Air- Adsorption Error in Micro-Dumas Method for Nitrogen H. ARRIIN PAGEL AND ITSUMI JACK OITAl Acery Laboratory, C-niversity of Nebraska, Lincoln, Keb.
T
HE air-adsorption error arising from the copper oxide in the
.
temporary filling of the combustion tube is given by Trautz ( 7 j as 0.006 nil. The authors have long believed this value to be much too large. I n order to measure such quantity accurately it is almost imperative that the generator and diffusion error should remain essent,ially constant throughout the entire investigation. Eight years ago a large-capacity, all-glass carbon dioxide generator (6) was constructed t,o meet this requirement. Air-adsorptiori values for the temporary filling were found to be 0.001 to 0.003 ml. As these values are of the same order of magnitude as the sum of the two unavoidable (&0.001 ml.) reading errors of the conventional microazotometer, these results were not published. I t was observed, however, that the same adsorption values were obtained with 10-gram portions of either fine or coarse air-ignited copper oxide as n ith 10-gram portions of copper oxide which had been deaerated by heating in high vacuum a t 550' C., cooled, and preserved in pure carbon dioxide in glass ampoules. I t thus beconies apparent that the observed air adsorption might arise from air adsorbed on the wall of the combustion tube, rather than from the copper oxide, since, in removing the old temporary filfing and inserting the new filling, most of the carbon dioxide in this part of the tube will be replaced by air. This research was recent,]!- continued using the same carbon dioside generator in conjunction with a specially constructed ultrsniicroazotonieter which rould be read to fO.OOO1 nil. Gener:ator and diffusion blanks and empty combustion t u h and copper oxide absorption hlnnks u.ere determined inciependentl>-. ULTRABIICRO.AZOTOMETER
Thi. was made of borosilicate g t o the same gt~rici~al dinienter, except that the gradusion5 a5 a conventional niicronzot ated section \vas replaced b>-n capillary bore tube (approriniately 7 111111. in outsiclr diameter ti>- 1-nini. bore) bearing a 50-nini. scale engraved ivith a precision engraving machine. Calibration of this tube using mercury showed 0,000881, and with 50y0 potassium hydroxide solution of knov-n density, 0.0008i5 nil. per mni. The was chosen as sufficiently accurate for this work. ne1 stopcock wa? eliminated by closing the top with ground, tapered borogilicate glass plug 3 mm. in red t o 2 mm. when seated 6 mm. deep into the capillary openin'g. The first scale graduation was approriniately nini. tjelorv the bottom of the plug; therefore, the air s1)ac-ew small enough to eliminate pressure-volume errors. The plug w sealed with metaphosphoric acid lubricant ( 1 ) and hcld in place with fixed spring tension. The three-way stopcock leading to the combustion tube was sraled to the azotometer t o eliminatr the wual rubber coupling. Fifty per cent, potassium hydroxide solution \vas prepared in 500-1nI. lots in a 1-liter glass-stoppered borosilicate glass Erlenmeyer flaqk. After being cooled to room temperature, the solution waF shaken vigorously for 5 minute.< to saturate with air, in order that the observed air blanks would reprewnt maximum rather than minimum values due t o possible soluhility of air in the solution. A small drop of isoamyl alcohol per 100 nil. of potassium hydroride solution completely elirninatrd fonni in the azotometer.
The air adsorbed in the bo~osillcateglass coupling was first removed by cautiously heating with a microburner and flushing with carbon dioxide. Equal volumes (100 ml.) of carbon dioxide were then passed into the azotometer a t different rates such that 10, 20, 40, and 60 minutes were required. Triplicate runs were made for each time value. The maximum observed deviation in any set of runs was 0.0002 ml. The generator and diffusion blanks were calculated independently by assuming that the air impurity in a given volume of carbon dioxide is constant and that the air-diffusion rate per minute is likewise constant. Thus, for the 10- and 40-minute rates, total air blanks of 0.0006 and 0.0014 ml., respectively, were found. B y solving the simultaneous equations:
+ 1Oy 0.0006 + 40y = 0.0014 x = 0.00033 ml. = generator blank for 100 ml. of carbon dioxide z z
=
z, = 0.000027 = diffusion rate in ml. per minute These values when inserted into the equations for the 20- and the GO-minute rates gave calculated values that agreed fO.OOO1 of the observed. Empty Combustion Tube Blanks. The tube was first freed of adsorbed air by heating to low red heat to within 5 cm. of the rubber connections while flushing with 100 ml. of carbon dioxide. After being cooled to room temperature under slight carbon dioxide pressure, the tube was disconnected and supported in a vertical position (capillary end down) in an essentially air-tight glass cabinet. ilfter the desired time of air exposure, the tubk was again connected and rapidly flushed a t room temperature with 100 ml. of carbon dioxide followed with about 50 ml. at the regular rate, a part of the latter being consumed in testing for microbubbles. The tube was then heated as described above and 100 ml. of carbon dioxide were passed through the tube into the azotometer. The tinie required t o carry out the latter step was recorded. The combustion tube air-adsorption blank n-as then calculated by deduct,ing the grnerator and diffusion blanks from the total air collected in the azotometer. Table I shows typical results of empty combustion tube airadsorption studies. In this case a new Pyrcx 172 tube, which had been r l ~ a n e dby soaking for 3 hours in dichromate cleaning solution, \\-asused. Identical studies n.ere carried out with new Pyrex 172 and Vj-cor 7900 tubes which were used without the preliminary treatment with dichromate cleaning solution. The air-adsorption values found were in general 0.0002 to 0,0005 ml. lover than given in Tablr I ; however, the same general increase of adsorption with
Table I.
J
0 0021 0.0018 0 0020
13 I4 11
0 0014 0 0011 0 0014
10
0 0023 0.0020
15
0.0016
14
0.0013
13
0.0018 0 0017
40
0.002; 0.0024 0.0031
00
0 0037
0.0036
12 14
0 0039 0 0041
1:
18
0 0032 0 0032
0 0046
11
0 0040
BLAYKS 20
Generator and Diffusion Blanks. In these detrrminations the azotometer was connected t o the generator by means of a borosilicate glass coupling having the same dimensions as a regular micro combustion tube, except that the barrel was only 5 em. long. The capillary end was connected to the azotometer with a piece of Butyl rubber microtuhing, while the opposite end was connected to the generator with a regular soft red rubber microstopper. These two connections (the only rubber parts in the entire assembly) were lubricated with a trace of glycerol.
Hour. 2
36 a
1
Present address, Standard Oil Co.. \Thitinp, Ind.
Air Blanks on P y r e x 1 i 2 Combustion Tube
(Treated with cleaning solution for 3 hours. 100 mi. of carbon diovide used f o r all runs) Time of Corretted Tiibe Time of Tube Obsd. Air Exposwe. Blank, Run. Blank. All." hll. hlin. Nin.
0.0033
Generator and diffusion blanks dediicted.
.l; 11 12
0,0025 0.0026 0 0030 0 0029
V O L U M E 2 4 , NO. 4, A P R I L 1 9 5 2
757
time was found. Results of less ext,ensive studies with old tubes which had become devitrified from prolonged heating were in good agreement with results found with new tubes. I n these determinations and others described later the combustion tube was supported on a conventional gas-fired combustion stand, 28.5 cm. long, fitted with Chrome1 wire gauze tunnels and combustion tube sleeves. T h e maximum temperature in the combustion tube in the long burner section was determined with a Bureau of Standards calibrated Chromel-illume1 thermocouple and a Leeds and Yorthrup potentiometer and found to be 664” C. Copper Oxide Adsorption Blanks. Some workers ( 2 , 4,5 ) have stated or implied that “porous” copper oxide in wire form retains air which is not removed by flushing with carbon dioxide a t room temperature. I n order t o investgiate t’he effect of porosity on air retention, four lots of copper oxide having different pore sizes and lengths were prepared. Copper oxide wire from four different sources was gently crushed in an agat,e mortar. Small portions were t,hen put on smooth, lint-free paper, where the pieces were separated by sliding those of desired length t o one side with a Monel metal microspatula. Over 100 grams of the desired length from each of the four sources were prepared. The lengths and pore sizes for each lot were then determined using a small spoonful of the well mixed material. Lengths of the individual pieces were measured t o the nearest 0.5 mm. and pore diameters t o the nearest 0.02 nim., the latter with a measuring microscope. Well over !OO pieces from each lot were measured. Pore diameters ranging from 0.02 t o 0.26 nim. were found in all four lots; hoivever, the weighted averages were very different. Table I1 gives the dimensions of the four types of copper oxide wire. Table 11. Diam., hIm.
Uimensions of Copper Oxide Wire
Length, N m . , Tgpe Range I 0.68 1 . 0 t o 3.0 I1 0.68 1.0to2.5 I11 0.68 1.0 t o 5 . 0 I\’ 0.51 1.0 t o 6.0 Weighted average.
Length, Mm., Dominant 72%; 1.5to2.j 80%; 1 . 0 t 0 2 . 0 5 2 % ; 3.0 t o 4 . 5 53%; 3.0t o 5 . 5
Pore Diam.a. hlm. 0.06 0.12 0.12 0.09
These lots of copper oxide were put into covered nickel crucibles :ind air ignited in an electric muffle furnace a t 700” C., and then allowed t o cool to room temperature in the closed furnace. T h e ignited oxides were then transferred into 25-ml. glass-stoppered Erlenmeyer flasks which had been ignited a t 600” C. in the muffle furnace. The copper oxide in these flasks was then stored in a large desiccator t o avoid further possible atmospheric contamination. -4ir-adsorption or occlusion blanks for the above copper oxide were determined as follows: The empty combustion tube (which noT7 contained a snugly fitting borosilicate glass rod 5 cm. long placed a t the constricted end) was first freed of air and cooled under slight carbon dioxide pressure as previously described. T h e tube was then removed and exposed t o air for definite time periods. Twenty grams of copper oxide were poured into t h e tube. T h e charged tube was then flushed with carbon dioxide a t room temperature and the run completed as described above for empty combustion tube blanks. T h e copper oxide air-adsorption (or occlusion) blanks were calculated by deducting the previously determined generator, diffusion, and empty tube blanks from the total volume of air found. A total of 14 determinations using the four types of copper oxide gave 0.0000, f 0.0002 ml. of air per 20 grams of copper oxide. Single det,erminations on each type, repeated a month later, gave the same values. These results confirmed the authors’ earlier belief that thc amount of air retained by porous copper oxide in the temporary filling is less than can be read (1-0.001 ml.) on the conventional niicroazotometer. (Twenty-gram portions of the copper oxide filled the combustion tube to a length of 12 to 13 cm.) Charged Combustion Tube Blanks. The permanent filling in a charged combustion tube always contains a section of metallic copper which was not included when the air-adsorption blanks
for the empty tubes (and also blanks for the copper oxide) were determined. I n an actual analysis the adsorbed-air blank for a given time exposure ought therefore to be even smaller than given in Table I, because a part of the oxygen in the “air” would be removed by formation of copper oxide. Four runs were made to confirm this.
R u m 1 AND 2. A T’ycor tube which had been used for 12 nitrogen determinations was chosen. The temporary filling was removed, followed by heating, flushing, and cooling in carbon dioxide. Purified air from a carbon-hydrogen determination preheater was then passed through t h e t,ube for 1 hour to allow the tube and copper oxide to adsorb air. a f t e r flushing with 150 nil. of carbon dioxide a t room temperature, the permanent filling \vas heated t o maximum temperature and the run completed in the usual manner. Another identical run was thcn made, using the same tube. RVNS3 AND 4. -4Pyrex 172 combustion tube, which had hccn used for empty tube air blanks, was charged with both permanent and temporary fillings using air-ignited coa oxide. The 5-cm. section of metallic copper so that i t extended t o within 5 em. of the exit end of the combustion st,and. The freshly charged t,ube was used directly without the customary prolonged “burn-out” in a carbon dioxide atmosphere. The charged tube was merely flushed at room temperature with 1.50 ml. of carbon dioxide and the air blank was then determined. Another run n-as then made in nhich purified air was passed through the tube for 30 minutes before the air blank \vas determined. After deducting generator and diffusion blanks the net adsorbed air blanks for the above four determinations ranged from 0.0006 to 0.0008 ml. As the air-adsorption blank for 1-hour air exposure for an empty Vycor tube was 0.0026 ml., while in runs 1 and 2, also for 1hour exposure, it is only 0.0006, it would appear that the adsorbed “air” is largely oxygen. CONTROL DETERiVlINATIONS
Because some organic substances are easily air-oxidized in strongly alkaline solution, the question arose whether the isoamyl alcohol added to the potassium hydroxide solution might cause IOK results. This possibility was checked by the method of control determinations in which accurately measured volumes of air and oxygen were used. Apparatus and Procedure. h capillary bore tube with a sidr arm, having the general shape of the lerter h, was sealed at right. angles to a standard dimension ordinary borosilicate glass micro combustion tube. The seal was made about 5 cm. from the stopper end of the combustion tube. The top arm of the capillary tube was fitted with a precision millimeter scale and was accurately calibrated. T h e curved capillary bore arm was fitted with a section of rubber tubing and closed n-ith a glass plug. This closed rubber tube served a? an air or oxygen reservoir. The adjacent capillary bore arm was likewise fitted and this reservoir was completely filled with mercury. A screiv pinchclamp was used on each. This special combustion tube was connected to the generator and azotometer in the usual manner. The mercury reservoir was then compressed, causing the mercury to rise vertically unt,il the graduated capillary bore arm was completely filled. The combustion tube was then heated and flushed v i t h carbon dioxide t o remove all air. T h e mercury Tras lowered to about 5 mm. above the air inlet and t h e air reservoir was compressed, whereby t h e mercury column was split by an air bubble. H v again compressing the mercury reservoir the entrained air bubble n-as forced u p into the graduated sect,ion \There the exact volume of air was measured. T h e combustion tube rvas then flushed rapidly with 50 ml. of carbon dioxide t o remove any air which had diffused through the connections during the above operation. The run was completed by gradually forcing the air bubble into the combustion tube vihile about 30 ml. of carbon dioxide passed into the azotometer. The tube was heated while 70 ml. more were passed through to ensure complete transfer of the air. Three control runs with air and two with oxygen were carried out, using volumes ranging from 0.006 to 0.010 ml. After deducting the generator and diffusion blanks from the observed volumes collected in the azot,ometer, the difference between used and
758
ANALYTICAL CHEMISTRY
found was less than 10.0002 nil. Thcsc rcwlts show that the isoamyl alcohol does not iritcrfere aitd a l ~ othat t l i c ovrr-all a(’c~iracyof mcasurements throughout this \\ u r l ~is 11rol):thly11 ithiri 10.0002 1111. cONcLusIo6
The results of this work indicate that deductions for adsorbtd or occluded air from air-ignited porous copper oxide are not justified. The practice of deaerating copper oxide by evacuating, saturating with carbon dioxide, and preserving in an atmosphere of carbon dioxide is likewise unnecessary. Upon long air exposure the air adsorbed in a combustion tube attains significant values. I n practice, however, the actual error will be very small, because the time of air exposure while placing the sample and temporary filling is comparatively short. and furthermore a major portion of the adsorbed air is held back by the metallic copper by formation of copper oxide. In an actual drterrnination the legitimate adsorbed air deduction appears to be not over 0.001 ml. It has long been recognized that control of the rate of burning the sample is a critical operation. The authors believe that high nitrogen results have been too frequently attributed to adsorbed air errors, rather than to incomplete combustion, in which case methane and similar products may be formed. The conditions for this combustion are far less favorable t,han in the determination of carbon and hydrogen, where platinum catalyst in presence of excess oxygen assures complete oxidation of suhstnricos that are not easily oxidized by coppcr oxide alonc.
Gonick et al. (3) statc that high nitrogen results can hc causcd release of adsorbed nitrogen on copper oxide when the latter is reduced to the metallic state by burning the sample. They do not state how large such errors are. I n the authors’ work the customary sugar blank was omitted to avoid errors due to the possible formation of methane. Theoretically, a 5 - m ~ . sample of sucrose will reduce about 28 mg. of copper oxide. A(atually, the amount of copper oxide reduced will no doubt be a sniall fraction of the calculated, because a major portion of the' rarbon will remain as charcoal. If the volume of nitrogen released by the reduction of such small amounts of copper oxide i$ significant, the values reported are not conclusive. In that caw, however, all blanks of this kind become rather meaningless bt,causc the amount of copper oxide reduced and, in turn, the volume of nitrogen released will depend not only on the fate of thc carbon but also on the size of the samplc and the rcduc-iiig equivnlcnt n cight of the particular substance used.
1)sthe
LITERATURE CITED
J . .4m. Chem. Soc., 52, 2813 (1930). (2) Flaschentriiger, B., Z. nngeu. Chem., 39, i 1 7 (1926). Asar,. ED.,17, 677-82 (1945). (3) Gonick, H., e t a l . , IND. ENG.CHEM., (4) Niederl, J. E., and Niederl, V., “Micromethods of Quantitative Organic Analysis,” 2nd ed.. p. 88, New York. John n’iley 8(1) Boughtoil, JV.
A\.,
Sons, 1942. ( 5 ) Ogawa. S . , Sci. Repts. T o k y o I m p . Uniz>., 16, 667 (1927). (6) Pagel, H. A , IND.EXG.CHEM.,ANAL.ED. 16,344 (1944). (7) Trauta, 0. P., hlikrochentie. 9, 300 (1931). ~ ~ E C E I I E for D
rcview .lugtist G , 1031. .iccepted Octoiier 30, 1951
Chemical Assay for Tocopherol in Animal Materials R . W . SWICK A N D C . A. BAUMANN College of Agriculture, Cnicersity of Wisconsin, Madison, Wis.
hlAJOR problem in the estimation of tocopherol in animal
A materials is t,he elimination of interfering substances.
In 1947 Wanntorp and Sordlund ( 1 9 ) used saponification and adsorpt,ion on Floridin for the removal of fat and the carotenoids, but recovered only 92%,pf added a-tocopherol. The more recent assay of Quaife and Dju (13)requires a molecular still and the use of hydrogenation, which is prohibited in certain hospitals (3). The present procedure combines elements of these two methods: it is easy to perform, requires only readily available equipment, and yields an over-all recovery of 98% of added a-tocopherol. PROCEDURE
Homogenization and Extraction. Ten grams or less of tissuv were cut into small pieces and homogenized with 40 to 50 ml. of cthyl alcohol in the microcup of the Waring Blendor. The homogenate was filtered through an extraction thimble and the cup mashed repeatedly with small amounts of alcohol; the washings also were filtered into a Soxhlet extractor. The thimble was dropped into the extractor, about 100 ml. of alcohol were added to the flask with a boiling chip, and the residur \vas rrfliised for at least 18 hours. Most tissues were homogenized completely by this procedure, except for small knots of connective tissue which remained intact but offered no hindrance to extraction. The intestine !vas the most resistant to homogenization, although ext’raction appeared to be complete, as shown by comparisons of the amounts of tocopherol in two samples of the same intestine, one of which was homogenized, the other frozen and ground (13j: 285 and 273 micrograms of tocopherol per sample, rePpPctivel>-. Extraction for periods longer than 18 hours f d e d to inrrrape the apparent tocopherol content of the tissuep. Saponification. Because tocopherols w e oxidizat)ln in t h r presence of alkali, ehborate equipment has htwi designed to
provide an oxygen-free atmosphere over saponifi(,ation inixtulcs (Z), and neutralization of the alkali before transfer from this atmosphere has been suggest,ed ( 2 ) . The use of antioxidants such as pyrogallol (18) or p-acetylaminophenol ( 7 ) has also bccn recommended. I n the present study the latter was used. The condenser of the Soxhlet apparatus was removed :inti 10 nil. of an alcoholic solution of 2% potassium hydroxide and 0.5% of p-acetylaminophenol were pipetted down the arm of the ext,ractor. The mixture was heated for 30 minutes, the extract being concentrated t o about 50 ml. during this period. Following saponification, the extract was cooled immediately, an equal volume of water was added t o t.he alcohol, and the solution was saturated with sodium sulfate. The nonsaponifiable matter was extracted into 25 ml. of purified petroleum ether (5) by shaking for 10 minutes in a separat.ory funnel (18). The layers separated rapidly and the pet,roleuni ether was dried over anhydrous sodium sulfate. Under these conditions a second petroleum ether extract of thc aqueous alcohol contained no measurable amount of tocopherol. For oils or very fatty tissues alkali a as added gram for gram of fat (ca. 5 times the theoretical amount required), and in the presence of the extra alkali further extraction into petroleum ether became necessary. Washing the petroleum ether layer Tvith water or dilute acid or alkali did not alter the final values for tocopherol. When 110 micrograms of a-tocopherol were added to tissue extracts, the amounts recovered ranged from 106 to 115 micrograms. p-hcetylaminophenol did not affect recovery in the presence of tissue substances, but in their absence the losses of tocopherol were serious unless a supplementary antioxidant was present. p-Acotyl:iminophenol was used routinely in preference to pyrogallol h x i w the ktttei, gave high recoveries---l04 to 181% (17). Chromatography. .In aliquot of the petroleum ether extract containing 20 to 1000 micrograms of tocopherol was evaporated