H NMR Spectra of Heavy Petroleum Fractio - American Chemical

Protons in 1H NMR Spectra of Heavy Petroleum. Fractions. G. S. Kapur*. Indian Oil Corporation Limited, R&D Centre, Faridabad-121 007, Haryana, India...
0 downloads 0 Views 125KB Size
508

Energy & Fuels 2005, 19, 508-511

Unambiguous Resolution of r-Methyl and r-Methylene Protons in 1H NMR Spectra of Heavy Petroleum Fractions G. S. Kapur* Indian Oil Corporation Limited, R&D Centre, Faridabad-121 007, Haryana, India

S. Berger Institute of Analytical Chemistry, University of Leipzig, Linnestrausse 3, D 4103, Leipzig, Germany Received July 31, 2004. Revised Manuscript Received December 8, 2004

In this paper, results of gradient selected (gs) NMR experiments for editing the 1H NMR spectra of wide range of petroleum fractions are presented. Two such experiments have been performed which edit the proton spectra, with respect to carbon multiplicity: one results in the generation of selective CHn (n ) 3, 2, 1) sub-spectra (1d-hmqc), and the other offers carbon-13 distortionless enhancement by polarization transfer (DEPT) type editing of the proton spectra (1d-hsqc). Both the experiments result in unambiguous resolution of R-methyl and R-methylene groups appearing in highly overlapped region from 2 ppm to 4.5 ppm, leading to better characterization of petroleum products. The proton editing experiments also provide the quantitative extent of overlap between various groups. The experiments are quite robust, simple to execute, and work well for the entire range of petroleum fractions.

Introduction Proton NMR spectra give a direct measurement of the distribution of protons present in different chemical shift environment in fossil fuel products. Generally, the proton spectra of such products are highly overlapped and crowded, because of the presence of a large number of components. However, average structural information can be obtained by assigning the spectra to the structural specificity of the hydrogen-type distribution associated with the chemical shift regions:1 for example, hydrogens of aromatic rings (6.5-9.0 ppm), olefinic protons (4.5-6.0 ppm), hydrogens on carbon alpha to aromatic rings (R-CH, R-CH2, and R-CH3; 2.0-4.5 ppm), hydrogens on β-CH/CH2 groups to aromatic rings, and -CH/CH2 groups of alkanes and cycloalkanes (1.0-2.0 ppm), and methyl hydrogens on γ, δ, to aromatic rings, and methyl hydrogens of alkanes and cycloalkanes (0.51.0 ppm). For more-detailed analysis, further subdivision of the chemical shift regions is desirable. Particularly, the 2.04.5 ppm region is one such region, where specific assignment to R-CH2 and R-CH3 groups is not straightforward. Because of severe overlapping, precise demarcation of the region, in terms of R-CH2 only and R-CH3 only protons, and, hence, their quantification had not been possible. Few propositions for subdividing this region are reported in the literature; most of these * Author to whom correspondence should be addressed. E-mail address: [email protected]. (1) S. W. Lee, S. W.; Glavincevski, B. Fuel Process. Technol. 1999, 60 (1), 81-86.

pertain to pure monoaromatics,2 diaromatics,3 or polyaromatic fractions,4 however, without sufficient experimental evidences. In an actual composite sample containing all types of aromatics, such a straightforward assignment is not valid, because the position of the chemical shift cut-off value between protons in the R-CH2 and R-CH3 groups changes considerably, depending on the type of aromatic and the nature of the sample.5 Two-dimensional (2D) NMR experiments such as 2D-HSQC/HMQC do help in assigning the overlapped 1H spectrum, using the high resolution and known assignment of the 13C NMR spectrum. However, determination of carbon multiplicities obtained by running a carbon-13 (distortionless enhancement by polarization transfer (DEPT) spectrum becomes essential for complete assignment, leading to increased experimentation time. Furthermore, in the case of highly overlapped regions in the 1H NMR spectra, such as those in heavier petroleum fractions, unambiguous assignment/demarcation is not possible, because of the spread and diffused nature of the correlations contours in the 2D spectra.6 This short communication describes two proton-NMR experiments, which help in the unambiguous assignment of signals, particularly those of R-CH2 and R-CH3 groups, by selective editing of 1H NMR spectra. Both the experiments give complimentary results and are easy to execute. The edited proton spectra, when (2) Abu-Dagga, F.; Ruegger, H. Fuel 1988, 67, 1255. (3) Haw, J. W.; Glass, T. E.; Dorn, H. C. Anal.Chem. 1983, 55, 22. (4) Dosseh, G.; Rousseau, B.; Fuchs, A. H. Fuel 1991, 70, 641. (5) Cerny, J.; Pavlikova, H. Fuel Sci. Technol. Int. 1994, 12 (10), 1377. (6) Kapur, G. S.; Berger, S. Fuel 2002, 81 (7), 883-892.

10.1021/ef049810c CCC: $30.25 © 2005 American Chemical Society Published on Web 01/06/2005

R-CH2/R-CH3 Protons in Heavy Petroleum Fractions

Energy & Fuels, Vol. 19, No. 2, 2005 509

compared to normal proton spectra, can also lead to quantitative information about the concentration of R-CH2 and R-CH3 groups in complex heavy petroleum fractions. Experimental Details Sample. Various samples were used in this study, including the following: commercial gasoline (95 octane), high-speed diesel (HSD), vacuum gas oil (VGO) and its aromatic fraction, heavy vacuum gas oil (HVGO), main column bottom (MCB), reduced crude oil (RCO), clarified oil (CLO), and rubber processor oil (RPO). These were obtained from different Indian refineries, excluding the gasoline and HSD samples, which were procured from a gasoline station in Germany. Proton Editing Experiments. Pulse Sequence A. Edited one-dimensional 1H NMR subspectra, according to the multiplicity (i.e., CH3, CH2, and CH only), were obtained using the pulse sequence and conditions described in our earlier work.6 The pulse sequence is referenced as 1d-hmqc (one-dimensional heteronuclear multiple quantum coherence), using a multiple quantum filter (MQF) to detect the desired level of the coherence, and results in CH3-only (heteronuclear quadruple quantum coherence, HQQC), CH2-only (heteronuclear triple quantum coherence, HTQC), and CH-only (heteronuclear double quantum coherence, HDQC) subspectra.7 Pulse Sequence B. Another experiment, called the gradient selected one-dimensional-multiplicity edited heteronuclear single quantum coherence (1d-hsqc), as described by Parella et al.,8 has been performed. This experiment edits the proton spectrum so that signals due to CH and CH3 groups appear anti-phase to those from CH2 groups, similar to that observed in a more commonly used carbon-13 DEPT experiment. Appropriate manipulation of the edited proton spectrum, with respect to normal proton spectrum (all positive signals), can also result in pure CH3 or CH2- or CH-only proton subspectra. The pulse sequence used is shown below:

The narrow and wide black bars indicate hard 90° and 180° pulses, respectively. All pulses are applied from the x-axis, unless otherwise indicated, whereas Φ1 is usually set to y. A minimum two-step phase cycle has been applied in which the first 90° carbon pulse and the receiver have been inverted in alternate scans. To obtain an edited spectrum with CH and CH3 protons that have positive intensity, whereas CH2 protons appear to have negative intensity, the delay ∆ is set to 1/4JCH, ∆′ is kept as 1/2JCH and Φ° ) 180°. (JCH denotes the coupling constant.) The delay ∆1 was kept at a value of 1/6JCH, to achieve overall improvement in sensitivity for all multiplicities. All the experiments were performed on a Bruker model AMX 400 spectrometer that was equipped with a 5-mm inverse broadband probe head incorporating a shielded Z-gradient coil. All the gradient pulses had a duration of δ ) 1-2 ms, followed (7) Parella, T.; Sanchez-Ferrando, F.; Virgili, A. J. Magn. Reson., Ser. A 1995, 117, 78. (8) Parella, T.; Sanchez-Ferrando, F.; Virgili, A. J. Magn. Reson. 1997, 126, 274.

Figure 1. Edited proton subspectra obtained using the pulse sequences; the terms “1d-hmqc” and “1d-hsqc” denote data for the samples of high-octane gasoline and high-speed diesel (HSD), respectively. by a recovery delay of 50 µs. The spectra were acquired using a relaxation delay of 5 s and 128-256 scans were accumulated. Details of other experimental conditions for 1d-hmqc spectra (pulse sequence A) have been provided in our previous work.6 To obtain 1d-hsqc spectra (pulse sequence B), the gradient strength for the two gradients G1 and G2 were in the ratio of 4:1. Details of both pulse sequences are available from the authors on request.

Results and Discussion Both the pulse sequencessviz, 1d-hmqc and 1d-hsqcs were first optimized on a standard sample of ethyl crotonate, and on a sample of high-octane gasoline and high-speed diesel (HSD), where the assignment of signals to various group types is already known. The results of these experiments are shown for the HSD sample in Figure 1a-c. The 1d-hmqc pulse sequence produces a CH3-only (heteronuclear quadruple quantum coherence, HQQC) spectrum of the diesel sample (see Figure 1b), where signals due to CH2 and CH groups have completely vanished. Similarly, the result of the CH2-only (heteronuclear triple quantum coherence, HTQC) experiment is shown in Figure 1c, where the only signals visible are due to CH2 groups and those due to other groups disappear, whereas the edited 1d-hsqc experiment results in a proton spectrum (see Figure 1d), in which signals due to CH and CH3 groups appear anti-phase to those from CH2 groups. A comparison of the edited spectra with the conventional proton spectrum (see Figure 1a) allows a clear separation of proton signals due to various group types. Similar results of the 1d-hsqc NMR experiment are shown for the gasoline sample, where signals of CH3 groups of oxygenate (MTBE), aromatics, olefins, and paraffins (normal + iso-) can be observed to be well-separated from those of methylene groups (see Figure 1e and 1f). After establishing the qualitative aspects of both experiments resulting in equivalent structural information, the quantitativeness of these experimentssin particular, 1d-hmqcswas ascertained on a standard sample of ethyl crotonate. Ethyl crotonate contains two methyl groups, one methylene group, and two methane groups, each showing a distinct chemical shift/splitting in the conventional proton spectrum. The edited CH3-

510

Energy & Fuels, Vol. 19, No. 2, 2005

Kapur and Berger

Table 1. Quantitative Estimation of Relative Intensity (I) of CH3, CH2, and CH Groups for a Standard Ethyl Crotonate Sample, Using the Edited One-Dimensional Spectra (1d-hmqc)

proton

conventional spectrum

CH3 only spectrum

CH2 only spectrum

CH only spectrum

H-5 (δ 7.0, m) H-2 (δ 5.9, d) H-3 (δ 4.2, q) H-2 (δ 2.0, d) H-1 (δ 1.2, t)

0.97 1.01 1.98 2.97 3.02

0.0 0.0 0.1 3.00 3.10

0.0 0.0 2.0 -0.07 -0.03

1.01 1.00 0.25 0.32 0.35

only, CH2-only, and CH3-only subspectra were recorded using the 1d-hmqc experiment, and the integral intensity of the signals obtained was measured. The data for the conventional spectra, versus the edited spectra, is reported in Table 1, which supports the quantitative nature of the 1d-hmqc experiments within the experimental errors. Only the CH-only proton spectrum shows an ∼10% signal intensity that is due to CH3 and CH2 groups. After optimization of the pulse sequences, the edited proton spectra of a variety of petroleum fractions that had different boiling ranges and structural complexities (see Experimental Details) have been obtained. For various fractions, the JCH value has been kept at ∼130135 Hz, except for the aromatic fraction, where a JCH value of 145 Hz was used. Figure 2a-d shows the 1d-hmqc edited spectra for the CFO sample. Clean subspectra have been obtained for selective CH3 and CH2 groups only, when compared to the corresponding conventional 1H NMR spectra. In the CH3-only spectrum, the signals due to CH groups (in the aromatic region), and those due to CH2 (in the aliphatic region), are completely suppressed for both the samples. The efficient suppression of the strongest CH2 signal at 1.26 ppm is indicative of the success of this experiment. Figure 2d shows the CH-only subspectrum of the sample, which clearly shows enhancement of the signals due to methyl groups only. The region of 2-4 ppm shows a very small concentration of the signals due to R-CH groups; thus, on the basis of this observation, it can be ascertained that the region of 2-4 ppm has contributions from mainly R-CH3 and R-CH2 groups only. Clear resolution of the highly overlapped 2-4 ppm region can be seen in the expanded subspectra, which are shown in Figure 3 for two of the samples, viz, VGO (Figure 3a-c) and CFO (Figure 3d-f). Changes in the signal pattern/intensity in the bands, demarcating different chemical shift regions, can be easily observed. Cut-off values for R-CH3 and R-CH2 chemical shifts can be conveniently and unambiguously established with the help of such edited subspectra, compared to the very time-consuming two-dimensional hsqc/hmqc spectra. Complementary results have been obtained in 1d-hsqc experiments and are shown in Figure 4 for both VGO and CFO samples. This establishes the authenticity and versatility of these pulse sequences and their applications to such complex materials, despite a range of JCH (coupling constants) values that exists in these products. Based on the results of these two experiments (1dhmqc and 1d-hswc) that were performed on the rest of

Figure 2. Edited proton NMR spectra (1d-hmqc) of the clarified oil (CFO) sample.

Figure 3. Expanded (0-5 ppm) 1d-hmqc edited proton subspectra for (a-c) vacuum gas oil (VGO) and (d-f) CFO samples.

the samples, the exact chemical shift of protons of R-CH3 and R-CH2 groups attached to aromatic rings have been obtained and summarized in Table 2. The table shows that the ranges of the chemical shifts for R-CH3 groups start at 2.0 ppm and extend up to 2.35 or even 3.0 ppm, depending on the structural complexity. Similarly, R-CH2 groups may appear, starting from 2.3 ppm and extending up to even 4.0 ppm. The region of 2.35-2.8 ppm is an overlapped region and has contributions from both R-CH3 and R-CH2 groups. The aforementioned results for several samples indicate that one cannot

R-CH2/R-CH3 Protons in Heavy Petroleum Fractions

Energy & Fuels, Vol. 19, No. 2, 2005 511 Table 3. Quantitative Estimation of Relative Intensity (I) of r-CH3 and r-CH2 Groups for Various Petroleum Fractions, Using the Edited One-Dimensional Spectra (1d-hmqc) spectrum

I4.5-2.0 ppm

I1.4-1.0 ppm

I1.0-0.5 ppm

HSD Sample conventional CH3 only CH2 only

34.3 16.7 20.1

conventional CH3 only CH2 only

46.6 27.0 18.7

conventional CH3 only CH2 only

39.0 24.2 16.1

174.0 10.3 174.0 (ref)

100.0 100.0 (ref) -1.5

RCO Sample

Figure 4. Edited proton NMR spectra (1d-hsqc) for VGO and CFO samples. Table 2. Division of Proton Chemical Shifts of r-CH3 and r-CH2 Groups for Various Petroleum Fractions, Using the Edited One-Dimensional Spectra Obtained from 1d-hmqc Experiments Proton-NMR Chemical Shift (δ ppm) sample

R-CH3 groups

R-CH2 groups

gasoline (95 octane) high-speed diesel (HSD) vacuum gas oil (VGO) aromatic fraction of VGO high VGO clarified oil (CFO) main column bottom (MCB) reduced crude oil (RCO) rubber processor oil (RPO)

2.05-2.35 2.05-2.6 2.05-2.9 2.05-3.0 2.1-2.8 2.0-3.0 2.05-3.0 2.0-3.0 2.0-3.0

2.4-2.7 2.5-3.2 2.5-3.5 2.4-3.5 2.3-3.5 2.35-4.0 2.5-3.6 2.4-3.5 2.35-4.0

define a fixed cut-off chemical shift for the R-CH3 and R-CH2 groups, which is dependent on the type of samples and structural complexities associated with a sample. However, the presented experiments offer a very reliable and quick way to distinguish individual groups in a highly overlapped proton NMR spectrum, leading to an unambiguous assignment of signals. Quantitative Extent of Overlap As shown previously, these pulse sequences work perfectly for knowing the qualitative extent of overlapping of various signals in a proton NMR spectrum. The quantitative aspect of the 1d-hmqc experiment is also shown in Table 1 for a standard sample of ethyl crotonate. It is also possible to measure the quantitative extent of overlap, i.e., the relative integral intensity of R-CH3 and R-CH2 groups in the present complex mixtures of hydrocarbons present in heavy petroleum fractions. For this purpose, the relative intensity of a set of signals in the normal (conventional) proton spectrum is compared with those in an edited 1d-hmqc spectrum. To estimate the contribution of R-CH3 groups, the relative intensity of signals in the region of 2.0-4.5 ppm (due to both R-CH3 and R-CH2 groups) and 0.5-1.0 ppm (due to only CH3) in the normal proton-NMR spectrum is compared with the intensity of signals in these two regions in an CH3-only edited spectrum. The integral intensity of signals in the region of 0.5-1.0 ppm is kept the same and used as a reference in both spectra. Similarly, to estimate the contribution of the R-CH2 groups, the relative intensity of signals in the region of 2.0-4.5 ppm (due to both R-CH3 and R-CH2 groups) and

244.2 6.1 244.2 (ref)

100.0 100.0 (ref) -2.5

VGO Sample 250.0 13.0 250.0 (ref)

100.0 100.0 (ref) -2.3

1.4-1.0 ppm (due to only CH2) in a normal proton-NMR spectrum is compared with the intensity of signals in these two regions in an CH2-only edited spectrum of the sample. The integral intensity of signals in the region of 1.4-1.0 ppm is kept the same and used as a reference in both spectra. Using the aforementioned methodology, the relative contribution of R-CH3 and R-CH2 groups has been calculated; Table 3 shows the representative values for a few of the samples, which indicate the authenticity of the aforementioned methodology within the experimental error. Conclusions Two simple pulse sequences have been presented for editing the 1H NMR spectra of wide range of petroleum fractions, resulting in unambiguous resolution of R-methyl and R-methylene groups appearing in highly overlapped regions, both in qualitative as well as quantitative terms. Both the experiments are robust and have equal ease of execution. The 1d-hsqc experiment results in a visually impressive proton spectrum, where the signals due to CH and CH3 groups appear anti-phase to those from CH2 groups. However, quantification of the spectrum is not straightforward, because of the appearance of positive/negative signals. Also, the sensitivity of a spin system (CHn, where n ) 3, 2, or 1) is dependent on the delay ∆1, which, in the present case, was kept as 1/6JCH to achieve overall improvement in sensitivity for all multiplicities. On the other hand, the 1d-hmqc experiment provides selective CHn (n ) 3, 2, or 1) edited proton spectra, each one requiring optimization of the gradient ratios for best results. The demarcation of the chemical shifts of methyl, methylene, or methane groups is clearer and more unambiguous. Quantification is also possible in the overlapping regions of the spectra by comparison of the conventional proton spectrum with the selective CHn spectra. Acknowledgment. One of the authors (G.S.K.) acknowledges the receipt of fellowship for the research stay from Alexander-von-Humboldt (AvH) Foundation, Germany. G.S.K. is also thankful to the management of Indian Oil R&D Centre, Faridabad, India, for granting study leave and permission to publish this work. EF049810C