Table 1. Carboxylic Acid Content of a Serratia Marcescens 0-Antigen Preparation Determined by Different Methods
Ester micro equivaMax. lent in deviation Method of 1 mg. of in 10 detn. material detns., % Titrimetric 0.80 =I=8 . 4 Hydroxylaminolytic 0.93 221.2 BFa, total acids 1.19 f 3.8 BFs, long-chain acids 0.86 =k 5 . 0
ratia marcescens 0-antigen preparation determined by different procedures is shown in Table I. The transesterification described converts the carboxylic acids into their methylesters, without any respect to their chain length. The differentiation between acetyl and long-chain (C, and higher) carboxylic acids is based simply on the different volatility of their methylester derivatives a t elevated temperature. In preliminary experiments, sam-
ples of pure methyl acetate and methyl caprylate and their equimolar mixtures were dissolved in hexane and heated in open test tubes in a 67’ C. water bath. After different time intervals, the ester content of the tubes was quantitatively measured by the Snyder and Stephens procedure (8). The results (Figure 2) show that methyl acetate completely escaped from the open tubes in 60 minutes, but the amount of methyl caprylate was unchanged after 3 hours. This method makes possible simultaneous determination of volatile and nonvolatile carboxylic acids. When the hexane extract of the transesterification mixture is divided into 2 aliquots, one may serve for the determination of the total carboxylic acids. The other aliquot after 60 minutes a t 67’ C. will contain only the methylesters of the higher carboxylic acids. Naturally other acids between Cz and Cs will not be removed completely during 60 minutes at 67’ C. Their residual amount can be measured as long-chain
methyl esters. However, the amount of these acids in bacterial lipopolysaccharides is very low. Whether the method described herein can be applied for the determination of fatty acid content of other lipides and lipide complexes is under further inyestigation. LITERATURE CITED
R.,J . Biol. Chem. 170, 671 (1947). ( 2 ) Bloor, W. R., Pelkan, K. F., iillen, D. M., Ibid., 5 2 , 191 (1922). (3) Haskins, W. T., AXAL. CHEW. 33, 1445 (1961). (4) Hill, U. T., IND.ENG.CHEM.,ANAL. ED. 18, 317 (1946). ( 5 ) Kibrick, .4. C., Skupp, S. J., Arch. Biochem. Biophys. 44, 134 (1953). ( 6 ) Nowotny, A., J . A m . Chem. SOC. (1) Bloor, W.
83,501 (inn" YUI).
(7) Page, E .. Michaud., L.., Canad. J . M . Sc. 29, 239 (1951). (8) Snyder, F., Stephens, K., Biochem. Biophys. Acta 33, 244 (1959). RECEIVEDfor review June 11, 1962. Accepted January 2, 1963. Work supported in part by Grant E-3849 from the U. S. Public Health Service.
Limited Area Flame Spectrometry Chemiluminescence BRUCE
E. BUELL
Union Research Center, Union Oil Co. of California, Brea, Calif.
Limited area techniques are utilized to investigate the excitation possibilities of the oxyhydrogen flame atomizing organic solvents. Chemiluminescence, or inner cone excitation greater than expected from thermal excitation, is studied by correlating excitation potentials and dissociation energies with height of maximum emission above the burner tip and solvent-enhancing factors. Over 600 atomic lines, with excitation potentials varying from 2.1 to 9.0 e.v., were recorded during this study, many for the first time in flames. Only selected lines are presented in tables and representative spectra. The tables establish conclusively that as excitation potentials increase, solvent-enhancing factors increase and height of maximum emission in the flame decreases until that for CO and CH emission is reached. Chemiluminescence and chemical reduction in the flame are offered as possible explanations for this and as a major contributor for large solvent-enhancing factors (established as exceeding 40,000-fold).
372
ANALYTICAL CHEMISTRY
C
has been established as occurring in hydrocarbon flames and hydrogen flames containing added organic substances and has been discussed and defined by Gaydon (9) as electronic excitation produced directly by chemical reaction. Chemical reduction in the flame followed by thermal excitation will not be strictly differentiated from chemiluminescence in this study and the approach used is to simplify rather than delve into details of complex-multiple reactions. To simplify further, the excitations studied are those occurring in the inner cone, which appear to be greater than expected from the thermal energy provided by the oxyhydrogen flame. Recent reviews by Gilbert ( l a ) on chemiluminescence and Mavrodineanu (17) on flame characteristics and emission discuss the history and development in this field, which appears to date back to 1877 but has been little used and poorly understood. Referring to the inner cone, Mavrodineanu says, “The analytical flame spectroscopist has found little or no interest for this HEMILUMINESCENCE
particular region of the flame, except for the efforts he is making to avoid it.” Perhaps this is partly due to limited knowledge of the inner cone excitations and poor dissemination of such knowledge to analytical spectroscopists. Inner cone excitation or cherniluminescence can be very useful not only for exciting elements such as tin, zinc, and nitrogen (via CN bands) which are not excited adequately otherwise, but also for increasing sensitivity. For example, upon atomization of cleaner’s naphtha into the oxyhydrogen flame, the height of maximum emission for lead a t 4057.8 drcps from 35 mm. to 17 mm. above the burner tip (in the top of the inner cone) and emission is enhanced 18-fold. The purpose of this investigation is to promote the use of and a better understanding of chemiluminescence. Previous work in this laboratory (4) reported a simple modification of a Beckman Model DU flame spectrophotometer for limited area flame spectrometry in conjunction with atomization of organic solvents into an
Io
or
80
-
60
-
40
-
z
o_ v) v)
-
E w
m
-
N
N
0 a,
8
20 4
2100
Figure
2300
2200
1.
2400
2500
ANGSTROMS Spectra of carbon and fourth positive cobalt Sensitivity 4- 9.5 Slit 0.03 mm.
oxyhydrogen flame. Because the solvent employed was primarily cleaner’s naphtha, this flame will be called the oxyhydrogen - naphtha flame. The specific purpose of this investigation is to determine the excitation possibilities provided by this flame and to correlate excitation potentials and dissociation energies with distribution in the flame and solvent-enhancing factors. Limited area techniques (as previously described) are extremely valuable in establishing heights of maximum or “peak” emission in the flame and for improving line to background ratios. EMISSION DISTRIBUTION IN FLAME
As shown previously, upon atomization of organic solvents as compared to atomization of aqueous solvents (4, emission distribution in the oxyhydrogen flame becomes lower, centering around the reaction zone in the inner cone. The cobalt spectra in Figure 4 vividly demonstrate this variation in distribution of emission. The two adjacent lines marked 1 and 2 a t 3745.5 A. and 3732 m A. have excitation potentials of 4.2 e.v. and an average of about 5.3 e.v. and have similar intensities in naphtha. At a height of 18 mm. in the flame (Figure 4,B) line 1 is more
sensitive, but a t 13 mm. (Figure 4 4 ) line 2 becomes more sensitive. This agrees with the predicted response for chemiluminescent lines with peak emission lower in the flame for the higher energy line. Upon atomization of aqueous solutions, retaining the same instrument sensitivity (Figure 4,E, height 35 mm.), the intensity for line 1 drops 25-fold and line 2 is not detected. On increasing the slit from 0.01 to 0.03 mm. (Figure 4,C) line 2 is just detected. Reducing the height from 35 to 18 mm. (Figure 4,D) reduces intensity and shows that the distribution of emission for aqueous solutions is much higher in the flame than for naphtha solutions as previously reported.
Table 1.
Although the oxyhydrogen flame theoretically provides up to 5.0 e.v. (116 kcal.), lines with lower excitation potentials appear to be influenced by chemiluminescent reactions. The large enhancing factors observed for the cobalt lines above and peak emission lower in the flame indicate this. TO emphasize this trend, Table I presents a summary of enhancing factors v8. excitation potentials for magnesium and cobalt (including the lines above). The enhancing factors observed represent extremes, with those for magnesium lower than equivalent excitation potential lines for cobalt. As a further aid in differentiating thermal excitation from chemiluminescence (or anomalous excitation) distribution of emission in the flame for selected lines us. excitation potentials and enhancing factors is given in Tables I1 and 111. As a n arbitrary borderline 4.0 e.v. was chosen; lines rated higher than this are given in Table I1 and those lower in Table 111. Distribution of emission is indicated by the height in millimeters above the burner tip for the point of peak emission. Enhancing factors given as mindicate no detection in aqueous solutions. Measurements and calculations made for zinc a t 4810.5 A. show enhancing factors exceeding 40,000-fold. Because dissociation energies also influence chemiluminescence, and to present a complete picture of distribution and enhancement of emission, equivalent data for molecular band heads are given in Table IV. Each table is arranged approximately according to increasing height of peak emission. EXPERIMENTAL OBSERVATIONS
As can be seen by close examination of the tables, excitation potentials and enhancing factors decrease as height of peak emission increases. The zinc doublet a t 2801 A. (8.5 e.v.) peaks lowest in the flame and correlates closely with peak CO and CH emissions a t 12 mm. above the burner tip. The sodium doublet a t 5893 A. (2.1 e.v.) peaks very high a t 21 mm. and is enhanced only threefold.
Enhancing Factor vs. Excitation Potential
Magnesium Wavelength, A. Excitation potential Enhancing factor Wavelength, A. Excitation potential Enhancing factor 4 Multiple.
2852 4.3 2.5
5184 5.1 12
3337 6.4 m
Cobalt
3595 3.6 13
3746 (No. 1) 4.2 25
3732” (No. 2) 5.3 250
3677
VOL 35, NO. 3, MARCH 1963
6.2 m
373
From these trends it appears that for a given element peak emission becomes lower and enhancing factors become larger as chemiluminescence increases. As a relative measure of the degree of chemiluminescence, the ratios of enhancing factors to that of the lowest energy atomic line for a given element appear logical. A comparison of enhancing factors among elements shows large differences, which may be due to a number of factors. The dissociation energy of the molecular species appears highly significant. For example, A10 and BOz with high dissociation energies, 6.0 and >7.6
Table II.
A.
2801a
Magnesium Cobalt Zinc Copper Lead
Table 111.
Element Titanium Molybdenum Vanadium Barium Allurninurn Cobalt Sodium Calcium Strontium Barium Cobalt Chromium Iron Manganese Calcium Strontium Barium Sodium Copper
e
374
0
Height of max. emission, mm. 12.5
Excitation potential, e.v. 8.5
Enhancing factor m
334,; XQ
_13 _
7. . 8_
m
3096.9 3336.7 3676.6 4810.5 2618.4
13 13 13 14 14 14 14 14 14 14 14 14 14 15 15 15
6.7 6.4 6.2 6.7 6.1 6.1
m
3572.7
Silver Maenesium Maiganese Silicon Boron Cobalt Magnesium Manganese Aluminum Calcium Barium Cobalt Iron Magnesium Strontium Lead a Double. Multiple.
Triple. Double. Ion lines.
Spectral Lines Rated Higher than 4.0 e.v.
Wavelength,
Element Zinc
e.v., respectively, peak lower in the flame than any other metallic oxide emitters. Atomic lines for such elements appear to be highly chemiluminescent but are different because they are enhanced more and peak lower in the flame. It is more likely that chemical reduction is occurring, followed by thermal excitation and/or chemiluminescence in such cases. Elements such as zinc with high exitation potentials (over 5 e.v.) and low oxide dissociation energies (less than 5 e.v.), on the other hand, appear to be more susceptible to chemiluminescence. The highest energy lines should not be excited by the
3740.0 5465.5 3838.3 4451 6 2881 6 2497" 3732b
5183 6 4823.5 2575 1
4454.8 3993.4 3745,5
3581.2 2352.1 4876 3 4057 8
15 15
15 16 16 17 17
6.0 6.0 5.9 5.7 5.1
5.0
m m
m m m m m m 33
m m
5.2 5.1
250 12 80
4.7
41
4.9 4.8
4.3 4.2
4.3 4.3 4.3 4.4
m
200
25 8 2.5
2.9
18
Spectral Lines Rated Lower than 4.0 e.v.
Wavelength, ,4. 3653.5 3903,O 31 m. .i. ~ I -
3961,5 3602.1 3303b
396S.5c 4215. 5c 4934.1c 4092; 4 4351 3475 5 4033 1" 4226.7 4607.3 5535.6 5893* 5218.2 3247.5
ANALYTICAL CHEMISTRY
Height of max . emission, mm. 14 14 15 16 16 17
18 18 18 18 19 19 ~.
19 19 19 19 20 21 20 22
Excitat,ion potential, e.v. 3.4 3.2 3.9 3.5
3.1 3.7 3 .7 3.1 2.9
2.5 4 0
3 9 3 7 3 1 2 9
27 2 2
2 1 3 8
3.8
Enhancing factor m
m m
76
30
12
7.1 7
3.8
13 10
6 7 2 3 *5 5
1 0
0
3 2 18 14
thermal energy provided by the oxyhydrogen-water flame even after reduction to the atomic state. Trends for molecular emission are also interesting. Except for the nonmetallic species, molecular band heads peak higher in the flame and are enhanced less. Emitting significantly higher in the flame than any other species is CuOH. Kider distribution also occurs for copper, as the atomic lines 2492.1 and 3274.0 A. peak a t 12 and 16 mm., CuO 6060 -1.peaks a t 18 mm., CuH 4280 -1.a t 20 mm., and CuOH 5370 -1.a t 27 mni. This may be involved with the CuH formation, which appears to be peculiar to copper. MnOH peaks much loner in the flame (at 20 mm.) than CuOH, although both their dissociation energies are low, 3.2 and 2.6 e.v., respectively. For comparison, sodium n-ith a low excitation potential (2.1 e.v.) peaks a t 21 mm. JIgOH 3702 A. peaks at 19 mm. and also has a low dissociation energy, 2.4 e.\'., but behaves differently, being depressed rather than enhanced by organic solvents. The corresponding enhancing factors for NnOH, CuOH, and IIgOH are 6.0, 2.5, and 0.54 and also do not appear to fit any particular pattern. N n O H and I I n O emissions are both enhanced more than the lowest excitation potential atomic line for manganese. Although most of the band heads are enhanced to only a small degree, probably because of increased temperature, atomization rates, and vaporization rates, oxides such as -\lo and BOz (again those with high dissociation energies) appear to be influenced by chemical reaction. This is indicated by the lower peak emission in the flame and the much larger enhancing factors shomm in Table IV. Perhaps vibrational-electronic exchange is involved. All of this seems to indicate that different reactions are occurring to produce the hydroxide-emitting species or similar reactions are influenced by flame conditions to highly different degrees. Other discrepancies are also noted for the alkaline earths. Both oxide and hydroxide emission are alternately depressed and enhanced for the series magnesium, calcium, strontium, and barium, as shown in Table Is'. Enhancing factors for atomic emission in the same series also appear to be higher for calcium and especially barium. Differences in dissociation energies may explain part of this. Values for dissociation energies and identity of emitting species for alkaline earths have been the subject of considerable research and controversy in recent years. Some of the recent data reporting lower values for SrO and XIgO may be more nearly correct. In this event the extra heat provided by the organic solvent could dissociate more
IO E
-
In
n 0
u
8 t-
2
.
N
0
z 0
6
u)
5!?
z W 4
P 2
-A
C
2710
2790
2950
3000
3050
ANGSTROMS
Figure 2.
Spectrum of chromium, 0.5%
A.
1 1 peaks, left to right: 2726.5, 2731.9, 2736.6, (2739.4 sh, 2741.11, 2748.3, 2752.9, 2757.8, 2761.8, 2764.4, 2769.9, 2780.7 Sensitivity 4-8 C. 17 peaks, left to right: 2967.6,2971,1,2975.5,2980.8, *2986.5m, (2991.9,2992.4),2995.1 t, (2998.8,3000.9),3005.1,3014.9m,3017.6 m, (3020.7, 3021.6),3024.4, *3030.2 m, 3034.2 sh, 3037.0, 3040.9 Sensitivity 3-7
molecules, resulting in a n enhancing factor less than 1. This also would mean that barium and calcium atomic emission (for equivalent excitation potentials) should be more susceptible to chemiluminescent reactions or chemical reduction. -4s shown in Table IV for the lines rated a t 4.3 e.v. this appears to be true, as emission for barium is lower in the flame and enhanced more than for magnesium and strontium. The order for barium and calcium also fits, as the former with a much higher oxide dissociation energy is enhanced more, indicating greater susceptibility to chemiluminescence. These are only offered as tentative possibilities, and other unknown reactions may be occurring. For example, CaOH emission (also atomic emission) in organic solvents is enhanced by the presence of barium and magnitude of enhancement is dependent on both barium and calcium concentrations. On the other hand, BaOH emission is enhanced by calcium but the degree of enhancement is the same for different barium concentrations. Again this indicates there is much to be learned about reactions in flames and that further quantitative studies of flame emission and correlation with atomic absorption should be made. Data for nonmetallic band heads are also presented in Table IV, since they occur in the reaction zone and may
sh.
Shoulder Main contribution m. Multiple t. Triple
*.
100-
-
80
Table IV.
Molecule CO CH C?
cs
PO
OH H2O CuH BO2 A10
MgO RIgO AlgOH CUO
VO FeO
MnOH RlnO
BaO BaOH CaO CaOH
CaOH Ti0 SrO
SrOH CuOH
Wavelength,
Molecular Band Heads
Height of max.
A.
emission, mm.
2315 43 1.5 5165 3851 2529 3089 9277 4005 5180 4842 5007 5206 3702 6060 5737 5647 3730 5586
12 12 13 13 14 16 15-25 17 17 18 18 1s 19 20 20 20 20 22 21 21 21 21 22 22 20 22 27
6644
5130 6006 6220 5340 5168 8272 6059 6370
Dissociation energy, e.v. 11.1 3.5 6.5 8.2 5.4 4.5 5.1 2.9 7.6 6.0 7.6 11.1 < 3.8
...
7.2
6.0
Oxide dissociation energy, e.v. 8.0 5.6
4.3 4.3
< 3.9,?4.3 1.2
5 0 < 4 3 8.0 1.4 3.6;? 4.6
5.6
.5 . 7
6.8
6.9 6.4
< 4.0
loor 80
m -
-
N
In
ui (D
m
t'I
40
3800
3600 ANGSTROMS
Figure 14.
Spectrum of titanium,
1.6% Slit 0.01 5 mm. Sensitivity 2 - 9
other instruments, relative amplifications for the settings are given below. Settings employed 1-1
2-1 3-1 4-1
4-5 4-10
Relative amplification 1
6 35 300 600 3000
In the n avelength assignments for Figures 1 to 17 and in the tables, peaks for multiple, unresolved lines are indicated by enclosure nithin parentheses or brackets. The folloning symbols are used: d = double, ni = multiple, sh = shoulder, t = triple, * = main contribution for multiple lines, I1 = ion lines. Ilolecular bands are followed by the emitting species and all other lines are atomic. CARBOYAND CO. The atom line for carbon 2478.6 -1.and some of the fourth positive system for CO is shoun in Figure 1. They n ere obtained n ith the oxyhydrogen-nalhtha flame and occur low in the flame. They are of interest practically as n ell as theoretically, as they appear aq background in Figure 13 for tin and for other spectra below 2500 -1. CHROMIUM. It is interesting to compare the chemiluminescent spectrum of chromium, shon-n in Figure 2, B , for the 3000-A. region, to oxycyanogen and inner-cone air-acetylene lines listed by Gilbert (13, 15) in his latest tables, the most complete available for flames. His table shows nine peaks for 13 VOL. 35, NO. 3, MARCH 1963
379
loor I
Ti 0 -
t
Figure 15. Titanium oxide band heads Titanium 1.670 Slit 0.01 mm. Sensitivity 2-9