5516
Ind. Eng. Chem. Res. 1997, 36, 5516-5519
Rapid Quantitative Determination of Aromatic Groups in Lubricant Oils Using Gel Permeation Chromatography Nikos Varotsis* and Nikos Pasadakis The Technical University of Crete, Department of Mineral Resources Engineering, PVT and Core Analysis Laboratory, Chania, 73100 Greece
An analytical procedure has been developed which enables rapid quantitative determination of the total aromatic components in the petroleum fraction, used in the production of lubricating motor oils. The method utilizes the ultraviolet-diode array (UV-DAD) and refractive index detector (RI) signals derived from a gel permeation ghromatography analysis (GPC). It provides information about the concentrations of the aromatic compound groups (mono-, di-, tri-, thioaromatics) based on the differences in their absorbances in the ultraviolet (UV) spectrum. The method can be also applied for the determination of the distribution of these groups along the molecular weight range. Introduction Knowledge of the chemical composition of the petroleum fractions, which are used for the production of lubricating oils, enables fast and reliable monitoring of the industrial process and optimization of the operating conditions. Characterizing these fractions using conventional methods is a difficult and time consuming procedure due to the complexity of their composition. Lubricating motor oils are complex mixtures consisting of paraffinic, naphthenic, and aromatic hydrocarbons, with carbon numbers ranging from 15 to 35 or higher. Their composition depends on the nature of the crude oil of origin and on the type and conditions of the industrial processes to which they were subjected. Although heterocompounds containing mainly S and N are present in relatively low concentrations, they influence the operational characteristics of the final product (color, oxidation stability, etc.). A commonly applied analytical procedure consists of separating the hydrocarbon samples into the saturated, aromatic, and polar aromatic (resins) compound classes using adsorption chromatography (Lundanes and Greibrokk, 1994). The separated groups can be further analyzed into subfractions using mass spectroscopy (MS) or gas chromatography-mass spectroscopy (GCMS) techniques (Ashe, 1991). This procedure, due to its requirements in analysis time, equipment and experienced personnel cannot be applied on a routine basis for the characterization of industrial lubricant oil streams. The high sensitivity of the UV-spectroscopy toward aromatic compounds was employed for the quantitative determination of the total aromatics present in hydrocarbon mixtures (Zerlia et al., 1990). It has also been applied for aromatic group type analysis, using an average absorption coefficient determined for each group of components (Burdett et al., 1954). The method is based on the fact that every type of nonsaturated component exhibits a unique UV absorption spectrum (Speight, 1994). The presence of substitutes in the aromatic molecules has a limited effect on their spectrum. An inherent condition of this method is that other groups, which also absorb in the given interval of the UV spectrum are absent or that they have been eliminated prior to measurement by a suitable experimental * Tel: 30-821-37468. Fax: 30-821-69554. E-mail: varotsis@ mred.tuc.gr. S0888-5885(97)00463-6 CCC: $14.00
procedure. This method has been widely used in the determination of naphthalenes and acenaphthenes in jet fuel and in white oils free of sulfur and oxygencontaining compounds with low concentrations of aromatics. It has also been used for the determination of aromatic groups in diesel and lubricating oil fractions where, due to overlapping UV spectra, the calibration procedure included the absorbances of the groups at several wavelengths (Dixit et al., 1994). The scope of this work was to adapt this approach for the determination of the aromatics using the UV diode array detector (UV-DAD) signal derived from gel permeation chromatography (GPC) analysis. The peak height monitored at a specific wavelength at any moment corresponds to the absorbance of the components eluting from the column at this time. These heights, monitored over the entire range of wavelengths, form the UV spectrum of the eluting components. As the components elute from the column in accordance to their molecular size, the derived spectra reflect the composition of the nonsaturated compounds belonging to a specific molecular weight range. The developed method was applied for the determination of the total aromatic content as well as of the composition of the aromatic fraction in lubricating motor oil fractions. Experimental Section The high-pressure liquid ghromatography (HPLC) system that was employed in this study consists of a pump (Waters model 600), a refractive index detector (Waters model 410), and a diode-array detector (Waters model 996) and uses the Millenium software for data acquisition and analysis. The analytical part of the system consists of two GPC columns, an Ultrastyragel 500A one from Waters and a PL gel 300 × 7.5, 10 µm, 50 A from Polymer Labs. A guard column 10 µm, 50 × 7.5 mm was also used. The mobile phase (chloroform, HPLC grade from Labscan) was filtered through a 20 µm membrane filter and degassed with helium. The flow rate was set at 0.75 mL/min and the oven and the RI unit were thermostatically controlled at 35 °C. The samples used during this study were oil fractions produced at different stages of a typical base oil production process in an oil refinery working with crudes of different origins (A, Arabian; R, Russian). The samples were produced under the same operating conditions and they are classified as follows: © 1997 American Chemical Society
Ind. Eng. Chem. Res., Vol. 36, No. 12, 1997 5517 Table 1. Physical Properties of the Samples sample AR ASa AD AB RR RSa RD RB a
density at 20 °C (g/mL)
refractive index at 20 °C
0.840
1.472
0.856 0.856 0.829
1.4760 1.4750 1.4725
0.857 0.852
1.4760 1.4741
mol wt 330 380 320 320 350 375 340 340
S content (% wt) 0.762 0.837 0.518 0.515 0.669 0.314
Solid at ambient temperature.
AR, RR ) raffinates produced by solvent extraction AD, RD ) dewaxed oils AS, RS ) produced slack waxes AB, RB ) final hydrotreatment products Some physical properties of the samples are presented in Table 1. Results and Discussion The signal of the GPC-DAD of an oil fraction at a given UV wavelength is a curve which represents the absorbance of the nonsaturated components along the molecular weight range. The total area under the GPC curve, which results from the sum of the absorbance intensities of the different compound groups of the oil weighted by their concentration, is a measure of the total absorbance of the sample at the selected wavelength. Consequently, the molar concentrations of each group are the unknowns of an equation system (37 equations) that is solved iteratively. The coefficients of the equations are the molar absorptivities of each aromatic group. Quantitative Determination of the Aromatic Functional Groups Present in Lubricating Oil Samples. Firstly, a method was developed for the quantitative determination of the aromatic functional groups present in the lubricating oil samples. Pure aromatic hydrocarbons (benzene, naphthalene, phenanthrene, and dibenzothiophene) were selected as representative members of the mono-, di-, tri-, and thiophenoaromatic compound groups. The selected components were injected separately into the GPC system in solutions exhibiting concentrations within the expected concentration range of the aromatic groups in the samples under study. The total peak area at each wavelength between 254 and 398 nm was recorded, in 4 nm steps. The molar absorptivity () of each pure component, at any recorded wavelength, was subsequently calculated according to Beer’s and Lambert’s law:
) A/cL
(1)
where is the molar absorptivity (L mol-1 cm-1), A is the measured absorbance, L is the path length of the optical cell (cm), and c is the concentration of the solute (mol L-1). The molar absorptivity values were found to be constant and independent of the solute concentration. Figure 1 shows the molar absorptivities of the pure aromatic components versus wavelength. The molar concentration of the selected pure components in the injected solutions is directly related to the number of aromatic functional groups present in the sample per
Figure 1. Molar absorptivity of the model aromatic components vs wavelength. Table 2. Quantities of the Aromatic Functional Groups Present in the Oil Samples per Unit Volume (Multiplied by N) sample
benzene groups naphthalene groups phenanthrene groups benzothiophene groups total
RR
RS
RD
RB
AR
AS
AD
AB
1.02 0.04 0.00 0.01 1.07
0.38 0.02 0.00 0.00 0.40
1.32 0.05 0.01 0.01 1.38
0.80 0.04 0.00 0.01 0.85
1.68 0.07 0.01 0.01 1.77
0.45 0.02 0.00 0.00 0.48
1.98 0.08 0.01 0.01 2.08
1.61 0.07 0.01 0.01 1.70
unit volume. Therefore, their molar absorptivities can be considered as the sum of the absorbances of the N functional groups contained in this volume (N ) Avogadro’s number). The molar absorptivity of the oil samples can then be expressed as the sum of the molar absorptivities of the four functional groups (benzene, naphthalene, phenanthrene, and dibenzothiophene) weighted by their numbers.
si )
∑mijcmj
(2)
where si is the molar absorptivity of the oil sample at wavelength i per unit volume, mij is the molar absorptivity of the pure component j at wavelength i per unit volume, and cmj is the number of aromatic functional groups per unit volume (multiplied by N). The molecular weights of the oils were determined from the RI signal using the analytical procedure described elsewhere (Varotsis and Pasadakis, 1997). A system of 37 equations with 4 unknowns (overdetermined) was composed and solved for cmj. The results obtained using the above methodology are presented in Table 2. As was expected, the number of aromatic functional groups is greater in the dewaxed oils than in the raffinates. From the results it is also clear that fewer aromatic functional groups are present in the base oils (RB, AB) than in the dewaxed oils (RD, AD) which were subjected to hydrotreatment. The method has been used with samples not exhibiting sulfur concentration higher than 1.0% wt. Determination of the Molar Concentration of the Aromatic Compound Groups in the Lubricating Oils. A new approach was subsequently adopted for the determination of the molar concentration of the
5518 Ind. Eng. Chem. Res., Vol. 36, No. 12, 1997
Figure 2. The RI and UV-DAD at 254 nm GPC signal of sample RB. Table 3. Molar Concentration (% mol) of the Aromatic Groups Present in the Oil Samples total aromatics sample
by column chromatography
determined by this method
AR AS AD AB RR RS RD RB
31 8 35 30 20 8 24 16
32 8 37 31 19 7 24 15
breakdown of the total aromatics (by this method) monoaromatics diaromatics triaromatics thioaromatics 87.9 85.8 87.6 87.7 87.5 86.6 87.6 86.1
9.5 11.0 9.4 9.2 9.2 10.2 9.1 10.7
1.2 1.8 1.5 1.5 1.7 1.7 1.6 1.5
1.4 1.4 1.5 1.6 1.6 1.5 1.7 1.7
aromatic compound groups based on the methodology described above. The model assumes that the aromatic fraction is composed of four compound classes, each one of which can be represented by a model component that contains the functional group of the class and exhibits a molecular weight equal to the mean molecular weight of the entire sample. The validity of this assumption was proved by injecting separately into the GPC system the saturated and aromatic fraction of the samples, produced by column chromatography. The curves of the two fractions as well as of the entire sample obtained from the RI detector were practically overlapping. The molar absorptivities of the oil samples can be approximated by the sum of the molar absorptivities of the aforementioned model components weighted by their molar concentration.
si )
∑mijcmj
(3)
where si is the molar absorptivity of the oil sample at wavelength i, mij is the molar absorptivity of the model component j at wavelength i, and cmj is the molar fraction of the model component j in the oil sample. The molar fractions of the four aromatic groups cmj were calculated as described in (a). The results obtained using the above methodology are presented in Table 3. To verify these results, the samples under study were separated into saturated and aromatic fractions using column chromatography. The determined mass of each fraction is expressed in mole percentage by using the mean molecular weight of the sample. As is is shown in Table 3, there is very good agreement between the total aromatics obtained by the two methods. Determination of the Distribution of the Aromatic Groups along the Molecular Weight Range. The developed method was then applied to the different
Figure 3. Concentration of the aromatic groups in the different molecular weight segments of the oil sample.
segments of the oil samples as they are eluting from the GPC column. The aim was to determine the distribution of the four model components throughout the molecular weight range. The UV-DAD elution curve at 254 nm presented in Figure 2 is divided into segments with a time step of 1 min. These segments correspond to fractions of the entire oil sample exhibiting different molecular weights (Figure 3). Subsequently, the spectrum of each segment, between 254 and 400 nm was calculated. The molar absorptivity of each segment was obtained under the following assumptions: (a) The molecular weight of each segment is taken as equal to the molecular weight value that is determined at the middle of the selected time interval from the calibrated RI signal. (b) The mass of the sample
Ind. Eng. Chem. Res., Vol. 36, No. 12, 1997 5519 Table 4. Determination of the Concentration of the Aromatic Groups in the Different Molecular Weight Segments of the Oil Samples 1 mass (%) mol wt monoaromatics diaromatics triaromatics thioaromatics total aromatics mass (%) monoaromatics diaromatics triaromatics thioaromatics total aromatics
19 560 0.431 0.028 0.004 0.003 0.466 18 0.317 0.025 0.002 0.002 0.346
2 41 420
3 Sample RD 26 325
4 10 250
Composition (Mole Fraction in Each Segment) 0.149 0.183 0.378 0.013 0.019 0.029 0.001 0.002 0.002 0.002 0.005 0.013 0.165 0.208 0.423 40
Sample RB 27
11
Composition (Mole Fraction in Each Segment) 0.090 0.096 0.185 0.010 0.014 0.018 0.001 0.001 0.002 0.001 0.003 0.008 0.102 0.114 0.213
that elutes during the time slice of 1 min is calculated as the fraction of the area under the RI curve, recorded during the same time. The mean molecular weights and mass percentages of the segments are presented in Table 4. This table also presents the concentrations of the aromatic groups obtained from the analysis of two oil samples (RD and RB). It can be seen that there is a decrease in the aromatic content of the lubricating oils, which has been provoked by the hydrotreatment throughout the entire molecular weight range. Generally, a more rapid decrease of the monoaromatic content is also observed in comparison to the higher order aromatic groups and it becomes more pronounced in the low molecular weight region. The proposed method requires 35 min of analysis time to determine for a lubricant oil the total aromatic content, the composition of the aromatic fraction, and the distribution of its subfractions versus molecular weight. On the contrary, similar information to be derived from column chromatography and GC-MS analysis requires several hours. Conclusions This work shows that extended information concerning the type and molar concentration of aromatic groups can be derived from a GPC separation of lubricant oils using RI and UV-DAD detection. The RI signal provides the mean molecular weight of the sample, whereas the signal of the DAD describes the abundance of the aromatic compound groups at different wavelengths. These groups can be quantified by exploiting the differences in their absorbances throughout the UV spectrum. The presented method does not require sample pretreatment (i.e., fractionation into separate compound
5 3 200
6 1 170
0.645 0.018 0.005 0.026 0.694
0.575 0.000 0.009 0.032 0.617
3
1
0.352 0.012 0.004 0.016 0.383
0.333 0.002 0.006 0.021 0.362
classes). It can be easily applied for the rapid determination of the changes that occur in the aromatic part of the petroleum fractions during the different industrial processes and for the monitoring of the changes in the distribution of the major aromatic groups at different molecular weights during a chemical process. Literature Cited Ashe, T. R. Petroleum Mass Spectral Hydrocarbon Compound Type Analysis. Energy Fuels 1991, 5, 356-360. Burdett, R. A.; Taylor, L. W.; Jones, L. C. Determination of Aromatic Hudrocarbons in Lubricating Oil Fractions by Far Ultra-violet Adsorption Spectroscopy. Mol. Spectrosc., Rep. Conf., Inst. Petrol. 1954, 30-41. Dixit, L.; Gypta, R. B.; Ram, S.; Kumar, P. Application of Normal and Derivative Ultraviolet Absorption Spectroscopy for the Determination of Aromatic Hydrocarbons in Gas Oils. Fuel Sci. Technol. Int. 1994, 12 (2), 171-189. Lundanes, E.; Geibrockk, T. Separation of Fuels, Heavy Fractions and Crude Oils into Compound classes: A Review. J. High Resolut. Chromatogr. 1994, 17, 197-202. Speight, J. G. Application of Spectroscopic Techniques to the Structural Analysis of Petroleum. Appl. Spectrosc. Rev. 1994, 19, 269-307 Varotsis, N.; Pasadakis, N. An Analytical Method for Rapid Monitoring of the Degree of Hydrogenation of Recycled Lubricating Motor Oils. Ind. Eng. Chem. Res. 1997, 36, 3915-3919. Zerlia, T.; Pinelli, G.; Zaghi, M.; Frignani, S. UV Spectrometry as a tool for the Rapid Screening of the Petroleum Products. Fuel 1990, 69, 1381-1385.
Received for review July 3, 1997 Revised manuscript received September 30, 1997 Accepted October 1, 1997X IE970463M
X Abstract published in Advance ACS Abstracts, November 1, 1997.