles obtained from MCR-ALS using the spectra of 1-chloroalkanes in the MIR (a) and NIR (b-d) regions. Red - midchain CH2, blue - CH3 + terminal CH2 (C2H5). 16
Table 1. Assignments of MIR (ν) and NIR (2ν, 3ν, 4ν) peaks in the groups spectra of n-alkanes (Fig. 7) and 1-chloroalkanes (Fig. 8) resolved by MCR-ALS. Peak positions are in [cm-1]. ___________________________________________________________________________ n-alkanes
1-chloroalkanes
___________________________________________________________________________ ν(CH2)m
2852 2921
2855 2924 2955
ν(CH2)t
2861 2927
2935
ν(CH3)
2874 2957
2875 2960
----------------------------------------------------------------------------------------------------------------2ν(CH2)m
5672 5785
5680 5789
2ν(CH2)t
5684 5812
5811
2ν(CH3)
5727 5874 5910
5727 5876 5918 5965
----------------------------------------------------------------------------------------------------------------3ν(CH2)
8239
8263
3ν(CH3)
8396
8393 8518
----------------------------------------------------------------------------------------------------------------4ν(CH2)
10736
10775
4ν(CH3)
10960
11028
___________________________________________________________________________ m
t
- midchain CH2
- terminal CH2 (resolved together with CH3)
17
The intensity changes for the methyl group and adjacent methylene group follow similar pattern; the relative intensities of both groups decrease on increase in the chain length. As a result, both groups contribute to the same spectral component that one can assign to the ethyl group. In contrast, the intensity of the midchain CH2 increases as the chain length increases. Hence, MCR-ALS provides two spectral components due to the midchain methylene and the ethyl group. As it can be seen from Fig. 8a, the group spectrum of CH2 for 1-chloroalkanes has an additional peak at 2955 cm-1 located in close vicinity of the CH3 peak at 2960 cm-1. Both peaks follow different pattern of intensity changes and therefore they are resolved in 2D correlation spectrum (Fig. 4b). The presence of the second peak near 2960 cm-1 is not expected since 1-chloroalknes possess only a single methyl group. In addition, the spectral changes for this peak correlate with those for the midchain CH2. At present, the origin of this additional peak is not obvious and requires additional studies. It is of particular note that the first overtone spectra of the methyl and methylene groups for n-alkanes (Fig. 7b) are identical with the group spectra presented by Parker et al. (Figs. 2A and 2B in ref. 15). This similarity proofs that the idea of the group spectra for nalkanes works well regardless of used basis data set and the computational procedure. In addition, our work extends this idea beyond the first overtone region. The group spectra are particularly helpful during the band assignments both in the MIR and NIR region.
3.4 Spectral predictions The basis set for spectral prediction consisted of the spectra of n-alkanes from nhexane to n-dodecane, while the predicted was the spectrum of n-tetradecane. For prediction we used the spectral and concentration profiles obtained from MCR-ALS. Since the spectral profiles are identical to the group spectra, this means that the prediction was based on the
18
group spectra. The values of the concentration profiles for the predicted samples were approximated by the polynomial of the second or third order. The predicted spectrum was a linear combination of the group spectra (Ai):
Apred = ∑ Ai ⋅ ci
(1)
i
where ci are the values of the concentration profiles calculated for the predicted sample. The quality of the prediction was estimated by the determination coefficient (R2), where R is the correlation between the original and the predicted spectrum. Other measure of the quality of prediction was the sum of relative differences between the original (Aorg) and the predicted (Apred) spectrum defined as follows:
D=
∑A −A ∑A org
pred
* 100%
(2)
org
Fig. 9 displays the original and predicted spectra together with the difference between these two spectra. As it can be seen, the best reproduction of the spectrum of n-tetradecane was obtained in the first and second overtone regions. This observation rationalizes very good performance of the chemometric analysis in the NIR region. An application of similar procedure for prediction of the spectrum of 1-chlorododecane from the basis spectra from 1chloropropane to 1-chlorooctane does not provide satisfactory results. Probably, it results from the fact that the molecule of 1-chlorododecane is appreciable larger as compared with the molecules included in the basis set. The differences between the successive spectra in the series of homologous compounds become more linear on increase in the chain length. On the other hand, for short chain compounds these differences significantly deviate from the linearity. One can expect that the presence of the functional groups in the chain increases the tendency for nonlinear spectral changes in the series of homologous compounds. Therefore, to improve the ability of spectral prediction, the basis set should be extended to include information about all specific structural units present in the samples.
19
Figure 9. Prediction of MIR (a) and NIR (b-d) spectra of n-tetradecane from the basis spectra
of n-alkanes from n-hexane to n-dodecane. Red - original spectrum (Aorg), blue - predicted spectrum (Apred), green - difference spectrum (Aorg - Apred). 20
4. Conclusions
The structure of the methylene band was clearly resolved into two components in 2D asynchronous contour plots constructed from the ATR-IR, Raman and NIR spectra of nalkanes and 1-chloroalkanes. These two components were assigned to the terminal and midchain methylene groups. 2D correlation analysis evidences the presence of three independent molecular fragments contributing to the total absorbance: the methyl group and two kinds of the methylene groups. For the first time, the methylene peaks were separated in the spectra of liquid n-alkanes and 1-chloroalkanes. The intensity changes for the terminal CH2 and the CH3 groups follow similar pattern, therefore both groups contribute to the same spectral profile. As a result, MCR-ALS provides two spectral components assigned to the midchain methylene and the ethyl group. The similarity between the spectral profiles obtained from MCR-ALS and the group spectra shown in ref. 18 evidences validity of the idea of the group spectra. Besides, this work extends this idea into different spectral ranges. The spectral prediction was successfully preformed for n-alkanes, particularly in the first and second overtone regions. In contrast, the prediction for 1-chloroalkanes did not provide satisfactory results, since the spectrum of the predicted compound was too different from the basis spectra. One has to state that our spectral prediction was based on two components only. To increase the predicting power one should increase the number of spectral components to represent the information about all specific structural units present in the system.
21
References
[1] D. M. Small, Aliphatic hydrocarbons. In D. M. Small, The Physical Chemistry of Lipids from Alkanes to Phospholipids, Handbook of Lipid Research Series, Vol. 4. ed. D. Hanahan, Plenum Press, New York 1984. Chapter 7. p. 183-232. [2] D. M. Small, J. Lipid Res. 25 (1984) 1490-1500. [3] L. Ricard-Lespade, G. Longhi, S. Abbate, Chem. Phys. 142 (1990), 245-259. [4] I. R. Hill, I. W. Levin, J. Chem. Phys. 70 (1979) 842-851. [5] R. A. MacPhail, R. G. Snyder, H. L. Strauss, C. A. Elliger, J.Phys. Chem. 88 (1984) 334341. [6] R. G. Snyder, H. L. Strauss, C. A. Elliger, J. Phys. Chem. 86 (1982) 5145-5150. [7] J. Šebek, R. Knaanie, B. Albee, E. O. Potma, R. B. Gerber, J. Phys. Chem. A 117 (2013) 7442-7452. [8] H. L. Fang, D. M. Meister, R. L. Swofford, J. Phys. Chem. 88 (1984), 410-416. [9] W. R. A. Greenlay, B. R. Henry, J. Chem Phys. 69 (1978) 82-91. [10] S. Burberry, J. A. Morrell, A. C. Albrecht, R. L. Swofford, J. Chem. Phys. 70 (1979) 5522-5526. [11] G. Kjaergaard, H. Yu, B. J. Schattka, B. R. Henry, A. W. Tarr, J. Chem Phys. 93 (1990) 6239-6248. [12] G. Kjaergaard, B. R. Henry, A. W. Tarr, J. Chem Phys. 94 (1991) 5844-5854. [13] H. G. Kjaergaard, B. R. Henry, J. Chem. Phys. 96 (1992) 4841-4851. [14] C. Sandorfy, R. Buchet, G. Lachenal, In Y. Ozaki, W. F. McClure, A. A. Christy, NearInfrared in Food Science and Technology, John Wiley & Sons, New Jersey 2006, 11-46. [15] R. Iwamoto, Appl. Spectrosc. 63 (2009) 354-362. [16] R. N. Clark, J. M. Curchin, T. M. Hoefen, G. A. Swayze, J. Geophys. Res. 114 (2009) 119.
22
[17] M. Klavarioti, K. Kostarelos, A. Pourjabbar, M. Ghandehari, Environ. Sci. Pollut. Res. 21 (2014) 5849-5860. [18] M. E. Parker, D. Steele, M. J. C. Smith, J. Phys. Chem. A, 101 (1997) 9618-9631. [19] I. Noda, Appl. Spectrosc. 47 (1993) 1329-1336. [20] I. Noda, Appl. Spectrosc. 54 (2000) 994-999. [21] M.A. Czarnecki, Appl. Spectrosc. Rev. 46 (2011) 67-103. [22] S. Wold, K. Esbensen, P. Geladi, Chemom. Intell. Lab. Syst. 2 (1987) 37-52. [23] H. R. Keller, D. L. Massart, Chemom. Intell. Lab. Syst. 12 (1992) 209-224. [24] R. Tauler, B. Kowalski, S. Fleming, Anal. Chem. 65 (1993) 2040-2047. [25] A. J. Rest, R. Warren, S. C. Murray, Appl. Spectrosc. 50 (1996) 517-520.
23
Highlights:
- We measured MIR, NIR and Raman spectra of 8 n-alkanes and 7 1-chloroalkanes in the liquid phase.
- The contributions from the terminal and midchain methylene units were separated in asynchronous 2D correlation spectra.
- From MCR-ALS were obtained the group spectra that represent the main structural units.
- The group spectra were used for the spectral prediction.
24
Graphical abstract
25
S1386-1425(15)00208-5 http://dx.doi.org/10.1016/j.saa.2015.01.134 SAA 13346
To appear in:
Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy
Received Date: Revised Date: Accepted Date:
9 November 2014 19 January 2015 29 January 2015
Please cite this article as: M. Kwaśniewicz, M.A. Czarnecki, MIR and NIR group spectra of n-alkanes and 1chloroalkanes, Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy (2015), doi: http:// dx.doi.org/10.1016/j.saa.2015.01.134
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
MIR and NIR group spectra of n-alkanes and 1-chloroalkanes
Michał Kwaśniewicz, Mirosław A. Czarnecki* Faculty of Chemistry, University of Wrocław, F. Joliot-Curie 14, 50-383 Wrocław, POLAND
Address for correspondence: Prof. Mirosław Czarnecki Faculty of Chemistry, University of Wrocław F. Joliot-Curie 14, 50-383 Wrocław, POLAND E-mail address: [email protected]
Tel. 48-71-3757238 Fax 48-71-3282348
Abstract Numerous attempts were undertaken to resolve the absorption originating from different parts of alkanes. The separation of the contributions from the terminal and midchain methylene units was observed only in the spectra of solid alkanes at low temperatures. On the other hand, for liquid alkanes this effect was not reported as yet. In this study, ATR-IR, Raman and NIR spectra of eight n-alkanes and seven 1-chloroalkanes in the liquid phase were measured from 1000 to 12000 cm-1. The spectra were analyzed by using two-dimensional (2D) correlation approach and chemometrics methods. It was shown that in 2D asynchronous contour plots, constructed from the spectra of n-alkanes and 1-chloroalkanes, the methylene band was resolved into two components. These two components were assigned to the terminal and midchain methylene groups. For the first time, the contributions from these two molecular fragments were resolved in the spectra of liquid n-alkanes and 1-chloroalkanes. MCR-ALS resolved these spectra into two components that were assigned to the ethyl and midchain methylene groups. These components represent the group spectra that can be used for assignment, spectral analysis and prediction of unknown spectra. The spectral prediction based on the group spectra provides very good results for n-alkanes, especially in the first and second overtone regions.
Keywords: n-alkanes, 1-chloroalkanes, group MIR/NIR spectra, midchain methylene groups, spectral prediction, two-dimensional correlation analysis, chemometrics.
2
1. Introduction Conformation of hydrocarbon chains determines the structure and properties of biological systems and polymers. In solid state the chains usually take all-trans conformation, while in the liquid state there exists a conformational disorder [1,2]. Many efforts were undertaken to resolve the absorption from different parts of hydrocarbons. Particularly, separation of the contribution from the terminal and midchain methylene units is not easy task. Ricard-Lespade et al. reported that the MIR spectrum of polymethylene chains in solid state is almost insensitive to the flexibility of the chain [3]. However, Raman spectra were complicated by Fermi resonances between the CH2 symmetric stretchings and the overtones and combinations of CH2 bendings. The authors did not find any differences in peak positions for the midchain CH2 and those adjacent to the CH3. On the other hand, Hill and Levin observed in the Raman and MIR spectra of n-hexadecane at liquid nitrogen temperatures two bands at 2855 and 2846 cm-1 that were assigned to the symmetric CH2 stretching of the terminal and midchain units, respectively [4]. MacPhail et al. also resolved these two bands at 2853 and 2846 cm-1 in the MIR spectrum of crystalline n-CD3-(CH2)20-CD3 at low temperatures (<10 K) [5]. Snyder et al. evidenced that the methylene band of the crystalline polyethylene (at 2922 cm-1) was affected by the chain conformation, whereas the methyl band (at 2938 cm-1) was influenced by the solvent [6]. Another conclusion was obtained by Šebek et al. from the studies of the CH stretching band in selected organic molecules in the liquid and gas phase [7]. It was shown that the CH3 symmetric and asymmetric frequencies are almost independent on the surroundings and the corresponding bands always appear at the same position. In contrast, the location of both symmetric and asymmetric CH2 stretching bands strongly depends on the neighboring groups. Fang et al. evidenced the conformational sensitivity of the overtones of methyl C-H stretching vibrations resulting from the influence of the heteroatom (N, O, S) [8]. This effect was explained in terms of donation of electron
3
density form the lone pair into the trans oriented antibonding orbital of a C-H bond. Numerous works demonstrated that the overtones region is also sensitive on the chain conformation [9-14]. The better peaks separation in the NIR spectra made possible to resolve the absorption from the axial and equatorial CH bonds in NIR spectrum of cyclohexane [13]. The first overtones have both the local and normal nature. As a result, the corresponding spectra are complex due to possibility of various resonances. In contrast, the higher overtones spectra are simple due to the local character of the vibrations [9-14]. Iwamoto suggested that NIR spectrum is a useful tool for separation of the characteristic spectral features of heptene from the spectra of solvents like heptane or toluene [15]. This separation can be performed down to concentrations of 0.1%. Reflectance spectra of alkanes from the ultraviolet to MIR show spectral similarity within the alkane family. On the other hand, the spectra retain sufficient diversity to distinguish among family members from methane to decane [16]. Klavarioti et al. demonstrated that NIR spectra can be used for discrimination between different classes of hydrocarbons (alkanes, aromatics, chlorinated alkanes) [17]. Discrimination within the class of alkanes is more difficult but still possible through the comparison of the peak ratios. Another interesting idea was proposed by Parker et al. [18]. The authors demonstrated that the spectra of liquid aliphatic hydrocarbons measured from 3900 to 6000 cm-1 can be represented as the sum of a limited number of group spectra that were obtained from principal factor analysis (PFA). Using PFA to the set of the spectra of related structure (e.g. homologous series) one can determine the factors necessary to reconstruct the original spectra with a high precision. For n-alkanes spectra from n-hexane to n-hexadecane the authors considered only two structural units: CH3 and CH2. However, it was later shown that this was an oversimplified picture. The authors concluded that in the absence of deuterated species at either the terminal CH2 group, it is not possible to separate the contribution from the terminal and midchain CH2.
4
This study evidences that the separation of the vibrational absorption from the terminal and midchain CH2 is possible also for non-deuterated samples in the liquid phase. In this purpose we recorded ATR-IR, Raman and NIR spectra of eight n-alkanes and seven 1chloroalkanes in the liquid phase. The spectra were analyzed by using 2D correlation approach and chemometrics methods. The main aim of present study was (a) evaluation of the number of groups contributing to the total MIR and NIR absorption, (b) determination of the group spectra for homologous series of n-alkanes and 1-chloroalkanes, (c) performing spectral prediction based on the group spectra. Besides, we examined the effect of substitution by Cl on the chain properties. In contrast to previous studies [18], we used a different computational approach and modified spectral basis to obtain the group spectra. Besides, our study covers both MIR and NIR region up to the third overtones.
2. Experimental and Computational Methods 2.1 Materials and spectroscopic measurements. All samples of eight n-alkanes (from n-hexane to n-dodecane and n-tetradecane) and seven 1-chloroalkanes (from 1-chloropropane to 1-chlorooctane and 1-chlorododekane) were distilled and dried under freshly activated molecular sieves (4A). The water content was monitored by inspection of NIR spectra in the 5000-5300 cm-1 region were absorb the ν2+ν3 band of water. The spectra were recorded on Nicolet Magna 860 spectrometer with DTGS or InGa detectors. NIR and Raman spectra were measured at resolution of 4 cm-1 (512 scans), while ATR-IR spectra were collected at resolution of 2 cm-1 (256 scans). NIR spectra were recorded in a variable-temperature quartz cells (Hellma) of 1 mm (4500-6200 cm-1), 5 mm (6200-9000 cm-1) and 50 mm (9000-12500 cm-1) thickness at 30 oC. ATR-IR spectra were measured with multi-reflection ZnSe crystal (PIKE). 2.2 Computational procedures.
5
At first, the baseline fluctuations were minimized by an offset at 3200 cm-1 (MIR, Raman) or 6200/7500 cm-1 (NIR). Next, to eliminate the density variations from sample to sample and the other effects (like different scattering in Raman spectra), all spectra were normalized. The normalization was achieved by division of the given spectrum by its total integrated intensity. This way, the normalized spectra include mostly the information on the relative changes at particular wavenumbers due to the chain length variation. In addition, ATR-IR spectra were subjected to advanced ATR correction (OMNIC 5.12). The sample refractive indexes were taken from www.sigmaaldrich.com/catalog. The generalized 2D correlation spectra were calculated according to Noda's algorithm [19, 20], using MATLAB 7.0.4 (The Math Works Inc.) based software [21] and an average spectrum was used as a reference. To facilitate interpretation of the asynchronous contour plots, the asynchronous intensity was multiplied by corresponding synchronous intensity. However, this procedure may lead to sudden change in sign of the asynchronous peaks. The rules for interpretation of 2D correlation spectra were presented elsewhere [19, 21]. The first step in chemometric analysis is an estimation of the number of species present in the data set. These values were primarily obtained from principal component analysis (PCA) [22], and then verified by evolving factor analysis (EFA) [23]. Initial estimates of the concentration profiles were obtained from EFA, or from selective regions of the spectra (where the absorption originates mostly from a single component). The actual concentration profiles and the pure component spectra were resolved by multivariate curve resolution-alternating least squares (MCR-ALS) method with constraints (non-negativity on concentrations and spectra) [24]. The chemometric calculations were realized by PLSToolbox 6.2 (Eigenvector Research Inc.) for use with MATLAB.
3. Results and Discussion
6
3.1 ATR-IR, Raman and NIR spectra Figs. 1 and 2 show ATR-IR and NIR spectra of n-alkanes and 1-chloroalkanes, respectively, while the corresponding Raman spectra are displayed in Fig. 3. In ATR-IR spectra of n-alkanes (Fig. 1a) appear four peaks assigned to the symmetric stretching vibrations of CH2 at 2860 cm-1 and CH3 at 2873 cm-1, and the corresponding asymmetric vibrations of CH2 at 2922 cm-1 and CH3 at 2958 cm-1. The bands due to the methyl groups appear at the same positions, while the methylene bands are more or less shifted, confirming previous observations [6, 7]. This shift suggests that the CH2 band consists of at least two components located at slightly different positions, however in the raw spectra these components are not resolved. The spectrum of 1-chloropropane (Fig. 2, dashed line) is significantly distinct from the remaining spectra in the data set. Therefore, this spectrum was not used in the further analysis. On the other hand, this difference reveals sensitivity of the chain structure and properties on the chain length. This effect is more evident for short chains, confirming the observation by Rest et al. [25]. The authors proved that significant differences are observed in NIR spectra of methane and ethane, while the differences for the longer chain alkanes are smaller. In 1-chloropropane atom of Cl affects all parts of molecule, including the terminal CH3 group. As expected, an increase in the chain length decreases this effect. Comparing the band shift, it is clear that the methyl group is less affected by Cl as compared with the methylene groups. In going from 1-chlorobutane to 1-chlorooctane the position of the CH3 band is red-shifted by 6 cm-1, whereas the analogous shift for the CH2 band is around 12 cm-1 (Fig. 2a). An inspection of Fig. 3 confirms the conclusion by Ricard-Lespade et al. that the Raman spectra are complicated by possibility of Fermi resonances [3]. In contrast, the complex appearance of the first overtone spectra (Figs. 1b, 2b) results from the coexistence of the normal and local vibrations. The second and third overtones (Figs. 1c-d, 2c-d) are relatively simple due to the local character of vibrations [9-13].
7
Figure 1. Normalized ATR-IR spectra in the range of 2780-3020 cm-1 (a) and NIR spectra in the range of 5300-6100 cm-1 (b), 6500-9000 cm-1 (c), 9000-11700 cm-1 (d) of liquid n-alkanes at 30 oC.
8
Figure 2. Normalized ATR-IR spectra in the range of 2780-3020 cm-1 (a) and NIR spectra in the range of 5300-6100 cm-1 (b), 6500-9000 cm-1 (c), 9000-11700 cm-1 (d) of liquid 1chloroalkanes at 30 oC. The spectrum of 1-chloropropane was drawn by dashed line.
9
Figure 3. Normalized Raman spectra of liquid n-alkanes (a) and 1-chloroalkanes (b). The spectrum of 1-chloropropane was drawn by dashed line.
3.2 2D correlation analysis. A simple inspection of the raw spectra does not provide detailed information, hence, we used more advanced methods of data analysis. Fig. 4 displays 2D asynchronous contour plots constructed from the normalized ATR-IR spectra of n-alkanes and 1-chloroalkanes. It is of particular note that the methylene peaks are resolved into two components, whereas the methyl peaks mostly do not show this splitting. The asynchronous peaks of the symmetric and asymmetric vibrations of CH2 for n-alkanes (Fig. 4a) are located at 2851/2857 and 2920/2925 cm-1, respectively. The corresponding peaks for 1-chloroalkanes (Fig. 4b) appear at 2853/2862 and 2922/2933 cm-1. In accordance with previous results [3], the high frequency component peaks were assigned to the terminal CH2, whereas the low frequency components
10
Figure 4. Asynchronous 2D correlation contour plots constructed from ATR-IR spectra of nalkanes (a) and 1-chloroalkanes (b). In red and blue are drawn the positive and negative peaks, respectively.
11
Figure 5. Asynchronous 2D correlation contour plots constructed from Raman spectra of nalkanes (a) and chloroalkanes (b). In red and blue are drawn the positive and negative peaks, respectively.
12
Figure 6. Asynchronous 2D correlation contour plot constructed from NIR spectra of nalkanes in the range of 5600-6000 cm-1. In red and blue are drawn the positive and negative peaks, respectively.
were assigned to the midchain CH2. The separation between the resolved peaks is larger for 1chloroalkanes, suggesting that Cl has more significant effect on the adjacent methylene group as compared with the methyl group. In the asynchronous spectrum of 1-chloroalkanes (Fig. 4b) is observed the splitting of the asymmetric stretching band of CH3 at 2955/2966 cm-1. The reason of this splitting is not clear as yet. The asynchronous contour plot constructed from the Raman spectra (Fig. 5) develops the splitting only for the symmetric stretching band of CH2. Probably it results from weak intensity of the asymmetric stretching band of CH2 in the Raman spectra. The absorption from the terminal and midchain methylene units in n-alkanes is also resolved in the first overtone region (Fig. 6) at 5771/5790 cm-1. The value of this splitting (19 cm-1) is close to that estimated by Parker et al. (22 cm-1) [18].
3.3 Chemometrics analysis. 13
The main aim of chemometric analysis was evaluation of the group spectra from the basis spectra of n-alkanes and 1-chloroalkanes. These group spectra represent basic structural units present in the data sets. In contrast to work by Parker et al. [18] we used a different computational procedure (MCR-ALS) and a modified spectral basis. Our basis set included the spectra of n-alkanes from n-hexane to n-dodecane and the spectrum of n-tetradecane, whereas Parker et al. used the spectra from n-hexane to n-decane and the spectrum of nhexadecane. Besides, previous studies were limited to the first overtones and the first order combination bands in the 3900-6000 cm-1 region, while our study covers MIR region from 1000 to 4000 cm-1 and NIR region from 5000 to 12000 cm-1. In Figs. 7 and 8 are displayed the results obtained from MCR-ALS for n-alkanes and 1-chloroalkanes, respectively. The results obtained from Raman spectra (not shown) are not as clear as those from ATR-IR spectra. It results from different sensitivity of the Raman and MIR intensities to vibrations of particular molecular fragments. In addition, the Raman spectra of alkanes are complicated by Fermi resonances [3]. As a result, the relative peak intensities in the Raman and MIR spectra are very different and difficult to compare. Assignments of MIR and NIR peaks in the group spectra of n-alkanes and 1-chloroalkanes are collected in Table 1. From 2D correlation analysis it results the presence of three independent molecular fragments contributing to the total absorbance: methyl, midchain methylene and terminal methylene groups. However, the reasonable results from MCR-ALS were obtained by using two components only. As it can be seen from Fig. 7a, the spectrum of the methyl group is more complex as compared with the spectrum of the methylene group. The spectral profile attributed to CH3 has two additional peaks at 2861 and 2927 cm-1. These peaks appear close to the CH2 peaks at 2852 and 2921 cm-1 and definitely do not originate from vibrations of the methyl group. Hence, one can conclude that the spectral profile of methyl group includes also contribution from the terminal methylene group, as was suggested by Parker et al. [18].
14
Figure 7. Spectral profiles obtained from MCR-ALS using the spectra of n-alkanes in the MIR (a) and NIR (b-d) regions. Red - midchain CH2, blue - CH3 + terminal CH2 (C2H5). 15
Figure 8. Spectral profiles obtained from MCR-ALS using the spectra of 1-chloroalkanes in the MIR (a) and NIR (b-d) regions. Red - midchain CH2, blue - CH3 + terminal CH2 (C2H5). 16
Table 1. Assignments of MIR (ν) and NIR (2ν, 3ν, 4ν) peaks in the groups spectra of n-alkanes (Fig. 7) and 1-chloroalkanes (Fig. 8) resolved by MCR-ALS. Peak positions are in [cm-1]. ___________________________________________________________________________ n-alkanes
1-chloroalkanes
___________________________________________________________________________ ν(CH2)m
2852 2921
2855 2924 2955
ν(CH2)t
2861 2927
2935
ν(CH3)
2874 2957
2875 2960
----------------------------------------------------------------------------------------------------------------2ν(CH2)m
5672 5785
5680 5789
2ν(CH2)t
5684 5812
5811
2ν(CH3)
5727 5874 5910
5727 5876 5918 5965
----------------------------------------------------------------------------------------------------------------3ν(CH2)
8239
8263
3ν(CH3)
8396
8393 8518
----------------------------------------------------------------------------------------------------------------4ν(CH2)
10736
10775
4ν(CH3)
10960
11028
___________________________________________________________________________ m
t
- midchain CH2
- terminal CH2 (resolved together with CH3)
17
The intensity changes for the methyl group and adjacent methylene group follow similar pattern; the relative intensities of both groups decrease on increase in the chain length. As a result, both groups contribute to the same spectral component that one can assign to the ethyl group. In contrast, the intensity of the midchain CH2 increases as the chain length increases. Hence, MCR-ALS provides two spectral components due to the midchain methylene and the ethyl group. As it can be seen from Fig. 8a, the group spectrum of CH2 for 1-chloroalkanes has an additional peak at 2955 cm-1 located in close vicinity of the CH3 peak at 2960 cm-1. Both peaks follow different pattern of intensity changes and therefore they are resolved in 2D correlation spectrum (Fig. 4b). The presence of the second peak near 2960 cm-1 is not expected since 1-chloroalknes possess only a single methyl group. In addition, the spectral changes for this peak correlate with those for the midchain CH2. At present, the origin of this additional peak is not obvious and requires additional studies. It is of particular note that the first overtone spectra of the methyl and methylene groups for n-alkanes (Fig. 7b) are identical with the group spectra presented by Parker et al. (Figs. 2A and 2B in ref. 15). This similarity proofs that the idea of the group spectra for nalkanes works well regardless of used basis data set and the computational procedure. In addition, our work extends this idea beyond the first overtone region. The group spectra are particularly helpful during the band assignments both in the MIR and NIR region.
3.4 Spectral predictions The basis set for spectral prediction consisted of the spectra of n-alkanes from nhexane to n-dodecane, while the predicted was the spectrum of n-tetradecane. For prediction we used the spectral and concentration profiles obtained from MCR-ALS. Since the spectral profiles are identical to the group spectra, this means that the prediction was based on the
18
group spectra. The values of the concentration profiles for the predicted samples were approximated by the polynomial of the second or third order. The predicted spectrum was a linear combination of the group spectra (Ai):
Apred = ∑ Ai ⋅ ci
(1)
i
where ci are the values of the concentration profiles calculated for the predicted sample. The quality of the prediction was estimated by the determination coefficient (R2), where R is the correlation between the original and the predicted spectrum. Other measure of the quality of prediction was the sum of relative differences between the original (Aorg) and the predicted (Apred) spectrum defined as follows:
D=
∑A −A ∑A org
pred
* 100%
(2)
org
Fig. 9 displays the original and predicted spectra together with the difference between these two spectra. As it can be seen, the best reproduction of the spectrum of n-tetradecane was obtained in the first and second overtone regions. This observation rationalizes very good performance of the chemometric analysis in the NIR region. An application of similar procedure for prediction of the spectrum of 1-chlorododecane from the basis spectra from 1chloropropane to 1-chlorooctane does not provide satisfactory results. Probably, it results from the fact that the molecule of 1-chlorododecane is appreciable larger as compared with the molecules included in the basis set. The differences between the successive spectra in the series of homologous compounds become more linear on increase in the chain length. On the other hand, for short chain compounds these differences significantly deviate from the linearity. One can expect that the presence of the functional groups in the chain increases the tendency for nonlinear spectral changes in the series of homologous compounds. Therefore, to improve the ability of spectral prediction, the basis set should be extended to include information about all specific structural units present in the samples.
19
Figure 9. Prediction of MIR (a) and NIR (b-d) spectra of n-tetradecane from the basis spectra
of n-alkanes from n-hexane to n-dodecane. Red - original spectrum (Aorg), blue - predicted spectrum (Apred), green - difference spectrum (Aorg - Apred). 20
4. Conclusions
The structure of the methylene band was clearly resolved into two components in 2D asynchronous contour plots constructed from the ATR-IR, Raman and NIR spectra of nalkanes and 1-chloroalkanes. These two components were assigned to the terminal and midchain methylene groups. 2D correlation analysis evidences the presence of three independent molecular fragments contributing to the total absorbance: the methyl group and two kinds of the methylene groups. For the first time, the methylene peaks were separated in the spectra of liquid n-alkanes and 1-chloroalkanes. The intensity changes for the terminal CH2 and the CH3 groups follow similar pattern, therefore both groups contribute to the same spectral profile. As a result, MCR-ALS provides two spectral components assigned to the midchain methylene and the ethyl group. The similarity between the spectral profiles obtained from MCR-ALS and the group spectra shown in ref. 18 evidences validity of the idea of the group spectra. Besides, this work extends this idea into different spectral ranges. The spectral prediction was successfully preformed for n-alkanes, particularly in the first and second overtone regions. In contrast, the prediction for 1-chloroalkanes did not provide satisfactory results, since the spectrum of the predicted compound was too different from the basis spectra. One has to state that our spectral prediction was based on two components only. To increase the predicting power one should increase the number of spectral components to represent the information about all specific structural units present in the system.
21
References
[1] D. M. Small, Aliphatic hydrocarbons. In D. M. Small, The Physical Chemistry of Lipids from Alkanes to Phospholipids, Handbook of Lipid Research Series, Vol. 4. ed. D. Hanahan, Plenum Press, New York 1984. Chapter 7. p. 183-232. [2] D. M. Small, J. Lipid Res. 25 (1984) 1490-1500. [3] L. Ricard-Lespade, G. Longhi, S. Abbate, Chem. Phys. 142 (1990), 245-259. [4] I. R. Hill, I. W. Levin, J. Chem. Phys. 70 (1979) 842-851. [5] R. A. MacPhail, R. G. Snyder, H. L. Strauss, C. A. Elliger, J.Phys. Chem. 88 (1984) 334341. [6] R. G. Snyder, H. L. Strauss, C. A. Elliger, J. Phys. Chem. 86 (1982) 5145-5150. [7] J. Šebek, R. Knaanie, B. Albee, E. O. Potma, R. B. Gerber, J. Phys. Chem. A 117 (2013) 7442-7452. [8] H. L. Fang, D. M. Meister, R. L. Swofford, J. Phys. Chem. 88 (1984), 410-416. [9] W. R. A. Greenlay, B. R. Henry, J. Chem Phys. 69 (1978) 82-91. [10] S. Burberry, J. A. Morrell, A. C. Albrecht, R. L. Swofford, J. Chem. Phys. 70 (1979) 5522-5526. [11] G. Kjaergaard, H. Yu, B. J. Schattka, B. R. Henry, A. W. Tarr, J. Chem Phys. 93 (1990) 6239-6248. [12] G. Kjaergaard, B. R. Henry, A. W. Tarr, J. Chem Phys. 94 (1991) 5844-5854. [13] H. G. Kjaergaard, B. R. Henry, J. Chem. Phys. 96 (1992) 4841-4851. [14] C. Sandorfy, R. Buchet, G. Lachenal, In Y. Ozaki, W. F. McClure, A. A. Christy, NearInfrared in Food Science and Technology, John Wiley & Sons, New Jersey 2006, 11-46. [15] R. Iwamoto, Appl. Spectrosc. 63 (2009) 354-362. [16] R. N. Clark, J. M. Curchin, T. M. Hoefen, G. A. Swayze, J. Geophys. Res. 114 (2009) 119.
22
[17] M. Klavarioti, K. Kostarelos, A. Pourjabbar, M. Ghandehari, Environ. Sci. Pollut. Res. 21 (2014) 5849-5860. [18] M. E. Parker, D. Steele, M. J. C. Smith, J. Phys. Chem. A, 101 (1997) 9618-9631. [19] I. Noda, Appl. Spectrosc. 47 (1993) 1329-1336. [20] I. Noda, Appl. Spectrosc. 54 (2000) 994-999. [21] M.A. Czarnecki, Appl. Spectrosc. Rev. 46 (2011) 67-103. [22] S. Wold, K. Esbensen, P. Geladi, Chemom. Intell. Lab. Syst. 2 (1987) 37-52. [23] H. R. Keller, D. L. Massart, Chemom. Intell. Lab. Syst. 12 (1992) 209-224. [24] R. Tauler, B. Kowalski, S. Fleming, Anal. Chem. 65 (1993) 2040-2047. [25] A. J. Rest, R. Warren, S. C. Murray, Appl. Spectrosc. 50 (1996) 517-520.
23
Highlights:
- We measured MIR, NIR and Raman spectra of 8 n-alkanes and 7 1-chloroalkanes in the liquid phase.
- The contributions from the terminal and midchain methylene units were separated in asynchronous 2D correlation spectra.
- From MCR-ALS were obtained the group spectra that represent the main structural units.
- The group spectra were used for the spectral prediction.
24
Graphical abstract
25