Iodometric Microdetermination of Organic Oxidants and Ozone Resolution of Mixtures by Kinetic Colorimetry BERNARD E. SALTZMAN Public Health Service, Bureau o f State Services, Occupational Health Program. U. Education and Welfare, Cincinnati 2, Ohio
S. Departmenf of Health,
NATHAN GILBERT University of Cincinnati, Cincinnati 2 7, Ohio
A basic technique of kinetic colorimetry was developed for determining oxidants in mixtures. An appropriate reagent to produce a slowly developed color was added. Components were resolved graphically using a plot of logarithm of undeveloped color (maximum absorbance minus that a t given time) vs. reaction time. lodometric microdeterminations were thus made of liquid dilutions of various oxidants. This technique was applied to study the stoichiometry of iodide reagents with low concentrations of ozone in air. Reactions appeared to involve both hypoiodite and perphosphate. The most satisfactory reagent was 1% potassium iodide in neutral phosphate buffer. Comparisons were made with an independent method found to b e specific for ozone in gaseous mixtures, based on: converting ozone to nitrogen dioxide by reaction with excess nitric oxide. These methods should be valuable for differentiating natural smog oxidant into its components; the kinetic colorimetry technique should have broad analytical applications in making specific determinations, even if the available reactions are non-
specific.
I
methods have been widely used for the microdetermination of smog oxidants and ozone in air pollution and associated toxicologic studies. However, interpretation of the results is restricted because of the nonspecificity of these methods and the uncertainty of the stoichiometry of iodine liberation. This report deals with an important part of a study (3,14-17) of the kinetics of smog oxidant generation by the reaction of ozone with 1-hexene, which required precise and specific methods for the determination of low concentrations of oxidants. Upon passing ozone through iodide reagents, whereas most of the iodine appeared to be liberODOMETRIC
1914
ANALYTICAL CHEMISTRY
ated instantly, the spectrophotometric readings continued to increase slowly until 45 minutes or more had elapsed, indicating other slow reactions. Slow liberation of iodine from such reagents had been reported (4) for hydrogen peroxide and a variety of organic hydroperoxides, the rates decreasing as the complexity of the organic molecules increased. A systematic investigation of these kinetic factors was therefore undertaken to determine the possibilities of differentiating between various organic oxidants as well as to clarify the chemistry of ozone analysis. A correlative study of ozone-iodine stoichiometry for various reagents was also conducted. KINETIC COLORIMETRY APPLIED TO OXIDANT MIXTURES
Iodornetric methods are standard procedures for analysis of organic peroxides (11, 60). The technique of analysis of binary mixtures by measurements at various times after addition of a slowreacting reagent was reviewed by Lee and Kolthoff (7). An involved inathematical calculation was required as well as knowledge of the kinetic constant for the reaction rate of each of the two components. Because the kinetic constants were not available for the various unknown oxidants to be studied, a simpler and more direct method of treating the data waa necessary. A simple graphic method was developed which permitted not only separation and determination of microgram quantities of oxidants in binary and possibly ternary mixtures, but also evaluation of their kinetic constants for iodine liberation. ’ This kinetic colorimetry method was applied to hydrogen peroxide, peracetic acid containing hydrogen peroxide impurity, and methyl ethyl ketone peroxide mixture, using neutral and alkaline iodide reagents (3), with results illustrated in Figures 1 and 2.
Reagents. Highest quality analytical grade chemicals were used. The reagents were stable for months. DOUBLE-DISTILLED WATER. Distilled water was redistilled from an all-borosilicate-glass still after addition of a single crystal of potassiuni permanganate aLid of barium hydroxide. This was used for all reagents. NEUTRALBUFFERED POTASSIUM 10DIDE, 13.61 grams of potassium dihydrogen phosphate, 14.20 grams of anhydrous dibasic sodium phosphate, and 10.0 grams of potassium iodide successively dissolved. The mixture was made to 1 liter. ALKALINEPOTASSIUM IODIDE.40.0 grams of sodium hydroxide, then 10.0 grams of potassium iodide dissolved. The mixture was made to 1 liter. ACIDIFYING REAGENT(for alkaline potassium iodide). Five grams of sulfamic acid were dissolved in 100 ml. of water, then 84 ml. of 85% phosphoric acid were added and the mixture was made to 200 ml. Two milliliters of this reagent were added to 10 ml. of alkaline iodide reagent after sampling to liberate the iodine, and the mixture was cooled to room temperature in a water bath. Procedure. Only a very small amount of oxidant, equivalent to about 5 y of active oxygen, was required for each test. Each substance was freshly diluted in water or glacial acetic acid and the required quantity was transferred to a dry colorimeter tube (2-em. light path) in a volume of a few hundredths of 1 ml. Ten milliliters of the reagent were then added and rapidly mixed with the sample. Spectrophotometric readings at 352 m+ were started immediately after sample treatment with the neutral reagent, and immediately after rapid acidification and cooling of the alkaline reagent, over a period of about an hour, until the maximum iodine color was obtained and fading began. Readings were made to 0.001 absorbance unit by using a niagnifying glass over the meter pointer of a Beckman DU spectrophotometer and working with highest possible accuracy.
Graphical Method. In the kinetic colorimetry plots of Figure 1, the
2
't
't
31
2t
-a 8
W
0 0.5 z
8
a 0 ul
m
4 0.2 0 W
n
sw
>
0.1
W
n
z 3
1
0 05
0.05
0.03
0.031 0
I
Figure 1 . Kinetic colorimetry using neutral potassium iodide reagent. A, B. Hydrogen peroxide C. Peracetic acid containing hydrogen peroxide as impurity D. Methyl ethyl ketone peroxide mixture (dilution in glacial acetic acid)
abscissa is the time in minutes after mixing the sample and reagent. The ordinate is the logarithm of the undeveloped iodine color absorbance (the maximum absorbance minus the absorbance a t the time plotted). Because of the large excess of iodide, the reaction of these oxidants with the reagent was found to follow a pseudounimolecular pattern in preliminary experiments. It was therefore expected that if a semilog plot were made for each unreacted oxidant us. time, a straight line would result. The ordinate selected waa the optical equivalent of the total unreacted oxidant. A straight-line plot for total unreacted oxidant was evidence of the presence of only a single component. The kinetic constant, k, and the half life, f l I 2 , of the iodine liberation reaction were computed from any two points on the line, with coordinates (al,tl) and (A&), where A and t denote undeveloped optical absorbance and time, respectively:
Two different dilutions of hydrogen peroxide yielded two parallel lines, A and B, Figure 1, having a slope equivalent to a half life of the reaction of 11.9 minutes a t 27" C. The intercepts on the vertical axis corresponded to the total quantities of peroxide taken for analysis. Binary patterns were noted for peracetic acid and methyl ethyl ketone peroxide, which yielded curved lines C and
I
IO
1
20
TIME,
I
& ,
0 IO MINUTES
I
4
20
30
Figure 2. Kinetic colorimetry using alkaline potassium iodide reagent A.
Hydrogen peroxide Methyl ethyl ketone peroxide mixture (dilution in glacial acetic acid)
B.
D, respectively, each having one elbow. The resolution of these lines into the sum of two straight lines representing the two components is illustrated by the daahed lines applied to curve D. This curve is linear after about 15 minutes, indicating that only a single component remained and that the fast-reacting component had been almost completely depleted. The straight portion was therefore extrapolated upward to the left to zero time. The values of undeveloped absorbance corresponding to the slow-reacting component were graphically obtained from this extrapolated line and subtracted f:om the line representing the total undeveloped color. It waa necessary bo make this subtraction arithmetically rather than graphically. because the gIaphical quantities were logarithmic. The difference, representing undeveloped absorbance values for the fmt-reacting component, is shown aa the dashdot line, which i s perfectly straight. Discussion. The quantity of each component was determined by the intercept of its line a t zero time. The component analyses for both the methyl ethyl ketone peroxide and the peracetic acid checked well with manufacturers' data. The components were identified from the slopes of their lines. The slow-reacting component of peracetic acid waa readily shown to be hydrogen peroxide. The hydrolysis of a 1 to IO00 aqueous solution of peracetic acid yielding hydrogen peroxide could be followed easily and accurately. After 4 days, while the total oxidant was 91% of the
original value, the fraction of slow color representing hydrogen peroxide increased from 18 to 34%. This illustrates the utility of the technique in treating what would otherwise be a difficult analytical problem for large samples, and an impossible problem for micro quantities. The reactions of these materials were also followed using the alkaline iodide reagent, with very interesting results. The abscissa shown in Figure 2 represents the time after the reagent was acidified. Curve A for hydrogen peroxide' shows the presence of two componenta when the hydrogen peroxide had been allowed to react with the alkaline iodide reagent for 5 minutes before acidifice tion. More strikingly, under similar conditions, no iodine a t all was obtained for the peracetic acid, in spite of its peroxide content. However, if the alkaline reagent contact time were reduced to 15 seconds, about one third of the expected iodine was obtained after acidification. This suggests that reactions occurring are complex in the strongly alkaline solution. Peroxides and peracids are known to show reducing properties with liberation of oxygen under such conditions. Apparently the peracetic acid decomposes with a half life of about 10 seconds under these conditions, and reduces the peroxide during the process. The presence of iodide in the alkaline solution appears to be necessary for these results: no such effect was observed when peracetic acid wm allowed to stay in contact with 1N sodium hydroxide alone, followed by acidification and addition of potaasium iodide, nor VOL 31, NO. 11, NOVEMBER 1959
1915
when peracetic acid was added to the previously acidified alkaline potassium iodide reagent. I n the case of the methyl ethyl ketone peroxide mixture, similar changes noted in Figure 2 indicate the appearance of a new component which is very fast-reacting. Further study is necessary to eliminate the possibility that these new components may be artifacts caused by errors in the reaction rate due to momentary heating after addition of acid and prior to cooling. The kinetic colorimetry method a p pears to offer very interesting possibilities. Its application to a gaseous concentration of hydrogen peroxide was another illustration of its efficacy. Because a perfectly straight line was obtained with the same slope as that found for liquid samples, it was demonstrated that no substantial decomposition or alteration had occurred during the vaporization to produce low concentrations in air. When this method is applied for sampling gaseous oxidants a t low concentrations, the question of absorption efficiency becomes important. Slowreacting components cannot be expected to be absorbed efficiently. However, the method appears to be suitable for samples such as would be obtained by a liquid air trap from gaseous pollutants. A catalog of kinetic reaction rates for various organic oxidants could be readily made, thus permitting the identification of unknown materiale from the slopes of their plots. A search was made to determine a
suitable diluent for non-water-soluble oxidants. Because great dilutions are required, only the most inert solvents may be used. Dioxane and Cellosolve had too much peroxide impurity for use, although this could be reduced by shaking with zinc dust and filtration. Glacial acetic acid appeared to be satisfactory. Reaction times can be brought into a convenient range by a suitable choice of reagents and reagent concentrations. By simultaneous use of several reagents, additional differentiation may be obtained, as illustrated in the cases of the neutral and alkaline iodide reagents. Some tests were made using neutral 1% potassium iodide reagent with higher hydroperoxides (diisopropyl benzene hydroperoxide, cumene hydroperoxide, m- and ptertbutyl isopropyl benzene hydroperoxide, and pmenthane h y d m peroxide) with which the half lives of iodine liberation were found to be about 2 hours. With these substances, a 5% potassium iodide reagent would be more suitable. This method may be applied not only for the analysis of oxidant mixtures with iodide reagents but for any colorimetric procedure whev there is a sufficient difference in the reaction rates of the components being analyzed. THE IODOMETRIC MICRODETERMINATION OF OZONE
Iodometric reagents for ozone were studied for both their stoichiometry and
Table I.
NO. % KI 1 20 2 20 20 3 4 20 5
5
G
5 5 5
7
8 9
12 13
5 5 2 2 2
14 15 16 17 18
1 1 1 1 1
10 11
Sampling Reagent Buffer 0 . 1 M KHiPO44.1M N&HPO4 1 % N&B407.10 HfO 0 . 1 N NaOH 1N NaOH
iodine liberation rates. Low concentrations of ozone in air have been determined with neutral phosphate-buflered (9,10, 12, 19) and alkaline (3,18) reagents. Although the amount of iodine liberated has been shown (I,,??, 6) to be equivalent to 1% = Osfor high ozone concentrations of several volume per cent, the stoichiometry has not been eatablished for low concentrations because of the difficulty of preparing accurately known low concentrations of this unstable and reactive substance. H e n berger (@ showed that the amounts of iodine liberated from buffered 0.01N potassium iodide varied from 90% at pH 9 to 113% a t pH 1 as compared to that at pH 7. The early findings (9) of the present investigation were that some stoichiometric variation occurred for neutral and alkaline iodide reagents when sample size and ozone concentration were varied over great ranges. DBculties were also noted because of instability of certain reagents both before and after sampling. This report deals with further studies to develop a stable and precise reagent, and to elucidate the chemistry of ozone reaction with iodide reagents.
Tests of Reagents. A dynamic flow system was used t o generate a steady low concentration (about 10 p.p.m.) of ozone. Ozone was generated by three 4 w a t t ozone ultraviolet bulbs in a brown 2-liter bottle, through which a metered stream of purified air (0.35 liter per minute) was paased;
Tests of Iodide Reogents for Ozone
pH Stability. 6.79 P G 9.00 G 13 14 G
1 % N&B&. 10 HzO, O.S'%NHr SOaH, 0 . 5 0 AcOH 0 . 1 M KH2Px4-0.1M N&HP04 1% N&B&. 10 HzO ' N&BIO~. 10 HzO 1%
3.35
F
6.71 9.00 9.00
F G G
1% N&B&. 10 HzO 1% N&B407.10 HtO 2% NaHCO.
9.00 9.00 8.42 9.00 9.00
G G G G G
6.80 9.04 9.04 9.04 14
G G G G G
Acidifying Reagent6
None 0 . 5 ml. 20% NHBOJ3 0 . 5 ml. 20% NH&O,H, 0 . 1 ml. AcOH 1 ml. 5% NH808H-5% AcOH None None 0 . 5 ml. 2070NHBOJ3 0 . 5 ml. 20% NH&OaH, 1 ml. 0.5M KHQOr 0 . 5 M NaHPO4 None None 0 . 5 ml. 20% NH80rH 1 ml. glycerol 2 ml. 36% H,P04 satd. with "BOSH
Final Color Stabil- Relative itya-c Lld PH 6.79 1.30 1.15 2.29
P+
P+ P+ P+
1.03 1.75 1.44 0.70
3.35
F+
0.92
6.71 1.52 1.55
FF+ F+
0.93 0.83 0.86
3.35 9.00 8.42 1.40 6.64
F+ GP-
0.70 0.64 0.69 0.93 0.69
6.80 9.04 1.68 7.16 2.28
G+
P+ G-
FF+ P-
FG
+
1 .OO 0.72 0.79 0.67 0.64
G good, F fair, P poor. Added to 10-ml. portion of sampling reagent after sample collection. Color increasing with time, - color decreasing with time, 1 Optical absorbance converted to approximate iodine values, baaed on iodine extinction coe5cienta of 95 for 1% KI, 100 for 5% KI, and 104.3 for 20% KI. b
0
+
1916
0
ANALYTICAL CHEMISTRY
"r
A
SAMPLING HEXENE KEY RATE LP.M. RRM. X 030 11.8 A 0.50 1.41
0 0
A
e
irz oj
0
I
I
0.5 TOTAL d;kORBANk5E
2.0
TOTAL
2 m a
c
b L 0. 0.2 $.Oo
BUFFER MOLARITY
I
SLOW ABSORBANCE
Figure 3. Kinetic colorimetry of ozone reaction with neutral potassium iodide reagents A.
Reagents with various iodlde concentrations in 0.1 M KHrPOc
0.1M NarHPO, buffer
B, C.
Reagents with various buffer concentrationsbut constant pH Data for replicate samples of one ozone cancentration
1% potassium iodide.
a portion of this output was meter9d into another air stream for further dilution. The lamps were operated for many hours to reach a stabilized condition, and teats showed the ozone output did not vary more than 1 or 27, during the course of a run, or 10% over a period of days. Only a single lamp waa lit for these tests. This system has been fully described (16, 16). Replicate samples were collected in rapid succession using 10 ml. of a variety of reagents in midget impingers, at. a flow rate of 1 liter per minute. The sample air stream was diverted from a ballast impinger of equal resistance, so that no change in flow occurred in the system during the sampling operation. An acidifying reagent was then added to certain reagents to liberate the iodide, and the relative amounts of iodine were determined photometrically, with the results shown in Table I. When the kinetic colorimetry technique was applied, a 1- or Zminute sampling period was used, and readings were made immediately and a t intervals timed after the mid-point of the sampling period.
All the reagents containing 20yo potassium iodide were unsatisfactory because of color instability; those containing 57, potassium iodide were moderately stable; those containing 1% potaasium iodide exhibited excellent stability. Reagent 1, Table I, is widely used (9) in oxidant recorders, where stability is not a problem because the reagent is continuously circulated through a carbon bed and readings are made a t a constant age after exposure. However,
13 0.006 0.01
0.03
0.06 0.1
O Z O N E FLOW/ T O T A L
Figure 4. Analyses by relative flowmeter values
neutral
FLOV 0.6
iodide
reagent
I
vs.
Sampler of fresh unreacted mixtures of azcne and 1 -hexene. Abscissa represents degree of dilution by purifled a i i o f outpvt of ozonlzer (operated under constant conditionrl
when it was used in midget impingers in the preliminary kinetic work, the data were not suitable for calculations because of excessive variations. Reagents containing 1% potassium iodide were most practical for manual sampling met hods. The influence of reagent pH was also shown by these experiments. When ozone was passed through strongly alkaline iodide reagent, no color was immediately evident. With weaker alkaline solutions, such as 1% borax, some color could be obtained. Almost all the iodine wa.s liberated from neutral reagents without acidification. When the alkaline reagents were acidified, additional iodine waa liberated; a pH of 3 was optimal for maximal iodine release. Higher pH resulted in a slow fading of the iodine color, whereas lower pH resulted in a slow gain in iodine color. It waa evident from the relative iodine values that the stoichiometry varied from one reagent to another and was not necessarily integral for any reagent. The relative iodine values listed in Table I were based on the photometric readings corrected for the iodine extinction coefficients. Thus the intensification of the iodine color by higher iodide concentrations hss been eliminated from these values. Higher iodide concentrations resulted in greater absolute quantities of iodine. Because a different amount of iodine was obtained for
each reagent for a fixed amount of ozone, these values cannot be relied upon as absolute ozone measurements. However, some of the reagents gave precise relative values. Kinetic Colorimetry of Ozone Analysis. When the kinetic colorimetry technique was applied to the analysis of ozone with the neutral potassium iodide reagent (also listed as reagent 14, Table I), it was noted that iodine was liberated by two components. About 90% was liberated in less than 1 minute and the remaining 10% appeared to be liberated by a single slow-reacting component with x half life of about 10 minutes. This immediately suggested that the ozone may have contained about 10% impurities. The previous anomalous results had already suggested impurities in the ozone, but the efforts to find such impurities had been unsuccessful. T e s t were made of ozone from different sources, a t different concentrations both before and after aging. In all cases the slow-reacting component remained at about 10%. The hypothesis of an ozone impurity was therefore rejected. A second hypothesis, which appeared to be supported by the data presented below, was that the ozone was pure but reacted with potassium iodide reagent t C J produce two products. The fraction of slowly developed color could be reduced by sampling with a fritted bubbler rather VOL. 31, NO. 1 1 , NOVEMBER 1959
1917
than a midget impinger. Changes could also be made in this fraction by using M e r e n t concentrations of potassium iodide and phosphate. The effect of potassium iodide concentration OD, the fraction of slow color in various ozone determinations by kinetic colorimetry is shown in Figure 3, A . The rate constant of the slow color reaction at 26" C. was found to be 0.054 An.-' times per cent potassium iodide. Thus to follow the reaction with 20% POtassium iodide, a recording spectrophotometer was necessary and a Beckman DK spectrophotometer was employed. The reaction was essentially completed in 3 minutes. The fraction of slow reaction was substantially increased with higher iodide concentrations. The total amount of iodine liberated for a fixed amount of ozone was also increased.
Table II.
No. 14 19 20 21 22 23
24 25 26 27 28
ated from each reagent, even though the pH of most of them, as well as the potsssium iodide concentrations, were carefully held close to fixed values. It is thus evident that the supporting media participated in the reactions in an important way. Other tests showed that the concentration of the supporting medium affected the amount of liberated iodine even though no slow color waa developed. Chemistry of Ozone Reaction with Iodide Reagents. Experimental evi-
dence suggested t h a t the primary product of ozone reaction with alkaline iodide is hypoiodite, and not iodate as previously assumed. When ozone was sampled in 1% potassium iodide in 1 N sodium hydroxide, the iodine released upon acidification to p H 6.2 with solid boric acid was 50% of that resulting from the usual acidifica-
Kinetic Colorimetry of Ozone Reaction with Iodide Reagents
Reagent
PH
1% KI, O.lMI(HzPO~, O.lMN%HPO, Rea en! 14 1 drop.HIPOl 1% %I m double-dIstdled water 1% NaI in doubledistilled water 1% K I in 0.1Mcitrate buffer 1% NaI in eth lene glycol 1% K I in O.l& acetate buffer 1% K I in 0.1M malate buffer 1% K I in 0.1M hthallate buffer 1% K I in 0.04h?arsenate buffer 0.3% K I in dioxane (on 'nal iodine de-
+
colorized by addition o?zjlsinc dust and filtration) 18 1% K I in 1N NaOH, acidified after sampling with phosphoric-sulfamic acids
A similar effect was noted with regard to the concentration of phosphate b d e r as shown in Figure 3, B and C. These are the results for equalaised samples of a fixed ozone concentration collected in a series of reagents where the iodide concentration was held constant a t 1% and the neutral phosphate buffer concentration was varied. The pH of these solutions was found to be essentially equal by checking with a glass electrode both before and after sampling. Increasing phosphate concentration resulted in an increasing amount of slow color aa well as total color. That these effects probably were not due to impurities in the phosphate buffer was indicated by adding 5 y of ferric iron,5 7 of vanadium as vanadate, and 10y of cupric copper in various tests without any substantial effects. The kinetic colorimetry technique waa applied to the ozone reaction with a variety of reagents (Table 11). Only reagents containing phosphate showed a slow color; furthermore, a difIerent quantity of iodine waa liber-
6.60 5.81 5.78 5.69 5.90
...
5.82 5.80 5.90 6.79
...
14, then 2
Relative Iodine Color 100 109 57 51 69 75
%
Slow Color 12 13 0 0 0 0
70
(E) 74
0 0 0 0
(35)
0
ca. 65
0
tion (reagent 18) to pH 2. However, no iodine was obtained by the same boric acid acidification from iodate added to the same medium. With added periodate, such acidification yielded 14 to 20% of the iodine obtained a t pH 2. This latter corresponds to the 25% reported for periodate b y w d a r d a n d Merritt(91). A reasonable explanation of these data would appear to be the formation of hypoiodite as the primary reaction product. The dismutation of hypoiodite to give iodate and iodide appears to be very slow. Calculations from available kinetic data (8) indicate that under these experimental conditions the half life of hypoiodite is a t least 4 days. Certain irreversible losses of micro quantities of iodine were noted during the acidification of alkaline iodide solutions. These losses may be related to the "ozone demand" previously described (3,14). Whenstandardiodine solution waa added to the alkaline iodide resgent (reagent 18,Table I) only part was recovered upon acidification. Very slow acidification with vigorous stirring
of the ozonized reagent, or of the reagent with added iodate, also yielded low results. The latter losses may be the result of the iodine liberated a t the surface of the acid droplets redissolving in the excess alkali. These errors were avoided by rapid acidification of the alkaline reagent, especially through the weakly alkaline region. The mechanism of these losses is not clear. This study showed that a nonintegral stoichiometry existed for the reaction of low concentrations of ozone with iodide reagents. However, a number of reagents were stable, and yielded stable iodine colors which were linear with respect to ozone quantities over a wide working range. The best of these reagents was the 1% potassium iodide, neutral phosphatebuffered (0.1M KH, PO,, 0.1M NaJIPO.) reagent, previously listed under reagents for kinetic colorimetry, and in the tables as reagent 14. This reagent was selected because the maximum color developed (after 45 minutes) was represented by stoichiometry very close to Os = I,, as shown below by comparison with an independent method. The chemistry of the reaction of ozone in low concentrations in air with neutral phosphate iodide reagent is evidently complex. It is reasonable to asmme the primary ozone reaction is with iodide ion. No chemical reaction directly between ozone and the water or phosphate was found when ozone samples were collected in double-distilled water, or in the phosphate buffer alone. The small amount of iodine liberated upon subsequent addition of iodide could be accounted for on the basis of ozone solubility alone. Secondary reactions between ozone and reaction products are possible, but are probably outstripped by the competitive rapid reaction with the much higher concentration of iodide ion. The following hypothetical scheme appears likely:
Os + I-
-P
IO-
+ 02'
(fast)
Primary reaction (3) IO-
+ I- + 2 H+F? I2
O,*
+M
-P
O2
+ HIO (fat) (4)
+ M (fast)
(5)
I n alkaline solution, no iodine is liberated, but in neutral or acid solutions. Reaction 4 proceeds. The oxygen liberated by the primary reaction is assumed to be in an activated form, but rapidly loses ita energy by molecular collisions (Reaction 5). This pattern would yield a stoichiometry of 03 = 11. The observed nonintegral stoichiometries (Tables I and 11) require simultaneous reactions resulting in losses of iodine. To account for the stoichiometry of reagent 14 (Os = 11) in the presence of these loases, there must be
70
-
40
-
20
I--
z
w
-I
U
KEY A
-
method Sampler of fresh unreacted mixtures of ozone and 1 -hexenc
X
s'
d a'
Figure 5. Analyses by neutral iodide reagent compared 4 with replicate analyses by nitrogen dioxide equivalent
10
+
-
?4-
L
2 *-
W
W
z
I -
0 0.74) 0.4
0.2
-
0.2
./ 0.4
0.7
I
2
7
4
10
20
40
NITROGEN D I O X I D E EQUIVALENT, P.P.M.
still other reactions with a higher yield, utilizing the oxidizing power of more than one oxygen atom in the ozone molecule. The high yield reactions occur only when phosphate is present during sampling, are favored by increased iodide and phosphate concentrations (see Figure 3), and are associated with equal increments of both slow and fast color (slope of curve in Figure 3, C). The addition of phosphate immediately after sampling vielded no slow color (reagent 18, Table 11). As there appears to be no primary reaction bet\\ een ozone and phosphate, it must be assumed that the extra oxidizing equiialent comes from the utilization of part of the activated oxygen liberated from Reaction 3: 02'
+ I- + HZPO4-4 IO- + HIPOI- (fat)
(6)
The perphosphoric acid, similar in properties to hydrogen peroxide, is regarded as the source of the slow color by Reaction 7 followed tw Reaction 4: HZPOb-
+ 1 - 4 HZPOd- + IO-(slow) (7)
A number of reactions resulting in losses of iodine are possible, some likely ones being:
The above scheme is, of course, highly hypothetical, but it appears t o explain the observed pattern. From a practical viewpoint, the neutral iodide reagent was selected because it had a
stoichiometw close to O3 = I*. For this reagent,-we may write the following over-all empirical reaction: O8 + H+
21--t
'I fLHzo
(lo)
To attain this relationship, 45 minutes should be allowed after sampling for maximiim color development. Much other work was carried out to find a more useful reagent based on other inorganic and organic reactions, but the best ones found had more serious shortcomings than this reagent. However, a radically dserent approach based upon an ozone gas phase reaction was successfully developed.
Recommended Procedure for Ozone. SAMPLING. Draw the air sample through 10 ml. of neutral iodide reagent in a n all-glass midget impinger a t the rate of 1 liter per minute, using a flowmeter downstream from the sampling device. Collect the proper size sample by means of a stopwatch. Allow 45 minutes for maximum color development, and read a t 352 mp in a spectrophotometer. STANDARDIZATION. Prepare 0.01N iodine solution (1.27 grams of iodine and 16 grams of potassium iodide per liter) and standardize with sodium thiosulfate by titration, using starch indicator. Dilute the iodine with neutral iodide reagent to form a standard series up to O.ooOo6N (for Zcm. cells) and read in a spectrophotometer at 352 mp. Ozone is computed on the basis 0 8 = Is. From the standardization, calculate standard sample volumes (at 25" C. and 760 mm. of mercury pressure) for which the optical absorbance (corrected for the blank) multiplied by an integral factor (1, 2, 5, or 10) gives parts per million of ozone directly. (For Zcm. cells and factor 1, the standard sample volume was 4.8 liters.)
INTERFERENCES. Sulfur dioxide is a particularly serious interference which causes low results. Many other reducing and oxidizing gases may cause errors. The fading of the weak iodine colors was greatly accelerated by dust. Even the dust ,settled in a single day from laboratory, air upon clean inipingers drying on a towel produced a noticeable increase in the fading rate for several hours. Hence all glass surfaces must be kept scrupuloudy clean and dustfree. Small losses of iodine even on c l a n glasa surfaces were observed when a weak iodine solution of measured color was poured from a spectrophotometer tube into an impinger and back. The loss did not occur again when the process was repeated, indicating that volatilization of iodine was unimportant. Similar losses of iodine on glass surfaces were reported by Wolfenden (22). For accurate work, sufficient sample to gi-:c a dark color is preferable to minimize errors due t o impurities on glass surfaces and in reagents. Where dust is a problem, it may be advisable to read the colors immediately, rather than after 45 minutes, and to add 10% to the results for the undeveloped slow color. Comparison with Other Methods. A series of relative concentrations of ozone prepared by accurate dilution in the flow system of the output of the ozonizer operated under constant conditions, was analyzed by the neutral iodide procedure. Extensive data from the kinetic study (16,16) are given in Figure 4. Although 1-hexene was mixed with the ozone in these determinations, it did not interfere with the analysis, nor appreciably react with the ozone during the 4aecond nuxing period prior to sampling. Excellent agreement was found between the analyses and flowmeter dilution factors over a wide range of ozone concentrations, 2 to 80 p.p.m. Below 2 p.p.m. a tendency for the analysis to be about 0.1 p.p.m. low was noted. T o check the absolute validity of the procedure, a widely different method of omne analysis was developed. Ozone was quantitatively converted to nitrcgen dioxide by addition of nitric oxide (from a tank containing 1% nitric oxide in nitrogen mixture) to the sample air stream. The nitrogen dioxide was then determined with GriessSaltzman reagent (13). A 10 p.p.m. excess of nitric oxide gave better than 95% conversion in the 4-second flow-time allowed for the following very rapid reaction : VOL 31, NO. 1 1 , NOVEMBER 1959
1919
01
+ NO
-C
NO*
+ 02
(11)
A capillary tip packed with g h wool was used to dilute the tank gas more rapidly than it could be oxidized by the oxygen of the sample air stream. This latter reaction is kinetically third order, and although it proceeds rapidly at higher concentrations, the rate is negligible at the h a 1 dilute concentrations. In this manner the nitrogen dioxide blanks were always kept at a small fraction of the total found. The complete details of this method using a simplified and improved apparatus are given elsewhere (13). Replicate analyses by the neutral iodide and the nitrogen dioxide equivalent methods are compared in Figure 5 for the same unaged ozone-1-hexene mixtures given in Figure 4. The smaller scatter of points indicates that most of the scatter in Figure 4 waa from variation in the ozonizer output or flowmeter errors. At very low values the iodine results are again about 0.1 p.p.m. low, while above 20 p.p.m. the nitrogen dioxide values are low. Similar results were obtained for pure ozone-air mixtures. The iodine values were again low by a constant small amount in the low range. These errors were probably due to dust and impurities on the glass surfaces and in the reagent. Considering the widely differing naturc of the two methods, the agreement is excellent through most of the range. These data established the validity of the neutral iodide method chosen, for the determination of ozone in the absence of other oxidants. The same two methods were found (16, f6) to diverge for the same ozone1-hexene mixtures after 5 to 30 minutes of flow reaction time had been allowed. The data indicated that the iodide method was not specific for ozone in the presence of oxidized hexene, whereas the nitrogen dioxide method waa Apparently the organic oxidants did not appreciably react with the nitric oxide in the minimal reaction time allowed. The difference between simultaneous analyses by the two procedures was interpreted as representing oxidants other than ozone, and was found to be a relatively constant fraction (17 to 25%) of the ozone previously consumed, over a hundredfold concentration range. These means of differentiating oxidants should be valuable for further study of the composition of natural smog oxidant mixtures. The two methods should also give widely divergent results in the presence of reducing gases. Sulfur dioxide and
1920
ANALYTCAL CHEMISTRY
hydrogen sulfide, each separately in ten- to hundredfold ratio to ozone, produced little or no interference with t,he nitrogen dioxide equivalent method (17). The same mixtures yielded no iodine whatsoever with the iodide reagent, indicating serious interference. The nitrogen dioxide equivalent method (with appropriate blank corrections) is therefore recommended for the specific determination of ozone in smog mixtures, and the iodide method for determination of total oxidant. CONCLUSIONS
A powerful and broadly applicable technique was developed for spectrophotometrically resolving and determining micro quantities of components in mixtures. A slow-reacting reagent was added and measurements of the developed color at progressive time intervals were made, until maximum color was obtained. The logarithm of the undeveloped color (maximum absorbance minus the absorbance a t the given time) waa plotted against the reaction time. Pure substances undergoing a single reaction with the reagent gave straight-line plots with characteristic slopes. Binary mixtures exhibited plots with one elbow, and were resolved into the arithmetic sum of tKo straight lines, one for each component. More complex mixtures theoretically could be handled by excension of this technique. This kinetic colorimetry technique was successfully applied for the iodometric microdetermination of various oxidants in liquid dilutions. Studks of various iodide reagents for the determination of low concentrations of ozone in air showed the chemistry of the reaction to be complex, and the stoichiometry nonintegral, although reasonably constant within certain ranges. The most stable and precise of the reagents tested was neutral phosphate buffered, 1% potassium iodide, with which reactions appeared to involve both hypoiodite and perphosphate. The stoichiometry of this reagent for pure ozone in air was determined by comparison with an independent method based upon conversion of the ozone to nitrogen dioxide by addition of excess nitric oxide. It, was found to be very close to O3 = I2 in the range of 2 to 80 p.p.m. Gaseous oxidant mixtures were successfully analyzed by simultaneous applications of the iodide and nitrogen dioxide methods. The latter method is recommended as specific for ozone, and free from interference by organic oxi-
dants and reducing gases. The iodide method is recommended for determining total oxidant. The difference between the determinations by the two methods represents oxidants other than ozone and should be valuable in the study of natural smog oxidants. ACKNOWLEDGMENT
Sincere thanks for encouragement and support of the work are due R. G. Keenan, D. H. Byers, H. E. Stokinger, L. J. Cralley, and W. C. Cooper of the U. S. Public Health Service, LITERATURE CITED
(1) Birdsall, G. M., Jenkins, A. C., Spadiiger, E., ANAL. CHEM.24, 662-4 (1952). (2) Boelter, E. D., Putnam, G. L., Laah, E. I., Zbid., 22, 1533-5 (1950). (3) Byers, D. H., Saltzman, B. E., d d uances in Chem. Sm.es, No. 21 (1959); Am. Znd. Hyg. Assoc. J . 19, 251-7 (1958). (4) Cadle, R. D., Huff, H., J. Phys. & Colloid C h m . 54, 1191-5 (1950). (5) Effenberger, E., 8.a d . Chem. 134, 106-9 (1951-2). (6) Lechner, G., 8.Elektrodrem 17,412-14 (1911). (7) Lee, T. S., Kolthoff, I. M., Ann. A‘. Y . A d . Sci. 53, 1093 1951).
(8) Li, C. H., White,
6. F., J . Am. Chem.
SOC.65, 335-8 (1943). (9) Littman, F. E., Benoliel, R. W., ANAL. CHEM.25, 1480-3 (1953). (10) Littman, F. E., Marynowski, C. W., Zbid., 2 8 , 819-25 (1956). (11) Nozaki, K., IND. ENG. CHEM., ANAL.ED.18,583 (1946). (12) Renzetti, N. A., Romanovsky, J. C., A . M . A . Arch. Znd. Health 14, 458-67 (19561. (13) Saltzman, B. E., ANAL.&EM. 26, 1949-55 (1954). (14) Saltzman, B. E., Znd. Eng. Chem. 50, 677-82 (1958). (15) Saltzman, B. E., Ph.D. thesis, University of Cincinnati, 1958. (16) Saltrman, B. E., Gilbert, S . , Id. E . Chem. 51, 1415 (1959). . - - - - I .
(17)%altzman, B. E., Gilbert, N., J.
Am. Znd. Hyg. Assoc. 20,379-86 (1959). (18) Smith, R. G., Diamond, P., A m . Znd. Hyg. Assoc. Quart. 13, 235-8 (1952). (19) Thorp, C. E., “Bibliography of
Ozone Technology,” Voi. I, Armour Research Foundation, Illinois Institute of Technology, Chicago, 1954. (20) Wagner, C. D., Smith, R. H., Peters, E. D., IND.ENG.CHEM.,ANAL. ED. 19, 976-9 (1947). (21) Willard, H. H., Merritt, L. M., Jr., Zbid., 14, 489-!W (1942). (22) Wolfenden, J. H., ANAL.CHELM. 29, 1098-100 (1957).
RECEIVED for review April 8, 1959. Accepted July 10, 1959. Division of Water, Sewage, and Sanitation Chemistry, Air Pollution Symposium, 136th Meeting, ACS, Atlantic City, N. J., September 1959. Baaed upon Ph.D. thesis of B. E. Saltzman, University of Cincinnati, 1958.
