Anal. Chem. 1980, 52, 537-541
537
Determination of Primary Amine Endgroups on Polymers with Fluorescamine in Nonaqueous Solvents Y. Eckstein and P.
Dreyfuss”
Institute of Polymer Science, The University of Akron, Akron, Ohio 44325
A procedure for determining the concentration of primary amine endgroups on polytetrahydrofurans and using fluorescamine reagent in nonaqueous solvents is described. The method is sensitive at very low concentrations, even below lo-’ M. It is fast and more accurate than other analytical methods for polymers of high molecular weight. Preparation of the polymer for analysis is simple. By comparison of the fluorescence intensities from polymers with known concentrations of amine groups with intensities from simple amines, 1,4-diaminobutane was found to be the most suitable for determination of a calibration curve. A method for detecting instability in stock solutions of various reactants is presented.
Recent studies by Dreyfuss and co-workers (1-3) have demonstrated the effectiveness of the alkyl halide/silver salt combination for inducing both the homopolymerization of heterocyclic monomers and the grafting of polymers of heterocyclic monomers from halogenated polymer backbones. On the basis of previous studies, it can be assumed that with 4and 5-membered cyclic ethers these polymerization reactions are “living” in the sense that the active chain ends do not undergo termination or transfer reactions. T h e specific goal of this study was to find a method to determine the concentration of these active chain ends. In 1968 Saegusa and Matsumoto ( 4 ) reported a method for determining the concentration of active oxonium ion chain ends by phenoxy end capping and UV spectrometry. T h e method seems t o give very good results in many cases. However, in the alkyl halide/silver salt systems, preparation of the sample for analysis is long and elaborate and, for unknown reasons does not work well with our preferred counter ions, PFs- and SbFs- ( 5 ) . Here we report a new rapid and simple method for determining the concentration of oxonium ions by quantitative termination t o give primary amino end groups, whose concentration is measured using fluorescamine and fluorescence spectrometry. Because of its high sensitivity, as low as M “,-groups, this method can be applied also to the determination of the number of branches of polyether grafted from a halogenated polymer backbone. Schmidt and Geckeler (6) were the first to report a similar method for the determination of the number of primary amine groups on water soluble polymers, copolymers of l-vinyl-2pyrrolidinone and alanine vinyl ester or allylamine. They treated the amino group with fluorescamine and measured the resulting fluorescence intensity of the reaction product. Their method has several disadvantages if it is to be used with alkyl halide/silver salt systems, especially with those leading t o graft copolymers. (1) Preparation of Calibration Curve. T h e polymer used by Schmidt and Geckeler to prepare a calibration curve is very different in chemical structure from the polymers being studied here. Their degradation method is not applicable and a more closely related calibration standard is needed. 0003-2700/80/0352-0537$01 O O / O
( 2 ) Solubility in Water. The presence of water is essential for determination of the amino group, when using fluorescamine as the reagent, since the unreacted fluorescamine is hydrolyzed by water ( 7 ) . Our polyethers are, unlike the polymers used in ref. 6, insoluble in aqueous solutions. Because of these problems we developed a new method using mainly nonaqueous solvents. T o establish a calibration curve, we used a polytetrahydrofuran ( P T H F ) with known concentration of amino endgroups or an amine whose spectral characteristics are similar to those of our systems and whose fluorescence intensities are in the range of those of the standard PTHF samples.
EXPERIMENTAL Reagents. Fluorescamine, Fluram (Roche, Division of Pierce, Rockford, Ill. 61105, Fluram) was used as a solution of 53 mg Fluram in 100 mL of chloroform (Spectrmalyzed, Fisher Scientific Company). Amines. Reagent grade propylamine and 1,4-diaminobutane were chosen for determination of the calibration curve. Prior to preparation of standard solutions, the amines were redistilled in glass. A series of standard solutions in chloroform were prepared for each amine by dilution of lo-’ M stock solutions from 1.64 p L / 1 mL of propylamine or 1 pL/1 mL of 1,4-diaminobutane. Polymers. Three samples of polytetrahydrofuran (PTHF) with known concentrations of amino groups were used to develop the method. They are designated PTHF-1, PTHF-2, and PTHF-3. These polymers were prepared and analyzed by the BFGoodrich Company. The molecular weights of the polymers determined by several different analytical techniques including content of nitrogen, amine equivalent, and GPC were in good agreement with each other. The general chemical formula of the polymers is:
H,N-CH2CH,CH2CH2-(OCHzCHZCHZCHZ)n-NHZ A series of standard solutions in chloroform was prepared for each PTHF sample by dilution of a stock solution containing 30 mg PTHFI5O mL chloroform. These polymers were used as “standard PTHFs”. The standard PTHFs were prepared by reacting the oxonium ion chain ends with a saturated solution of gaseous ammonia in THF. This procedure is somewhat troublesome and secondary amines are sometimes formed in side reactions (8). Also there was some possibility that the fluorescence might have been influenced by the presence of two amine groups in each molecule. Therefore, PTHFs with an oxonium ion a t only one end were prepared in our laboratory using triethyloxonium hexafluorophosphate as initiator and terminating in nitrogen with a NH4OH-NH&l buffer solution at pH 9 (0.087 g NH,Cl/mL of concentrated “,OH). After standing ,30 min with occasional shaking, the polymers were washed with methanol. The remaining solvent was evaporated and the polymers were dried in a vacuum oven at room temperature. Since ammonia does not interfere with the analysis (6),no special care was taken to remove adsorbed NH4C1salt. PTHF sample B was further purified by dissolving in T H F and reprecipitating into ice water. Stock solutions of the polymers in chloroform contained 20-40 mg PTHF/50 mL of solution. Apparatus. A Perkin-Elmer 512 double beam fluorescence spectrophotometer, with two monochromators, a high pressure xenon source and quartz cuvettes, was used to obtain the fluorescence spectra. The calibration curves were obtained using an excitation X of 395 nm; the emission was measured at 485 nm. C 1980 American Chemical Society
538
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
To compare the spectral characteristics of the amines and the PTHFs, and to check the stability of various stock solutions, fluorescence spectra were measured using excitation at 290 nm, and the excitation spectra were monitored using the 485-nm emission of the various compounds. Procedure. For analysis of PTHF solutions, 20-50 mg of polymer were dissolved in 50 mL of chloroform. The larger weights were used for polymers of higher molecular weight. Stock solutions of amine and of fluorescamine were prepared by successive dilution. The concentrations of the final stock solutions M for the amine and 2 X M for the fluorescamine. were Further mixing and dilutions were carried out in 10-mL volumetric flasks. The solutions were thoroughly mixed with the aid of a Vortex stirrer before chloroform was added to bring the total volume t o 10 mL. A calibration curve was established using solutions containing 200 pL of fluorescamine stock solution, 400 pL of 95% ethanol, and amine solution amounts ranging from 10 to 100 pL. Polymer solutions for analysis contained 200 pL of fluorescamine stock solution, 400 pL of 95% ethanol, and 10-50 pL of polymer solution. The larger volumes were used for polymers of higher molecular weights. Fluorescence spectra were measured as described above. The amine concentration in the polymer solution was obtained by comparison with the calibration curve. n-Propylamine, 1,4-diaminobutane,and the standard PTHFs were each examined as possible standards for determination of the calibration curve. Both 1,4-diaminobutane and the standard PTHFs are suitable in this mixture of reagents. The 400-pL amount of 95% ethanol was selected as optimum after carrying out experiments on solutions containing 100 pL of amine stock solution, 200 pL of fluorescamine stock solution, and amounts of 95% ethanol ranging from 100 pL to 1 mL. At this concentration the solutions remained homogeneous and the emission spectra were reproducible and stable. Examination of t.wo other mixtures in the same way showed that all the ingredients in the above recipe are essential arid that the concentrations given are preferred. Mixture 1 consisted of amine and fluorescamine in chloroform. The purposes of studying this mixture were to determine the range of concentration of aniine in which linearity of concentration vs. fluorescence intensity OCCUTS and to determine the ratio of fluorescamine to amine required to give complete reaction. Two kinds of experiments were carried out. In both, the final concentrations of the amine and fluorescamine stock solutions were lo4 M. In the first, equimolar amounts of amine and fluorescamine were used but the amounts varied from 50 to 100 pL of each stock solution per 10 mL of chloroform. In the second, 100 pL of amine stock solution was mixed with variable amounts (100 pL to 5 mL) of fluorescaniirie stock solution before the final dilution. Mixture 2 consisted of amine, fluorescamine, and water in chloroform. These experiments were run to verify that water is needed to hydrolyze excess fluorescamine and t o determine how much water is required. The final concentrations of the stock for the fluoresfor the amine and 2 X solutions were camine. Each solution for fluorescence measurement contained 100 pL of amine stock solution, 200 pL of fluorescamine stock solution and was diluted with chloroform to 10 mL. The water content was varied from 100 pL to 1.5 mL. A calibration curve of relative fluorescence vs. amine concentration was obtained using solutions with 200 pL of fluorescamine, 500 pL of water, and amine amounts ranging from 50 t o 300 pL of stock solution. The optimum concentrations established for this mixture led t o emulsification and to concern that the emission spectra would be affected by scattering due to water particles. Ethanol was therefore introduced into the mixture of reagents used in the preferred procedure. Calculations. The calibration curve of relative fluorescence vs. concentration of 1,4-diaminobutane is linear. The concentration of amino groups in moles per 1 gram of polymer was calculated from the following equation:
and the molecular weight is given by l/[NH,] assuming one amine group per molecule. Here c represents Concentration of the diluted sample (in moles), obtained by comparison with calibration c u n e .
VI represents the volume in mL of stock solution in which W grams of polymer were dissolved; V, represents the volume in mL of the final solution in which the reaction with fluorescamine was carried on; IJ represents the volume in mL of solution taken from VI which reacted with fluorescamine in V,;W represents weight in grams of the polymer dissolved in VI.
RESULTS AND DISCUSSION Methods. The objectives of this research were threefold. (1) Development of a general procedure in nonaqueous solvents for determination of primary amino groups in polymers. ( 2 ) Search for a suitable amine which can be used for determination of a calibration curve. This was accomplished by comparison of the fluorescence intensities and features of the emission and excitation spectra of the fluorophor derived from the amine with t h a t from standard PTHF. ( 3 ) Use of the calibration curve obtained from the fluorophor derived from the amine chosen in (2) to determine the number of amine groups in P T H F prepared by termination of the polymerization using NH40H-NH4C1buffer solution. A brief discussion of the experiments which preceded adoption of the preferred procedure follows. We believe that such information will be of interest to other investigators. Mixture 1 . Amine Fluorescamine in Chloroform. At equimolar concentrations, the fluorescence intensity was a measure of fluorescamine content rather than of the amine curlcentration. When the concentration of fluorescamine exceeded that of the amine by about three orders of magnitude, differences in fluorescence intensities as a function of atiiine Concentration were observed; however, they were small. The results confirmed that water is important for hydrolysis of the unreacted fluorescamine. Mixture 2. Amine + Fluorescarnine + Water in Chloroform. T h e optimum conditions for measurement of' fluorescence intensity were achieved when the concentration of the different reactants were 200 pL of fluorescamine stock solution (2 X M), 500 pL of water, and 50 t o 300 pL of amine stock solution ( l W 5 M). A linear relationship between fluorescence intensities and amine concentration was then observed. Coinparison of the excitation spectra and intensities of spectra obtained from a solution of standard PTHF-1 with those obtained from solutions of the amines indicated that with this mixture n-propylamine can be used as a standard for obtaining a calibration curve. T h e fluorescence intensities from 1,4diarninobutane were lower and changed constantly in intensity as a function of time. T h e readings of the n-propylamine standard were stable for 2-3 h. T h e disadvantages of this method include inhomogeneity of the solution (dispersion of water in chloroform) and long reaction time of the aniine with fluorescamine (50-60 min). Preft.1-red Procedure. Amine Fluorescutnitze + 95 7 G E t h a t d i t i Chloroform. This mixture of reactants produced the most reproducible results and the reaction was very fast (1 3 inill). The resulting fluorescence was proportional to the a i l h e coilcentration and the fluorophor was stable for 3--4 h. Apparently, addition of 95% ethanol brings the reaction betweeii fluorescarnine and the amine to cunipletioii. Immediately after addition of the ethanol to the reaction mixture containing amine and fluorescamine only, a n increase of fluoresceiice occurs (see also Figure 1). 'I'he fluorescence intensities obtained from n-propylamine were eight times higher than those from 1,4-diaminobutane a t a similar concentration of amino groups. Comparison of these fluorescence intensities with those of PTHF standards indicated that 1,4-diaminobutane is the better standard in this mixture. Moreover, the excitation spectrum of the diamine monitored a t 485-nni emission had spectral features resembling those of the P'I'HF standards. 'I'he spectral features of iL-propjlmiine h e i e different (see Figure 2 ) .
+
+-
,
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
539
,
---
I -
1.0
EXC!TATION A T 485nm
SPECTRUM
EMISSION
EMISSION
,
,'
SPECTRUM
%
',
,I,
5
1
'
0.01
-
._..._._
250
Figure 1. Fluorescence spectra of chloroform solution containing: 1,4diaminobutane fluorescamine or (- - -) 1.4diaminobutane fluorescamine ethanol
(-)
+
+
+
310 WAIELENGTH
370
-0
nm
Figure 3. Excitation spectra of chloroform solution containing PTHF-1, fluorescamine, and ethanol monitored at 485-nm emission. (-) and (- - -) correspond to 2 days and 3 weeks after preparation of the stock solution of PTHF, respectively
Figure 4. Emission spectra of chloroform solution containing PTHF-1, fluorescamine, and ethanol excited at 285 nm and measured immediately after mixing of the reactants. (-) and (- - -) in presence of freshly prepared and month-old solutions of fluorescamine, respectively
250
3 IO
370
430
.... nrn
WAVELENGTH
Figure 2. Excitation spectra of chloroform solution of amines ifluorescamine ethanol monitored at 485 nm emission. (. (-), and (---) correspond to n-propylamine, 1,4diaminobutane,and PTHF-1, respectively
+
m),
Sources of Error. T h e order of a d d i t i o n of reactants is important. T h e correct order is amine stock solution, fluorescamine stock solution, 95% ethanol, and finally chloroform. For reasons yet unknown, this order of addition and thorough mixing after each addition are essential for reproducibility of the measurements. Another factor that can affect reproducibility of the measurements is the stability of stock solutions of t h e reactants. The solutions can be stored for almost 1 month; the exact time is unknown. Changes in the stock solution usually can be detected by measurements of emission and excitation spectra of the reaction mixture. Changes with time in the excitation spectrum of the P T H F stock solution are shown in Figure 3. I n t h e earlier stages of instability, lack of proportionality between fluorescence intensity and concentration of P T H F occurs. Instability in the stock solution of fluorescamine can be detected by changes in the fluorescence spectrum of the reaction mixture excited a t 285 nm, as shown in Figure 4. Changes in the stock solution of t h e amine can be detected by measurement of excitation spectra of the reaction mixture at 485-nm emission. Typical changes in 1,4-diaminobutane
Figure 5. Excitation spectra of chloroform solution containing 1,4diaminobutane, fluorescamine, and ethanol monitored at 485-nm emission and taken immediately after mixing of the reactants. (-), are fresh, one-month, and two-month old 1,4-di(---), and (. aminobutane, respectively e)
solutions stored for one and two months are presented in Figure 5. While instability in the fluorescamine stock solution does not cause problems in the determination of N H 2 end group concentration as long as the same fluorescamine solution is used in preparation of the samples of the polymers and of the standard solutions, instabilities in stock solutions of the polymers or of the 1,4-diaminobutane are crucial for final determinations. Lack of proportionality between the fluorescence and concentration can be caused by quenching of the fluorescence due to ion-ion interaction a t high concentrations of amino groups. When the concentration of NH2's on P T H F is higher than 10%M, quenching occurs. Lack of proportionality detected a t concentrations much lower than lo4 M of NH2 is probably caused by instability of the stock solution of PTHF.
540
ANALYTICAL CHEMISTRY, VOL. 5 2 , NO. 3, MARCH 1980
Table I. -Comparison of Number Average Molecular Weight (iM, ) of PTHF Polymers Calculated from Different Data -
method of determination fluorescence spectroscopy
%N
amine equivalent GPC
&VI,,
-
PTHF-1
PTHF-2
PTHF-3
13050
27 580
7 900
11700 11 400 11 080
19300
..-
20 000
6 600 7 600
.__
Table 11. Determination of Concentration of Amino Endgroups on PTHF by Fluorescence Spectrometry reaction sample time, min no. A B C D E F G
3
9 11
13 9 9 9
of buffer,
reaction time with buffer,
polymer
mL
min
2.67 1.38 1.06 0.98 0.176
10 10 10 10
10 10 10 10
1
10
0.180
10 10
10 30
[",I
volume
lo',
x eq/g
0.161
-
0
a3
t
j
.
,
.
4
TIME
Figure 6. Variation of ",-endgroup reaction time
8 (min.1
12
I
concentration as a function of
Table 111. Endgroup Analyses of PTHF-B Other worthwhile precautions include the following. Since PTHF itself reacts very slowly with fluorescamine (5G80 min), a n d its fluorescence is stable over 24 h, it is advisable to prepare the reaction mixture for a n unknown P T H F 16-22 h prior to t h e measurement of its fluorescence. T h e calibration curve with 1,4-diaminobutane as the standard must be prepared close to the time of measurement. If a calibration curve is based on standard P T H F s with known amino content, it is wise to examine three to five samples, each of a different concentration, and to measure the fluorescence intensities vs. concentration just prior to measurement of the intensity of unknown sample. This is a test of the stability of t h e stock solutions. Traces of amine in the reaction flask are a potentially large source of error. Generally, two to three samples should be prepared for each value t o be determined, and the average of t h e observed fluorescence intensity should be used for calibration or determination.
no. of end-
groups x 10'
% of
total
endpolymer groups' eq/g
endgroup C1
NH, NH, OH OH
i
NH,
method
%
elemental analysisb fluorescence
0.065
0.18
10
---
1.38
75
element a1
0.18
1.29
anal ysisb
NMR'
.__
.__
--.
1.65 0.27"
15
a The t.otal number of endgroups was taken as t h e s u m of the OH . NH, determined by NMR + the C1. By Galhraith Laboratories, Inc. Total "hydroxyl" n u m b e r = 9 . 3 determined by J. C. Westfahl of the BFGoodrich Co. The difference between the OH - N H , determined b y S M R a n d the N H , determined by fluorescence spec-
'
Determination of Amino End Group Concentration of Various PTHF Samples. All the results given in this section
trometry.
were obtained using t h e preferred procedure. Table I shows a comparison between molecular weights of t h e three P T H F standards as calculated from different data a n d by use of our method. At low molecular weights, good agreement exists between t h e different methods (PTHF-3). However, a t higher molecular weights the agreement is not as good. We believe t h a t the determination of amino group concentration by fluorescence measurements is more accurate t h a n t h e value obtained by other methods for the following reasons. The fluorescence method is sensitive a t very low NH, M. T h e sensitivity of the concentrations, even below amine equivalent methods decreases with increasing molecular weight of the polymer; a molecular weight of 20 000 is usually t h e upper limit for its accuracy. T h e fluorescence method measures only the primary amine end groups. As stated above, some molecules of the polymer may contain secondary or tertiary amines. When this occurs, t h e molecular weights calculated from 90 N will be too low. T h e GPC trace of PTHF-3 showed a symmetrical distribution of molecular weight; the GPC traces of PTHF-1 and PTHF-2 each showed a long tail on the side of high molecular weight. This is consistent with t h e presence of dimers and trimers, secondary and tertiary amines, and the lower values of mo-
lecular weight calculated from the % N in these samples.
Efficiency of Chain End Termination by N H 4 0 H NH,C1 Buffer Solution. As stated in t h e Experimental section, some P T H F samples were terminated by an N H 4 0 H-NH4Cl buffer solution. In Table I1 are listed concentrations of amino end groups per 1 g of polymer, determined by the fluorescence method. The concentration of amino end groups vs. reaction time are plotted in Figure 6. As the molecular weights of the polymers increase, the concentration of the NH, end groups first decreases and finally reaches a constant value 12-13 min after t h e beginning of the reaction. In order t o test t h e accuracy of our method and the efficiency of t h e termination reaction, analyses for all possible end groups in sample B were performed. The results are given in Table 111. Good agreement exists between our method and the NH, concentration calculated from nitrogen analysis. Clearly, only 75% of the chains are terminated by NH2 groups and the remainder have either C1 or OH end groups. However, this percentage is constant, provided the same buffer is used each time, and this method of termination can be used. Further evidence for the efficiency of t h e termination was obtained from molecular weight calculations and GPC analysis of PTHF-C. T h e number average molecular weight deter-
Anal. Chem. 1980, 52, 541-546
mined by vapor phase osmometry was 8068 and that calculated from the total number of end groups was 7645. The GPC trace indicated a symmetrical distribution with no high molecular weight tails. T h e results obtained from measurement of the number of amine groups in samples E, F. and G (Table 11) indicate that the termination reaction is fast and efficient, since the concentration of amino end groups is independent of the amount of buffer solution and of time of contact with the buffer solution.
541
on unsaturated polymer backbones and therefore of the number of polyether branches grown from a halogenated polymer after addition of silver salt.
ACKNOWLEDGMENT We thank F. Hutterer of Northeastern Ohio Universities College of Medicine for the use of the fluorescence spectrometer and M. P. Dreyfuss of the BFGoodrich Co. for providing the standard P T H F samples.
LITERATURE CITED
CONCLUSIONS
( 1 ) Lee, K. I.; Dreyfuss, P. "Ring-Opening Polymerization", ACS Symp. Ser.
T h e fluorescamine method has the advantage over other analytical methods of determination of end groups on polymers in its high sensitivity even a t high molecular weights of the polymers. We have found that this method provided quantitative results in less time, with less sample, and with greater sensitivity than obtained by other analytical methods. In addition the method can be utilized in further investigation of the mechanism of initiation of P T H F polymerization and quantitative determination of oxonium ion concentration. Preliminary experiments show that this method is also applicable to the determination of the number of amine groups
1977, 59, 24. (2) Lehmann, J.; Dreyfuss, P. "Multiphase Polymers", Adv. Chem. 1979, 776, 587. (3) Eckstein, Y.; Dreyfuss, P. J . Inorg. N m I . Chem., in press. (4) Saegusa, T.; Matsumoto, S. J . Po/ym. Sci., Par7 A - 7 1968, 6 , 1559. (5) Saegusa, T. Kyoto University, personal communication, 1976. (6) Schmidt, K.;Geckeler, K. Anal. Chim. Acta 1974, 77, 79. (7) Udenfriend, S.; Stein, S.; Bohlen, P.; Dairman, W.; Leimgruber, W.; Weigele, M. Science. 1972, 778, 871 (8) Hubin, A . J.; Smith, S. U.S. Patent 3 8 2 4 197, July 16, 1974.
RECEIVED for review March 15,1979. Accepted December 20, 1979. Work supported by the
U.S.Army Research Office.
Improved Neutral Buffered Potassium Iodide Method for Ozone in Air Gijsbertus Bergshoeff
and Roelof W. Lanting
TNO Research Institute for Environmental Hygiene, Postbox 2 14, Delft, The Netherlands
Jozef M. G. Prop Central Laboratory of Dutch State Mines, Postbox 18, Geleen, The Netherlands
Hans F. R. Reynders National Institute of Public Health, Postbox 7, Bilthoven, The Netherlands
This paper describes a neutral buffered potassium iodidepotassium bromide-thiosulfate (KIBRT) method for ozone in air, proposed in The Netherlands as a standard reference procedure for calibration purposes. I n this new variant, ozone is absorbed in a 1 % potassium iodide solution containing a known amount of thiosulfate and 2 % potassium bromide. The thiosulfate prevents losses of iodine, resulting in a substantial improvement of the collection efficiency and sample tenability. After sampling, a known excess of iodine is added to the absorber and this excess is determined simply by photometry at 352 nm. Subtracting a blank (obtained with ozone-free air) eliminates errors by instability of the solutions and by the possible presence of reducing impurities in the reagents. The time between sampling and further treatment of the sample is not critical. The need for high purity water and careful conditioning of glassware is demonstrated. Owing to lowering of the phosphate concentration and addition of 2 YO potassium bromide to the absorbing solution, the influence of sampling flow rate and relative humidity on stoichiometry has been eliminated. Comparisons with gas phase titration (GPT) in an interlaboratory study yielded the relation: KIBRT = 0.96 ( & 0.02). GPT ( 9 5 % confidence interval).
T h e neutral buffered potassium iodide (NBKI) method is well known as a standard reference procedure for ozone measurements (1-9) and has been officially recommended by 0003-2700/80/0352-0541$0 1 O O / O
such agencies as the US.EPA ( I O , II), the WHO ( 1 2 ) ,and, in a modified version, by the VDI in the Federal Republic of Germany ( 2 3 , 2 0 ) . It is based on the assumption (1) that 1.0 mol of I, is liberated per mol of O3 absorbed. This 1:1 stoichiometry has indeed been reported by several investigators ( 4 , 5, 14-16) using quite different methods. However, with the introduction of absolute instrumental calibration techniques like ultraviolet photometry (UV), gas phase titration (GPT) and long path infrared spectrometry in the area of ozone calibration, i t was established that the stoichiometry of the NBKI method (10, 1 1 ) was dependent on relative humidity of the sample (6, 17). At the same time several other inconsistencies with regard to the method arose (5, 7 , 8, 17, 18). Nevertheless, we believe that the simple and cheap manual potassium iodide method will retain a place beside the instrumental calibration techniques, provided some of its drawbacks can be eliminated. We therefore started with detailed interlaboratory studies of the K I T (neutral buffered potassiium iodide thiosulfate) method proposed by Bergshoeff (14, 19) and obtained very good results pertaining to loss of iodine, repeatability, limit of detection, sample keepability, and collection efficiency. These results show the favorable effect of thiosulfate and are presented in the first parts of this paper. Since the stoichiometry of the KIT method was dependent on the relative humidity and the flow rate of the sample, we also studied the neutral buffered potassium iodide-bromide-thiosulfate variant. This KIBRT method combines the C 1980 American Chemical Soctetv