Table I. Isotopic Content of COz from sec-Butyl Phthalate Column Sample a Mass 46/(44 45) ratio Enriched standard CO, 0.0285956 Normal standard CO, O.Oo4047 Enriched standard Cot 0.028614 Enriched standard CO, at normal flow rate 0.028555 Two samples of normal standard CO, ... Enriched standard CO, 0.028556 Samples listed in the order passed through the column. Ratio before treatment on column: enriched standard = 0.028595; normal standard = 0.004031.
With a helium flow rate of 30 cc per minute and the column at room temperature, the retention times relative to N2 as 0 were: COZ,0.5 minutes; ( C W , 4 minutes; S a , 9 minutes. Thus the COzis well separated from these impurities and it is quite easy to trap CO, free from SO2 and (C-. These examples suggest that in general the retention time of the gas on the phthalate ester columns depends on the boiling point of the gas, and may serve as a guide for applications involving the removal of other nonvolatile impurities from carbon dioxide. The impurity HCI was not encountered in any of our samples either as a peak in the gas chromatogram or in the mass spectrum. It is not known if it was absent in the pyrolyzed samples or whether it may have been removed'by reaction with the column materials--e.g., copper tubing. However, if HCl is a major impurity in other samples, it might be difficult to remove from COZon the sec-butyl phthalate column. The HCI could be conveniently removed by passing the sample over zinc amalgam as originally suggested by Anbar and Guttman (2).
using C-22 firebrick solid phase and a silicone oil liquid phase, employed a similar method of C02 purification but report no details on the accuracy of the results. Heating oxygen-containing compounds with Hg(CN)1 is a common method of converting the oxygen to COz (2). It was found that the cyanogen impurity produced by this method could be readily separated from the desired COZ product. Also SO2 and HzS produced from the pyrolysis of benzidine sulfate with HgClrHg(CN)9, can be readily separated from the COz by chromatography on the secbutyl phthalate column.
The authors gratefully acknowledge the helpful discussions and encouragement of Henry Taube. The gas chromatographic unit was made available by the Center for Materials Research of Stanford University.
(2) M. Anbar and S. Guttman, J. Appl. Rad. Isofopes, 5, 223 (1959).
RECEIVED for review November 14, 1966. Accepted March 9,1967. This work was supported by the A. E. C. under contract No. AT(04-3)-236, Stanford University.
Analysis of Complex Mixtures of Aromatic Compounds by High-Resolution Mass Spectrometry at Low-lonizi'ngVoltages B. H. Johnson and Thomas Aczel Esso Research and Ehgineering Co., Baytown Chemicals Research Laboratory, Bay town, Texas
THECONTINUOUS EXPANSION and progress of petroleum technology requires more and more sophisticated means of characterizing the composition of petroleum products. Mass spectrometry has played a dominant role in this effort since the early nineteen-fifties. The introduction of commercially available high-resolution mass spectrometers in recent years h a s further extended the scope of the mass spectrometric technique. The use of low-ionizing voltages restricts the mass spectra of aromatic and olefinic materials to the molecular ions (2-3). The spectra are thus greatly simplified and interferences among ihe components of the mixture are reduced to a minimum. This permits the determination of the carbon number distribution within each compound type and, in general, makes the applicability of the method less dependent on sample origin and history than in the case of high voltage methods. The use of high-resolution instruments eliminates one .najor limitation of low-resolution low-voltage techniques,
that is, the inability to distinguish more than seven compound types. At a resolution of about 10,OOO, pairs of compound types such as alkylbenzenes and benzothiophenes, naphthenobenzenes and pyrenes, naphthenonaphthalenes and dibenzofurans, etc. are resolved and can be analyzed separately. The potential of the combination of high resolution lowvoltage technique was recognized by Lumpkin (4) and Reid and coworkers (5). The peak matching mass measurement (6) techniques applied by these authors is, however, extremely lengthy when applied to very complex mixtures, and is not applicable to the determination of minor and trace components. This paper describes a method for making mass measurements directly from the recorded m a s s spectrum which greatly reduces the time required as compared to the
(4) H. E. Lumpkin, ibid.,p. 2399. Reid, W. L. Mead, and K. M. Bowen, paper presented
( 5 ) W. K.
Field and S. H. Hastings, ANAL.CHEM., 28, 1948 (1956). (2) H. E. Lumpkin, Ibid.,30, 321 (1958). (3) H. E. Lumpkin and Thomas Aczel, Ibid., 36, 181 (1964).
(1) F. H.
at the Institute of Petroleum/ASTM Spectrometry Symposium, Paris, September 1964. (6) K. S . Quisenberry, T.T. Scolman, and A. 0. Nier, Phys. Reo., 102, 1071 f 1956).
instrumental peak matching techniques and which permits the identification of trace components. As a consequence, this technique is now iised to characterize petroleum materials on a routine basis. EXPERIMENTAL
The apparatus used in this work is an Associated Electrical Industries Ltd., MS-9 high-resolution instrument. This is a double focusing apparatus, of the Nier-Johnson geometry. The output of the instrument is recorded with a MinneapolisHoneywell UV visicorder. The MS-9 is essentially a qualitative instrument, and its use for quantitative purposes requires a careful standardization of the operating conditions. The main variables are the position of the magnet, the electron beam and repeller voliages, ion beam fociising, and the widths of the source and collector slits. lnterpretation of charts requires a repeatable spectrum which is adequately resolved for the whole mass interval investigated, usually from mje 600 to m/e 100. This can best be accomplished by optimizing the position of the magnet at the high end of the mass interval to be studied. This is necessary because the resolution decreases rapidly above the mass at which the magnet is optimized, but only slowly below. In addition, one obviously needs the highest resolution at the high mass end of the spectrum. The effective electi-on beam energy is kept constant at 12.0 electron volts which essentially eliminates fragments from the types of compounds likely to be present in petroleum fractions. This effe,:tive energy is due both to the electron beam voltage and the repeller voltage, according to the expression : Effective V
E,lectron beam V
Any deviation of the repeller voltage from zero, desired sometimes for increasing sensitivity and repeatability has to be compensated by a corresponding change in the electron beam voltage . The ion beam focusing is optimized at the effective electron beam voltage used in each experiment. A set of typical instrumental conditions is: source slit width, 0.0018 inch; collector slit width, 0.0016 inch; effective electron voltage, 12.13electron volts; multiplier gain, 4400; scan speed, 8.7 minutesjoctave; chart speed, 0.4 inchj second; and sample size, 3-10 ma. A resolution of about 10,000 is obtained at these conditions. The maximum mje ratios, at which the doublets usually encountered in petroleum can be completely separated with a resolving power of about 10,OOO are listed. Maximum m/e at which doublet
is resolved AM
with 10,ooO resolving power
0.0905 0.0364 0.0165
'3CH-N C r *SH,
Doublet c-12w GH?,-a*S CHI-0
364 165 81
In effect, this resolution allows the identification of the first three doublets at masses higher than listed because partially resolved peaks can also be used. On the other hand, going to much higher resolving powers would reduce the intensity of the peaks without significantly improvirig the resolution of the last two doublets. The total ionizatio?, expressed as the sum of the monoisotopic peak heights of all parent peaks is about IO to 20
Figure 1. Portion of a high-resolution low-voltage spectrum of a petroleuni mixture
thousand millimeters. Base line noise is about 2 to 3 mm, and thus individual components can be detected in amounts of 0.05% or less of the total ionization. A portion of a typical spectrum obtained under these conditions is illustrated in Figure 1. DISCUSSION
Chart Reading Technique. Mass measurement from the recorded spectra is based on the relation between the masses and the position of the corresponding peaks on the record. This relationship is expressed by the following equation:
where Mx = Ma, Mb = ib - Ia = I, - iz = Ib - I, =
unknown mass known masses distance between Ma and M a distance between Ma and M , distance between Mb and M,
The relationship assumes a uniform logarithmic scanning rate and a constant chart speed. This was shown experimentally to be valid only for mass intervals of less than four. Thus for a low voltage spectrum covering several hundred integral mass numbers, a large number of individual external standards would be required. However, this difficulty can be circumvented in petroleum and similar materials which consist of homologous series of compounds. In these cases throughout the spectrum there are generally one or more parent peaks at every even mass number followed by their 13C isotopes. The distance between the pairs corresponds to the known mass difference AM('3C - IZC)= 1.0034 and therefore provides internal mass-distance standards at every second mass interval. Over these short intervals a linear relationship is assumed between mass and distance. It is furthermore assumed that M ( H ) = AM(13C - l2C) = AM(**CH2- 13C). The error introduced by the latter assurnp tion, (of the order of 0.004 amu), is partially compensate
Table I. High ResoiutiokLow Voltage Analysis of the Aromatic Portion of a Virgin Distillate (4lXWSO"p)
12.31 16.35 13.55 25.36 10.83 6.33 5.09 1.24 1.01 0.12 1.91 0.63 1.49 0.26 0.16 0.13 0.02
mol. wt. 180.2 198.8 218.3 184.8 219.1 234.1 232.6 266.6 282.0 305.3 240.9 298.6 249.5 290.2 331.5 293.9 365.0
0.38 0.40 0.26 1.93 0.20 0.10 0.04
206.2 202.5 212.6 208.0 271.4 251.7 278.0
Typical compound type Benzene NaphthenoDinaphthobenzene Naphthalene Naphthenonaphthalene Fluorene Phenanthrene Naphthenophenantb Pyrene ChrySene
Benzothiophene Dinaphthenobenzothiophene Dibenzothiophene Naphthenodibenzothiaphene Fluorenothiophene Phenanthrenothiophene Naphthenophenanthrenathiophene Naphthenodienofuran Benzofuran
Naphthenobenzofuran Dibenzofuran Naphthenodibenzofuran Phenanthrenofuran Naphthenophenant b o furan
provides the initial mass-distance calibration. The largest peak at mje 260 is a distance of 2 Io from the G&,. This peak is either CyoHfo,corresponding to the addition of two
hydrogen atoms, andior CI7H2& which cannot be resolved from C,oH,o. The peak of lowest mass at m/e 260 is separated from C,oH,o/C17H~rS by 0.04 Io which is equivalent to AM(0 - CH.,) = -0.036. Thus the peak is attributed to Clr H:,O and/or CI6H&O with the former being the most probable. The highest mass peak at nile 260 is separated from C,oHyoiCIjH,,S by 0.1 lo AM(12H - C) = f0.094. This peak can only be C19H32since C16HJ+Sdoes not exist. The procedure is continued for the next set of peaks at 262 using the l.CC.?oHL.o peak as the known and so on throughout the spectrum. The simplicity of the identifications is due to the fact that most of the hetero-compounds in petroleum contain only one heteroatom, usually either sulfur, oxygen, o r nitrogen. The parent peaks of compounds containing one nitrogen atom cannot be resolved from '3c isotope peaks of aromatics above about mje 100 whenever the mass difference is 4M( ' V H - I4N). In this case the nitrogen compounds can be identified and the peak intensities obtained by isotope corrections. Thus all major types can be determined quantitatively except those corresponding to the pairs which differ in mass by A M ( 5 H 4 - I C 3 ) . Determination of the latter compound types can, however, be partially achieved since the molecular weight of the first homolog of the sulfur types is 42 or 56 unit masses iower than that of the interfering hydrocarbon type, e.2.
Thus the first three or four homoiogs of a d u r compound type can be identified unequivocally and provide a basis for estimating the amounts of the higher molecular weight sulfur types in each Series. The technique can be corroborated by the obsemtion of the '5 isotope peak of abundant sulfur types. The doublet AM("G can be resolved completely up to m/e 165 with a resolving power of 1/1O,ooO. In cases where there are peaks in the spectrum which do not fit the above discussed simple pattern, more accurate mass measurements can be made from the chart by bracketing the unknown peak between the two nearest known peaks and using a ruled magnifying glass. The accuracy of this type of measurement is usually of the order of 10 to 20 p p m The method described in this paper has been used for the detailed characterization of a wide variety of petroleum materials. The analysis of the aromatic portion of a virgin distillate is reported in Table I. The amounts of sulfurantaining types listed in this analysis were calculated using only unequivocally identilied sulfur peaks. This procedure resuits in low sulfur content, but the deviation is very small.. Elemental sulfur calculated from mass spectral data was 0.57% against 0.65x found by x-ray fluorescence. The calibration c&cients used to calculate the data reported in Table I were obtained by dividing the relative sensitivity of the aromatic nucleus in a given series by the ratio of the molecular weight of the peak to the molecular weight of the nucleus. sensitivity data for the aromatic nuclei were derived from experimental data, wherever available, and by extrapolation techniques assuming that-.the nuclear sensitivitiesare directly proportional to the number of double bonds in the structure. Most of the experimental sensitivities were those obtained previously on a Consolidated E3ectrodynamics Company Model 21-103C instrument (3). it has been shown in these laboratories and by Reid (5) that the relative sensitivities on the two instruments are interchangeable. The calculations are carried out on a computer. The data output includes the weight per cent corresponding to each individual peak, the weight per cent of each compound type, the average molecular weight of each cornpound type, and the average number of C atoms in side chains. Elemental S and 0 content is also calculated. As evidenced from Table 1, some of the oxygenated and suifur compounds found have not been previously identified in petroleum materials. i n addition, compound types as condensed as CnHZI-,4, CnH2.-& and C.H,,-390 have been identified in higher-boiling materials. The d c i e n t of variation on replicate analyses is about 10% of the amount present and about 2% of the average molecular weight for compound types present in quantities higher than 1 of the total sample. Coochsion. Techniques have been developed which permit $he routine application of high-resolution mass spectrometry to the characterization of complex aromatic petroleum mixtures. The instrument time required for identification is reduced from weeks to about two hours, as the timeconsuming peak matching technique is eliminated. The data work-up requires only about six hours. Other features of the technique inciude simultaneous identification and intensity measurement, and analysis of trace compounds. The techruque is most applicable to petroleum or similar &materialswhich consist of homologous series, and contain only hetemcompounds with one N, S, or 0 atom. The
method can be readit? adapted to automated data acquisition svctem with a tnnsquent additional reductior! in data handh?.
J. H . Harding who was respnstble for the maintenance of t h instrument and D. E. Allan for assisting with the computer pr0gEiW.
The authors thank H. E. Lumpkin for many nelpfui s u w rions and G. F.. litylor and J. L. Tayior for arrying out most of the experimental work described in this paper; also
RECEIVED for review August 31, 1966. Acsepted February 24, 1%7. Presented at May 1966 Meeting of ASTM Committee E14 on Mass Spectrometry, Dallas, Texas.
Application of Deposited Thin Metal Films as Optically Transparent Electrodes lor internat Reflection Spectrometric Observation of
Electrode Sdution interfaces SIR: The feasibility of using in$ernal refkxtance specprometry @Its!as a nlethod for monitoring electrochemical reaci.ions spectrophotometrically at the electrode surface has k n demonstrated previously (I, 2). Since ihat time, it h x been found that it would be highly advantageous to produce IKS crystal electrodes which have surfaces with better eiectrochemic21 characteristics than those Llsrd previously-i.e., those which employed doped tin oxide in the visible range of‘ ihe electromagneric spectrum (I) and germanium in the infrared region (2). The semiconductor properties of tin oxide and
germanium make these materials quite unsuitable for many experiments as their electrode properties are quite complex ( 3 , 4 ) . This, “rerefore, makes the characterization of this combination eiectrolysiss~trophotonletri~technique exrremrlv dirliciilt. Experimentalli;, it has been observed t h ; ! ~ the ebsorption base line of tin oxide coated glass eiectrock daes noi ixmain consttant from one electro1,ysisIZI t:ir next (2;4) Clther investigators have also cbserved the same changes in tl1.r: opiicai properties OR electroiysis (5> 6 ) shich in55ates that the snrra:.: i s cirangir.~ IP, sxme way. k-or t n t w :asxi:;, thc possi%lhtyof using thin pktirwni SIt? p;j:&ii:m meid :ii;ns OR opticaliy ti-ansparerrl substrate.> v.hi~:-! a x riren 0 2 empioyed as IRS e i e c t m k ~has b e s
sufficiently high temperatures. It should be noted that this material also contains gold, as well as other metais, so tk electrode is not pure pl3:inurn To lessen the viscosity of the availabie iiquid plat-inum (which enables a thin platinam film IO be produced). a smali amount (0.1 to 0.2 mi, depending on the thickness of film desired) is dissolved in 3 ml of dichloromethane, and the resulting soiution is painted on to the IRS plate using brush strokes parallel to the light path. Several coatings may be applied, again depending o n the thickness of film desired. The plate is then allowed to cure in air for about 1 hour. a: room temperature. It is then fired in an open oven for a b u i 4 hours. The firing temperature for borosilicate giass plares is between 650”C and 680” C (approximately the fusion point of the glass). The cell was essentially the same as employed by hiansen et a!. (I, 6:. Constar?t current chronopotentiometry was carried out on the previously characterized (I) o-tolidine system. The a g paratus is similar to that described earlier (!, 7). A 55.V solution oi’ o-toiidine in a pM 2.00 KCL-HC! buHer was piare.; in the reacticja 41, and on eiccrroiysis the f d i o w i In i , ~7t‘aiC?!on
3 5 like I(> report a relativeiy ranid and inexper rnerrod of prrxloc;np buck f i l m s and present some preiimi eiei:trour optis;;i c.:sractenstics oi these elecirodcs EXPERUMkXT.41.
Platioum G b IRS E h o d e . The method o i producing the platinum giass 1R.F eiectrodtt employs the use or ii sohtiun of an organic ligand complex of platinum, (Liquid Platinum No. :Lp Engelhard Industries Inc.. Hanovia LiqEid Goid Divisioc; East Newark, N.J.) which is readily reduceci to platinum mtal or, an inert substrate, such as glass, at ___
( I ) W. S . Hansen, K A. Osterywng, and T. Kuwac8.J. Alr. C/ieni. S O C . , 88, 1062 (I!&>>” 12) ;-1. 8. Mark, Jr., and 3.S . Fons, ANAL.CHM.,38,119 (1966:. i i) “The Elecfrochec&xry 0:‘ Semiconductors,” Y. 3. Ho1mL.s. -.1., Academic pres:;, N% York. 19t.z. :4t 3. 5. Roc.;, L.. 0. Winstron. J . Mattson. and H. ’d. Mark, :;. unpubiished data. 9M. :Si 7 . kuwane. Cqerniswy De& Case institute of Techoiup,
Spec19 I is wlo-less In the visible range, but II atnorb? < + The. Cary Model IJ was set at t h ~ bwave 4380 -4 (A&) length and srrnultaneuus absorbance-time and potenual-tme plots of tne oxiciatior reaction were carried mt Tne resuit, are sh;twn :n figure Thristie ( 8 ) and Hansw et ai. ( l j have shown fur iorwzr? .
unpublished data, 15C.f.. ( 5 ) i? A . Osreryoung. 24onIi American Avia-;ialionCo.. Ine.. Ttou-
VOL 39, NO. 6 , MAY 1967