Kjeldahl Determination of Nitrogen with Sealed Tube Digestion Microgram Analysis BENJAMIN V'. GRUNBAUM, FREDERICK L. SCHA4FFER,A N D PALTI, L. KIRK University of California, Berkeley, Calif.
Y CONXECTIO?; ivith a st,udy of small needle biopsy samples '-of human liver from patients suffering from infectious hepatitis, it was necessary to have a rapid and dependable procedure for the deterininat.ion of microgram quantities of total nitrogen in fractions of liver. The method is also being applied in studies of liver metabolism in burn shock. X number of procedures in this range and for similar uses has been described by Seedham and Boell ( I O ) , Briiel, Hoker, Linderstr@m-Lang, and Rozits ( I ) , Tompkins and Kirk ( 1 1 ) ) and ot,hers ( 6 ) . The procedures previously described have all shown some limitations or deficiencies which made desirable additional investigation. Furthermore, it is a curious fact that the mechanism of the Kjeldahl digestion, one of the oldest methods in which organic material is destroyed by heating with strong acid, is almost completely unknoan from a chemical standpoint. While much investigation has been carried out, it has nearly all been purely empirical from necessity and, more unfortunate, either poorly controlled or not comparable in conditiolis v i t h other work with Jq-hich it might be compared. Some light is thron-n on one important phase of t'he digest>ionby this communication. The procedure of Briiel et a ! . ( 1 ) is far too time-consuming and laborious for routine application, though a high degree of accuracy and precision i.= claimed. Similar considerations are pertinent to t h e method of Xeedham and Boell, u-hile that of Tompkins and Kirk is not ideal, inasmuch as the procedure described for the digestion is difficult to control with respect' to temperature which has bem shon-n to br critical in t,he digestion (8). Open-tube digestion, as generally employed, presents certain liniitations and difficultics in the microgram range. The recent work of JVhitr and Long ( 1 2 ) seems to offer a means of avoiding some of the uncertainties knov-n to cause error and failure of complete recovery in the conventional Kjeldahl digestion. These authow digest,ed their miiples in sealed tubes a t 470' C. As described in their procedure, a considerable amount of mercuric oxide catxlyst was used. However, they showed with nicotinic acid, a compound that it; known to be difficult to digest, that the eat,alyst n-as complete]\- uiiriecessarj- if sufficient digestion time (30 miiiutes) were allo~vetl. They also suggested the use of the se:ileti-tul)e method \vli(an other constiturnts such as phosphorus were also to be dett~i~niiiiod. If ratalyst mid otlier c2str:tneous materials can be eliminated from the Kjeldahl digestion, a very important simplification results. In addition, nothing can escape from or be absorbed by the digest in the sealed tulle, thus giving the opportunity to make cei,tain definitive studies of the end products of t,he digestion, and t o contribute indirectly to our knowledge of the niechanisni of the digestion. The sealed-tuhe technique of White and Long, modified to much smaller quantities, v a s combined with a new diffusion proccdure sonien-hat similar to that of Briiel el ul. in this study. Thcx aiiinionia was collected in boric acid and determined by titration Xvith standard hydrochloric acid. EXPERIMEhTAL
Apparatus. DIGE~TIOS was carried out in borosilicate glass tubes with an outside dinmeter of 7 mm., which were thoroughly cleaned with chromic acid solution and cut into sections 45 mm. in length. Each section itas sealed at one end, after n-hich the edge of the ooen end w n 5 not touched. to avoid contamination. .41rsamples ahd reagents were measured with the standard type of capillary transfer pipets ( 6 ) modified by bending the tip to an angle of about 120'. Heating was performed in a small niuffle furnace equipped with a thermocouple. Inside the fur-
nace was placed a brass blocli with a number of chambers drilled into the metal, each of which served to hold one glass tube and protect the others and the furnace in case of any explosion. A later development allowed for separate brass tubes, with a screw plug and vent, each tube carrying a glass tube in order t,o make more flexible the handling of the latter. DIFFUSIOX was carried out in a vessel shown in Figure 1 . The interior of both sections of the vessel was treated with Desicote t o make the surface hydrophobic. The manner of treating the vessels was similar to that of Gilbert ( 3 ) . The vessels and other similar sniall pieces of glassware were cleaned thoroughly with chromic acid, rinsed, and dried in an oven a t 100" C. After cooling, Desicote was drawn into the vessel and allowed to stand for 10 minutes. The material was then expelled and the remainder centrifuged to an absorbent wad of cotton or paper. The vessels were rinsed with distilled water and again dried a t 100" C. The lower bulb served to contain the digest, the upper held the tioric acid, and served as titration vessel.
A
B
C
Figure 1. Diffusion Vessel A. B. C.
Diffusion bulb Titration buIb Assembled vessel
A glass syringe of 1-ml. volume, graduated in 10 fil. divisions, \vas employed t o add caustic to neutralize the digest. It was equipped with a No. 24 needle ground to a square end and bent a t a 90" angle as shown in Figure 2. The needle was coated on the outside with a thin layer of vaseline, after which alkali could be readily introduced under the layer of acid digest without loss of the latter. All stirring was carried out with a rotating magnetic stirrer and micro-stirring bar placed in t h e liquid. Titrations were performed with the conventional capillary buret (6) equipped with a n attached tip bent t o the vertical and held in place with small Tygon or gum rubber tubing. A centrifuge, vacuum desiccator, and a simple rack and pinion type micromanipulator were other necessary accessories. REAGENTS
Hydrochloric acid, standard 0.01 and 0.02 N , made by dilution of a carefully standardized 0.1 S hydrochloric acid solution. Boric acid, 2y0,to which a mixed indicator according to Ma and Zuazaga (9) was added. The indicator consisted of a mixt,ure of bromocresol green and methyl red, the final concentrations of which were approximately 0.0025% bromocresol green and 0.000570 methyl red. Sulfuric acid, c . P . , concentrated, was redistilled and protected from atmospheric contamination. Sodium hydroxide, saturated, from which the carbonate had been allowed to settle. PROCEDURE
The sample, ordinarily containing 1 t o 15 micrograms of nitrogen, was measured n i t h a pipet coated with Desicote, or
1487
ANALYTICAL CHEMISTRY
1488 weighed on a quartz helix balance if solid. Liquid samples were deposited in the bottom of the digestion tube with the pipet which did not need to be rinsed if pioperly coated. The tube was placed in the vacuum desiccator and dried, which required about 10 to 15 minutes with a 25 11. volume. After drying, the vacuum was released, the tube removed, and without touchin the sample with the pipet, 10 pl. of concentrated sulfuric aci8 were added. Drying was omitted with solid samples. Dried liver solid was digested in amounts several times as large as the sample size indicated and aliquots of digest were removed for diffusion after dilution to volume. With these larger samples, 25 pl. of sulfuric acid were used for the digestion to ensure an adequate excess.
o
i
2 cm.
i"l
vessel was tilted and rotated so as to alkalinize the interior of the bulb and neutralize any droplets of acid that might .have spattered to the walls. Each vessel 'ivas then placed vertically over the stirrer in a small rack and allowed to diffuse with constant stirring for 90 minutes. After diffusion was complete, the top bulb was withdrawn, any excess vaseline wiped from its edge, and the opening covered with,a small piece of Parafilm pressed firmly to the edge of the opening. The boric acid was then centrifuged into the bottom of the inverted bulb and a 3- or 4-mm. stirring bar added to it. The ammonium borate was then titrated carefully with standard hydrochloric acid in the presence of the mixed indicator. The end points were reproduced more accurately by comparing the color of the solution a t the end point with that of another aliquot of the boric acid containing the indicator. The vessel containing the standard was held close to the titration vessel as shown in Figure 3. Blank controls were treated in the same manner but with omission of the sample. RESULTS
Figure 2.
Syringe for Addition of Alkali to Digestion Tube in 3lanipulator a. Clamp t o manipulator b . Syringe holder
The open end of the tube was heated with a gas-oxygen torch until it was white hot; it was then removed from the flame and the edge's \yere sealed by pressing them together with hot forceps. The sealed end was annealed briefly in a gas flame to relieve any strain and the sulfuric acid was thrown to the bottom of the tube by a quick jerk which caused it to wet the dry sample. The tube was then encased in the brass block which had been placed a t an angle in the muffle furnace and preheated to 450' but not above 470' C. Asbestos pads in the bottom of each hole served as cushions. After being heated for 30 minutes the hot tubes were withdrawn with forceps and placed in brass centrifuge cups also carrying asbestos cushions to protect t,he rubber pads. The tubes were centrifuged immediately, which cooled them and forced the sulfuric acid to the bottom of the vessels. A brass casing for each digestion tube was adopted in later analyses. These were placed in a furnace preheated to 600' C. and were allowed 10 minutes to reach digestion temperature. The entire assembly was centrifuged after digestion. Each digestion tube was then scratched with a sharp file or glass knife and broken into two pieces, which were placed in a drying oven at 90' C. for a few minutes. This allowed escape of that portion of the sulfur dioxide and carbon dioxide that had been forced into solution during the digest,ion. Failure to remove these gases invariably led to low and erratic results through formation of ammonium salts of the weak acids formed by their absorption in boric acid. The digest was transferred with a suitable transfer pipet to the bottom bulb of the diffusion vessel which had been treated with Desicote. Care was necessary in expelling the liquid from the pipet to avoid spattering of the acid to the sides of the diffusion vessel. Both halves of the digestion tube were then rinsed with about 50 HI. of water which were also transferred to the diffusion c,ell. A 5-mm. stirring bar was introduced into the diluted digest and a thin layer of vaseline was spread around the round neck of the lower portion. The bulb was then suspenfed on a manipulator as shall-n in Figure 2 . The upper part of the diffusion vessel was then charged with 50 PI. of boric acid solution added so as to form a seal across the neck about 2 mm. above the end of the ground neck. The upper bulb was held in one hand while the bottom bulb was raised over the syringe needle until the end of the needle penetrated the acid and nearly touched the bottom of the bulb. By applying a slow pressure t o the syringe, a 2- to 2.5-fold excess of alkali n-as added, the bulb was lowered, and the upper bulb was immediately seated in it. After tightening the ground joint, the entire diffusion vessel was held over a rotating magnet until the digest and alkali were thoroughly mixed. While still over the stirrer, the
Early analyses of knoa n compounds showed variable results with occasional recoveries of zero. It was suspected that the temperature, which was not under close control, might exert a large effect. Tests of this were made with ammonium sulfate solutions of known strength, adhering to the conditions of temperature specified by White and Long. Erratic and low recoveries were obtained. Calibration of the thermocouple shon ed it to register 40' to 50" lower than the true temperature. The results of digesting ammonium sulfate in the absence of organic matter at different temperatures controlled to f 5 " C. are shown in Figure 4. It is clear that when the temperature is high enough, sulfuric acid, or more probably sulfur trioxide, is a strong olidizing agent capable of destroying ammonium ion. The nature of the end product was not determined but was probably elementary nitrogen, since this is a more stable form than the alternative oxides that might be formed. Addition of sucrose to provide reducing agent during the digestion as in an ordinary Kjeldahl analysis delayed the onset of destruction of ammonium slightly so that 0.5 hour a t 470" C. produced no loss. At higher temperatures, the destruction was as rapid as in the absence of sucrose. This result also would be expected if sulfuric acid can behave as a strong oxidizing agent a t raised temperatures. This critical role of temperature is not indicated in the article of Jj-hite and Long, but is confirmed in open tube digestions
f
Figure 3. Arrangement for Titration over Stirrer with Comparison Solution a. Comparison vessel b . Titration vessel C.
Buret
d. Rotating magnetic stirrer e. Rubber connection to buret tip f. Rlanipulator clamp
V O L U M E 24, N O
9, S E P T E M B E R 1 9 5 2
I
I
450
500
1489
, 6
550
Temperature in "G Figure 4. Recovery of Ammonia from Ammonium Sulfate with Varying Temperatures of Digestion for 0.5 Hour (8) without any explanation of its nature. It has been noted previously that too much potassium sulfate added to the sulfuric acid of an open tube digestion also leads to loss of nitrogen, but this was assumed to be due to the formation of bisulfate (6). Since longer time a t a favorable digestion temperature probably has a similar but slower effect, this finding may have an important bearing on the matter of very long digestions (2, 4 ) which may lose ammonia by oxidation as contrasted with shorter digestions in which decomposition may not be complete. The sealed tube makes possible such studies without interference by variations in catalyst, oxidizing agent, added salts, etc., on the one hand, and on the other, loss of material during the digestion, or absorption of ammonia from the air. The rate and completeness of diffusion were tested by diffusing the ammonia of an ammonium sulfate solution as described. The results are shown in Figure 5. Nearly all the ammonia was transferred in the first 50 minutes and after 70 minutes there was no increase in quantity, and the recovery was approximately
100%. Increasing the volume of solution from which the ammonia was diffused increased the time necessary for complete recovery. On the basis of this fact and the results shovn, the standard time of 90 minutes was chosen for all diffusions. The design of the diffusion vessel was such as t o avoid many of the occasional difficulties of other designs. It contained no dead spaces; the diffusion distance was minimal; it was simple to alkalinize its entire inner surface without danger of mixing alkali with the absorbing solution: and it was suitable for simple and effective stirring during diffusion. These advantages led particularly to great uniformity in results throughout the entire investigation. Having determined the conditions of digestion and diffusion under which complete recovery of ammonia was obtained from ammonium sulfate solutions, it remained to show that the procedure, particularly the digestion, was capable of giving complete recovery when decomposition of the sample was required. It was also necessary t o determine the range of application to various forms of nitrogen, some of which are very difficult to convert quantitatively to ammonia. The results of this study are shown in Table I. Simple aliphatic compounds like glutamic acid were included as well as aromatics like acetanilide and ring nitrogen compounds such as tryptophan and nicotinic acid which are known to be difficult to digest. I t is evident that sulfuric acid without catalyst, salt, or oxidizing agent is capable of converting all the nitrogen of these compounds quantitatively to ammonia when the conditions are controlled. Recovery of nitrogen from the compounds that are difficult to digest is not significantly different from that of easily digested compounds. The simplicity of the reagent inspires confidence in the fact that unknown sources of error are not present.
IO0
2
aJ
>
8
50
I
c
I I
C
Table I.
Analyses of Nitrogen-Containing Compounds
Compound Glutamic acid
Kitrogen Calcd.. 'I 9.69 6.90
Cystine
Acetanilide
Nitrogen Detd., y 9.62 9.62 9.68 7.01 7.00 6.87 6.80
Average Recovery, % 99.5 100.3
Standard Error, y
0.00
5.18
5.10 5.22 5 13 5.13
99.4
0.03
Uracil
5,13
5.12 5.16 5.07 5.07
100.6
0.02
Hydroxyproline
6.63
98,s
0.02
Tryptophan
6.00
97.8
0.03
99.0 98.4
Kicotinic acidd
12.7 4.51
12.57 4.44
.d solutions.
I
I
I
I
I
50
IO0
150
180
Time in Minutes 99.3
8.54 12.20
I I
I
0 05
1.40 1.40
Tryptophand
u,
8
0.02
1.41
6.50 6.61 6.54 6.56 5,885 5 . 88= 5.78" 5.93b 5.90C 8.36 11.95
aJ
Figure 5. Recovery of Ammonia from Ammonium Sulfate with Varying Diffusion Times The perfected method was applied to the study of the nitrogen content of lyophilized lamb liver, the results of which are shown in Table 11. I t n-ill be noted that excellent reproducibility was obtained, though the total nitrogen content was known only very approximately. Since these samples were dry initially, it was not necessary to dry them before adding the sulfuric acid.
Table 11. Nitrogen and Phosphorus Content of Dry Lamb Liver 97.9 98.0
Weight of Liver. y
Nitrogen Found, Aliquot 1/20 Digest, y
700 463 432 488 481 385
3.58 2.37 2.19 2.47 2.38 1.95
Average Standard error
Nitrogen Found, 70
Phosphorus Found, %
10.2 10.2
1.19 1.16 1.17 1.17 1.17
10.1 10.1 9.9 10.1 10.1
10.1
1.19
1.18 +0.01
ANALYTICAL CHEMISTRY
1490 Each sample was weighed carefully on a delicate quartz helix balance ( 7 ) and transferred to the digestion tube. A major advantage of the sealed-tube digestion with sulfuric acid is that it allows complete digestion in the absence of any catalyst and also prevents the loss of such volatile constituents as phosphoric acid which would be formed simultaneously. Thus, the sample may be digested and an aliquot for nitrogen determination taken while anothrr may be used for determination of phosphorus or other constituent. The procedure was actually followed in determination of phosphorus, as shown by the data of the last column of Table 11, in which good reproducibility is shown. As compared with the method of Tompkins and Kirk, this method suffers from the necessity of transferring the digest to the diffusion vessel, but the technique involved is not in any way difficult. The direct use of the digestion vessel as a diffusion vessel t o avoid transfer has been tried u-ith some success, though additional problems were introduced. The technique has received considerable study (by Wolfgang Kirsten while visiting this laboratory) and may be made the subject of a separate publication. The use of Desicote is almost essential in making operations smooth in the determination, not only in performing pipet measurements without rinsing but also in control of the neutralization of digest, and in suspending the boric acid as a seal in the neck of the diffusion vessel. In neutralizing the digest, the latter does not spread on the surface. When alkali is added underneath it, there is still little spreading until after the bulb is sealed and stirring is started. As soon as the alkaline solution is spread over the surface, it removes the Desicote and the surface wets, allowing the liquid t o form a thin film from which diffusion is rapid and stirring is effective. The method is sufficiently reliable and rapid for an operator to perform many analyses in a day without losing one of them. Digestions are performed in multiple (10 or more simultaneously), and diffusions are likewise made simultaneously, a single rotating
magnet serving t o stir as many as 13 a t once The range of analyzable nitrogen has not been shown to be lower than with other methods. Experience indicates that 1’1ith certain modifications of technique or method, the loner limit of range may be lonered very significantly. Studies are being made of this possihilitv. ACKNOWLEDGMEAT
Phosphorus analyses were performed by Jean Fong by a procedure t o be published. LITERATURE CITED
(1) Bruel, D., Holter, H., Linderstr$m-Lang, K., and Rozits, K., Compt. rend. trar. lab. Cadsberg, Ser. chim., 25, 289 (1946). (2)
Chibnall, A. C., Rees, h1. IT.,and Williams, E. F., Biochem. J . , 37, 354 (1943).
(3) Gilbert, P. T., Jr., Science, 114, 637 (1951). (4) Jonnard, R., IND. ENG.CHEM.,ANAL.ED., 17, 246 (1945). (5) Kirk, P. L., “Advances in Protein Chemistry,” edited by .4nSon, M. L., and Edsall, J. T., Vol. 111, p. 147, Kew T o r k ,
Academic Press, 1947. Kirk, P. L., “Quantitative Ultramicroanalysis,” Sew T o r k , John Wiley & Sons, 1950. (7) Kirk, P. L., and Schaffer, F. L., Rev. Sci. Instmments, 19, 785 (6)
(1948).
(8) Lake, G. R., hlcCutchan, P., Van Meter, R., and Neel, J. C., ANAL.CHEM.,23, 1634 (1951). (9) Ma, T. e., and Zuazaga, G., ISD. EN. CHEM.,~ A L ED., . 14, 280 (1942).
(10) (11)
Keedham, J., and Boell, E. J., Biqchem. J . , 33, 149 (1939). Tompkins, E. R., and Kirk, P. L., J . B i d . Chem., 142, 477
(12)
White, L. M., and Long, 31. C., ANAL.CHEM.,23, 363 (1951).
(1942).
RECEIVED for review February 2 5 , 1952. Accepted M a y 23, 1962. This investigation was made under contractual support b y t h e Veteran’s Administration a n d with aid of the University of California Committee o n Research, a n d t h e Office of S a v a l Research. The opinions contained herein are the private ones of the writer a n d are not to be construed as official or reflecting the views of the Navy Department or the naval service a t large.
Anhydrous Alumina as Adsorbent in Constituent Analysis of Asphalt RETHEL L. HUBBARD, K . E. STAKFIELD, AND W. C. KOMJIES Bureau of Mines, U . S . Department of the Interior, Laramie, F’yo. KHYDROUS aluminum oxide has been used as an adsorbent
A in constituent analyses of several hundred samples of as-
phalts from petroleum ( l a ) ,shale oil (8),native bitumen, and bituminous sandstone (6) by the method of Hubbard and Stanfield (4). This method involved removing pentane-insoluble asphaltenes from approximately 1.5 grams of asphalt, then dispersing the petrolenes on 25 grams of alumina to adsorb the resins and leave an oil fraction. The three different batches of alumina used in these analyses had similar adsorption properties, presumably owing t o the same method of preparation and t o comparable contents of alpha- and gamma-alumina. However, several subsequent batches of alumina procured under the same lot number and from the same supplier had variable adsorption characteristics. Accordingly, a study was made of several different batches of alumina and their use in the constituent analysis of asphalt. The study showed that different batches of anhydrous, crystalline alumina may have different adsorption characteristics. Therefore, samples of the same alumina adsorbent should be used to compare constituent analyses of a series of different asphalts. Preliminary attempts, involving blending or treating different samples of alumina, were unsuccessful for the purpose of preparing an adsorbent having adsorption characteristics which were intermediate between those of alpha and gamma forms of alumina. I n these tests, the utilization of adsorbentgrade aluminas commonly used for chromatographic analyses was not inveetigated.
Anhydrous alumina may consist of several crystalline forms, particularly the metastable gamma-alumina, which is highly adsorptive, and alpha-alumina, which is stable but less adsorptive ( 2 , 6, 13). These two crystalline forms were determined qualitatively in seven alumina samples by means of x-ray diffraction and by petrographic examination. By the latter method, i t was possible to distinguish the isotropic alpha-alumina particles from the anisotropic gamma-alumina particles. Asphalt constituent anal>-sesusing the different aluminas as adsorbents (each adsorbent \$as graded to pass a 100-mesh-per-inch sieve and be retained on a 200-mesh-per-inch sieve) showed that the yield of resins was increased in proportion to the gamma-alumina content of the adsorbent. The x-ray and petrographic examinations showed the presence of alpha- and gamma-alumina in the adsorbents but were not a measure of adsorption. Accordingly, several methods of directly measuring adsorption were tried. The adsorbent was exposed to a benzene-saturated atmosphere for periods of 24 and 72 hours, and the adsorbed benzene was determined b y the increase in weight in accordance Lvith the method of Mair, Westhaver, and Rossini (IO). By the method of 1Iair and Forziati ( 8 ) , columns of the adsorbent m r e treated n ith n-heptane solutions of anthracene and of phenanthrene. The adsorption of anthracene and phenanthrene mas folloived by fluoreseence under ultraviolet light. Although the percentages of henzene, anthracene, and phenanthrene adsorbed per 100 grams of adsorbent ranged from 7.5 to