Spectrometric identification of organic compounds

---

Robert W. Silverstein ond G. Clayton Bassler

Stanford Research Institute Menlo Park, California

Spectrometric Identification of Organic Compounds

Consider the plight of the chemist confronted mit,h a few milligrams of a completely unknown organic liquid which he has isolated by means of gas chromatography from a complex mixture. Gas chromatography, though fantastically successful as a tool for isolation, affords practically no help in identification, and has the further characteristic of being most effectivewith small samples. Let us look at the classical procedures in which the analyst has been trained. First, he smells the liquid. Then he carries out a fusion and qualitative tests for the elements. This is followed by a series of chemical reactions designed to establish a class based on a functional group, Solubility tests,.a boiling point, and a refractive index narrow down the possibilities to several in one of the standard tables. A crystallme derivative, whose melting point agrees with the literature and is not depressed by an authentic sample, usually serves for "conclusive" identification. If no satisfactory fit can he obtained, recourse is then had to combustion analyses, to a molecular weight determination, and to degrada tions. I t takes a skilled and determined analyst to get by on much less than 20 milligram of a compound that has been described in the literature. A good deal more may he necessary for a compound that has not been described. Infrared spectrometry can probably now be considered as a classical tool, and it, of course, saves much chemical probing to establish functional groups. Actually the classical methodology is an extremely useful device for teaching laboratory skills and organic chemistry per se, and students develop their first "feel" for organic chemistry in the qualitative organic analysis course. But, in contrast with methods now available and used in research hut not systematically taught in universities, the techniques just described are pretty feeble. Over the past few years, we have been engaged in isolati~lgsmall amounts of organic compounds from complex mixtures, and identifying these compounds spectrometrically. At the suggestion of Dr. A. J. Castro of San .Jose State College, we developed a one unit course entitled "Spectrometric Identification of Organic Compounds," and presented it to a class of first year graduate students and industrial chemists during, the 1962 spring semester. The invitation to teach a t San Jose State College was extended by Dr. Bert & Morris, I. Head, Department of Chemistry, who kindly arranged the administrative details. A book1 bearing the same title as this paper has evolved from the material gathered for the course.

' SILVERSTEIN, R. M., AND BASSLER, G. C., "Spectrometric Idendification of Organic Compounds," John Wiley and Sons, Inc., New York, in press. 546

/

Journal of Chemical Education

Our presentation was based on our experience in identifying organic molecules by their response to four energy probes; the responses of the molecules are recorded as spectra. We are concerned with the following kinds of spectra: mass, infrared, nuclear magnetic resonance, and to a lesser extent, ultraviolet. Only a rather modest level of sophistication and expertise in each of these areas of spectrometry is required to solve a gratifying number of identification problems. Extension of the methodology from identification of rather simple compounds, about which little or no informa tion is available, to elucidation of structural details of complex molecules, about which quite a bit is known, should be obvious. A higher level of competence in any of the areas of spectrometry is probably best achieved in connection with specific research problems. In a large number of cases, identification of a completely unknown compound can be made from mass, infrared and ultraviolet spectra obtained on a tenth of a milligram or less; and all except the mass spectrometer sample can be recovered. If a milligram sample is available, the unique data afforded by a nuclear magnetic resonance spectrum can be obtained. These data extend the range of identifications manyfold; again, all save the mass spectrometer sample can he recovered. The orientation throughout the course was on the rationalizations involved in translating spectra into chemical structures. We dealt only with identification of pure organic compounds. "Pure," in this context, is a relative term, and all we can say is:. the purer, the better. Probably the ultimate practical criterion of purity (for a sufficiently volatile compound) is chromatographic homogeneity on two capillary columns (several hundred feet in length), one containing a nonpolar substrate, the other, a polar. Another test is effusion through the micro-leak of a mass spectrometer. Various forms of liquid phase chromatography (adsorption and liquid-liquid columns, paper, thiilayer) are applicable to relatively nonvolatile compounds. All of the spectra presented in the course were obtained on samples that were purified by recrystallization to constant melting point or by gas chromatography. There is one limitation to the methodology we espouse. A mass spectrum is dependent on a degree of volatility and of thermal stability. And since mass spectrometry is our primary tool, this limitation can be serious. However, mass spectra have been obtained on a large number of high molecular weight compound, e.g., steroids, terpenoids, and alkaloids. Techniques for inserting a sample directly into the ionizing beam of the mass spectrometer promise to overcome the limitations of volatility and stability.

The problem of cost of necessary instrumentation will be raised, and answered by pointing to the amazing evolution of commercial instruments. The time saved, the smaller sample required, and the information made available far overbalance the cost. Infrared and ultraviolet spectrometers have been developed beyond the stage of reliable instrnments in the hands of a trained technician. They are now cheap, rugged, and simple enough to be used as a bench tool by the organic chemist. A nuclear magnetic resonance spectrometer is still a fairly expensive, complicated instrument that requires the service of a trained technician. Even here, the recent reduction in cost (by about one-half) and complexity indicates the trend toward use by relatively unskilled personnel, backed by a network of factorytrained servicemen. Almost from its inception, the utility of nuclear magnetic resonance spectrometry to the organic chemist has been evident, though its utility in identification in combination with mass, infrared and ultraviolet spectrometry has not been stressed. Mass spectrometry has had a somewhat different history. Developed by the physicist and utilized extensively by the petroleum chemist, it has been ignored almost completely by the organic chemist concerned with identification and structure determination. Even today there is only a handful of laboratories in which the application of mass spectrometry to these problems is appreciated. I t is still an expensive, very complex instrument which requires considerable skill in its use and maintenance. And yet, as we shall show it is nndoubtedly the most powerful tool of the four we use. The sequence of spectra chosen would depend on circumstances. If less than a tenth of a milligram is available, infrared and ultraviolet spectra can he run in solution in rnicrocells, the solvent removed, and a mass spectrum obtained. Avolatile sample from which solvent removal is difficult could be handled in an infrared gas cell and transferred directly to the mass spectrometer. The course given a t San Jose State College consisted of 11 hours of lectures and 3 hours of discussion. No laboratory work was involved. Three hours of lecture were devoted to mass-, IR-, and NMR-spectrometry. Two hours were devoted to ultraviolet, spectrometry. The discussions were concerned with translation of sets of spectra into organic structures; ten sets of spectra were covered. The final examination (open book) consisted of the translation of a single set of spectra. Our lectures on IR-, UV-, and NMR-spectrometry require no elaboration here. A brief summary of our presentation of mass spectrometry to the organic chemist may be of interest, followed by discussions of two sets of spectra. A bibliography of selected general references is appended. Mass Spectrometry

A mass spectrometer bombards the substance under

molecular radical ion (parent ion) is formed by interaction with the beam electrons. In effect, a single electron has been removed from the molecule. I t is important to he able to recognize the parent ion because it gives us the molecular weight of the compound. This is an exact numerical molecular weight, not merely the approximation obtained by all other molecular weight procedures familiar to the organic chemist. The parent peak is usually the peak of highest mass except for isotope contributions. The mass spectrometer is usually operated a t an electron beam energy of 70 ev, and under these conditions, numerous fragment ions are formed. A presentation of the masses of the fragment ions (including the parent ion) versus their relative concentrations constitutes the mass spectrum of the sample. The largest peak in the spectrum, called the "base" peak, is assigned a value of loo%, and the other peaks are reported as percentages of the base peak. The mass spectrum is presented as a table of two columns: one headed mle; and the other, % of base peak. In a separate table, the parent peak is set at 100% and the isotope peaks are reported relative to the parent 1 peak. In most cases, these will be the parent and the parent 2 peaks. Parent 4, parent 6, parent 8, etc., peaks are given for compounds whirh contain several chlorine or bromine atoms. These isotope peaks arise, of course, because a certain number of molecules contain heavier isotopes than the common isotopes. The mass spectrometer can distinguish between the molecule containing, for example, the I2Cisotope and the molecule containing the 13C. We shall limit ourselves to compounds containing C, H, 0, N, S, CI, and Br. Suppose a compound contains a single carbon atom. Then the parent 1 peak will be about 1.1% of the intensity of the parent peak. If a compound contains one sulfur atom, the parent 2 peak will be about 4.4% of the parent peak. The first step in the identification of a compound is to attempt to establish an empirical formula from the parent mass and the isotope contributions. We shall not always succeed. In some cases, the parent peak is so small that the isotope contributions cannot he accurately measured. We settle for the molecular weight. In some cases, the parent peak may be missing; we then try to establish the molecular weight from the fragmentation pattern and from the other spectra a t hand. In some cases, we may resort to preparation of appropriate derivatives, or to other methods of obtaining molecular weights. Selection of likely empirical formulas is greatly facilitated by a table constructed by Beynon. The table is limited to compounds containing C, H, 0, and N. The presence of S,C1, or Br is usually readily apparent

+

+

+ +

+

+

+

Principal Stable Isotopes and Relative Abundances Isotopes

Percent of isotope of lowest mass

investigation (in the gas phase) with an electron beam, and records the damage as a spectrum of positive ion fragments and their relative abundance. The positive ions are separated on the basis of mass. The usual sample size ranges from several milligrams to less than 0.1 milligram. At an electron beam energy of about 9 to 15 ev or so, dependmg on the molecule involved, a Volume 39, Number 1 1 , November 1962

/

547

Infrared Spectrum FREQUENCY ( c M - ' ~ 0

C e l l thickness 0 . 0 2 mm.

WAVELENGTH IMICRONSI

Mass Spectral Data (Relative Intensifier)

m/e

27 38 39 11 42 43 50 51 52

62 63 64 65 77

7, of haae peak 6. 5. 20.

4. 3. 73. 11. 23. 6. 3. 9. 3. 18 22.

m/e 77 78 79 89 90 91 !I2 105 106 107 108 109 150 151 152

% of base peak 6.

26. 13. 47. 71. 6. 6. 3. 20. 100 8. 28.70 2.84 0.26

NMR Spectrum

PPM Figure 1.

548

/

Data for benzyl acetate.

lournol o f Chemkol Education

Isotope Abundances

22.

(6)

% of I'

m/e 150 (P) 151 (P 1 ) 152 (P 2)

100. 9.9 0.9

+ +

Ultraviolet Data h ~ l ~ t r m=

268 264 202 257 252 248 (shoulder) 243 (shoulder)

~II,",

101 1.58 1-17 101

153 109 78

Infrared Spectrum FREQUENCY (cM-')

C e l l thickness 0.01 mm.

WAVELENGTH (MICRONS)

Mars Spectral Data (Relative Intensities)

% of base pcnk 11. 33. li.

28. 64. 9.

12. 8.

m/e

46 47 48 49 50 57 58 59 GO

Isotope Abundance6 %of P

% of base peak 25. 100. 65. 10. 3. 6. 11. 36. 98.

m/e

Ultraviolet Data AECO'

","'

r,.

232 ( i d )

infl.

=

136

inflection

NMR Spectrum

Solvent CDCI,

PPM Figure 2.

( 6 1

Dmta for 2-mercaptoothmml.

Volume 39, Number 1 1 , November 1962

/

549

from a large parent +2 peak. The number and combinations of chlorine and bromine atoms can be determined by means of another The procedure for obtaining an empirical formula will be demonstrated as we work through the sets of spectra presented below. We found it advisable to remind students of the "nitrogen rule": an odd-numbered molecular weight permits only an odd number of nitrogen atoms, and an even-numbered molecular weight permits only an even number of nitrogen atoms (including zero). Now let us consider the fragmentation pattern. A number of general rules for predicting prominent peaks can be written and rationalized using concepts of statistics, resonance, hyperconjugation, polarizability, and inductive and steric effects. For example: 1. Cleavage is favored at branched carbon atoms 2. Aromatic compounds generally give a larger parent peak than do aliphatic compounds 3. Double bonds favor allylic cleavage 4. Saturated rings lose side chains a t the or-carbon; special case of branching 5. In alkyl substituted aromatic compounds, cleavage is most probable a t the bond beta to the ring 6. A heteroatom will induce cleavage a t the bond beta to it. A feeling for these modes of cleavage, plus a reference library, form the basis for use of mass spectrometry for identification purposes. Identification is complicated, however, by rearrangements. These, too, can usually he rationalized on the basis of low energy transition pathways, and increased product stability. Examples

We are now in a position to interpret the first set of spectra (Fig. 1). The first step in translating the set of four spectra in Figure 1 into a molecular structure is to establish an empirical formula. The parent peak in the mass spectrum is 150; thus we have the molecular weight. The parent peak is an even number. We are therefore permitted either no nitrogen atoms or an even 2 peak obviously does not number of them. The P allow for the presence of sulfur or halogen atoms. We now look in Beynon's table under molecular weight 150. We are faced with twenty nine empirical formulas of molecular weight 150 containing only C, H, N and 0. Our P 1 peak is 9.9% of the parent peak. We list the empirical formulas whose calculated 1 peak falls-to be isotopic contribution to the P arbitrary-between 9.0 and 10.7 (our isotope contribution peaks are often somewhat higher than the calculated values) : we also list their P 2 values. We immediately eliminate three of these formylas because they contain an odd number of nitrogen ato@s. The P 2 peak is 0.9% of the parent; this hestfits C8Hlo02which we shall tentatively designate as kur empirical formula. We make a mental note that both the intensity of the parent peak and the C to H ratio of the empirical'formula indicate aromaticity. We turn now to the infrared spectrum and note thkl C=C band a t about 1730 cm.-' (5.76 p). This,together with the presence of two 0 atoms in the empirical formula, suggests an ester. We look for con-

+

+

+

+

+

BEYNON, J. H., "Mass Spectrometry and ita Applications to Organic Chemistry," Elsevier Publishing Co., Amsterdam, 1960.

550

/

Journal o f Chemicol Education

firmation in the C-O-C stretching region and note the large broad band at about 1225 em-' (8.15 p) characteristic of an acetate. Two large bands a t about 740 cm-' (13.5 p) and 700 cm-' (14.3 p) suggest a singly substituted benzene ring. Further evidence may be adduced from the aromatic C-H stretching peak at 3065 cm-' (3.28 p) and from the ring-stretching peak a t 1502 em-' (6.65 p). We have tentatively established the presence of a bennene ring, and are quite confident about the acetate group. Furthermore, we note from the position of the carbonyl band that the C=O moiety is not conjugated with the ring. This is confirmed by the wavelengths and intensities of the ultraviolet absorption peaks, which also eliminate a ketone from considerat,ion. If we subtract a singly substituted benzene ring and an acetate group from the empirical formula we obtain Empirical formula: 0

CsHmO~

I1

CeHs

+Remaining: CH&O:

CAO, CHP

It takes no great imagination to insert the CH2between the ring and acetate group, and write heuzyl acetate

The NMR spectrum provides almost conclusive confirmation for the above structure. We see three sharp unsplit peaks in the following positions and with the following integrated intensities. 7

6

Intensitv

The five protons a t r 2.75, 6 7.25 are the five benzene ring protons. The singlet of two protons a t r 5.00, 6 5.00 represents the methylene group substituted by a phenyl and an ester group. The singlet of three protons at T 7.98,6 2.02 represents the methyl group. We can obtain additional confirmation by returning to the mass spectrum and considering the fragmentation pattern in view of the information at hand. The base peak a t 108 is a rearrangement peak representing cleavage of an acetyl group (43) and rearrangement of a single hydrogen atom. The large peak a t mass 91 is the benzyl (or tropylium) ion formed by cleavage beta to the ring. The large peak a t mass 43, of course, represents the acetyl fragment. The peaks at 77, 78, and 79 are additional evidence for the benaene ring. We can state with a high degree of confidence that t,he compound is benzyl acetate.

Infrared Spectrum 10000

5000

3000

2000

1600

1400

FREQUENCY ICM-'1 1200

1000

900

BOO

700

750

650

00

W 0

*mz 0.2 e

0 Y)

$

04 0.6 0.8 1.0 1.5 I

2

3

4

5

7

6

8 9 WAVELENGTHIMICRONSI

II

10

12

13

14 15 Cell lhlchnerr 0.01 mm.

Mass Spectral Data (Relative Intensities)

7,of b s e per&

Isotope ~bundanees

3. 13. 10. 12. 29.

mle

% of P

150 ( P ) 151 ( P 1) 2) 152 ( P

100. 10.2 0.88

+ +

Ultraviolet D d a X"OA mnz

229 272 280

log

em.

4.08 2.90 2.85

NMR Snectrum 300

200

CPS

PPM Figure 3.

I61

Dot. for unkmxn.

Volume 39, Number

I 1, November 1962

/

551

There are a number of other sequences through which we might arrive at the identity of this compound. Having established the empirical formula, we could note at once the characteristic benzene ring peak a t r 2.75, 6 7.25 in the NMR spectrum. We could confirm this by the typical "benzenoid" fine-structure absorption in the ultraviolet spectrum. The base peak in the mass spectrum is treacherous because it is a rather unusual rearrangement peak, but the mass 91 peak immediately calls to mind the benzyl (or tropylium) structure. The large mass 43 peak strongly suggests the CH,CO group, in view of the C=O peak in the infrared. Subtraction of a benzyl and an acetyl group from the empirical formula leaves a mass of 16; consideration of the infrared spectrum leaves very little question as to how to handle this oxygen atom. The student will find it instructive to write the possible isomer structures, and to eliminate them on spectrometric grounds. One more example will be given to point up other aspects of the general procedure. Let us consider the set of Fpectra given in Figure 2. We immediately note the large P 2 peak which suggests that one sulfur atom is present. We then list the possible empirical formulas under mass 46 (7832). We should list all except the trivial formulas and those containing an odd number of N atoms. We also subtract the 33S contribution (0.78) from the P 1 peak. But this turns out to leave us with only a single choice, CpHO. The empirical formula, therefore, is C2HsOS. The infrared spectrum shows a strong, rather broad band a t 3367 cm-' (2.97 p ) . Our impression is that we are dealing with an alcohol, and the very broad hand at; about 1050 em-' (9.5 *) suggests a primary alcohol. Our attention is then caught by a rather weak band a t 2558 cm-' (3.91 r r ) which practically spells out a mercaptan group. In this case. the infrared spectrometer is a t some disadvantage with respect to the nose. Had this been a thin film spectrum, we might have missed the S-H stretching band. We now have the fragments: CH20H and SH. This only leaves a CH2 group to fit in; and we write HOCHZCHQSH,2-mercaptoethanol. Some of the major fragmentation peak^ can be assigned as follows :

downfield a t a given concentration. Second, the OH proton will undergo rapid exchange under normal conditions and will usually appear as a single peak; the SH proton, under the same conditions, will not exchange rapidly (at least in a non-aqueous solvent) and the peak will be split by the adjacent methylene group. We see then that the upfield triplet, with slight second order splitting of the middle peak, represents the SH proton coupled with (i.e., split by) the adjacent methylene group; the coupling constant is 8 cps. The adjacent methylene group is split into a doublet by the SH proton (conpling constant, of course, is 8 cps), and again into a triplet by the other methylene group with a coupling constant of 6 cps. This is somewhat distorted A,M,X system with two coupling constants. An idealized diagram is as follows.

+

+

The low field peaks must then consist of a triplet with a coupling constant of 6 cps, and it must also contain the OH proton. We do indeed see a triplet with the upfield peaklet distorted by the peak of the OH proton with which it almost coincides. The proton peak can be shifted by change of concentration, solvent, or temperature. The final examination for the one unit course consisted of the set of spectra presented here in Figure 3. The interested reader can work through this problem using the same possible empirical formulas tabulated in the first problem given above.3 Acknowledgment

The authors are indebted to the Perkin-Elmer Corporation and to Stanford Research Institute for financial support, to Varian Associates for the NMR spectra, and to their colleagues a t SRI for the other spectra and for numerous helpful discussions. Bibliography of Selected General References Infiawd Spectr&ry BELUMY, L. J., "The Infra-red Spectra of Complex Organic Molecules," 2nd ed. John Wiley and Sons, Inc., New York

.,"-.

IOU

The NMR spectrum provides exhaustive confirmation for the structure written. It also shows a number of interesting features. The starting point is the distributions of protohs as shown by the integration curve. If we assume that the triplet a t the high field position contains one proton, then the next cluster of peaks contains 2 protons, and the low field peaks account for 3 protons. At first glance, this does not seem reasonable. But we must bear several things in mind. First, the position of the OH peak and the SH peak depends on concentration, and since the OH group hydrogen bonds more strongly than SH, it is likely to be further 552

/

Journol of Chemicol Education

Cnos4, .4. I)., "Introduction to Practical Infra-red Spectroacopy," hft~rworth Scientific l'uhlications, London, 1960. JONES.R. K..AS" JANUURPY. C.. 'The Ao~liearlonof Infra-red to 'the ~luffddationof Molecular and' ami in ~pectrometr~' Structure," in "Technique of Organic Chemistry," edited by A. WEISSBERQER,Volume 10, Interscience Publishera, New York, 1956, chap. 4, pp 247-580. Nuclear Magnetic Resmnee Speelronaetry JACKMAN, L. M., "Applications of Nuclear Magnetic Resonance Spectroacopy in Organic Chemistry," Pergamon Press, New York, 1959. ROBERTS,J. D., "Nuclear Magnetic Resonance Applications to Organic Chemistry," McGraw-Hill Book Co., Inc., New York, 1959. ROBERTS, J. D., J. CHEM.EDUC.,38,581(1961). A. L., J. CEEM.Enuc., 34,618(1957). RICATER, The compound represented by Figure 3 is ethyl benzoate.

UUraviolet Spedromclry GILLAM,A. E., AND STERN,E S., "An Introduction to Electronic Absorption Spectroscopy in Organic Chemistry," Edward Arnold (Publishers), Ltd., London, 1957. DUNCAN, A. B. F., AND MATSEN,F. A,, "Electronic Spectra in the Visible and Ultraviolet," in "Technique of Organic Chemistry," edited by A. WEISSRERGER, Volume 9, Interscience Publishers, New York, 1956, pp. 581-706. Spedromet~y BEYNON, J. H . , "Mass Spectrometry and its Applications to

&fa88

Organic Chemistry," Eleevier Publishing Co., Amsterdam. 1960.

BIEMANN, K., "MUS Spectrometry, Applications to Organic Chemistry," McGraw-Hill Book Go., Inc., New York, 1962. MCLAFFERTY, F. W., "Ma= Bpeetrometry," in "Debermination of Organic Structures by Physical Methods," edited by F. C. Volume 2, Academic Press, Nacrro~and W. D. PHILLIPS, New York, 1962, chap. 2. BIEMANN, K , Angew. Chem., internat. edit., 1,98 (1962). ELIEL, E. L., PROSSER,T. L., AND YOUNG,G. W., J. CHEM. EDUC.,34, 72(1957).

Volume

39, Number 1 1 , November 1962

/ 553