Energy & Fuels 1994,8, 618-628
618
Structural Analysis by NMR and FIMS of the Tar-Sand Bitumen of Bemolanga (Malagasy) A. Rafenomanantsoa Service de Chimie I , Etablissement d'Enseignement Suphrieur des Sciences, BP 906, Antananarivo, Madagascar
D. Nicole* and P. Rubini Laboratoire d'Etude des Syst2mes Organiques et Colloidaux, UA CNRS 406, Universith de Nancy I, BP 239, 54506 Vandoeuvre-12s-Nancy Chdex, France
J . 4 . Lauer Centre de Recherche et de Dhveloppement de l'Est, Elf- Atochem, BP 1005, 57501 Saint-Avold Ckdex, France Received July 7, 1993. Revised Manuscript Received January 3, 1994"
The detailed characterization of the tar-sand bitumen of Bemolanga (Malagasy) was performed with the aid of NMR and FIMS techniques. After precipitation of the asphaltenes an elaborate separation of maltenes was carried out by HPLC. The number averagemolecular weight measurements indicate that this bitumen is much heavier than American bitumens (MW = 900 against 534 for the Athabasca bitumen for example). The modeling shows that the average length of the alkyl chains and the ratio of the number of naphthenic rings to that of aromatic rings increase from the aromatic fractions to the resin fraction and then decrease for the asphaltenes. Moreover, the average length of the alkyl chains is always larger than that in the corresponding fraction of American bitumens. All the results indicate that the saturated fraction does not contain many linear paraffins and is constituted of naphthenic rings on which alkyl chains are branched. The change from the saturated fraction to the aromatic fraction with the same molecular weight is made by substituting the naphthenic rings by aromatic rings. The more saturated character of the Malagasy bitumen with respect to other bitumens is explained by the difference in their geological origin; the Bemolanga bitumen, coming from a Rift basin, is relatively protected and, consequently, less subjected to degradation by microorganisms.
-
Introduction The characterization of tar-sand bitumens which are an alternative energy source to petroleum has been an important and difficult task. There have been many p a p e r ~ l -published ~ on the topic. These research works were mainly devoted to bitumens coming from the American continent, specially those originating from two geographic areas: Alberta (Canada) and Orinoco (eastern Venezuela) which represent 91% of the total known tar volume in the world.6-'6 However, only two studies, one on the physicochemical properties17 and one on the
aromaticity determination of the aromatic fraction and the asphaltenes,l8 have been published on the tar-sand bitumens of Bemolanga (Malagasy). The Bemolanga accumulation was emplaced as an oil in a liassic paleodelta converging over a regional arch separating the two main basins of Malagasy, the Morondava and Majunga basins. The aim of this work is the characterization of this bitumen by NMR spectroscopy and field ionization mass spectrometry (FIMS) in order to be able to compare its chemical structure to that of the American bitumens which were much studied by these techniques,lS2l but, in many cases, involving only some fractions of the bitumen. After
Abstract published in Advance ACS Abstracts, March 1, 1994. (12) Payzant, J. D.; Hogg, A. M.; Montgomery, D. S.; Strausz, 0. P. (1) Ignasiak, T.;Kemp-Jones, A. V.; Strausz, 0. P. J.Org. Chem. 1977, AOSTRA, J. Res. 1985, I, 203-210. 42, 312-320. (13) Kotlyar, L. S.; Morat, C.; Ripmeester, J. A. Bull. Magn. Reson. (2) Selucky, M. L.; Chu, Y.; Ruo, T. C. S.; Strausz, 0. P. Fuel 1978, 1989. 11., 324-326. .., -. - - . - -. . 57, 9-16. (14) Guzman, C.; Montero, C.; Briceno, M.; Chirinos, M.; Layrisse, I. (3) Tominaga, H.; Itoh, S.; Yashiro, M. Bull. Jpn. Petr. Inst. 1977,19, Prepr. Pap.-Am. Chem. SOC., Diu. Pet. Chem. 1988,33, 315-329. 50-55. (15) Altgelt, K. H.; Boduszinski, M. M. Energy Fuels 1992,6,68-72. (4) Yoshida,R.;Yoshida,T.;Ikawa,Y.;Okutani,T.;Hirama,Y.;Nakata, (16) Boduszinski, M. M.; Altgelt, K. H. Energy Fuels 1992,6,72-76. Y.;Yokoyama, S.; Makabe, M.;Hasegawa,Y.Bull. Chem. SOC. Jpn. 1979, (17) Andrianasolo, H. R.; Raveloson, E. D.; Rakotoarison, S.; Lala52,1464-1467. harisaina, J. V. UNITAR-Int Conf. Heavy Crude Tar Sands, 3rd 1985, (5) Speight, J. G.; Moschopedis, S. E. Fuel 1980, 59, 440-442. 2, 846-857. (6) Poirier, M.-A.; Das, B. S.Fuel 1984, 63, 361-365. (18) Raveloson,E. A.; Rouviere,F.;Ruiz, J. M.;Lena,L. Collect. Colloq. (7) Bunger, J. W.; Thomas, K. P.; Dorrence, S. M. Fuel 1979,58,183195. Semin. (ZNST.FR. PET) 1984,40 (Caract. Huiles Lourdes Residus Pet.), (8)Malhotra, V. M.; Graham, W. R. M. Fuel 1983, 62, 1255-1264. 117-121. (19) Boduszinski, M. M. Energy Fuels 1988,2, 597-613. (9) George, A. E.; Beshai, J. E. Fuel 1983, 62, 345-349. (20) Ueda, K.; Mataui, H.; Malhotora, R.; Nomura, M. Sekiyu Gakkai (10) Tsai, C. H.; Deo, M. D.; Hanson, F. !I Oblad, .; A. G. Fuel Sci. Shi 1991, 34; 53-61. Technol. Znt. 1991,9, 1259-1286. (11) Bukka, K.; Miller, J.-D.; Hanson, F. V.; Oblad, A. G. Energy Fuels (21) Ueda, T.; Tao, K.; Kinomoto, T.; Yasutake, A.; Hiraki, A. 1992,6, 160-166. Mitsubishi Juko Giho 1985,22, 786-789. '0
0887-0624/94/2508-0618$04.50/00 1994 American Chemical Society
Tar-Sand Bitumen of Bemolanga precipitation of the asphaltenes, the different fractions were obtained from an elaborate separatian of the m a l t " using HPLC. The modeling of the aromatic and polar fractions was performed by improving the method which we worked out previously for the analysis of petroleum asphalts and coal liquids.22The modeling of the saturated fractionswas carried out by combiningFIMS and l3C NMR techniques (using multiple-pulse sequence as DEPT in order to characterize and quantify the carbon types (CH,; n = 0-3) in each fraction. Experimental Section Preparation of Bitumen Samples. The sandstone was crushed and powdered to facilitate extraction. The bitumen was prepared by Soxhlet extraction with 300 mL of chloroform in a 500-mLflask immersed in an oil bath a t 70-80 "C for 6 h. The bitumen accounts for 9.3 5% of the initial tar sand. The physical characteristics of the bitumen defined by SpeightZ3were as follows: specific gravity 60"/60O F 0.967;API gravity 14.7;kinetic viscosity at 100 O F 1800cSt; UOP characterization factor K 11.6. The specific gravity was lower compared with American bitumens especially the Athabasca bitumens (0.989). Otherwise, the UOP characterization factor permitted to classify the Malagasy bitumen as rich in naphthalene compounds. The true boiling point (TBP) distillation of the bitumen gives 12.4 w t % of distillate a t 350 "C. The residue 350+ represents 87.6 w t % of the initial bitumen, indicating the presence of large concentration of heavy compounds. This bitumen may be considered as a Lloydminster Alberta heavy oi1,a the API density of which is 16,rather than an actual bitumen. For the precipitation of the asphaltenes by n-heptane, a bitumen to solvent ratio of 1:40 (wt/vol) was always maintained.% The procedure for the precipitation of asphaltenes by an alkane solvent at room temperature was carried out by dissolving the bitumen in an equal quantity of toluene, adding dropwise to 40 volumes of n-heptane and stirring magnetically for 1 h. The residue was filtered quantitatively, washed with 3 X 10 mL portions of solvent and dried in vacuo to constant weight. The asphaltene weight percentage was found to be 14.0%. For the precipitation a t reflux temperature, the bitumen was stirred magnetically with the requisite amount of n-heptane in a 500mL flask and refluxed in an oil bath for 1 h. The mixture was filtered while still warm and washed with 3 X 10 mL of warm solvent before drying and weighing. The experiments carried out through six samples were reproducible; the asphaltene weight percentage was 15.6% on average. The maltenes used in the chromatographic fractionation were obtained by mixing the six extracts. The chromatographic fractionation was carried out using a Waters Association preparative liquid chromatograph (Model 500) equipped with a differential refractometer Waters Model R 401 and UV Waters Model 441 detectors, recorder, and solvent delivery system. The semipreparative separation was performed on a trial column system (40 cm by 7.5 cm) packed vhth 10-pm Merck LICHOSORB silica. Maltenes were dissolved in hexane by sonication (0.25 g/mL), and 2 mL of solution was injected onto the columns. Then the sample was first eluted with hexane at a flow rate of 6 mL/min and eluent was monitored at 254 nm. The primary saturated fraction, defined as SAT 1', was collected in the first 6 min. At this time the UV response pointed out the elution of aromatic compounds which were collected and defined as ARO I' fraction. The collection of the ARO I1 fraction was started when a deflection on refractive index recorder, coinciding with a change in eluate color from yellow to orange, indicated the (22)Delpuech, J.-J.; Nicole, D.; Daubenfeld, J.-M.; Boubel, J.-C. Fuel 1985,64,325-334. (23)Speight, J. G. The Chemistry and Technology of Petroleum; Marcel Dekker: New York 1980. (24)Demaison, G. J. Am. Assoc. Pet. Geol. Bull. 1977,61,1950-1961. (25)Ali, L. H.;Al-Ghannam, K. A. Fuel 1981, 60, 1043-1046.
Energy & Fuels, Vol. 8, No. 3, 1994 619
I
SAT!
,\
ARO I
Figure 1. HPLC flow diagram. (a) Primary separation allowing to obtain the ARO Il and resin fractions. (b) Further separation for the collection of the SAT and ARO I fractions. Table 1. Molecular Weights and Elemental Analyses Americanu and Malagasy Tar-Sand Bitumens elemental analysis (wt %)* bitumens wtandfractions % ' MW" C H N S 0 Athabasca bitumen 534 82.6 10.5 0.5 4.9 (1.5) saturatefracn 29.7 324 86.9 12.8 - - aromaticfracn 34.7 465 84.3 9.9 0.3 5.4 (0.1) resin fracn 19.1 739 80.5 9.9 0.8 6.5 (2.3) asphaltene 16.5 2750 80.0 8.2 1.0 7.9 (3.3) Cold Lake bitumen 500 82.8 10.8 0.3 4.6 (1.5) saturatefracn 30.3 331 87.0 13.4 - - aromaticfracn 40.6 517 83.4 10.2 0.2 6.4 (0) resin fracn 13.5 1010 81.8 9.5 1.3 5.6 (1.9) asphaltene 15.6 2030 82.2 8.2 1.6 7.5 (0.5) Orinoco bitumen 630 83.9 10.5 0.8 4.5 (0.3) saturatefracn 16.6 380 86.5 13.7 - - aromaticfracn 38.4 530 84.2 10.3 0.2 5.4 (0) resin fracn 21.0 910 82.7 9.7 1.1 4.8 (1.7) asphaltene 24.0 2270 82.7 8.4 1.2 5.4 (2.4) BemolangaBitumen 900 85.4 10.9 0.6 1.0 2.4 saturateIfracn 14.3 470 86.5 13.4 - - saturate IIfracn 11.6 485 86.5 13.3 - - aromaticIfracn 14.8 455 87.2 10.1 0.3 2.0 0.5 aromatic IIfracn 17.5 820 83.0 9.6 0.8 1.0 3.0 26.2 1800 84.1 10.9 1.4 0.8 3.8 resin fracn asphaltene 15.6 4100 79.2 8.9 1.8 1.5 5.7
1
of
H/C 1.52 1.75 1.40 1.47 1.23 1.55 1.83 1.46 1.38 1.19 1.49 1.89 1.46 1.39 1.20 1.53 1.86 1.84 1.75 1.38 1.55 1.34
Number-averagemolecular weight determined by vapor pressure osmometry in benzene at 45 OC for the American bitumens and pyridine or toluene for the Bemolanga bitumen. Values in parentheses are by difference. elution of aromatics by methylene chloride. The collection of the resin fraction was started when the new three-component solvent can out of the columns as shown by the deflection of the UV response and the change of the eluate color from colorless to brown. The solvent was changed again to methylene chloride and then to hexane in order to clean and equilibrate the columns prior to the analysis of the following sample (Figure la). The first two fractions SAT I' and ARO I' which both contain saturate and aromatic compounds were submitted to further HPLC separation (Figure Ib). The saturate fraction SAT I' was dissolved in hexane. Elution of SAT I was finished 7.8 min after injection. A t this time solvent flow was reversed by the backflush valve (BF) and the collection of aromatics ARO Ia was completely carried out 13 min after injection. In a similar manner the ARO I' fraction was separated into SAT I1 and ARO Ib fractions. SAT I and SAT I1 fractions can be considered under the general designation SAT but were analysed separately whereas ARO Ia and ARO Ib were mixed and further called ARO I. Solvents were completely removed from each fraction under partial vacuum in a rotary evaporator and the residues weighed to determine the composition of maltenes. The weight percentages of each fraction are shown in Table 1 (they were forced to 84.6% of the total bitumen, a small lost of matter being assigned to retention of resins on to the columns). These resulta differ from the values obtained by Raveloson.% This is due, in
620 Energy & Fuels, Vol. 8, No. 3, 1994 particular, to an insufficient separation of the aromatic fraction which certainly contains saturate compounds as already noted in other works. Elemental Analyses a n d Molecular Weight Determinations. The elemental analyses were performed out by the Service Central d'Analyse du CNRS in Vernaison (France). The molecular weights of the bitumens and the corresponding fractions were determined by the vapor pressure osmometry (VPO) method on a Knauer apparatus after dissolution of the fraction in pyridine and (or) toluene. NMR Spectroscopy. Samples are prepared by dissolving about 0.4 g of each fraction in 3 cm3 of CzDzCL and 0.15 cm3of hexamethyldisiloxane(HMDS) used as an internal reference (13C chemical shift 1.98 ppm/TMS). The solutions are divided in two parts. In the former part a relaxation agent Fe(acac)s is added at a concentration of 5 X le2M, reducing the relaxation time 21 ' and allowing the recording of quantitative l3C NMRspectran-29 with a 90° pulse and a time between consecutive pulses of 15 s. The second part is used for lH spectrum. All the spectra were recorded at 25°C on a Bruker AM400 spectrometer equipped with an Aspect 3000 computer. The GASPE spectra30931were generated using the spectrometer microprogramming facility and were acquired with a spectral width of 20 000 Hz (ca. 200 ppm), 64 K data acquisition points and a delay between the 90 and 180" pulses of 8 ms (J= 125 Hz). The DEPT experiments, necessary for generation of CH, (n = 1-3) subspectra, were performed by using the pulse sequence described by Bendall and co-workers.32ss The quaternary carbon signals were obtained using t9m DEPT sequence with a 90"phase shift of the receiver as determined by Cleon.34 The 13Cand lH 90° pulse widths were 18.5 and 32 ps, respectively. Since we were only interested in the aliphatic region of the spectra, the carbon-hydrogen spin coupling was set to 125 Hz. The same amount of relaxation agent as for normal quantitative 13Cspectra was used; a pulse delay of 12 s was applied for classical DEPT sequences and a delay of 35 s for the "modifiedn DEPT sequence to observe the quaternary carbons. DEPT subspectra were generated by the appropriate linear combination of the spectra acquired for 0 = 45O, 90°, and 135". The different coefficients were adjusted with the help of a DEPT experiment on a test compound: 2,4-dimethylhexane. In order to have comparable absolute intensities, twice as many transients were taken for the DEPT 890 experiments as for the other 0 values. The quantification of each type of CH, group was made by the measurement of the respective integrals with respect to that of reference compounds added to the sample. Mass Spectrometry. Mass spectral analyses were made with a FINNIGAN MAT-95 Q mass spectrometer. For the field ionization analyses, the sample material was introduced via the direct insertion probe with sample in crucible. The probe was heated from ambient temperature to 400 "C (heating rate 100 "C/min). The emitter heating current was 3.5 mA and thevoltage 6000 V. A mass range of m/z of 150-1500 was scanned a t a rate of 5 s/decade. The source temperature was kept a t about 200 OC. The mass scale was calibrated using ULTRAMARK 3200 in electron impact mode at 70 eV at a resolution of 1250.
Results and Discussion
Molecular Weights and Elemental Analyses of the Bemolanga Bitumen and Its Fractions: Comparison (26) Raveloson, E. A. Ph.D. Thesis, University of Aix-Marseille 3, France, 1987. (27) Gillet, S.; Delpuech, J.-J. J. Magn. Reson. 1980, 38, 433-445. (28) Gillet, S.; Delpuech, J.-J.;Valentin, P.;Escalier,J.-C.Anal. Chem. 1980,52, 813-817. (29) Gillet, S.; Rubini, P.; Delpuech, J.-J.; Escalier, J.-C.; Valentin, P. Fuel 1981,60,226-230. (30) Cookson, D. J.; Smith, B. E. Fuel 1983,62, 34-43. (31) Cookson, D. J.; Smith, B. E. Fuel 1983,62,987-988. (32) Bendall, M. R.; Pegg, D. T. J. Magn. Reson. 1983,53,272-296. (33) Barron, P. F.;Bendall, M. R.; Armstrong, L. G.; Atkine, A. R. Fuel 1984,63, 1276-1279. (34) Cleon, P. Ph.D. Thesis, University of Metz, France, 1984.
Rafenomanantsoa et al. with American Bitumens. It is shown in Table 1that the Bemolanga bitumen is heavier than those comingfrom American deposits.% Its number average molecularweight is 900 against 534 for the Athabasca bitumen for example. This difference is present in each fraction, particularly in resins and asphaltenes, for which the molecular weights are about 2-fold higher. Moreover, the weight percentage of the resin fraction is always higher than that of the other bitumens. Furthermore, from the results of the elemental analyses, the high value of the HIC ratio for these two fractions indicates a more saturated character compared to the American bitumens, in agreement with the value of the UOP characterization factor. Regarding the heteroatoms, the sulfur content of the Bemolanga bitumen is lower than that of the other bitumens whereas the nitrogen and oxygen content (determined directly in this work) is higher. FIMS Analysis of the SAT and ARO I Fractions. With the FIMS method molecular ion peaks without fragmentation are mainly observed. Hence, the respective mlz peaks correspond to the mass number of the molecules and the mlz distribution represents the profile of the molecular weight distribution. The identification of compound types was performed as usual, referring to the general formula CnH2n+z(NH)y Only pure hydrocarbon compounds are taken into account; no hetero elements (N, S, 01, the detection of which would require high-resolution mass spectrometry, have been considered. Consequently, the general formula simplifies to CnHan+~. The 2 value, which indicates hydrogen deficiency compared to alkanes with the same carbon number, is related to the aromatic carbon number (CA)and the total number which means the aromatic rings (RA)plus the of rings (R), naphthenic rings (RN)in the following equation: 2 = 2 - [CA + 2(RA + RN)]
(1)
The spectra in Figure 2 show that, for the SAT fractions, only compounds with mlz < 800 are present. These fractions display major intensities at mlz = 384,398,412, 426, 440, 454, and 468 corresponding to to CM compounds in the 2 = -8 series (these masses suggest the presence of pentacycloalkanes and pentacyclictriterpanes) but also at mlz = 400, 414, 428, 442, 456, and 470 corresponding to Cm-C34 compounds in the 2 = -6 series (these masses are typical of tetracyclic steranes). These compounds are known to be biological markers which are not readily affected by biodegradation and are also present in the Asphalt Ridge bitumen.37 In the series of alkanes (2 = 2) the presence of phytane (C20H42), an isoprenoid having a head-to-head junction which is characteristic for microorganisms like Archaebacteries, is also noted in a small amount. The molar mass distribution allows us to calculate the number - average molecular weight of the SAT fractions: MW = 473 and 493 for SAT I and SAT 11, respectively; these values are close to those obtained by the VPO method. From the intensities of the molecular peaks corresponding to each family and after correction for I3C isotope contributions, the weight percentage can be calculated: (35) Suzuki, T.; Itoh, M.; Takegami, Y.; Watanabe, Y. Fuel 1982,61, 402-410. (36) Uchino, H.; Yokoyama, S.; Satou, M.; Sanada, Y. Fuel 1985,64, 842-848. (37) Holmes, S. A.; Raska, K. A. Fuel 1986, 65, 1539-1545.
Tar-Sand Bitumen of Bemolanga
“1
Energy & Fuels, Vol. 8, No. 3, 1994 621
I
SAT I
80
414
Table 2. Z-Series Analysis of SAT Fractions from Bemolanp Bitumen SAT I SAT I1 Z series alkanes +2 monocycloalkanes 0 dicycloalkanes -2 tricycloalkanes -4 tetracycloalkanes -6 pentacycloalkanes -8 hexacycloalkanes -10 hydrocarbon
type
40-
MW 645.5 554.8 473.6 463.8 440.9 454.9 571.1
mol
W, 7% 3.9 5.8 11.6 16.3 25.7 28.3 8.4
2.9 4.9 11.6 16.7 27.6 29.4 6.9
-
MW 644.6 591.3 496.6 475.7 458.5 459.4 580.0
mol
W, % 6.1 6.7 10.5 14.5 23.4 28.5 10.3
4.6 5.5 10.4 15.0 25.1 30.6 8.8
20-
zoo
400
600
1000
800
is found to be 1.85 (Table 1)corresponding to an average formula CnH2n-s (2= -6). The weight average number of rings per molecule RN is found to be 4.3 and 4.4 for SAT I and SAT 11,respectively, by using the formula of Netzel and al.,38
100
c=~ C , 2 X i / ~ C i X i R N = 1+ ( C - 6 ) / 4
SAT 11 80
60
40
20
260
460
660
moo
Mi0
ARO I
(3) (4)
where C is the weight average number of ring carbon atoms per polycondensed cycloalkane, Ci is the number of ring carbon atoms per cycloalkane i and Xi is the weight or volume percent of cycloalkane i from Table 2. Concerning the aromatic hydrocarbons (ARO I fraction) the formulas of which can be represented by C n H ~ nlike +~ all the other hydrocarbons, the calculation of 2 can also be done from eq 1. The determination of the number of the aromatic and naphthenic rings is only possible in the case of the first three series (2= -6, -8, and -10); the first members of these series are benzene, tetralin, and octahydrophenanthrene (or anthracene), respectively, if only the six-membered rings are considered. From the fourth series (2= -12) this determination becomes difficult and the low-resolution FIMS does not allow the assignment of each mass in the spectrogram to a particular series of compounds. In order to solve this difficulty, it should be necessary to separate again by HPLC the ARO I fraction into four subfractions, i.e., into fractions including one, two, three, or more aromatic rings. However, starting with the eighth series (2= -20), the use of the high-resolution FIMS is needed (R 10 000)to reveal the more condensed hydrocarbons (presenting 14 hydrogen atoms less when compared with the first series). Nevertheless, the calculation leads for the ARO I fraction to an average number of aromatic rings RA= 2.8 and of naphthenic rings RN = 0.9. Structural NMR Analysis of Aromatic, Resin, and Asphaltene Fractions. (i) Method. First, for the determination of an average structure the knowledge of the aromaticity factor which gives the percentage, denoted as (CA), of aromatic carbons CA with respect to the total carbon: (CA) = CA/C is primordial. Then the amount of protonated aromatic carbons is calculated from ‘H NMR spectra and the overall hydrogen to carbon atomic ratio using the following formula:
-
Figure 2. Molar mass distributionby FIMS analysis of the SAT and ARO I fractions (from top to bottom, respectively).
CMziIzi
where I is the relative intensity of the molecular ion, M, the molecular mass represented by the value of m/z of the molecular ion, and i the peak number of the ith molecule. The results of Table 2 where the Z-series type distribution, as discussed before, is also reported, indicate that SAT fractions include mainly cycloalkanes with four or five naphthenic units. This result is in agreement with that obtained from elemental analysis where the ratio H/C
(CAH) = HA/H X H/C (5) where HA represents the aromatic hydrogens. This method was found to be more accurate than that using DEPT sequence because it is possible to consider, particularly in the case of polar fractions, the carbons (38)Netzel, D.A.;McKay, D. R.; Heppner, R. A.; Guffey, F.D.;Cooke, S. D.; Varie, D. A.; Linn, D. E. Fuel 1981, 60, 307-320.
Rafenomanantsoa et al.
622 Energy & Fuels, Vol. 8, No. 3, 1994 Table 3. Nomenclatureand Chemical Shift Correlation Charts of IF and IH Atoms in Hydrocarbons ~
shift range 6 (PPm from
carbon tme
TMS)
CA (aromatic 118-130.5 carbons) 129-137 123.5-136.5 132-137
137-150
150-200
CS (saturated carbons)
10-15
15-17.5
18-22
22-23
22-60 27-60
nomenclature
assignment CAH CACHa
protonated methyl-substituted bridgehead (or internal) benzonaphthenic (at the junction of an aromatic and a six-membered hydroaromatic ring) alkyl substituted (not methylic) benzonaphthenic (at the junction of an aromatic and a five-membered ring) carbons substituted by polar groups terminal methyl group in aliphatic chain in y, 6 or further position from aromatic ring (except the case where two methyl groups are terminal) methyl of an ethyl group attached to an aromatic ring methyl group a to an aromatic ring methyl group branched to an alkyl chain or a naphthenic ring, in y position from the aromatic ring methyl group branched to CH3@b an alkyl chain or a naphthenic ring, in j3 position from the aromatic ring methylene carbon of alkyl CH2 chain or naphthenic ring CH methine carbon of alkyl chain or naphthenic ring
substituted by phenol groups as peripheral carbons (see next paragraph). The quaternary aromatic carbons CAL which are substituted by an alkyl chain other than a methyl group are obtained from the terminal methyl carbons of alkyl chains at 14.23 ppm. The use of the Williams’ method39 is needed for the determination of the other quaternary carbons: internal CAI and benzonaphthenic CAN (CAN(6) + CAN(5))carbons (Table 3). This method is based on the determination of the average number of carbons per alkyl chain from the knowledge of hydrogens HS which are in a,p, and y positions relative to aromatic rings: ii =
[(HS,) + (HS& + (HS,)
X
Z/3l/(HS,)
case, the benzonaphthenic ring is considered to result from the closure of two alkyl chains, e.g., ii = 2 in tetralin. The ii value may be largely overestimated in the case of coal liquids with short substituents,40 e.g., in ethylbenzene (ii = 2.5 instead of 2). Multiple fused benzonaphthenic rings may also result in underestimated or overestimated ii values, e.g., in unsymmetrical octahydroanthracene or phenanthrene (ii = 3.5 and 4.66 instead of 4.0). The ii value should be correct in the case of petroleum products which generally contain long open alkyl chains and a small number of fused benzonaphthenic rings. Under these conditions, the peripheral CAP and internal CAI carbons may be computed according to (CAP) = (CAH) + (CS)/ii
(CAI) = (CA) - (CAP) (9) The benzonaphthenic carbons CAN can be determined only if the aromatic carbons CACH3 substituted by a methyl group are known first, since (CAN) = (CS)/ii - (CAL) - (CACH,)
(39) Williams, R. B. ASTMSpectrosc. Tech.Publ. 1958,224,168-194,
(10)
where CSlii is the percentage of the aromatic carbons substituted by an alkyl chain. The methyl carbons in the y position or further relative to an aromatic ring are obtained with quite good accuracy from the ‘Hspectra (0-1.0 ppm): (CH3,) = ‘/,HS,/H with
X
(CH3,) = (CH,,)
H/C
(11)
+ (CH3,b)
where CH3,, represents the terminal methyl groups of the alkyl chains, evaluated from the intensity of the signal at 14.23ppm in the 13Cspectrum. Consequently it is possible to deduce the nonterminal carbons CH3, and thence, from the 18-22 ppm region, CACH3 carbons, the amount of which is found to be pratically negligible in all the fractions. The chemical shift of the carbons CH3,9,, like l-methyltetralin, is about 23 ppm and their concentration was considered as negligible. Finally, naphthenic carbons CN are obtained approximately by subtracting the prominent resonances from the overall integral of the 27-60 ppm area. Average structure parameters (ASPS) can be transformed in molecular parameters (AMPs) if the average molecular weight is known (Table 1). The number of aromatic rings RAper average molecule together with the number of naphthenic rings are calculated using the following equations: R,=l+CA1/2
(6)
where FA = (CAI is the aromaticity factor and 2 the average number of hydrogens per saturated carbon, by assuming that this average is the same for CY and @ carbons. It should be remembered that, in this method, the term ”alkyl substituents” applies to an alkyl chain, a methyl substituent, an alkyl chain containing one (or more) naphthenic ring or to a benzonaphthenic ring. In the latter
(8)
with
r = ii (1-%/2)
(12)
+ 0.5
(14)
r represents the number of ring closures per aliphatic chain, FS = CS/HS is the saturated carbon to saturated hydrogen atomic ratio, and Rs = @/A is the number of alkyl substituents per average molecule. (40) Delpuech, J.-J.;Nicole, D.; Le Roux, M.; Chiche, P.; Pregermain, S . Fuel 1986,65,1600-1607.
Energy & Fuels, Vol. 8,No. 3, 1994 623
Tar-Sand Bitumen of Bemolanga For the fractions as resins and asphaltenes with a high molecular weight, the average molecule can be represented by a series of identical elementary aromatic subunits, the unit structure (US) or the unit structure weight (USW). The number G of unit structures per average molecule can be found by using the Williams' semiempirical formula39
-
--
CA = 7(CA/CAP)' - 1 -with G = CAICA (16) where CA is the number of aromatic carbons per unit structure hence the other carbons of the unit structure are thus obtained by dividing the average molecular parameters by G. (ii)Comparison of the Fractions. The 13CNMRspectra of the two aromatic fractions as well as those of resins and asphaltenes have the same characteristics (Figures 3 and 4); particularly they exhibit signals at 14.2, 22.6, 29.7, 31.6, and 37.3 ppm due to the carbons of the linear alkyl chains:41
e
29.7
I
8.0
0.0
7.0
6.0
6.0
3.0
4.0
"
2.0
.
,
1.0
.
,
0.0
m
31.6 22.6 14.2
&>H&CH,)&H&H&H,
Other resonances appear in the spectrum, principally at 32.7 and 19.7 ppm, corresponding to a methine carbon with a branching point and, at least, three bonds removed from a terminal point and to the attached methyl group. The signal at 27.8 ppm can be attributed to a methine carbon in an isopropyl group. Finally, the signal at 24.3 ppm could be due to amethylene carbon situated between two branching sites which are three bonds removed.
19.7
CH3
I ' - l 1 1 1 1 , , , , ,
,"""","' 40
A first indication on the averagelength of the alkyl chains can be obtained from the 129.7/114.~ value which represents the ratio of the intensities of the peaks at 29.7 and 14.2ppm. This ratio is maximum for the resin fraction (2.3, 3.7, and 11.5 for ARO I, ARO 11, and resins, respectively) and decreases for the asphaltene fraction (8.3). These values are much higher than those obtained for the American bitumens35 (3.55 and 4.77 for the resin and asphaltene fractions of the Athabasca bitumen for example). The calculation of the average length of the alkyl chains rid can be performed more accurately42 by adding to the previous ratio, 5 plus the area of the branched methyl at 32.7 ppm: ad = 7.8, 8.9, 16.6, and 13.3 for the four fractions, respectively. These results can be confirmed by those which are deduced from the calculation of the structural parameters by using the above mentioned method. However, it is rid
(41)Kotlyar, L.S.;Morat, C.; Ripmeester, J. A. Fuel 1991, 70,90-94. (42)Cyr, N.;McIntyre, D. D.; Toth, G.; Strausz, 0. P. Fuel 1987,66, 1709-1714.
I
30
.
I
I
I
,
,
I
20
,
,
,
,
I
,
,
,
,,,,,,,,, IO
, , I
Oppn
Figure 3. 'H, I3C quantitative,and GASPE spectra of the ARO I1 fraction (from top to bottom, respectively). Concerning the GASPE spectrum only the aliphatic region is shown with CH2 and C peaks positive and CHs and CH peaks negative. necessary to take eventually into account the amount of heteroatoms. For the aromatic fractions, the incorporation of sulphur is generally assumed to take place in the aromatic system under the form of thiophene-like ring. The absolute number of aromatic carbons CA per unit structure should be calculated according to39 I -
-
with
\ -
(15a) G=
(a + 2 S ) / ( a + 2s)
(16a)
For the fraction ARO I, there is only 0.28 sulfur atom per average molecule and the G value, calculated from eqs 15and 15a, i.e., with and without inclusion of sulfur atoms, is nearly equal to 1 (0.97 and 1.00, respectively). The
Rafenomanantsoa et al.
624 Energy &Fuels, Vol. 8, No. 3, 1994
F 'F6
w- ab
P
p- 4,
RESINS at,
BESIN ASPHALTENE Figure 5. Average molecular formulas of ARO I and ARO I1
fractions and average unit structural formulas of resins and asphaltenes.
0 0
LO
1 0
6 0
S O
4 0
2 0
3 0
10
0 0
ASPH ALTENES
i 1m
180
IU)
120
ica
m
.
80
,
I
40
.
20
o
PPm spectra of the resin fraction and the
Figure 4. 1H and 13C asphaltenes. fraction ARO I1 contains 1.54 oxygen atoms per average molecule which can be situated in a furan-like system; in
this case the G value calculated from eq 15a (by replacing S by 0) is a little higher than that obtained when only carbon atoms are taken into account (1.59 and 1.37, respectively). But these heteroatoms can be also found in cyclic or acyclic ethers. To make those simpler, the average molecular formulas of these two fractions are represented without inclusion of heteroatoms (Figure 51, keeping in mind that they can also contain cyclic terpenoids. The polar fractions, resins and asphaltenes, contain 1.93 and 2.14 oxygen atoms, respectively, per unit structure. For such fractions it has been shown that oxygen, the amount of which is the highest among the heteroatoms, is essentially found in phenols, carboxylic acids, and ketones when nitrogen occurs in N-H functions as in ~ t major ~ ~ part of the hydrogens pyrroles or i n d ~ l e s . ~The bound to these heteroatoms, except those of the ester and thiophenols functions, are situated in the aromatic region of the 'H spectrum which was integrated up to 10 ppm. It can be concluded that the il value and the ratio FS = CS/HS are not much modified by the presence of heteroatoms. On the contrary, since the determination of the protonated aromatic carbons was made from the lH spectrum, the value of CAH, together with that of peripheral carbons CAP, takes into account the eventual substitution of a H atom by an OH,COOH, or NH function. The percentage of the internal carbons, calculated from eq 9, is right when the aromatic ring is substituted by a phenol function. This value can be overestimated when the ring is substituted by a carboxylic function or underestimated in the case of the substitution by a NH group. Therefore, one can consider that this percentage is approximately correct as well as that of benzonaphthenic (43)Frakrnan,Z.;Ignasiak, T.M.; Lown,E.M.; Strausz, 0.P.Energy Fuels 1990,4, 263-270.
Energy & Fuels, Vol. 8, No.3, 1994 625
Tar-Sand Bitumen of Bemolanga
Table 4. Average Structure (ASPS)and Molecular (AMPs) Parameters of the Aromatic Fractions ARO I and ARO I1 ASPS
AMPs
mo I C/H CA CAP CAH CAL CACH3 CAN CAI
NMR EA
0.712 0.716
ARO I1 0.699 0.722
CN CHI,, (343% CT
38.2 27.0 13.5 3.9 0.0 9.6 11.2 61.8 19.9 12.6 9.3 100.0
33.3 22.3 11.4 3.8 0.0 7.1 11.0 66.7 28.2 11.0 7.2 100.0
HA HS HSa HSR HS, HT
9.6 90.4 17.4 46.1 26.9 100.0
8.0 92.0 13.5 55.4 23.1 100.0
cs
FA FS
z n r
FC % AS
0.38 0.49 1.81 4.59 0.38 0.27 49.90
mo I
mo I1
MW C (wt %) H (wt %)
455 87.23 10.15
820 83.05 9.58
CA CAP CAH CAL CACH, CAN CAI cs -
exptl 12.6 8.9 4.5 1.3 0.0 3.2 3.7 20.4 6.6 4.2 3.1 33.1
12.6 9.3 4.6 1.3 0.0 3.3 3.3 20.5 6.6 4.3 3.0 33.1
exptl 18.9 13.0 6.7 2.2 0.0 4.0 6.2 37.8 16.0 6.4 4.2 56.7
18 12 6 2 0 4 6 38 16 6 4 56
4.4 41.8 8.1 21.3 12.3 46.2
4.6 42.4 8.3 21.2 12.9 47.0
6.3 72.3 10.6 43.5 18.1 78.6
6 73 11 44 18 79
2.8 4.4 1.7
2.7 4.6 1.7
4.1 6.2 3.6
4 6 4
-
CN
(333%
E,,, CT
HA HS e
a
HS, HS
-7
0.33 0.51 1.77 6.11 0.58 0.54 49.00
carbons according to eq 10 since aromatic carbons CAL were calculated from the terminal CH3 group and not from the region >137 ppm where are also present the quaternary aromatic carbons substituted by polar groups. However, as for the two aromatic fractions, a part of these heteroatoms can be situated in aromatic or saturated rings as it has been shown by the analysis of the structures stemming from the thermal degradation and ruthenium ions catalysed oxidation of a s p h a l t e n e ~ . ~ >Thus ~ 5 the molecular formulas in Figure 5 are average formulas in which oxygen can be found in manifold positions. It can be noted that, from the results reported in Tables 4 and 5, the variation of A, calculated according to the Williams’ method, is similar to rial. The ratio RdRN which represents the number of aromatic rings with respect to that of naphthenic rings per average molecule is always higher than unity, except for the resin fraction for which it is 0.66. The results show clearly that this fraction contains the longest alkyl chains and the highest number of naphthenic rings as well. Finally, if the aromaticity factors of these four fractions are compared with those of the other bitumens, it can be seen that the lowest values are obtained for the Malagasy bitumen (FA = 0.27,0.25, and 0.41 for the total bitumen, the resin fraction and the asphaltenes, respectively, against FA = 0.31,0.36,and 0.48 for the Athabasca bitumen and the two corresponding fractions). Once again the naphthenic character of the Bemolanga bitumen is confirmed by these results. (44)Mojelsky, T.W.;Ignaaiak, T. M.; Frakman, Z.; McIntyre, D. D.; Lown, E. M.; Montgomery, D. S.; Strausz, 0. P. Energy Fuels 1992,6,
83-96. (45)Strausz, 0.P.;Mojelsky, T. W.; Lown,E. M. Fuel 1992,71,13551363.
calcd
calcd
Structural NMR Analysis of the SAT Fractions. No residual signals are visible in the aromatic region of the I3C and lH NMR spectra as proof of the good separation of the different fractions performed by the twofold HPLC separation. The l3C NMR spectra exhibit signals at 14.2, 22.7,31.6, and 29.7 ppm which are usually assigned to C1, CZ,C3, and C4 carbons of n-alkanes, respectively (Figure 6, totalspectrum). However, the results obtained by FIMS demonstrate unambiguously that these peaks are only due to the carbons of the side chains residing on naphthenic rings. Although the influence of these rings on the chemical shifts of the long chain carbons is not exactly known, it can be assumed that a ring is equivalent to a branching as in the following compounds:%
Thus the carbon of the chain in an a-position relative to the branching site is strongly deshielded when the carbon in j3-position is slightly shielded (the ring does not act upon chemical shift anymore beyond the third carbon). Moreover, a signal with an intensity approximately equal to that of the terminal methyl group, is observed at 37.3 ppm and an unresolved band is also present at 27 ppm. It can be concluded (see aromatic fractions part) that the chemical shifts of a CH2 group of a side chain in an a-position relative to an aromatic ring on the one hand or
Rafenomanantsoa et al.
626 Energy & Fuels, Vol. 8, No. 3, 1994
Table 5. Average Structure (ASPs), Molecular (AMPs) and Unit Structure (USPs) Parameters of the Resin and Asphaltene Fractions ASPs resin C'H
NMR EA
0.659 0.655
CT
24.9 16.9 7.5 3.1 0.0 6.4 8.0 74.1 28.8 12.6 7.0 100.0
HA HS HSa HSB HS7 HT
4.9 95.1 10.8 61.8 22.5 100.0
CA CAP CAH CAL CACH3 CAN CAI
cs
CN CH37, CH37b
FA FS
0.25 0.52 1.73 7.93 0.80 0.45 57.50
z a r FC % AS
AMPs resin 1800 84.14 10.85
asphaltene 0.744 0.740
asphaltene 4100 79.17 8.87
41.5 26.3 16.7 4.1
31.4 21.3 9.4 3.9
0.0
0.0
11.1 0.0
5.5 15.2 58.5 9.1 6.5 100.0
8.1 10.0 94.8 36.3 15.9 8.8 126.2
14.9 41.1 158.2 34.8 24.6 17.6 270.5
12.4 87.6 13.1 54.1 20.4 100.0
9.6 185.7 21.1 120.7 43.9 195.3
45.1 318.6 47.6 196.7 74.2 363.7
6.0 12.0 9.0
21.6 26.0 12.0
12.1
0.415 0.50 1.83 6.08 0.46 0.49 36.60
to a naphthenic ring on the other hand are nearly the same as well as for a methyl group in the same situation. It can be also noted that the intensity of the peak at 29.7 ppm is not much higher (as in the fraction ARO I) than that of the terminal methyl group, indicating that the side chains contain six and seven carbon atoms on average. As for the aromatic fractions, the signals at 19.7, 24.4, and 32.7 ppm with roughly equal intensities are useful to reveal the branchings, as described in the following scheme:
112.3 71.1 45.2
G
814.5
USPS resin asphdtene 6.83 600
exptl
cdcd
exptl
calcd
14.2 9.6 4.3
14 10 4
1.8 0.0
2 0
16.4 10.4 6.6 1.6
16.0 10.8 7.0 1.8
4 4 43 16 7 4 57
0.0 2.2
0.0
3.7 4.5 42.9 16.4 7.2 4.0 57.1 4.3 84 9.5 54.6 19.9 88.4
4 84 10 53
2.7 5.5 4.3
2.21
usw CA CAP CAH CAL
CACH, -
CAN CAI -
cs CN -3% CH -3'1b CH
CT
HA HS HS
-a
HS, HS HT
7
-A R
5 RN
6.0 23.1 5.1 3.6 2.6 39.6
2.4 5.2 23.2 6.0 3.8 2.0 39.2
88
6.6 46.7 7.0 28.8 10.9 53.2
7.0 47.0 7.0 28.6 11.4 54.0
3 6 4
3.2 3.8 1.8
3.6 3.8 1.6
21
n=0-3
(17)
with (CH,) = CHJCT, where CT represents all the saturated carbons. The number of saturated carbons for the average molecule is calculated using the following formula
m,
X (CH,) (18) CH, = and the respective average numbers of branching sites BB, branches RBand saturated rings RNper average molecule were obtained using the following equations:
-
B, = CT X FB Other resonances with a lower intensity are present at 39.7, 18.7, and 11.5 ppm and can be attributed to the carbons of the following molecule type:46
e
11.5
C H ~ C H ~ ~ ~ H - W T W 18.7 CH3
From the DEPT subspectra (Figure 6) which allow us to calculate the percentage of carbons CH, ( n = 0-31, the AMPs parameters can be deduced since the average molecular weight is known. The application of the equations used by Netzel and Guffe97 leads to the determination of an average number of carbon atoms per molecule
E:
Burnham, A. K. Fuel 1984,63,404-914. (47) Netzel, D. A.; Guffey, F. D. Energy Fuels 1989, 3, 455-460. (46) Ward,R. L.;
(19)
where FB = (CH) + (CQ) is the branching factor,
N B = CT[2(CQ)
(20)
+ (CH)]
R , = 0.5CT[2(CQ) + (CH) - (CH,)]
(21)
+1
(22)
In our case, CQ i= 0 and RB= Bs. The total number of carbons per average molecule is found to be 34 for the SAT I fraction and the number of naphthenic rings 4, a value close to that deduced from FIMS measurements. It means that 16 carbons are contained in the side chains, representing 47 % of all the carbon atoms. This result can be approximately verified by assuming that the unresolved region between 22 and
Energy & Fuels, Vol. 8, No. 3, 1994 627
Tar-Sand Bitumen of Bemolanga
C"3
40%
60%
SAT i
CH
I
1 1
40%
C
60%
Total spectrum
SAT II Figure 7. Average molecular formulas of the SAT fractions. Table 6. Average Structure (ASPS)and Molecular Parameters (AMPs) of the SAT Fractions b ,
..
I . . . .
I , . . . . .
50.0
,.. .I.....
40.0
...
I,...
30.0
.. ...
.I...
10.0
..
.
.
.
.
,
.
10.0
.
.
,
.
.
.
.
.
,
I
AMPs
.
0.0
PPm
Figure 6. quantitative spectrum and DEPT subspectra of the SAT I fraction.
C/H
ASPS SAT SAT I I1 NMR 0.527 0.532 E EA 0.538 0.542 C (wt %) H (wt %) o.. CQ
experimental calculated
SAT SAT SAT SAT I I1 I I1 473
493
86.5 86.5 60 ppm is due to the naphthenic carbons CN and by 13.4 13.3 subtracting from this region the intensity of the prominent CQ 0 0 0 0 peaks; in these conditions a value of about 50% is found 11.3 11.8 11.2 12.0 CH 33.1 38.2 for the CN carbons. The number of side chains is equal CH2 50.9 50.0 CH, 17.4 18.0 17.8 18.0 to 2 from t b integral of the signals at 14.2 and 11.5 ppm CH2N 24.4 24.6 CH N 8.3 8.8 8.0 8.4 corresponding to terminal methyl carbons CHa (Table 6) CH3 16.0 17.2 5.5 6.2 5.2 6.0 and the number of methyl groups branched on these side CH3t 5.9 6.1 2.0 2.2 2.0 2.0 chains (or eventually on the naphthenic rings) is 3-4 since CN 50.0 50.2 &t 17.1 18.1 18.0 18.0 the total number of CH3 carbons is 5.5. However, if the 100 100 34.2 36.0 34.2 36.0 CT side chains are isoprenoid type chains and if they are 11.3 11.8 11.2 12.0 4 constituted on average by seven carbon atoms, the number 11.3 11.8 11.2 12.0 EB of branchings cannot exceed 2. As in the case of the 3.8 4.0 4.0 4.0 RN aromatic fractions, many methyl groups are connected to the naphthenic rings. This result may be ~ e r i f i e d ~ 8 * ~ ~ f r o m against the theoretical value 45% from the average the determination of the percentage of the prominent moleculeon Figure 7 calculated from eqs 17-22. Moreover, carbons CH; in the DEPT subspectrum CH2 and by this result is confirmed by the presence of an unresolved subtracting these peaks from the 22-60 ppm region; a value band between 20 and 22 ppm due to the methyl groups of 48% is then obtained for the naphthenic carbons CH2N bound to naphthenic rings. The same principle can be applied to the SAT I1 fraction the molecular formula of (48)Cookson,D.J.; Smith, B. E. Anal. Chem. 1985,57,864-871. which is nearly identical to that of SAT I. (49)Cookson, D.J.; Rolls, C.; Smith, B. E. Fuel 1989,68,'788-792.
z
z2 z3
628 Energy & Fuels, Vol. 8, No. 3, 1994
Conclusion The modeling of each bitumen fraction indicates that their chemicalstructure is quite homogeneous. The change from the SAT fractions to the ARO I fraction can be done by substituting two or three naphthenic rings to aromatic rings keeping more or less the same number and the same length of the alkyl chains. The change from the ARO I fraction to the resin fraction can be performed by increasing the number of naphthenic rings relative to that of aromatic rings together with the length of the alkyl chains whereas the opposite effect is observed for the asphaltene fraction. Finally, the difference between resins and asphaltenes is a question of greater pericondensation of the last ones. The more condensed and saturated character of the Bemolanga bitumen compared with American bitumens is probably due to their geological origin. The Athabasca and Orinoco bitumens belong to Foreland basins whereas the Bemolanga bitumen is situated in a Rift basin. As pointed out by D e m a i ~ o n the ? ~ longer distance migrations in the case of the Foreland basin generate a degradation of the oil by water washing and bacterial action. Water
Rafenomanantsoa et al. washing removes the more water-soluble, light hydrocarbons, especially the aromatics, whereas biodegradation removes preferentially normal paraffins. This difference explains that the Bemolanga bitumen, relatively protected, presents the longest alkyl chains and has a higher number of naphthenic rings than the Athabasca bitumen. Moreover, this conclusion is confirmed by the high level of the sulfur content in American bitumens which are tied to the more bacteria-resistant heavy and complex cyclic organic compounds. All these considerations show that the Bemolanga bitumen could be exploitable and a study on the oil obtained by a postcombustion pyrolysis process is in progress. Acknowledgment. The authors are grateful to the CNRS-PIRSEM and to the Cellule des Relations Internationales of the university of Nancy I for financial support. All NMR spectra were recorded on the spectrometers of the NMR Center of the University of Nancy I. The authors thank Mrs. Eppiger and Mr. Fringant for their technical assistance.
