ciency of gel filtration columns, most evidence indicates a typical N of about 400 (2). A typical column can therefore be expected to separate a maximum of only about five peaks. As mentioned before, other forms of chromatography have no clear maximum retention volume; additional peaks can be eluted until time becomes excessive or until their dilution hinders proper detection. In gas chromatography, with its considerable speed and sensitivity, it is practical to work with peaks whose retention volumes are 50 or so times that of the air or initial peak (in many such cases, programmed temperature would actually be preferred). Thus V,/VI in Equation 5 can be replaced by a number the order of 50. This shows, very roughly, that the peak capacity of a gas chromatographic column is 5-fold greater than that of a gel filtration column of 1 "12. Peak capacities based on equal plates- i.e., n this assumption are also shown in Table I. If we limit Vn/Vlto 10, as would be reasonable in many gas and most liquid chromatographic columns, the peak capacity would be only 3 times better-i.e., n 1 0.6 N1I2(Table I). The above indicates that for a separation of comparable overall complexity, gel filtration and permeation columns must have many more plates than conventional columns. Again using Equation 5, with a = m/N1/2,we see that to fully offset the 3- to 5-fold increase in peak capacity of conventional columns, plate numbers of gel filtration and related columns
-
+
-+
(2) J. C . Giddings and K . L. Mallik, ANAL.CHEM., 38, 997 (1966).
+
would need increasing by 3 and 52, respectively-roughly one order of magnitude. While such increases may not be easy to achieve, the enormous importance of the gel techniques in complex biological and macromolecular separations makes an effort worthwhile. A preliminary discussion of factors affecting plate height in such systems may provide a basis for improvement (2). While the foregoing conclusions are based on highly simplified assumptions, they are probably valid as a general rule. For instance, while N is not constant for all components in a given column, as assumed, values generally lie within a twofold range. An average value for several peaks should be adequate. Also the two approximations leading from Equation 4 to 5 give exceptionallygood results. It so happens that the errors of the two steps cancel one another to a first order; even for N = 100 and thus a = 0.4, the approximations together are valid to nearly 1%. The question of separation time has not been considered above. Clearly the limited elution range of gel chromatography is to its advantage timewise. This would compensate to some extent for increases in time which might accompany the search for more theoretical plates. RECEIVED for review February 6, 1967. Accepted May 1, 1967. Investigation supported by Public Health Service Research Grant G M 10851-10 from the National Institutes of Health.
Determination of Trace Amounts of Phosphorus in a Composite Propellant by Fast Neutron Activation Analysis M. H. Rison, W. H. Barber, and Peter Wilkniss Research and Development Department, U.S . Naval Propellant Plant, Indian Head, M d . A RAPID, SENSITIVE METHOD to determine small amounts of phosphorus in a composite propellant was needed at this laboratory. The propellant consisted of approximately 60% ammonium perchlorate, 20 % aluminum, and 20 % binder. The small amounts of phosphorus to be determined were known to be in the binder phase of the propellant. Fast neutron activation analysis seemed to be a promising approach to the problem, and the results obtained with the method are described. To obtain the best sensitivity the 31P(n,a)BAl reaction was used. This reaction had already been used by Lbov and Naumova (1) and Maen To-on et al. (2) to perform fast neutron activation analysis of phosphorus in different materials. Interfering reactions are 28Si(n,p) %A1and nAl(n,y)"Al. In this investigation small amounts of silicon were detected in aluminum, but no silicon could be detected in the other propellant ingredients and materials used. The interference of the reaction 27Al(n,y)28Al which has a cross section of 0.53 mb for 14.5-MeV neutrons is serious. This compares with 150 mb for 31P(n,~~)~~Al for neutrons of the same (1) A. A. Lbov and I. I. Naumova, Atomnaya Energ., 6,468 (1959), (In Russian), NSA 13, 13335 (1959). (2) Maen To-on,F. Sicilio, and R. E. Wainerdi, Trans. Am. Nucl. SOC.,7 , 328 (1964). 1028
ANALYTICAL CHEMISTRY
energy [Gillespie and Hill (3)]. Using these cross sections one can calculate that the same amount of %A1will be produced during the irradiation in the same flux of 14.5-Mev .neutrons when the ratio of aluminurn to phosphorus is the order of 300. This was the case with samples which contained 200 mg of aluminum and approximately 0.4 mg of phosphorus per gram of propellant. Two methods were used to overcome this interference: chemical separation of the phosphorus and aluminum, and spectrum stripping. EXPERIMENTAL
Neutron Generator. The generator used was a Kaman Nuclear 1001 200-KV unit; 14.5-MeV neutrons are produced by 3H(d,n)4He. With a new target (4 curies 3H/in~h2) output is 1011 n/sec, usable flux several times lOgn/cm2-sec at 1 cm from the target. The flux decreases, of course, with use of the target, and, therefore, all results are adjusted to lo9 n/cm2 sec. Flux Monitoring. A piece of copper wire was included with each sample irradiated. The reaction 63Cu(n,2n)6 2Cu was employed. The annihilation radiation ofe2Cu was counted using a 1.75 X 2-inch NaI(T1) well crystal and a single channel (3) A. S. Gillespie and W. W. Hill, Nucleonics, 19, (No. ll), 170
(1961).
analyzer. An RCA 531 program was used to calculate the fast neutron flux frora the measured 82Cu activity, the irradiation time, the copper weight, and the (n,2n) cross section. Sample Packing. T'he samples (0.5 ml) were packed in small polyethylene capiiules: 0.6 cm in diameter, 1.3 cm long. Sample Transfer. A, pneumatic transfer system moved the samples from the irrad ation position to the counting position in less than 3 seconds. The irradiation position is directly in front of the ceider of the target with the center of the sample 1 cm from it. During irradiation, the polyethylene capsules were spun at approximately 300 rpm by an electric spinner. Sample Counting. After irradiation, the samples were transferred to the counting position, where the copper monitor wire was removed and counted. The samples were counted using a 3 x &inch NaI(T1) well crystal and a 400channel analyzer. If 110 interference occurred, only the 28Al from 31P(n,a) was counted under the 1.78-MeV photopeak. Where there was interference from nAl, the counting procedure was different and is described below. The data accumulated in the analyzer were punched on paper tape and fed into the RCA 501 computer. A program was available to integrate under tht: desired photopeak after subtraction of the background. All activities were calculated for irradiation end and the samples were compared with each other via the copper flux monitors. Interference from 4 1 ( n , ~ ) ~ ~and A l 28Si(n,p)28Al. To determine the interference from nAl(n,y)28Al 1.0 gram of the pure aluminum used in the propellant was irradiated for 30 seconds. After a decay time of 30 seconds (16N) the 1.78MeV y-radiation was counted for 4 minutes under the photopeak as described The average 28A1activity of 5 runs was found to be 3600 cpm/gram A1 at irradiation end (flux adjusted to l o 9 n/cm2 sec). Half life determinations with the same setup in the multiscaler mode gave 2.3 minutes. The propellant ingredients, other chemicals used, and the polyethylene capsules were checked separately for any interference from silicon via 28Si(n,p)28Al. The tests were run in the same way as those described for the aluminum. After subtraction of the background an insignificant number of counts were found under the 1.78-MeV photopeak indicating the absence of' detectable amounts of silicon in these materials. A different method was employed to detect silicon in aluminum. Emission spectrometry revealed the presence of approximately 0.02 silicon in aluminum. At this silicon level, the contribution of 28Si(n,p)28Alto nAl(n,y)28A1 is of the order of 10%. This contribution has to be subtracted from the value obtained for nAl(n,y)28A1which reduces then to 3240 cpm/gram Al. Determination of Phosphorus in Propellant after Separation from Aluminum. According to the preceding discussion the only interference in the determination of phosphorus in the propellant comes from aluminum. This interference can easily be eliminated by chemical separation of the phosphorus from the aluminum. To achieve this, the hydrocarbon polymer binder which contains the phosphorus is extracted from the propellant leaving the aluminum and ammonium perchlorate as the residue. The experimental procedure is as follows: A. Approximately ti grams of propellant are accurately weighed with an analyi.ica1balance into a centrifuge tube. B. The propellant is then stirred with 10-15 ml of pentane for several minutes, C. The mixture is ccmtrifuged, and the supernatant binder solution is decanted. This step is repeated several times. D. The solvent is evaporated from the binder solution. E. The pure binder is irradiated in a polyethylene capsule, and the 28A1from 31P(ri,a) is counted as described. F. A standard is prepared from phosphorus-free propellant binder and triethyl phosphate and irradiated and counted the same way as are the samples. G . From the integmted counts under the photopeak and
:z
Table I. Determination of Phosphorus in a Propellant after Chemical Separation Deviation, Phosphorus,*rng 28A1,cprnn Added Found Standard Binder 2070 66 0.88 ... ... Extracted Binder I 5760 i 169 2.36 2.41 +2.1 Extracted Binder I1 5800 & 107 2.36 2.46 $4.2 a All of the reported values are the average of at least three determinations. * Concentration of phosphorus in propellant sample was 0.05
z
*
the known amount of phosphorus in the standard, the unknown phosphorus content in the extracted binder and in the propellant is calculated by standard methods. Note: This procedure is carried out with an uncured propellant to assure complete extraction of the binder. Determination of Phosphorus in Propellant Using Spectrum Stripping. When a propellant sample is irradiated, the counts under the 1.78-MeV photopeak are derived from 31P (n,a)28A1, and nAl(n,y)28Al, and 28Si(n,p)28A1. Direct spectrum stripping in the analyzer failed mainly because of the differences in neutron flux for the different irradiations. Radiation counts from nAl(n,y)z8Al can be subtracted in a different way. Fast neutron activation also gives rise to the nuclear reaction 27Al(n,p)nMg. nMg has a half life of 9.5 min., main gamma energy of 0.84 MeV. For a given neutron flux, the ratio of *Mg/28Al in A1 is constant. When the amount of A1 present has been determined by counting nMg, the amount of 28A1 can be calculated by dividing the nMg activity by the ratio 27Mg/28A1,The procedure for the nondestructive determination of phosphorus in the propellant is then as follows: A. A standard propellant without phosphorus contamination is prepared from all the ingredients. B. The standard propellant is irradiated, and the ratio of 27Mg to 28A1is determined by counting under the different photopeaks as described. C. A standard propellant with a known amount of phosphorus is prepared, irradiated, and the activities of nMg and 28A1are determined. D. The unknown propellant is irradiated, and the nMg and 28A1activities are determined. E. The nMg activities obtained in Steps C and D are divided by the 27Mg/28A1ratio obtained in Step B. These steps give the 28A1activity derived from Al. F. The 28A1activities derived from A1 are subtracted from the 28A1determined in Steps C and D which gives the 28A1 activities due to P. G . From the 28A1activities obtained in Step F and the known amount of phosphorus in standard propellant Step C, the unknown phosphorus content of the sample propellant can be determined. Usually, 1 gram of propellant is irradiated for 0.5 to 2 min depending on the flux level. Decay time is 45 sec ( IsN). Counting time is 5 min for 28A1 under the 1.78-MeV peak. Because of its higher initial activity, nMg is counted 20-40 min after irradiation. All activities are calculated for irradiation end. RESULTS AND DISCUSSION
Determination of Phosphorus in Propellant after Separation from Aluminum. For these experiments a propellant sample was prepared with a known amount of triethyl phosphate. The binder was then extracted from 5 grams of this propellant as described and compared after irradiation with the results VOL. 39, NO. 8, JULY 1967
1029
Table 11. Determination of Phosphorus in a Propellant Using Spectrum Stripping %A1- -27Mg I 41 Phosphorus, mgb 2*A1,cpm= 27Mg,cpma *7Mg/28A1 cpm Added Found “Standard” “Unknown” Standard without phosphorus
x x
I
1720 =t75 2160 f 61
3.98 X lo4 f 1 . 2 6.08 X lo4 =tz 2 . 6
I1
2260 f 32
6.55 X l o 4 + 6 X 102
1450 f 60
5.92 X l o ‘ & 4 . 6 X lo2
108 108
Deviation,
... ...
750 f 89 680 f 102
0.62 0.52
0.56
+7.7
...
660f
0.52
0.55
1-5.8
58
...
41 f 1.65
All reported values are average of at least three determinations. b Concentration of phosphorus was 0.056x. 5
obtained from a standard binder prepared with triethyl phosphate. Two propellant extracts were made, and the results obtained are shown in Table I. Samples were irradiated for 45 sec in a flux of lo0 n/cmZ sec (En. 14.7 MeV), After a decay time of 45 sec, the %A1was counted 5 min under the 1.78 peak. All count rates are calculated back to irradiation end. Errors of the method are due to chemical processing and counting statistics. The errors found in weighing of the samples are considered to be less than 0.5%. Considerable error comes from the difficulty of preparing good standard propellant samples. Small, 30-gram propellant mixes were made, and it was known from other data that a deviation of the triethyl phosphate concentration of 3 % from the calculated concentration can be expected. The error encountered in the extraction procedure is estimated to be less than 1%, because this procedure is straight forward and quantitative recovery of the extracted binder poses no special problem. The absence of the interfering aluminum in the extracted binder can easily be checked by the disappearance of the strong nMg peak. To obtain complete separation of aluminum and ammonium perchlorate from the binder solution a high speed centrifuge (max. 18,000 rpm) had to be used, because of the micron-sized particles used in the propellant. The standard deviations calculated for the 28A1count rates are given in Table I. Relative standard deviations of these count rates are seen to be on the order of 3 %. Determination of Phosphorus in Propellant Using Spectrum Stripping. For these experiments, two propellant mixes were prepared which contained triethyl phosphate in different amounts. One of these was designated “standard,” the other was “unknown.” A third mixture was prepared containing the propellant ingredients without phosphorus contamination. One gram each of these different mixtures was irradiated for 30 sec in a flux of 109 n/cmzsec. After 45-sec decay time, the BA1 was counted for 5 minutes. The nMg was counted after 2C-40-min decay time for 3 minutes. All count rates are adjusted back to irradiation end. Results are shown in Table 11. The same error considerations as for the chemical separation procedure above apply in this case. Errors due to chemical processing stem from sample weight, considered to be less than 0.5%and from the nonhomogeneous distribution of the triethyl phosphate in the standard propellants, which introduces an estimated error of 3 %. The largest errors are due to the “spectrum stripping” procedure. The standard deviation of the 28A1 and nMg count rates is given in Table 11. The relative standard deviation of these count rates is seen to be around 2 to 4 %. Much 1030
ANALYTICAL CHEMISTRY
higher (10 to 15%) is the relative standard deviation of the %A1 count rate due to phosphorus, after “spectrum stripping” of the =A1 count rate due to aluminum. The relative sfandard deviation of the “Al count rate due to phosphorus was calculated according to Lyon (4). Usefulness of Fast Neutron Activation Analysis for Phosphorus in Propellants and Overall Error of the Method. In the above estimation of errors the difficulties and sources of errors of fast neutron activation in general were not included. Detailed discussions of these problems have been made by Anders and Briden (5) and Mott and Orange (6). Using the results of these authors an overall relative error of 5 % was estimated for the determination of phosphorus after chemical separation, while a relative error of 15% was estimated for the determination of phosphorus after spectrum stripping. It should be noted that it is important that the silicon content in all the aluminum contained in standards and samples must be the same in the case of spectrum stripping. Furthermore, the samples used to determine the nMg/ **A1ratio in aluminum should closely resemble the propellant formulation, because this ratio differs with the composition (41 for the propellant, 53 for pure aluminum containing 0.02 % silicon). From Tables I and 11, it is apparent that the results obtained are too high, especially when spectrum stripping is used. No explanation can be offered for this fact. In conclusion it can be said that the determination of phosphorus in a composite propellant by fast neutron activation analysis has a good potential for use in quality control of propellant production. Because the method is fast, it is useful in the continuous propellant mixing processes now under development. Only small samples are required, which makes it applicable to propellant mixing studies. Finally, the spectrum stripping method is nondestructive and can be used for cured propellant samples, which pose a serious problem for other methods of analysis. ACKNOWLEDGMENT
The authors are grateful to George J. Wynn for writing the computer programs used for the experiments. RECEIVED for review February 3, 1967. Accepted April 27, 1967. Work supported by the Foundational Research Program of the Naval Ordnance Systems Command. (4) . , W. S. Lyon, “Guide to Activation Analysis,” Van Nostrand, Princeton; N.-J., 1964, p. 75. ( 5 ) 0. U. Anders and D. W. Briden, ANAL.CHEM., 36, 287 (1964). (6) W. E. Mott and J. M. Orange, Ibid., 37, 1339 (1965).
