that they would not prevent adsorption of nitrogen compounds. Sitrobenzene was determined by a modified digestion procedure (4) employing o-mercaptobenzoic acid. The salutary effect of iodine on the digestion of pyridine is clearly indicated in Table 111.
Table IV. Comparison of UOP Hydrogenation and Silica Gel Percolation Methods
Total Nitrogen,
P.P.RI. LTOP Si gel method method
4
1 2 3
Approx. Blank 0.01N Acid, RI1.
5
0.8 0.8 0.8
T 0
Table VI. Check Determinations on Reformer Charge Stocks
(Sample containing 92%, acid-treated naphtha, 27, diisobutylene, 2%, toluene, 475 iso-octane, and pyridine. Theoretical nitrogen content 0.0131 wt. %.) Total Sitrogen Found, Conditions Kt. % No iodine, regular digestion 0.0096 0,0079 No iodine, 3-hour slow diges- 0.0121 tion, 4-hour rapid digestion 0.0118 Iodine added before percola- 0.0122 tion, regular digestion 0.0122
86
Blend
Analysis of Blends of l o w Nitrogen Content
Approx. Volume 0.01N Acid for Titration, 111.
Nitrogen, P.P.M. Theory Found 97 91 4.9 5.0 2.4 2.5
ACKNOWLEDGMENT
Table 111. Effect of Iodine on Determination of Pyridine
Sample Blend of visbreaker virgin naphtha Middle East virgin nnphtha Coker distillate Blend coker distillate :md virgin naphtha Coker naphtha
Table V.
Volume Sample, hI1. 10 225 300
68, 70 3
29
34
24 82
29 100
Sample A
B C
n
Nitrogen, p.p,m. 2.2 2.3 2.2 2.1 16 1 6
OT
0.7
The authors wish to thank the Gniversa1 Oil Products Co. for permission to include their cooperative data. They also wish to express appreciation to the Houdry Process Corp. for permission to publish this method.
LITERATURE CITED
A comparison was made between samples of unknown but very low nitrogen content which had been analyzed by the Universal Oil Products hydrogenation procedure ( 8 ) and by this method. Results are shown in Ta'rile IT'. As a further check of the method a t very low nitrogen concentrations, three synthetic blends of quinoline in acid washed naphtha were prepared and analyzed as shown in Table V. In each case, the total titration was much greater than the blank. I n blend 3 a difference of 0.16 ml. in the titration would only make a difference of 0.1 p.p.ni. of nitrogen in the results, a further proof of the advantage of the ability to use large samples. Duplicate determinations s h o ~e.;cellent agreement, as illustrated by results on reformer charge stocks in Table TI.
(1) Fanale, D. T., Fricioni, R. B., Hutton, D. R., Snyder, R. E., Clark, R . O., Am. Petroleum Inst. Svmnosium. AIay 1955. King, R. IT., Faulconer, W. B. RI., ANAL.CHEW2 8 , 255 (1956). Lake, G. R., IIcCutchan, P., Van Meter, R., Keel, J. C.. Ibid., 23, 1634 11951). XcCutrhan. P.. Roth. W.F.. Ibid.. 24. 369 (1952). ( 5 ) Mills) G. A., Boedeker, E. R., Oblad, A . G., J . Am. Chem. SOC.72, 1554 (1950). (6) Koble, E. D., A N A L .CHEW 2 7 , 1413 (1955). ( 7 ) Schall, J. W.,Dart. J. C., Petrolezun ReJnsr 31, 101 (March 1952). (8) Wankat, C., Gatsis, J. G., ANAL. CHEJI. 25, 1631 (1953). "
)
RECEIVED for review June 11, 1956. Accepted September 19, 1956. Division of Refining, American Petroleum Institute, llontrenl, Canada, M n y 1950.
Determination of Trace Amounts of Carbon Monoxide in Guseous Hydrocarbons KURT H. NELSON, M. D. GRIMES, D. E. SMITH, and B. J. HEINRICH Research Division, Phillips Petroleum Co., Bartlesville, Okla.
F A method for the determination of 0 to 100 p.p.m. of carbon monoxide in saturated and unsaturated hydrocarbon gases i s based on the reaction of carbon monoxide in methane with heated iodine pentoxide to form iodine and carbon dioxide. Methane does not react with heated iodine pentoxide but other hydrocarbons do react. In conducting the analysis, methane i s first added to the sample to be analyzed. low temperature
180
ANALYTICAL CHEMISTRY
fractional distillation removes the carbon monoxide from the sample and concentrates it in a methane fraction. The methane-carbon monoxide mixture i s passed through a tube filled with activated charcoal to remove any traces of other hydrocarbons, then through a drying tube for removal of moisture, and finally through a heated combustion tube filled with alternate layers of iodine pentoxide and glass wool. The liberated iodine
i s collected in a potassium iodide solution and measured spectrophotometrically. In the range of 0 to 100 p.p.m., the accuracy of the method i s within 1 to 2 p.p.m.
I
on the amount of carbon monoxide in hydrocarbon gases can be of great importance in many petroleum processes. The yield of products and the reactivity of the cataNFORJIATIOK
A - CONNECTION TO DISTILLATION UNIT B,C-MEASURING BULBS D- CHARCOAL TUBE
A
G - KI SCRUBBER H - FURNACE I - FLOW CONTROLLER J NEEDLE VALVE K- FLOWMETER L - THERMOCOUPLE
-
NITROGEN
=G2!==J
-
Figure 1 .
I
J
Combustion unit for carbon monoxide determination
lysts may be affected by trace quaiitities of this contaminant in the reacting gases. There are numerous analytical procedures for the determination of carbon monoxide, but these procedures are not necessarily applicable to hydrocar bon gases containing only a few parts per million of this contaminant. Generally, the methods either lack sufficient sensitivity, or the hydrocarbons react with the reagents. I n the iodine pentoxide procedure, the interfering constituents in a gas are first removed. The gas is then passed through a heated iodine pentoxide tube to oxidize the carbon monoxide quantitatively t o carbon dioxide. During this process, iodine is stoichiometrically liberated. Either the carbon dioxide or the iodine may be collected and measuied. The absorption of the iodine in potassium iodide solution, followed by titration with thiosulfate, :ippears to be thc more sensitive of the two techniques ( 4 , 9, 10). As little as 20 p.p.ni. carbon monoxide has been determined in this manner ( 9 ) . Recently, this loner limit of detectability has been reduced hy measuring the liberated iodine spectrophotometrically (5, 7 ) . Because gaseous hydrocarbons, other than methane, react n ith heated iodine pentoxide (1, 9 ) , a separation of the carbon monoxide is required. To accomplish this, procedures based on condensing the hydrocarbons in a liquid nitrogen trap (S), or by absorbing them in concentrated sulfuric acid ( 9 , 10) have been used. However, these two methods are not practical in the determination of a few parts per million of carbon monoxide in hydrocarbon gases. Booth and Campbell ( 2 ) employed a distillation technique in the analysis of ethylene containing about 0.017,
carbon iiioiioside. The tlist~illatioiiconcentrated the carbon monoxide in an ethylene fraction for subsequent iiieasurement by the pyrotannic acid--blood method. This work, along with an inspection of the boiling points of carbon monoxide and gaseous hydrocarbons, suggested the separation of carbon monoxide from the intcrfwing hydrocarbons by lon- temperature fractional dist,illation. Furthermore, the boiling point of methane is hetiveen those of carbon monoxide and the h i g h boiling hydrocarllons. Also, niethane tlocs not react wit'h hot iodine pentoxide. These facts suggested the addition of nipthane to the gas sample before distillation. This paper describes a metliod based on tlie separation of the carbon monoxide from the interfering gztscous hydrocarbons by low temperature fractional distillation after the addition of methane. The carbon monoxide is then measured using the iodine pentosidespectrophotometric technique. APPARATUS A N D REAGENTS
Potassium Iodide Solution, 2.57,. Dissolve 50 grains of reagent grade potassium iodide in distilled water and dilute t o 2 liters. Iodine Stock Solution A. Dissolve 0.0'300 gram of reagent grade iodine in several milliliters of 2.5% potassium iodide solution. Transfer t o a 100nil. volumetric flask and dilute t o volume u-ith the potassium iodide solution. Iodine Stock Solution B. Pipet 10 inl. of iodine stock solution A into a 1-liter volumetric flask and dilute t o volunie with the potassiuni iodide solution. Magnesium Perchlorate, anhydrous. (Handle and use this reagent with precautions. From the standpoint of
safety, acidic materials should not. be allowed t o contact magnesium perchlorate. I n laboratories where this reagent is not used, anhydrous calcium sulfate follon-ed by a mixture of fine sand ant1 phosphorus pentoxide can be subst'ituted.) Activated Charcoal. H e a t 40- t o BO-mesh charcoal t o 200" C. in a stream of nitrogen. Iodine Pentoxide, 40- t o 200-mesh. Nitrogen, oil-pumped. Methane. Phillips Petroleum Co. research grade. Distillation Unit. d low temperat,ure fractional distillation apparatus such as sold by Podbielniak, Inc., Chicago, Ill. (6). Combustion Unit. 1 dual borosilicate glass combustion unit is assembled as illustrated in Figure 1. At -4,the apparatus is connected t o the "fraction withdrawal" of the distillation unit diagrammed by Preston and Podbielniak (6). The volunies of bulbs B and C, which are approximately 260 nil. each, are measured before assembling the apparatus. The conibustion tube, F , 1.2 X 33.0 mi., and the adsorption tubes, D and E , 1.2 X 19.0 cni., are connected nith 11/35 ground-glass joints. The combustion tube and therniocouple L are enclosed in furnace H . Scrubber G is joined t o the combustion tube by means of a 7!25 ground-glass joint. The inlet tube and cap of the scrubber are coniiected to the scruhliei. bulb by a 19/38 groundglaes joint. A reference mark corresponding to 10-mI. volume is placed on tlie scrubber Iiulb for ease of filling. The flow rate of t'lie nitrogen is regulated by the differential-tJ-pe flow contidler I (Moore Pi,oduc+s Co.. Philadelpliia, Pa.). needle valve J . and flonmeter I
