Table I. Conditions for Optimum Signal/noisea Column inlet pressure, psig
Pulse Standing period, current, amperes Carrier gas PSec ArCHl 50 60 1.29 x 1.21 x 10-8 120 1.20x 10-8 180 1.08x 10-8 330 550 0.88X 0.58 x 940 23 60 0.71 x Nitrogen 40 0.64 x 120 0.57 X 180 0.41 x 10-8 330 550 0.26 X 10% 0.16 X 10-8 940 5A molecular sieve column, 190-210 microns. a 6-ft x (1) Sample concentration = 1 :lo11 SFe in nitrogen. (2) Sample concentration = 1:1OloSFe in nitrogen. Temperature, "C 23
of the electron absorption detector itself has a limited dynamic range, and at higher concentrations a nonlinearity in the calibration curve would be predicted due to a nonlinear detector response. This factor should be taken into account in the preparation of calibration curves over a wide dynamic range. An additional problem is the preparation of calibration standards in low concentration ranges. Extreme precision and accuracy are required and the system must be scrupulously clean and inert to minimize errors from contaminants and loss of SF6due to irreversible adsorption in the calibration system. The linearity of the calibration curve indicated that irreversible adsorption of SFs in the chromatographic column is not a serious problem at the concentration levels investigated. Under ideal conditions the response of the electron capture detector approaches 100% efficiency. Thus, if coulometric conditions are assumed, the theoretical sensitivity may be calculated for a captured species such as SFe. The instruampere; therement noise level was observed to be 4 X fore, the minimum detectable signal, defined as twice the noise level, would be 8 X 10-13 ampere. For a 5-ml sample
Noise level, amperes (peak to peak)
7 x 10-13
Signal, amperes at peak maximum (1) 0.20 x 10-10 0.35 x 0.49x 0.68 X 1.03 X
Signal/ noise
x (2) 0.40 x 10-9 0.48 x 10-g 0.56 x 10-g 0.44x 10-g ...
150
0.80
1400
...
and a SFs peak width at half height equal to 3 seconds, this would correspond to a SF6 concentration of 1 part per 1013 by volume. Under practical conditions, however, it is generally not possible to achieve coulometric response particularly when using high carrier gas flow rates. The relatively high carrier gas flow rate of 100 ml/min used in this work was necessary to ensure rapid injection from the 5-ml sample loop. Under these conditions detector efficiency seldom exceeds 70 %, which limits the concentration detectivity to approximately 2 parts per 1013. This sensitivity is approaching the background level of SF6in ambient air which was recently measured by Lovelock in the lO-I4range (IO). ACKNOWLEDGMENT The authors wish to thank R. K. Stevens for 'many helpful discussions.
RECEIVED for review August 30, 1971. Accepted November 9,1971. (10) J. E. Lovelock, Nature, 230, 379 (1971).
Saltzman Method for Determination of Low Concentrations of Oxides of Nitrogen in Automotive Exhaust G . E. Fisher and D. E. Becknell Scientific Research Staf, Ford Motor Campany, Dearborn, Mich.48121 THEANALYSIS of ambient air for NOzconcentrations of 1 ppm and below was originally reported by Saltzman in 1954 ( I ) . Since that time the method has been applied to the analysis of NO, NO2, and/or NO, in a variety of gaseous samples. In 1960, Saltzrnan ( 2 ) reported a modification of the original dye forming reagent, and in 1965 (3) published an extension of his original method to determine N O after being converted to NOz. In this laboratory the method was applied to the determination of oxides of nitrogen in automotive exhaust (1)B. E.Saltzman, ANAL.CHEM.,26, 1949 (1954). (2) Zbid., 32, 135 (1960). (3) B. E. Saltzman, US.Pub. Health Rep., 999-AP-11, p (21427.
( 4 ) at NO, levels ratlging from 200 to 2000 ppm. In this case the oxidizing agent utilized for the conversion of N O to NO2 was a 1 :1 volume ratio of Oz to sample. In this relatively high concentration range, accurate (f5 %) results were obtained with this modification of Saltman's original method. As the amount of NO/NOz present in automotive exhaust emissions decreased, there arose the necessity for a method to analyze for these lower concentrations (10-200 ppm), In this range the modified Saltzman method proved to be successful for the analysis of NOz but was inadequate for NO.
(4) G . E. Fisher, Ford Motor Company, Technical Memorandum NO,RSM 65-2 (1965). ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
863
(5) both in 1965, reported substantiation of the 0.72 figure; whereas, two years earlier Kooiker et al., (6)had claimed that the factor varied with NOz concentrations. In 1966 Stratmann and Buck (7) reported a factor of unity but were disputed one year later by Shaw (8) who pointed out that their reagent, which contained acetone, was not the same as Saltzman had originally utilized. In the same article Shaw refuted Kooiker et al.% claim by noting that their calibration curve had not passed through the origin. Shaw also found the factor to be 0.72 as did Newark (9) late in 1967. It seems as though, at best, there is still room for conjecture on the exact value of the Saltzman factor. To avoid the problem of NaNOz calibration, primary standard NO and NOz cylinders were obtained, which were guaranteed by the vendor to within i l ppm of the stated value, and which were analyzed by chemiluminescence (IO) and found to be within this tolerance. Through the use of blended gas standards the uncertainty in the calibration factor as well as some systematic errors are eliminated. In order to recognize possible interferences in the Saltzman procedure, the reactions which occur must be understood. The reactions have been discussed by several authors ( I , 7 , I I ) .
CALIBRATION CURVE D i m ? Nitrogen Dioxide Gao Direct Nitric Oxidc Gas
PPM
Figure 1. Typical direct calibration curve
400
-
300
-
CALIBRATION CURVE Using 3 0 c c Oxygon 3Hourr OIidation l i m o
EXPERIMENTAL
0
$a K W
6
200-
c W Y
100
-
0-
50
100 PPM NO
I50
200
Figure 2. Typical calibration curve using oxygen
The major problem was that a straight line calibration curve of ppm NO us. optical absorbance could not be obtained. In general, at below 100 ppm NO, absorbances on the order of 30% less than those from comparable concentrations of NOz were obtained. In the belief that this was due to the kinetics of the oxidation reaction a study was undertaken to evaluate various oxidizing agents. Several oxidizing agents including 03,Mn04-, Cr207*-,and HzOz (12%) were tested but Oa produced the most reproducible absorbance values. In addition, the optical absorbances obtained for N O were identical to those from the same concentrations of NOz. The Saltzman method was therefore further modifled to utilize O 3to oxidize NO to NOn, the latter of which was reacted with Saltzman reagent in the usual manner. It was determined that at 200 ppm NO, complete oxidation occurred with 15 ml of 0.1% 0 3 in oxygen. In addition it was established that 30 ml of NO in N2, together with 20 ml of Saltzman reagent and 15 ml of 0.1 % 03, provided a straight line calibration curve from 200 to 25 ppm NO; this curve passed through the origin. In his original paper, Saltman reported the use of N a N 0 2 calibration for the analysis of NO2. In order to use this external calibration, the relationship between NOz- and NOz had to be established. This “Saltzman Factor” was originally reported ( I ) to be 0.72. Since that time much controversy has resulted, some confirming and some refuting the 0.72 relationship. Saltzman (2) and Saltzman and Wartburg 864
ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
Apparatus. During the course of this investigation, 100-ml Hamilton syringes with Kel-F Luer-Lok tips were employed. The syringes were fitted with 18 gauge (Hamilton KF-718) needles with Kel-F hubs which fit into inert plug valves (Hamilton I-M-F-1). The colorimeter was a Klett-Sumerson Model 8971 with a KS-54 filter. The colorimeter cells employed were Klett No. 9006, equipped with a Klett No. 9013 reduction plate. The cell was inserted to provide a 40-mm effective path length. The ozonator was an Orec model 03 VI set to produce 0.1 % O3 generated from a pure oxygen stream. The nitric oxide tanks were fitted with stainless steel regulators (Matheson model 3500). Reagents. The color forming Saltzman reagent was prepared by dissolving 5 grams of sulfanilic acid (reagent grade ACS crystals) in 940 ml of 2.5M acetic acid. After dissolution, 5 ml of a 1 %aqueous solution of N-(1-naphthy1)ethylenediamine dihydrochloride were added and the solution was diluted to volume in a 1-liter volumetric flask. The reagent, when stored in a brown bottle and in the dark, had a shelf life of about one week. Immediately before using, 10 ml of 1% aqueous boric acid were added to each 100 ml of Saltzman reagent. Sample Collection and Analysis. The samples were collected in clean, dry, 100-ml syringes which were conditioned by flushing three times with the gas to be analyzed prior to collection of the actual sample. After conditioning, 20 ml of Saltzman reagent, 30 ml of sample gas, and 15 ml of 0.1 % O3 were drawn into the syringe. Twenty minutes were allowed for the NO, to react with the Saltzman reagent after which the absorbance of the solution was determined. Calibration, Calibration curves were prepared by using primary standard NO and NOz cylinders (Matheson Scientific) containing 25, 43, 60, 100, and 200 ppm NO or NOz in nitrogen, These tanks were analyzed by the vendor using ( 5 ) B. E. Saltzman and A . F. Wartburg, ANAL.CHEM.,37, 1261 ( 1965).
(6) R. H. Kooiker, L. M. Schuman, and Y . K. Chan, Arch. Enoiron. Health, 7 (Part l), 13 (1963). (7) H. Stratmann and M. Buck, Air WaterPollut. J., 10,313 (1966). (8) J. T. Shaw, Atmos. Enciron., 1, 81 (1967). (9) P. Newark, Syringe Method for Measuring NO and NO2 in 200-2000 ppm Range, California Department of Public Health, Vehicle Pollution Laboratory. (10) A. Niki, A. Warnick, and R. R. Lord, SAE, 710072. (11) I. C. Huygen, ANAL,CHEM., 42,407 (1970).
Table I. Comparison of NO and NOz Tank Values with Chemiluminescence and Saltzman Analysis Tank ChemiluSaltzman, minescence, value, PPm PPm PPm 620 f 2.0 626 591 503 i~ 5.8 494 500 198 f 2.0 191 200 95 f 4 . 0 91 100 4 2 f 1.5 43 42 23 i 0.6 23 22 420 f 5.0 425 435 200 i. 5.8 200 200 97 f 2.9 95 100 60 i. 3.0 61 60 38 i.2.9 42 31 24 Z!Z 1 . 7 23 24
Table 11. Analysis of NO, NOz, and N O / N 0 2 Mixtures Prepared by Blending with Nitrogen Blend concentration,“ppm Saltzman NO,, ppm 90 NO2 81.1 f 0.6 60 NOz 62.0 f 1.0 30 NO2 30.1 f 2.3 10 NO2 10.3 f 1.5 85 NO 85.3 3Z 2.5 90 NO 92.0 =t2.0 63 NO 62.3 3Z 1.5 49 NO 48.3 f 2.9 10 NO 9.3 i 1 . 5 30 NO2 21 NO 50.1 f 2.3 13.7 f 1 . 5 ION02 f 3 NO a Analyzed by chemiluminescence.
+
infrared, mass spectrometry, and phenoldisulfonic acid and were verified by chemiluminescence to be within fl ppm of the stated values. Caution had to be taken to flush the regulator valves thoroughly prior to collection of samples. The NO calibration curve was established by analyzing the standard gases as described above; in the case of NO,, the 0 3 was omitted. A typical calibration curve is presented in Figure 1. Samples of exhaust gas were analyzed using the same procedure. In each case triplicate determinations were performed as a check on precision.
Table 111. Comparison of Salbman with ChemiluminescenceAnalysis of Exhaust Gas from CVS Bag Chemiluminescence NO=, PPm Saltzman NO,, ppm 125 127.3 f 1.2 120 120.0 i. 2 . 0 92 95.0 i. 3.0 80 11.7 + 1.5 35 32.3 i 2.5 21 25.3 Z!Z 2.3 24 25.3 j=2.5 26 26.3 f 2.9 22 23.0 Z!Z 1.0
RESULTS AND DISCUSSION
The primary advantage in the use of O3 rather than Oz as the oxidizing agent for the production of NO2 from NO is that the reaction NO
+ + NO, + 0 3
0 2
(1)
is fast, having a half-life of 25 seconds at 0.1 ppm each of NO and 03. Thus at the concentrations of NO found in exhaust gas, and O3 being used as oxidant, the reaction is virtually instantaneous. The only time restriction then is that required for NOz to react and form the characteristic dye (20 minutes). The typical calibration curve obtained using O3as the oxidizing agent (Figure 1) can be compared with one obtained utilizing 0 2 (Figure 2). The curve derived by using 0, is not a straight line and fails to pass through the origin, whereas the one obtained utilizing Oa is a straight line and intersects the axes at the origin. In order to test the procedure on pure NO and NOZ,several commercially available tanks which had been analyzed and certified by the vendor were analyzed. Typical data are listed in Table I; all Saltzman data are averages of triplicate determinations. In addition, a large number of analyses were performed on NO, NOZ,and NO/NOZmixtures prepared with a blending system and analyzed by chemiluminescence immediately prior to sample collection. From these data (Table II), it can be concluded that the method yields accurate results with high precision. Several authors have reported inaccuracies due to the use of O3as the oxidant in Saltzman type analyses. It is believed that problems of the type previously reported are not occurring in this case. If 0 3 were causing the conversion of NOz to nitric oxide vapors, for example, then the agreement between NO and NOz calibration curves as shown in Figure 1 could not be obtained (the NO curve was generated after NO was converted to NOZwith 0 3 , the NOz curve was from NO2 only). The agreement between absorbance values from NO which has been oxidized with 0 3 and that obtained from pure NOz
Table IV. Effect of NH3on Saltzman Analysis for NO, Chemiluminescence NH8 added, ppm NO, present, ppm Saltzman NO,, ppm 200 90 130 133 41 55 100 90 101 19 90 56 63 41 51 30 29 83 *8 9 71 81 50 76 70 51 36 50 50 50 25 40 53 47 25 63 63 60 60 55 53 52 51 46 46 15 53 53 49 49 51 50 10 53 53 49 50 49 48 0 55 54 12 11
seems to be strong evidence that O3 interferences are not occurring in this case. The most stringent tests of the method’s practicality were performed on automotive exhaust samples obtained according to the Federal test procedure. In these tests the Saltzman method was compared with instrumental analyses on the same bag samples. The Saltzman data (Table 111) agreed closely with the chemiluminescence method, varying not more ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
865
Table V. Effect of 1 % Boric Acid on Saltzman Determinations with Samples Containing Controlled NHa Additions Chemiluminesence “3 NO, Saltzman with added, present, Saltzman without Boric acid PPm ppm boric acid NO,, ppm NO,, ppm 100 56 65 f 1 . 0 55 i 2 . 0 90 79 90 i: 3 . 0 79.7 f 0.6 80 85 92.3 f 5 . 0 85.3 i 2 . 5 80 8 8.7 2= 1 . 2 10.0 f 2 . 0 50 52 62.3 i 3 . 1 55.0 k 2.0 15 51 52.3 f 2 . 5 50.0 f 2 . 3 than 3 ppm from chemiluminescence analysis, the latter being a Federally recognized analytical method for NO,. One difficulty encountered in the use of the Saltzmanmethod for NO, was the interference due to NH,. The NH, interference was investigated by blending known amounts of N H , with N O and/or NO2. The Saltzman analyses for NO, was compared to the analyses by chemiluminescence. From the data obtained (Table IV), it has been shown that pp to approximately 25 ppm N H 3 little or no interference occurs. At concentrations above 25 ppin NH,, erratic results were obtained and both the precision and accuracy of the analysis suffered. Two approaches to avoid N H 3 interference are to selectively eliminate N H 3 prior to its introduction into the sampling syringe or to chemically inhibit its interference. Of these, the latter was chosen because of the high probability that any filter which would remove N H 3 would also tend to remove some of the extremely reactive NO2. Since H3BOahas long been recognized as an efficient collection agent for NH,, its use seemed to be a logical further modification of the Saltzman reagent. In order to establish whether or not would effectively remove NH,, six samples were drawn consecutively. Of these the first three utilized the usual Saltzman reagent whereas the last three employed the Saltzman reagent with 1.0% H 3 B 0 3added. From the data obtained (Table V) it can be concluded that the N H 3 interference was
eliminated when H3BOawas added to the Saltzman reagent. This is to be expected since H3BOsforms a poorly dissociated salt with ammonia. Precision of the Modified Saltzman Method. The precision of the method was tested by repeatedly measuring the NO, concentration in a NOIN2 blend where the NO had been analyzed by chemiluminescence to be 100 ppm. The average of fifteen measurements was 104.3 f 2.1 ppm. The reproducibility was also determined on automotive exhaust gas with eight analyses yielding 95.1 =t2.0 ppm; chemiluminescence analysis was 97 ppm. Detection Limits. When the method is used as herein described, a detection limit of 3 ppm can be expected. A lower concentration could be detected by using a larger volume of sample with the same amount of modified Saltzman reagent. CONCLUSIONS
The Saltzman method has been modified to analyze for
low concentrations (0-200 ppm) of NO, in synthetic blends and exhaust gases. The method provides good precision and accuracy over the specified range when compared to the chemiluminescence method. In addition, it furnishes a reliable method for analyzing blends used for calibration of chemiluminescence, NDIR, and N D U V instruments. It can also be used as an occasional spot check of continuous monitoring methods in the analyses of exhaust gas. The Saltzman method involves less sophisticated analytical techniques than other wet chemical procedures and is better adapted for use at test sites. Six samples can be analyzed in about 30 minutes, once the calibration curve has been established. ACKNOWLEDGMENT
The authors wish to acknowledge the valuable contributions of L. P. Breitenbach for his many helpful suggestions and for the preparation of the synthetic blends used in obtaining much of the data in this article.
RECEIVED for review July 26, 1971. Accepted October 21, 1971.
Estimation of the Equivalent Weights of Celluloses Vernon L. Frampton’
Basic Cotton Research Laboratory, The University of Texas, Austin, Texas
GENERALLY A REDUCING GROUP is assumed to form one end of the cellulose molecular chain, and a variety of methods have been proposed for the determination of this carbonyl group Many of the methods proposed for this determina-
(1-11). 1
Present address, Southern Regional Research Laboratory,
U. S . Department of Agriculture, 1100 Robert E. Lee Blvd., New Orleans, La. 70119
(1) A. R. Martin, L. Smith, R. L. Whistler, and M. Harris, J. Res. Nut. Bur. Stand., 27, 449 (1941). (2) H. F. Lamer, W. K. Wilson, and J . H. Flynn, ibid., 51, 237 (1953).
(3) K. H. Meyer, G. Noetling, and P . Bernfield, Hela. Chim. Acta, 31,103 (1948). (4) S. Neusenbaum and W. Z. Hassid, ANAL.CHEM., 24,501 (1952). ( 5 ) A. C. Ellington and C. B. Purves, Can. J . Chem., 31,801 (1953). 866
e
ANALYTICAL CHEMISTRY, VOL. 44, NO. 4, APRIL 1972
tion, such as the “copper number,” the “silver number,” etc., are of doubtful usefulness, since they are not specific and are based on empirical methods for reducing sugars. The experimental conditions under which H C N adds quantitatively to simple aldoses to form cyanohydrins, which may then be hydrolyzed to yield NH,, have been established (6) C. G. Schwalb, Ber., 40, 1347 (1907). (7) M. Bergman and H. Machemer, ibid., 63, 316 (1930). (8) E. Geiger and A. Wissler, Hela. Chim. Acta, 28, 1638 (1945). (9) M. L. Wolfrorn and L. W. Georges, J . Amrr. Chem. Soc., 59, 282 (1937). (10) M. L. Wolfrom, L. W. Georges, and J . C . Snowden, ibid., 60, 1026 (1938). (11) M . L. Wolfrom, J. C. Snowden, and E. N. Lassettre, ibid., 61, 1072 (1939).
