Nagy E. MOUSTAFA, Kout El-Kloub Far MAHMOUD(1. , 1177 , ; . , ' Art, , Sham Univerity, 11341 , )

Several studies have reported that separating tiny concentrations of heteroatom-containing compounds (HACC) in petroleum distillates is challenging owing to the complexity of these mixtures [1-4]. The previously developed belief detector design contributes to this problem because it has the ability to detect all compounds. Researchers have attempted to design selective detectors that can detect specific atoms, such as sulphur and nitrogen. However, selective detector shave not yet been designed for oxygen atoms [5,6]. Consequently, oxygenated compounds (OCs) have been reformulated so that they can be selectively detected [7]. Many analysts and non-specialists believe that selective detectors could solve this problem. This is a misinterpretation and, unfortunately, many previous research studies and projects have failed to disprove this because large concentrations of hydrocarbons disrupt the mechanics of these detectors and render them unable to detect even minor concentrations of HACC compounds (quenching effect) [8]. Many companies that manufacture selective detectors have, intentionally or not, ignored the studies that have investigated the effect of hydrocarbons on the selectivity and stability of detectors.

It is vital to define selectivity limits for each detector. Selectivity limits based on the concentration of overlapping hydrocarbons as well as the lack of study by detector designers are unacceptable shortcomings. Selective detectors designing began with the flame photometric detector (FPD) [9] and ended with the atomic emission detector (AED) [1,5], wherein each detector had a lower selectivity limit than what was specified in the petroleum industry. Thus, it is necessary to resort to complex and expensive technologies, such as chemiluminescence [10,11]. However, HACC compounds should be separated as a group from the crude oil using column chromatographic technologies before they are analysed. The solution to the problem lies in improving ion exchangers and stationary phases, which can be prepared using simple laboratory techniques. A sharp separation of these compounds from the oil matrix makes their analysis feasible, even when using a normal detector [8].

The ordinary column chromatographic fractionation scheme for crude oil is based on two steps. In the first step, the crude oil fractionates into four primary fractions: saturates, aromatics, resins, and asphaltenes (SARAs) [12-14]. Commonly, this technique of ordinary column chromatography is primarily based on the adsorption chromatography usage of silica and/or alumina. In the second step, the desired HACC is typically isolated from the pre-isolated aromatic fraction using a selective adsorbent. Generally, the preparation of these selective adsorbents is based on a certain transition metal [15]. The use of nanotechnologies is one of the simplest approaches for extracting tiny concentrations of HACC from the oil matrix. The isolation properties of nanoscale metal particles are advanced to the metal aggregates, which results from direct blending preparations with a solid support [15]. One of the risks in using metallic aggregates is the non-reversible adsorption and/or reaction that occur with many S-PAHs congeners. These nanoscale-range approaches reduce the atom overlaps that cause their dispersion and orientation in the solution (such as sol colloid) resulting in a sharp extraction of the tiny concentrations. The evolution through reformulation of PbCl2-silica with the insertion of the mercaptopropyl chain transforms PbCl2aggregates into a caped PdS-nanoparticle dispersion. These nanoparticles interact weakly with heteroatoms according to the chromatographic isolation demand [8,16,17]. Previous studies have investigated ionic exchanger containing metals [15], but few have focused on nanoscale preparations [8,6]. Generally, Pd is used in the petroleum industry as a catalyst.

Crude oil is a super complex mixture of hydrocarbons with trace amounts of sulphur, nitrogen, and oxygen as well as metals. In the petroleum refining industry, OCs are undesirable compounds because they are corrosive and block the refining units through gum formation. Previous GC research within the scheme of OC analysis has relied on esterification of the hydroxyl groups to overcome their high polarities and assigning them to selective detection [7]. Distinctive analytical strategies have developed based on different OC formulas. These are classified as phenols [18,19], carboxylic acids [20,21], alcohols[22], and ketones [23]. Typically, the strategies used to isolate OCs from petroleum are based on solvent extraction [24] or solid phase extraction because strong, modified polymeric phases and ion pair reagents as well as different adsorbents for phenols [25,26] and KOH-impregnated silica gels and ion-exchanger for acids [27] are often characterized by complex and solvent-consuming techniques.

ESI(-) and ESI(+) [28,29] were used to efficiently detect OCCs [7,30] and S-PAHs [31], respectively, in different matrices of petroleum crude oil and its deposits. To date, there has not been a single analytical approach developed to isolate combined OCs from petroleum distillates. According to our literature survey, there are no studies on the isolation of OCs from Saudi Arabian crude oil. ESI-MS produces a complicated chromatogram of gas oil [28]. The new here is using nano-Pd to separate tiny concentrations of OC molecules from Saudi Arabian crude oil by column chromatography. Therefore, LC-ESI(+)-MS/MS can be used to assign molecular weight distributions and to identify isolated S-PAHs and OCs. These were previously known as nano-Pd form exchangeable complex with sulphur atoms and non-exchangeable complex with nitrogen atoms [1,8]. However, there has not been any studies on nano-Pd and oxygen complex-ability thus far.

1 Experimental

1.1 Apparatus and chemicals

The petroleum crude oil samples were provided by the Saudi Arabian petroleum company. The chemicals for standard and HPLC grade solvents were provided by Sigma-Aldrich.

1.1.1LC-MS/MS

Agilent 6400 Series Triple Quadrupole LC/MS; Agilent 1290 Infinity Ⅱ high speed pump (G7120A); Agilent 1290 Infinity Ⅱ multisampler (G7167B); Agilent 1290 Infinity Ⅱ multicolumn thermostat (G7116B); Aglient HPLC; column: Agilent Poroshell 120 EC-C18, 100 mm×2.1 mm, 2.7 μm (p/n 695775-902). Column temperature: 40 ℃; mobile phase A: 0.1% (v/v) formic acid in water; mobile phase B: 0.1% (v/v) formic acid in methanol; flow rate: 0.3 mL/min; injection volume: 5 μL.

1.1.2MS conditions

Instrument: Agilent QQQ; source positive ESI; drying gas flow: 4 L/min; nebulizer: 0.3 bar; drying gas temperature: 200 ℃; set capillary: 4 500 V; scanm/z50 to 1 500; set end plate voltage 2 000 V; ion polarity: positive.

1.1.3GC-MS

The GC-MS analyses were carried out on an Agilent GC 7890b and a MSD 5977. The initial temperature was set to 60 ℃ for 1 min, and was then increased at a rate of 5 ℃/min to 300 ℃, where it was then held for 20 min. The capillary column was a HP-ms 5 (30 m×0.25 mm×0.25 μm).

1.1.4Infrared instrument

IR measurements were carried out at room temperature for the pre-isolated S-PAH fraction by Fourier transform infrared spectroscopy (FTIR).

1.1.5Nuclear magnetic resonance

1H NMR spectra were detected on a Bruker AscendTM850 H1spectrometer.

1.2 Chromatographic procedure

1.2.1Nano-Pd(II)-mercaptopropano silica gel synthesis

Briefly (see [1]), dried silica gel (60 mesh) was refluxed with 5 mL 3-mercaptopropanotrimethoxysilane (3-trimethoxysilyl-1-propanethiol) in 20 mL dry toluene for 5 h. The obtained mercaptopropano silica gel (MPS) was further treated with a 250 mL aqueous palladium chloride solution (0.01 mol/L) for 12 h.

1.2.2Column chromatography

Firstly, the aromatic fraction of the crude oil was separated by column chromatography [14] as follows. A 20 cm×0.8 cm glass column was filled with 2 g of the silica gel (about 3 cm long). The saturates of the oil were eluted with 40 mL cyclohexane and then an aromatic fraction with cyclohexane-dichloromethane (3∶1). Resins and asphaltenes were retained on the column. The separation of OC and S-PAHs by Pd nanoparticles from pre-isolated aromatic fraction was described previously [1]. Briefly, 1.5 g of Pd(II)-MPS gel was packed into a 20 cm×0.8 cm glass column. A sulfur-free fraction was eluted with 40 mL cyclohexane-dichloromethane (9∶1) and the S-PAHs and OCs were eluted with 40 mL cyclohexane-dichloromethane (2∶1) containing 1% (v/v) isopropanol. The sulphides were eluted with 40 mL cyclohexane-dichloromethane (2∶1) containing 1.5% (v/v) isopropanol saturated with NH3.

Fig. 1 Investigated chemical structures

2 Results and Discussion

The chemical formulas that are of interest to this study are shown in Fig. 1.

First, the aromatic fraction of the investigated crude oil was separated from saturates, asphaltenes and the resin by using a silica column. S-PAHs and OCs were isolated from the pre-isolated aromatic fraction by bonded Pd nanoparticles, which selectively retained those congeners through an exchange complex with sulphur or oxygen heteroatoms. As a result, the retained S-PAHs and OCs can be eluted using a more polar solvent. Fig. 2 illustrates the analysis scheme for the S-PAHs and OCs obtained from the crude oil sample.

Fig. 2 Analysis scheme of the investigated crude oil sample

Usually, nano-Pd is used to isolate S-PAHs from petroleum distillates [1]. Analysis schemes of preceding research were based on chromatographic strategies. In addition, the absence of diverse analytical strategies did not provide insight toward the isolation capability of Pd nanoparticles in regard to the other functional groups. The functional groups were investigated to determine pre-isolated aromatic fractions using FTIR spectroscopy. The FTIR spectrum is characterized by strong absorption near 2 920, 2 850, 1 455, 1 375, and 730 cm-1due to the CH2and CH3groups (Fig. 3). The bands near 1 646, 810 and 760 cm-1were ascribed to aromatic CH2, which is due to aromaticity [32]. The absorbance at 1 259 cm-1was ascribed to a sulfoxide group [13]. The FTIR spectrum for the isolated fraction by Pd nanoparticles is illustrated in Fig. 4. The more intensive absorption bands observed at 1 646, 810 and 760 cm-1were due to aromatic CH2. The absorption band at 703-731 cm-1was ascribed to aromatic CH bending vibrations. This absorption pattern is related to the polyaromatic configuration corresponding to BTs and DBTs [13].

Fig. 3 FTIR spectrum of aromatic fraction

Fig. 4 FTIR spectrum of the fraction isolated by Pd nanoparticles

It can be observed that the Pd nanoparticles isolate and enrich the OCs whose concentrations are undetectable in FTIR. Fig. 4 illustrates the existence of absorption bands at 3 323, 1 747, 1 038 cm-1corresponding to -OH, C=O and C-O groups, which cannot be detected and observed in the FTIR spectrum of the aromatic fraction. The existence of these intensive bands indicates that the nano-Pd atoms are able to extract tiny concentrations of OCs from the aromatic matrix. Evidently, the formed complex between the Pd nanoparticles and -OH, C=O and C-O is reversible, and can be exchanged by a more polar eluent.

Table 1 Hydrogens content of fraction isolated by nano-Pd [32,33]

Fig. 5 1H-NMR spectrum of isolated fraction by Pd nanoparticles

The1H NMR spectrum of the isolated fraction by Pd nanoparticles is shown in Fig. 5. The following protons types can be identified: methyl (0.841-0.899 ppm (10-6)), methylene (1.245 ppm), naphthenic (1.56 ppm) and alpha (2-2.4 ppm) whereas the band at 7.262 ppm was ascribed to mono- and di-aromatic CH [20,32]. The three bands in the 4.3-5.4 ppm range of the chemical shift region were ascribed to phenolic and alcoholic hydroxyls. The13C band at 143.25 was ascribed to the hydroxyl group bonded to the carbon atom (see supplemental information at http://www.chrom-China.com/UserFiles/File/1707017SI.doc) [20]. Table 1 illustrates the percent-hydrogen contents, which were calculated from the integral ratio consistent with ASTM D 5292. The calculated percentage of aromatic hydrogens is 29.02% whereas the reported value of aromatic fraction isolated by silica column equals ca. 8.53% [32,33]. It can be observed that this hydrogen distribution differs from that of the aromatic fraction. The extra percentage of hydrogen in the aromatic and naphthenic moieties corresponds to a condensed heterocyclic distribution. It can be concluded that Pd nanoparticles efficiently isolate OCs from the aromatic fraction. This is confirmed by decreasing both the NMR percentage value and the IR intensity for methyl hydrogens.

The absence of proton bands in the 10-13 and 9.7 ppm regions indicates an absence or undetectable concentration of carboxylic and carbonyl compounds. The presence of carboxylic acids in undetectable concentrations is the most probable according to preceding studies [20]. This is confirmed by the13C band at 183.27 ppm, which was ascribed to the carboxylic group bonded to carbon atoms (see supplemental information at http://www.chrom-China.com/UserFiles/File/1707017SI.doc) [20] and to the relative IR bands intensities of both hydroxyl and carbonyl groups. Previous studies have agreed that carboxylic acids are higher molecular weight naphthenic acid derivatives that existed in minor concentrations [20]. IR band appears at ca. 1 750 cm-1indicating that the acids are free monomeric and non-dimeric species (1 710 cm-1) [34]. The existence of protons bands at 2.0-2.3 ppm can be ascribed to the methyl adjoining the carbonyl group. It can be concluded that OCs are composed of hydroxyl (most abundant) and carbonyl containing compounds.

Fig. 6 GC-MS spectrum of OC authentic standard mixture

Furthermore, the isolation of an authentic standard mixture by nano-Pd was investigated in order to confirm the obtained results. The standard mixture included phenol (m/z=94), benzaldehyde (m/z=105) and benzylalcohol (m/z=106) in petroleum ether (400 ppm). Fig. 6 shows the GC-MS spectrum of the isolated standard by nano-Pd column. The elution of three oxygenates on the nano-Pd column by iso-butanol containing eluent can be observed, in addition to the impurities, benzoic acid, anthrol, and others of high molecular weights. This confirms that the formed complex between nano-Pd and OCs is reversible. GC-MS cannot be used for the analysis of isolated OCs from the petroleum matrix due to their tiny concentrations and higher reactivity. Therefore, LC is used as an alternate and a direct analytical technique.

Fig. 7 ESI(+) spectrum in scan range m/z 190-245 of the fraction isolated by Pd nanoparticles

Fig. 8 MS/MS precursor ion scan to selectively detect [DBT-H]+(m/z=185) and C1-DBTs (m/z=198)

A current coupling between LC and MS/MS simplifies the identification of the distributed S-PAHs and OCs in a selected trace of the given spectrum. Our previous results reveal the presence of two heterocyclic groups, one containing sulphur (BTs and DBTs) and the other containing oxygen atoms (mainly hydroxyl compounds). Accordingly, LC-ESI(+)-MS/MS was used to assign the distribution of molecular weights for both groups as a single mixture. Previous studies precise that two MS/MS scan modes can simplify the spectrum; (i) a selected mass trace, which is characterized by molecular weights distribution and by a difficulty to identify the target analytes due to ion interference (Fig. 7) and (ii) precursor ion scan, which selectively detects the targets the analyte only. The co-eluting ion interferences can be reduced as the scan proceeds from multi to a single component scan (Figs. 7 and 8). DBTs and BTs have known molecular weight distributions [1]. Thus, LC-ESI(+)-MS/MS identification based on different scan modes was investigated for these congeners. Furthermore, it is easier to assign traditional molecular weight distributions of S-PAHs than it is for hydroxyl compounds. In addition, the S-PAHs exhibited protonated molecular ions [-S+H]+.

Fig. 7 illustrates the distribution of molecular weights recorded in a selected mass trace (m/z184-240) used to identify DBTs congeners. It is clear that DBTs congeners are difficult to identify due to the interfering protonated and pseudo molecular ions. A more focused scan simplifies the spectrum trace and reduces the ion interferences(see supplemental information at http://www.chrom-China.com/UserFiles/File/1707017SI.doc). Finally, MS/MS precursor ion scans identified only the target DBT as follows: DBT (m/z=184), C1-DBT (m/z=198), C2-DBT (m/z=212), C3-DBT (m/z=226) and C4-DBT (m/z=240). This distribution is the same as that assigned via routine atomic emission techniques [1,2]. It can be observed that the precursor ion scan spectrum is characterized by the existence of protonated ions and the absence of differentm/z+2,m/z+3, etc. ions. Figs. 7 and 8 illustrate the existence of pseudo molecular ions for C1-DBT and C4-DBT, [-S+H]+for DBT and C3-DBTs and both ions for C2-DBTs. C1-DBTs spectrum includes 4-, 2-, 3- and 1-positional isomers. Lower peaks intensities of C2-DBTs compared to protonated ions are illustrated in Fig. 7 (see supplemental information at http://www.chrom-China.com/UserFiles/File/1707017SI.doc).

DBT identification indicates an ESI(+) formation mechanism of tiny charged droplets and [-S+H]+. As mentioned previously, this [-S+H]+is not constant [28]. In addition to the presence of sulphur atoms as a nucleophilic centre [35] illustrates the formation mechanism of [-S+H]+. The components of these droplets before solvent vaporization are solvent, analyte (DBT) and a growing conductivity component (formic acid). It can be concluded that the formation of a charged adduct or not is based on the solvation orientation of the given congener under the effect of an electric field and analyte basicity. All the previous research, within the discipline of ESI liquid chromatographic identification of phenols and carboxylic acids, were based on the ESI(-) technique[20,31]. ESI(-) mass spectra for carboxylic acids were characterized by fragments (the removal of CO2, CO, and H2O) and also by phenol derivatives [30,31]. In our case, ESI(+) detected OCs as pseudo molecular ions through a non-ionized technique for these species. The differences between the ESI(+) mass spectra of sulphur and hydroxyl compounds are represented by the formation of protonated ions.

LC-ESI(+)-MS/MS was used to identify the isolated OCs based on distinctive scan modes. The hydroxyl compound identification was primarily based on selected mass trace (m/z90-180). These molecular weights are less than the molecular weights of the carboxylic acids [20] and the peak overlaps with S-PAHs can be identified via protonated ion formations. Fig. 9 illustrates the MS/MS selected trace for molecular weights distribution of isolated hydroxyl compounds. The spectrum trace is characterized by ion interferences with the target analyte and by a different distribution compared to S-PAHs. As always, the precursor ion scan selectively identifies the target analyte from the co-eluting interferences seen in the rest of the spectrum. MS/MS precursor ion scan can be used to identify the fingerprint of the hydroxyl compound (see supplemental information at http://www.chrom-China.com/UserFiles/File/1707017SI.doc), such asm/z=94 phenol,m/z=108, 122, 136 for methyl substituted phenols andm/z=170 phenylphenol. The existence of cyclohexanol (m/z=100), its alkylated derivatives (m/z=114.9 and 128) and cyclohexylphenol (m/z=176) can be observed.

Fig. 9 ESI(+) spectrum in scan range m/z 0-180 of the fraction isolated by Pd nanoparticles to identify OCs (see supplemental information at http://www.chrom-China.com/UserFiles/File/1707017SI.doc)

3 Conclusions

All OC formulae can be isolated from complex crude oils and comparable matrices by Pd nanoparticles. The presence of palladium in the nanoscale precipitated the formation of an exchangeable Pd-oxygen complex comparable to a nanoPd-sulfur complex. OC isolation facilitates the identification of these compounds by LC-ESI (+)-MS/MS. ESI(+) distinguishes between OCs and S-PAHs via protonated ions. Three kinds of molecular ions are detected by ESI(+) random protonation mechanism: (i) protonated ions (some S-PAHs); (ii) massive ions and (iii) ions ofm/zvalues corresponding to molecular weights. The first ion type can be identified with selected mass trace. A precursor ion scan mode can be used to identify the other types of ions.

[1] Moustafa N E, Andresson J T. Fuel Process Technol, 2011, 92: 547

[2] Al-Malki A. [MS Dissertation]. Saudi Arabian: King Fahad University of Petroleum and Minerals, 2004

[3] Payzant J D, Mojelsky T W, Strausz O P. Energ Fuel, 1989, 3: 449

[4] Liu P, Xu C G, Shi Q, et al. Anal Chem, 2010, 82: 6601

[5] Stump A, Tolva K, Juhas M. J Chromatogr A, 1998, 819: 67

[6] Rolfes J, Andersson J T. Anal Chem, 2001, 73: 3073

[7] Wasinski F A H, Andersson J T. J Chromatogr A, 2007, 1157: 376

[8] Moustafa N E, Mahmoud K F. Chromatographia, 2014, 77: 829

[9] Bradley C, Schiller D J. Anal Chem, 1986, 58: 3017

[10] Pana Y, Li Z, Zhou X, et al. Chinese Chemical Letters, 2017, 28: 1670

[11] Toramana H E, Franza K, Ronsseb F, et al. J Chromatogr A, 2016, 1460: 135

[12] Speight J G. Handbook of Petroleum Analysis. New York: John Wiley & Sons, 2001

[13] Fan T, Buckley J S. Energ Fuel, 2002, 16: 1571

[14] Aske N, Kallevik H, Sjoblom J. Energ Fuel, 2001, 15: 1304

[15] Kishore S. [PhD Dissertation]. Germany: Westfalische Wilhelms-Universitat Munster, 2005

[16] Moustafa N E, Mahmoud K F. Fuel Process Technol, 2017, 156: 376

[17] Sripada K, Andersson J T. Bioanal Chem, 2005, 382: 735

[18] Guenther F R, Parris R M, Chesler S N, et al. Chromatographia, 1981, 207: 256

[19] MacCrehan W, Brown-Thomas J M. Anal Chem, 1987, 59: 477

[20] Rikk P. [MS Dissertation]. San Marcos: Texas State University, 2007

[21] Jones D M, Watson J S, Meredith W, et al. Anal Chem, 2001, 73: 703

[22] Luong J, Gras R, Yang G, et al. Chromatogr Sci, 2007, 45: 674

[23] Alhassan A, Andersson J T. Energ Fuel, 2013, 27: 5770

[24] Ioppolo M, Alexander R, Kagi R I. Org Geochem, 1992, 5: 603

[25] Laespad M E F, Pavon J L P, Cordero B M. J Chromatogr A, 1999, 852: 395

[26] Wissiack R, Rosenberg E, Grasserbauer M. J Chromatogr A, 2000, 896: 159

[27] Jones D M, Watson J S, Meredith W, et al. Anal Chem, 2001, 73: 703

[28] Zhan D, Fenn J B. Int J Mass Spectrom, 2000, 194: 197

[29] de Hoffmann E. J Mass Spectrom, 1996, 31: 129

[30] Diehl G, Wasinski F H, Roberz B, et al. Microchim Acta, 2004, 146: 137

[31] Lobodin V, Juyal P, McKenna A M, et al. Energ Fuel, 2014, 28: 447

[32] Akmaz S, Iscan O, Gurkaynak M A, et al. Petrol Sci Technol, 2011, 29: 160

[33] Silv S L, Silva A M S, Ribeirob J C, et al. Anal Chim Acta, 2011, 707: 18

[34] Yu S K T, Green J B. Anal Chem, 1989, 61: 1260

[35] Li M, Wang T G, Simoneit B R T, et al. J Chromatogr A, 2012, 1233: 126