tween added and found is from 10 to 25%. Higher sulfur values result in a much lower percentage error. Up to this point the data were gathered by research grade chemists. To demonstrate applicability of this technique to control procedures, the data in Table V are presented. These samples \\-ere submitted on a daily basis, disguised as routine samples. The choice of sample depended on the expected concentration level as well as the type of fuel predominantly present in the day’s sample schedule. These data illustrate the excellent repeatability obtainable in a control laboratory by this method. No explanation is offered of the higher standard deviation of sample 11; however, the higher value for sample I is probably attributable to the higher volatility of the naphtha samples.
Table V. Repeatability of Disguised Samples Submitted to Control Laboratory for Routine Analyses
Sulfur Found in Different Samples, P.P. I f . _ _ _ . ~ _ _ _ Ia 1 1 6 IIIb IV* 105 189 204 189 204 212 196 203 200
148 12G 136 183 133 125 124 117 130
168 1Gl 185 1C,G 167 163 166 1Gl 1G5
143 1-10 144 13G
1:37 130 140 138 135
Average 199 129 164 139 Standard deviation, 3t7 f 8 f 2 &3 p.p.m. (7) Naphtha sample. Diesel fuel samples.
SUMMARY
Excellent results are obtained for sulfur in excess of 100 p.p.m. by using a Beckman atomizer-burner to aspirate and burn combustible petroleum samples in a n open combustion tube, drawing the products of combustion through a hydrogen peroxide scrubber, and determining sulfate in the scrubbing solution by a n EDTA titrimetric procedure. The method is rapid, furnishes selective results, and is adaptable to
a control program. The data gathered in this laboratory indicate that this method can be used on a wide variety of sample types covering a wide boiling range and containing high concentrations of unsaturates and aromatirs. If the precautions noted are followed, a wide variation of sulfur concentrations are analyzable and all samples are burned completely with no smoking
and without the necessity of dilution. The work done indicaates that all types of sulfur can be determined; the time of combustion of a 5- t o 7-gram sample is 3 to 10 minutes, depending on volatility and viscosity of the sample. LITERATURE CITED
(1) Am. Soc. Testing %Taterials, Phila-
delphia, Pa., “Standards on Petroleum Products and Lubricants,” p. 15,1952.
( 2 ) Ibid., p. 77. ( 3 ) Ihzd., part 5, pp. lG7-73, 1954. ( 4 ) Bersworth Cheniica! Co., Framingham, Mass., “The Versenes,” Tech. Bi111. 2, (1951). (5) Granatelli, L., ANAL.CHEW27, 2669 (1955). (6) Hidy, J. A , , Rlair, R. D., Ibid., 27, 802-5 (1955). ( 7 ) Hoel, Paul G., “Introduction to
Mathematical Statistics,” Wiley, New York, 1948. (8) Laboratory Equipment Corp., St. Joseph, Mich., private communication, 1954. ( 9 ) Lindberg Engineering Co., Chicago 12, Ill., private communication,
1955. (10) Manns, T. J., Resahovsky. h l . U., Certa, A. J., ANAL. CHEM. 24, 908 (1952). (11) Toennies, J., Bakay, B., Ibid., 25, 160 (1953). (12) Walter, R. N., Ibid., 22, 1332 (1950).
RECEIVED for review June 11, 1956. Accepted April 27, 1957. Group Sewion on Analytical Research, Division of Refining, American Petroleum Institute, Montreal, Canada, May 14, 1956.
Direct Determination of Oxygen in Organic Compounds IRVING SHEFT and JOSEPH J. KATZ Chemistry Division, Argonne National laboratory, lemonf, 111.
> Oxygen can b e obtained quantitatively from numerous organic compounds by heating with BrFnSbFe a t 500’ C. It is essential that the reaction vessel b e shaken vigorously to assure completion of the reaction. Oxygen has been quantitatively recovered from alcohols, aliphatic and aromatic acids, ethers, ketones, phenols, phosphate esters, sulfones, and organic nitrogen-containing compounds in which the oxygen is not bound directly to the nitrogen. This method of direct oxygen determination can be applied to solid, liquid, or gaseous samples.
A
many methods for the direct determination of oxygen in organic compounds have been described, generally oxygen is obtained by difference. The earlier literature on direct oxygen determination has received a LTHOUGH
1322
ANALYTICAL CHEMISTRY
critical review by Elving and Ligett ( 3 ) . More recently several new methods have been described. One method is based on the conversion of the oxygen in the organic compound to an insoluble carbonate by means of strontium oxide (6). Another is a n isotope dilution procedure in which the dilution of oxygen18 in enriched oxygen gac that is used t o oxidize the sample is measured ( 4 ) . The development of simple fluorination procedures for the quantitative release of molecular oxygen from a wide variety of inorganic materials suggests the use of these procediires for the analysis of oxygen in organic compounds. Bromine trifluoride has 1)wn used in the direct determination of oxygen in oxides and oxygen-containing compounds which are fluorinated a t or near room temperature ( 5 ) . Recent work indicates the feasibility of using addition compounds of bromine trifluoride for the determination of oxygen
in inorganic compounds ~ h c hdo not react completely with tmmine trifluoride ( 8 ) . Most inorganic oxides are completely fluorinated and molecular oxygen is r c ~ v e r e dquantitatively by heating a t 500” C. with RrFzSbFe or potassium fluohromite (potassium bromotetrafluoridc). This paper estends the successful use of BrFQSbFB t o the direct determination of oxygm in organic compounds. Because molecular oxygen is obtained and measured tensimetrically, the method is particularly useful for oxygen isotopp studies of orqanic compounds nhere t h r dilution due to the combustion methodc of analysis is a disadvantage. CHEMISTRY A N D PROPERTIES OF FLUORINATION REAGENTS
Bromine trifluoride is a n ionizing solvent with a high electrical conduc-
\ll-
I
TO CONTROLLER 81 RECORDER
'b
--
1
I
POL1 ti NICKEL REACTION POT
I
FURNACE ~~
L--_ - _._ T E r V S i - E OVEN ON SHAKER 140' C
Figure 1 .
tivity (a, 9). It may act both as a fluoride ion donor and an acceptor in an acid-base type solvent system in which the fluoride ion is analogous to the proton in water though reversed in rharge. I n bromine trifluoride solutions, antimony pentafluoride acts as an acid while potassium fluoride shows basic properties.
+ BrF4K + + BrF4-
213rF1s I3rF2-
+ BrF8
KF SbFd
+
BrFa
+
&
1154 STRIP hEATEQS
Reaction system
fluoride in vacuum at 130' C. It is obtained usually as a brick red solid which is stable in dry air. Although it reacts vigorously with water, the reaotion is much more moderate than with bromine trifluoride itself. Because the fluorination reaction involving the release of oxygen occur:, best in the liquid phase, BrF2SbFG should be used above its melting point of 130' C. The vapor pressurr (8) of BrF2SbF6 is given as log
BrF*+ -t SbFe-
Potassium fluobroniite can be considered to be a basic flux and so would be most useful for acidic samples, while the acid flux, BrF2SbFe. is best used for I)asic samples, Because the latter comllound reacts faster to liberate oxygen quantitatively from organic compounds, it was used for all the analyses reported here. RrFzSbF6 may be prepared by dissolving antimony pentafluoride in a 10% inole excess of bromine trifluoride and by removing the excess bromine tri-
€'mn,.
7.66 - (3030/7'I
MATERIAL
The bromine trifluoride used in these experiments was obtained from the General Chemical Co. and was purified by vacuum distillation. The major impurities (bromille, bromine Pentafluoride, "ydrogen and nonvolatile fluorides) are easily separable by distillation. The fraction used (boiling point 95-5.5' c. a t 250 mm.) was pale yellow in color and RraS stored in a nickel vessel. For use in the preparation of the addition compound for rou-
Comer Furnace
Miller Pressure Gage
Manometer
Toepler Pump
Figure 2.
=
Vacuum system
tine analysis, the bromine trifluoride need not be so highly purified. Pumping on a container of bromine trifluoride a t room temperature will remove the volatile impurities without loss of significant amounts of bromine trifluoride. Then, collecting a fraction a t 125' to 135' C. through a simple overhead take-off will give bromine trifluoride of sufficient purity. Antimony pentafluoride was a commercial product obtained from the Harshaw Chemical Co. and was used as received. The organic samples were of reagent grade or the best grade available from The hlatheson Co., Inc., or Eastman Organic Chemicals. Fluorochemical 07 5 , a completely fluorinated cyclicether with empirical formula C8FI6O believed t o contain a five- or sixmembered oxygen ring with a fluorinated side chain, was an experimental sample received from the Minnesota Nining and AIanufacturing Co. The picryl chloride and tetramethyl glucose were prepared and recrystallized by members of the Arponne Kational Laboratory. APPARATUS
Kickel is usable with pure l3rF,SbFG up to a temperature of 500' C., and, therefore, it is a suitable material of construction for the high temperature fluorination reaction vessels. Because thorough mixing of the gas with the liquid phase is essential, the reaction tube used for the inorganic samples (8) was replaced by a nickel pot 2'/2 inches in inside diameter and 4 inches long welded to a 3/4-in~hnickel tube with a 3/4-inch SAE flare a t the top. The nickel pot is attached to thp nickel head containing the Teflongasketed sample addition port arid t o nhich is silver soldered a Hoke diaphragm valve (KO.413). The nickel head is constructed by mclding together a t right angles two pieces of nickel bar and carrying out the appropriate machining (Figure 1). To increase the effectiveness of the shaker, the reactor is shaken on its side and the riglit angle shape is used to prevent splashing of the molten fluorinating reagent into the top and plugging the v a h e and addition port. The reactor to be shaken is disconnected from the vacuum line, placed in a Transite oven. and shaken horizontally. The lower section of the reaction v e d is heated by an electric furnace; inserted into the furnace is a thermocouple which serve> as the sensing element for an electronic controller. The temperature cycle is followed on a Brown potentiometer recorder. Because nickel is a relatively poor conductor of heat, the temperature of the addition port remains lorn enough to prevent leakage through the Teflon gasket even when the reaction tube is heated as high as 500" C. Pu'evertheless, thermal conditions require a new Teflon gasket, usually after six heating cycles. Thorough mixing of the liquid fluorinating reagent with the oxygen containing compounds in the gas phase is essential for complete oxygen hberation. For reaction pots of the dimenVOL. 2 9 , NO. 9, SEPTEMBER 1957
1323
Table I.
Oxygen Analysis of Organic Compounds"
Compound
Oxygen, 3 'Calcd. Found
Formula Aliphatic
Ethyl ether Pentaer ythritol Tartaric acid Tetramethyl glucose
CzHsOCzHs C(CHz0H)a HOOC(CHOH)2COOH
21.59 47.00 63.96
21.49 47.08 63.40
Camphor
RESULTS A N D DISCUSSION
Aromatic Benzoic acid Hydroquinone Xanthone Urea
C.)
Picryl chloride Di-p-tolyl sulfone Tri-p-cresyl phosphate C8F16Ob Carbon dioxide a
cyclic
26.20 29.06 16,31
26.22 21) 07 16.39
26.64 46. 65c 38.78 16.97' 12.99
26.60 46. 82c 19.93 1.2P 13.03
17.3i
17.33
3.84 72. 71
3.78 72.35
Shaking 3 hr. a t 500' C. with BrFzSbF,. hlinnesota Mining Br. Manufacturing Co. Fluorochemicsl o-75. Nitrogen.
sions used, a n oscillating shaker operating at 215 r.p.m. gives the necessary mixing and reaction rate to complete the oxygen liberation in 3 hours. The vacuum line used to collect and measure the oxygen is shown in Figure 2. A Toepler pump transfers the noncondensable gases to a calibrated glass bulb connected to a differential manometer. Kel-F and glass traps immersed in liquid nitrogen remove condensable gases and keep corrosive materials out of the glass measuring system. A copper furnace containing finely divided copper on infusorial earth heated a t 200" C. (7) is used to absorb the oxygen and distinguish the oxygen from the small amount of carbon tetrafluoride which may pass through the traps. PROCEDURE
Forty to fifty grams of fluorinating reagent is poured into the clean, dry nickel pot (Figure l ) , which is then attached to the nickel head containing the addition port with a 3/4-inch flare nut. The system is evacuated and the two valves between the reaction pot and the vacuum line are shut. The reactor is disconnected and placed on its side in the oven on the shaker. When t h e oven temperature reaches about 100' C., the furnace and shaker (215 r.p.m.) are turned on and the system is prefluorinated at 500" C. for 5 hours. Prefluorination is essential to elimi1324
ANALYTICAL CHEMISTRY
in a gold capsule about 2l/, inches long made from tubing 3,/y2 inch in diameter with a 1/8a-inch wall. The bottom is flame-welded shut, and the tube is weighed and filled to the top with the liquid to be analyzed. The top of the tube is crimped shut and while the bottom of the tube is imniersed in water to keep the sample cool, the top is f l a m e - d d e d above the criinping tool. The capsule is reweighed to obtain the weight of the sample, placed in a n evacuated chamber fcr about 15 minutes, and rewighed to cnsure vacuum tight seals on the capsule. Gas samples can be measured in a calibrated volume and then condensed into the rcnction vessel.
nate from the reaction system impurities and contaminants which would be certain to confuse the interpretation of Rubsequent analyses. After the reactor cools, i t is connected to the vacuum line and evacuated. The system is filled with dry nitrogen at 800- to 1000-mm. pressure, the sample is added, and the system is evacuated and heated on the shaker as before. \Then the reactor is cool again, it is attached to the line and the noncondensable gases are transferred t o the calibrated volume and measured. The trapping system is shut off from the Toepler pump and the gas is pumped through the previously evacuated copper furnace to remove the oxygen. The volume of the nonoxygen residue (carbon tetrafluoride which has leaked through the traps) is subtracted from the initial reading to give the amount of oxygen collected. If the sample is a nitrogen-containing compound, the nonoxygen residue may contain molecular nitrogen. I n this case the infrared absorption (1) a t 1280 cm.-l is measured to determine the amount of carbon tetrafluoride and nitrogen obtained by difference. If the oxygen is to be used for isotopic analysis, the carbon tetrafluoride correction can be obtained from the mass spectrometer yields. Solid nonvolatile samples are added in a small gold cup made by rolling a 1 inch X 1.3 inches X 0.2 mil gold sheet around a '/r-inch rod and crimping the bottom. Liquid samples are added
The results of direct oxygen determinations on a wide variety of organic compounds are shown in Table I. The accuracy of the method is within 1%. Part of the discrepancy may be due to the difficulty of obtaining highly purified organic compounds to use for test samples. A series of five analyses on a single compound gave a precision within ~k0.275. (-4 blank of 0.05 cc. obtained with hydrocarbons is used.) If lorn oxygen-containing samples are analyzed, a smaller calibrated volume should be used to increase the accuracy of the pressure measurement. Adsorbed moisture becomes much more significant if the sample is low in oxygen. Complete oxygen recovery is achieved from all compounds except picryl chloride. Kitrogen is not fluorinated under these reaction conditions, and the nitric oxide or nitrosyl fluoride formed is resistant to further reaction. Even after a 16-hour heating, only 51% of the oxygen is recovered. I n the case of urea, which contains nitrogen not directly bound to oxygen, molecular nitrogen is quantitatively recovered in addition to the complete recovery of the oxygen. Complete oxygen recovery is obtained even from Fluorochemical 0-75. This fluoroether is reported by the producer to have unique thermal stability. It can be heated to a temperature as high as 600' C. without evidence of decomposition. Because carbon tetrafluoride is a major product of the reaction and is not completely condensable by the liquid nitrogen used to refrigerate the traps, a correction for residual carbon tetrafluoride in the collected oxygen must be made. Carbon tetrafluoride is a colorless gas condensing at - 124" C. and freezing at -184" C. with a vapor pressure at the boiling point of liquid nitrogen of 0.035 mm. of mercury. The amount of carbon tetrafluoride coming through the liquid nitrogen trapping system is usually between 0.1 and 0.2 cc. Attempts to reduce the carbon tetrafluoride content of the collected oxygen by pumping on the liquid nitrogen and
Table IL Carbon Balance in Fluorination of Pentaerythritol"
Found Compound CF,
CZFG n-CaF8
Cc. 29.7 1.5 0.9
Converted to CF,, Carbon, Cc. 29.7 3.0 2.7
52 5 5
T o h l carbon in sample as CF1
=
.57.42 cc.
'
lowering the effective trap temperature have not been successful. The amount of carbon tetrafluoride found seems to be affected more by the method and rate of opening the reactor to the vacuum system than by the equilibrium vapor concentration. However, a correction for the carbon tetrafluoride content can be made rather easily as previously described. An attempt was made in one of the runs in which all of the oxygen mas recovered from pentaerythritol to obtain a carbon balance. The temperature of the traps wvas increased to -73" C.
and several fractions 15 ere collected. The fractions were analyzed by infrared absorption using a Perkin-Elmer double-beam spcctrometer and by making comparisons with spectra of the pure compounds available in the literature. Table I1 shows that 62% of the carbon was recovered in the fractions collected. The rest of the carbon undoubtedly occurs as less volatile compounds. Because of possible damage to the glass measuring system by the corrosive fluorinating materials, i t was not feasible to increase the trap temperatures to obtain a more complete carbon recovery. Even under the severe conditions of fluorination, the carbon skeleton had not been completely disrupted. This method for direct determination of oxygen in organic compounds is easily carried out in relatively simple apparatus. Since volatile carbonyl fluoride is a frequent intermediate in the fluorination of organic compounds, it is essential t h a t the reaction vessel be shaken vigorously to assure completion of the reaction. Carbonyl fluoride is not readily fluorinated in the gas phase and thorough mixing with the molten
fluorinating reagent is necessary. Because the sample oxygen is not diluted with other oxygen, the method should have particular utility in oxygcn exchange studies for oxygen not readily exchangeable with carbon dioxide. LITERATURE CITED
Ayscough, P. B.,Can. J . Chem. 33, 1566 (1055). Banks, A. A , , Emrleus, H. ,J., Koolf. -4. A,, J . Che/?i.,Soc.1949, 2861. Elving, P. J., Ligett, IT, R., Chem. Rem. 34, 129 (19-14).
Grosse, .4.V., Hindin, S.C , Kirshenbaum. A . D.. A X A L . CIIEM. 21, 386 (1949).
'
Hoekstra, H. R., I h t z , J. J., Zbid.,
25,
1608 f19.531.
Lee; - T . ' S . i - ~ e y e r ,R , - ~ n a l .Chim. Acta 1 3 , 3 4 0 (1959).
lleyer, F. R., Rouge, G., 2. angezo. C h e m 52, 637 (1939).
Sheft. I.. LIartin. A . F . I h t z . J . J.. J. Am: Chem. SOC. 78.' 1557 (1956): ( 9 ) TVoolf, A. A,, Emeleus, H. J., J . C h e m SOC.1949, 2865.
RECEIVED for review Xovemher 16, 1956. Accepted April 20, 1957. Division of .4nalytical Chemistry, 130th Xeeting, ACS, Atlantic City, N. J , September 1956. Based on work performed under the auspi,ces of the E. S. Atomic Energy Commission.
Quantitative Determination of Histamine in Presence of Certain Interfering Metallic Ions A. C. ANDREWS and T. D. LYONS Department of Chemistry, Kansas State College, Manhattan, Kan.
b Various metallic ions interfere with colorimetric methods used to determine histamine concentrations. A method is presented to nullify this interference utilizing diethylenetriaminepentaacetic acid. A comparison of the stability constants for histamine-metal ion complexes with corresponding diethylenetriaminepentaacetic acid-metal ion complexes indicates that the metal ions are almost completely held in the latter complex. This complex allows the histamine to b e determined without detectable error.
A
the chemical methods utilized for the quantitative determination of histamine, @-imidazoyl-4-ethylamine, colorimetric procedures appear to be the most sensitive. These methods depend chiefly upon the light absorbance of the coupling product of histamine and a diazonium ion in alkaline media or the reaction product of histaVOKG
mine R ith dinitrofluorobenzene (5, 7 ) . Substituted phenyldiazonium ions t h a t have been used successfully include p-sulfonic acid ( 2 , S ) , p-bromo- ( I j 63, and p-nitro- ( I O , 11). Certain metallic ions, principally cobalt(II), copper(I), copper(II), nickel(II), and iron(III), cause interference in each of the above methods by forming a stable complex with histamine (8, 9). This metallic complex resists the coupling reaction with a resulting decrease in the concentration of histamine which niay undergo these reactions. A procedure has been outlined ( 7 ) to nullify the interference of trace amounts (microgram ion) of these metals in the analysis involving dinitrofluorobenzene which utilizes 8-quinolinol or sodium diethyldithiocarbamate to eliminate the interference. The present paper introduces a new method to nullify the interference as i t occurs in the coupling procedure involving the p-sulfonic acid diazonium ion.
Preliminary studies were carried out on various chelating agents including 8-quinolinol, sodium diethyldithiocarbamate, sodium cyanide, and diethylenetriaminepentaacetic acid (DTPA). 8Quinolinol was unsatisfactory because it undern-ent a reaction with the diazonium ion 'i\ hich r e d t e d in a high blank reading. Sodium diethgldithiocarbamate was unsatisfactory because it formed a cloudy solution with high absorbance n i t h the diazonium ion. The cyanide ion proved moderately satisfactory for lo^ concentrations of histaDiethylenetriaminepentaacetic mine, acid proved entirely satisfactory, showing no interaction with the diazonium ion and completely nullifying the metal ion interference over a wide range of histamine concentrations. EXPERIMENTAL
Reagent grade materials were used throughout and solutions were prepared with distilled water. Histamine diVOL. 29, NO. 9, SEPTEMBER 1957
1325