1347
V O L U M E 28, NO. 8, A U G U S T 1 9 5 6 the 1:trge volumes or. the nieazuring system. The manometer rratling may take as miicli as 15 minutes to come to its equilihriuni v:due after an amhicnt temperature change of 1" C.
Table 11. Gas Pressure Correction Factors at Various Temperatures" Temp..
c.
Correction Factor
Temp.,
1.000 0,993 0,990 0.988
27 28
20 21 22 23
24
c.
29 30 31 32
0.980 0,97J
25 26 a
0
Correction Factor 0.965 0.961 0.956 0.951 0.946 0,941
0.970
Total correction fact,or is O.j';o/degree. -
.
. ~ _ _ _ _ _~ __ -_-~_
_
'l'lir empirical additloid correction is an average of several obQervations of this factoi \\ ith amounts of tank carbon diolide in t tie $\ stem giving almit I O - and 25-cm difference in readings on thP tno-liquid manomrtcr Figure 1 shows the change in these pi essiire readings, and 111 thr, zero reading, a t various temperatiires The differential redings a t each temperature are corrected by the gas lav to 20" C.; then the additional correction necessary is noted to t i e 0.2 =k 0.05%. The total correction used to calculate the valuee 111 Tahle I is 0.50% per degree. There is some variation in this wilue with the amount of carbon dioxide present, but an average v d u e gives sufficiently good rcsults and i q niiich more convenient to use For extreme accuracy, a sliding
scale of factors might be used, to vary with the actual amount of carbon present. The use of these correction factors rather than those calculated by the gas law alone has led to greater precision, both in carbon and hydrogen percentages and in carbon-14 activity results (because of the greater accuracy in measuring the amount of carbon dioxide placed in a counting tube). Carbon values determined by this method are regularly within 0.5% of the correct value. The average deviation of carbon results previously reported (2) was 0.33% (only the gas law correction being used), while a comparable set of analyses using the present correction factors showed an average deviation of 0.24%from the theoretical value. I n the case of the carbon-14 results, the error introduced from uncertainty in the amount of gas in the counting tribe is calculated to be less t h m 0.1% r h e n these correction factors are used, a percentage rvhirh ir negligible compared to the statistical and other systemitic errors affecting the activity determinations (3). LITERATURE CITED
(1) Belcher, R., Ingram, G., -4naZ. Chirn. Acta 4 , 124. 401 (1950). (2) Christman, D. R., Day, N. E., Hansell. P. H . . Anderson, R. C., ANAL.CHEM.2 7 , 1935 (1955). (3) Christman, D. R.. Wolf, d.P., Ibid., 27, 1939 (1955). (4) Hodgman, C. D., "Handbook of Chemistry and Physics," 35th ed., p. 2241, Chemical Rubber Publ., Cleveland, Ohio. 1953. ( 5 ) Ingram, G., Milcrochim. Acta 1953, 71. (6) Latimer, W. AI., Hildehrand, J. H., "Reference Book of Inorganic Chemistry," p. 199, Alacmillan, New York, 1940. RECEIVEDfor review February 6. 1956. Accepted M a s 8. 1956. Rewarch performed under auspices of t h e U. S. Atomic Energy Cornniiision.
Improved Rapid Colorimetric% of Dissolved Oxygen W.
F.
LOOMIS
The Loomis Laboratory, Greenwich, Conn.
(:otnparison of a colorimetric method for determining dissolved oxygen i n O..i-inl. samples with the standard Finkler method rmealed that the calibration curve did riot p u arctiratel) through the origin and was slightly concave do^ I I W ard. This paper describes three modifications of the method, whereby a strictly linear ciirve passing accurately through the origin can be obtained; other improvements make possible its use as :I rapid micmi alternatke to the standard Winkler method.
( '( )I,(
)RIkII~CTTtlC'tleterniiriation of dissolved oxygen in (l..j-inl. m p I e s ( Z j , which could he completed in about 1 n i i r i i i t t , . mnsisted of measwring the red color produced by partially osLlizing reduced indigo r:trmine wit,h the oxygen dissolved in the t w t sample of water. At'mospheric oxygen was excluded by out the reaction in an airtight syringe that could he c~:ti~i~ying pI:ic,ed directly in il. Becknian spectrophotometer. Iii R subsequent cwniparison of this method and the standard \\.itikler method ( I , 6'); it ivas found that the resulting calibratioii curve did not pass wciirately through the origin and was not .iti,icatly linear but slightly concave downward. Conseqiiently,
three modifications of the original method were devised, whereby a strictly linear curve passing through the origin was obtained. This paper describes these modifications, together with various other improvements developed while using this method as a rapid micro alternative to the standard Winkler method. The coefficient of variation between the two methods was 2.57, (Table I). APPARATUS
A 1-ml. Tuberculin syringe, graduated in 0.01 nil., with S o . 22 hypodermic needle (A. H. Thomas Co., 9404) is used. -4small steel ball (or nail head) is inserted into the barrel of the syringe, so that shaking the syringe effectively mixes its contents. A I-cm. piece of tape is wound around the barrel of the syringe between the 0.75 and 0.90 markings, as illustrated in Figure 1. This bushing of tape serves to hold the syringe snugly inside the borosilicate glass absorption rrll (10-mm. light path) of a Beckman spectrophotometer. If the tape is wound around four small pieces of wire set equidistantly around the barrel of the syringe, the resulting cross section is approximately square and the syringe may he locked into position with a slight twist. The syringe should always be placed in one predetermined position, with the graduations a t one side, so as to give as clear a ight path as possible. If the plunger descends during a determination, the spring clip a t the end of the barrel should he tightened.
1348
ANALYTICAL CHEMISTRY
A Beckman spectrophotometer is adapted to receive the syringe, with its needle still attached, by drilling a I-em. hole in the floor of the sliding absorption cell carrier immediately beneath the fourth square compartment near the back of the instrument; and drilling a second 3-mm. hole through the main housing of the sam le compartment. This second hole is centered with the first, so txat the needle of the syringe can pierce the housing and project beneath the spectrophotometer (Figure 1). These two holes may be drilled with a drill press in a few minutes; they do not interfere with the normal operation of the spectrophotometer. Only under exceptional conditions of illumination is i t necessary t o place a small piece of black tape over the 3-mm. hole to make i t completely lighttight.
Table I. Variation of Absorbance at 586 M p of Present Method and Concentration of Dissolved Oxygen as Determined by Winkler Method Winkler 1\f et hod, Concn. of Dissolved Oxygen, Mg./L. O?
Present Method. Absorbance at 586 MAl
0.97 1.30 1.94 2.74 3.28 33.45 4.00 4.68 4.96 5.10 j,72 1 : . 50 7.40 7.58 8.80 9.21 9.80 10.50
0.189 0.241 0.330 0.479 0.564 0.500 0.721 0.807 0.865 0.913 0.995 1,138 1,252 1,352 1.541 1.543 1.670 1.813
Ratio" 3.15 ;1. 4 .j. 9 5.75
reagent and filled t o about, the 1.00 mark, the needle pointing upward during all such operations. The syringe is wiped with a piece of Kleenex and shaken about 50 times, following which the blank absorbance of the reagent alone is measured at 586 niM. The plunger of the syringe is advanced t o exactly the 0.40 mark. A strong light, behind the syringe considerably increases the accuracy of this placement. The syringe is then lowered into the sample of water to be tested anh the barrel is filled to exactly the 1.00 mark. After wiping and shaking, the experimental absorbance of t,he sample is dekrmined a t 586 nip withoii! delay. CALCULATION
The net absorbance of the sam le is obtained by subtrarting 50% of the blank absorbance Kom the experimental value. The concentration of dissolved oxygen in milligrams er liter is then obtained by dividing the net absorbance by a cabhation factor obtained as described below. Results in either milliliter< of oxygen per liter or per cent saturation may be obtained graphically, if desired, with the aid of Rawson's nomogram (3-6) Example Experimental absorbance Blank absorbance 1.630 -
S e t absorbance
5,8
j.8j L.79
5.8 .j.75
Oxygen concentration
1.630 0.176
=
net absorbance calibration factor
0.176
-= =
1.542
1.542
__ = 9 . 0 mg. per 0.171 liter o1
.? . 6
4.75 5,75 p.9 17.6
5.7 5.95 5.85 5.75
a Mean of ratios of two methods (omitting two at top of list) is 5.85. Standard deviation is 0.15 and coefficient of variation is 2.5%.
DETERMINATION OF CALIBRATION FACTOR
4 s the net absorbance of the colorimetric method varie.: linearly with t,he oxygen content of the sample, the de.sired calibration factor may be obtained hy determining the net absorbance of any sample of water, if the oxygen content is already known. Although the standard Winkler method TWP used to determine oxygen content in this work, it is usually more convenient to standardize the reaction against a sample of water, when the oxygen content can be calculated from the t,eniperature a t ivhich it was equilibrated with air.
The bottom plate of a borosilicate glass absorption cell (10mm. light path) is removed by tapping it with a small metal rod. This allow the syringe and needle to be placed directly in the bottomless absorption cell and lowered until the butt of the needle rests on the bottom plate of the compartment housing. As the height of the syringe precludes closing the sample compartment with its usual cover, a substitute is made by stapling 2 inches of black cloth t o the edges of a 4 X 5 X 5 inch light-tight cardboard box, which is placed over the sample compartment dui ing a determination. A 16-ounce narrow-mouthed acid bottle (Fisher Scientific Co., 2-922), fitted with a rubber cap (3-225), is used t o store the reagent, Hypodermic needles may be inserted repeatedly through such rubber caps without leakage of air. Small 60-ml. serum bottles (3-220) may also be used; they make convenient s a s t e bottles, into which t o eject the dye following a determination. Osvgen-free nitrogen, "prepurified nitrogen" (Matheson), ma1 be used, Commercial nitrogen must be purified before use l)v hubblinp it through alkaline pyrogallol.
A bottle is half-filled with water and shaken vigorously for several minutes t o ensure complete equilibration of the lvater with the air in the bottle. The temperature of the water is then determined to the nearest 0.5' C. and its oxygen content is obtained from the standard tables published by the American Public Health Association ( 1 ) . The net absorbance of the sample of water is then determined by the method described in this paper. The required calibration factor is obtained by dividing the net absorbance of the sample by its oxygen content in milligrams of oxygen per liter, as obtained from the standard tables. Thus, for example, if the teniperature of a shaken sample of water is found to be 21' C., its oxygen content is 8.99 according t o the tables. If its net absorbance is n o x determined and found to be 1.537, the required 1.537 calibration factor is -= 0.171, Duplicate determinations,, 8.90 as well as determinations of samples equilibrated at different temperatures, should check each other nithin f l yc.
REAGENTS
DISCUSSIOY
Two grams of indigo carmine (Sational), glucose, and anhydrous potassium carbonate are aced in a 16-ounce reagent bottle and 200 ml. of water are a d t d . The rubber cap is wired in place. The air space within the bottle is flushed with oxygenfree nitrogen for 10 minutes through two No. 22 hypodermic needles, after which the bottle is left under 5-pound pressure of nitrogen. The reagent is reduced in about an hour in an 80" C. water bath, or in about a week a t room temperature. Additional purified nitrogen is introduced through a needle from time t o time to maintain the internal pressure at about 5 pounds.
As originally described, this method was calibrated against a graded series of samples obtained by mixing different proportions of reagent (Oy0 oxygen) and fully saturated (10070 oxygen) water whose oxygen content could be calculated from the teniperature a t which i t had pieviously been equilibrated ll-ith air ( 1 ) . This simple and rapid method appeared adequate for most purposes, especially in view of the practical difficulties involved in preparing a graduated series of x-ater samples of known oxygen content, protected from the air, and in sufficient quantity for determination by the Winkler method ( 2 ) . Subsequent work has shown that such a graduated series of water samples may be prepared by siphoning oxygen-free water into a series of B.O.D. bottIes (Fisher Scientific Co. 2-926)
PROCEDURE
The spectrophotometer is first adjusted t o zero with water in the syringe. Following this, the syringe is rinsed twice with
V O L U M E 2 8 , NO. 8, A U G U S T 1 9 5 6 containing increasing amounts of fully saturated water; the resulting mixtures are protected from the air before use by the ground-glass stoppers of the especially designed B.O.D. bottles. A large supply of completely oxygen-free water can be o b tained conveniently as follows:
A 2-gdlon, wide-mouthed bottle is filled two thirds full with water and its remaining air space is flushed with purified nitrogen. A 50-ml. beaker is then hune with C O D D ~wire ~ from the bottom
20 mT. of 80%),pot&ium hydroxide are then pipetted carefully into the beaker, the rubber stopper is lowered into place, and the bottle is st.irred with z magnetic stirrer for about a week. Within this time, d l the oxygen dissolved in the water distilled into the beaker of alkaline pyrogallol. This method was used, because the direct addition of reducing agents interferes subsequently with the Winklor method, while boiling under v m u m does not completely remove dissolved, oxygen unless continued for longer than overnight with consequent excessive evaporntion. When the graduated series of B.O.D. bottles prepared in this way were .analyzed simultaneous1.v by both the Winkler and colorimetric methods, the resulting edihratiotion curve did not pass accurately through the origin and was slightly concave downward, instead of being stricti>- linear. Consequently, three modifications of the original method were instituted 60 that the linear curve passed aecwat,ely through the origin. Modification. The spectrophotometer was adapted to receive the syringe even when it.s needle was still attsehed (Figure 1). This seeminglj- small change, besides making the method more convenient and rapid in practice, eliminated an error of between 1 and 2 mg. per liter of oxygen by preventing contamination of the sample with about 1 to 2 y of atmospheric oxygen during the removal and suhsequent replacement of the needle on the syringe. The voliime of reagent within the bore of the needle itself was taken into aeroimt. It, was found that the absorbance of a sample of oxJ-gen-free water was exactly 50% of the blank only when the syringe was filled with reagent to the 0.40-ml. mark before it was filled with the sample, and not to the 0.50-ml. mark a8 previously dererihed. T h e operative wave length w a raked from 580 to 586 mg. This corrected the slight downward concavity in the calibration curve present in the original method and, together with the other changes, yielded a strictly linear curve that passed accurately through the origin. The general reliability, ease of reduction, and stability (severd months) of the reagent were markedly improved, when it was made n i t h oxygen-free rather than commercial nitrogen gas. Furthermore, slow reduotion a t room temperature yielded a reagent whose color did not fade the 1to 2% per minute observed with reagent reduced rapidly a t 80' C. I n practice, either method of reduetion may be used, as it is usually practical to measure the absorbance within 30 seconds after mixing. Water sampler containing above 10 mg. of oxygen per liter may be analyzed by the present method by reducing the ratio of sample to reagent, so that the absorbance of the mixture falls within the limits of the liness part of the curve. I n such cases, the percentage of the blank to be subtracted from the experimental value is determined by taking a sample of the desired
1349
size of oxygen-free w-ater into the syringe and finding the percentage of the blank that remains. A convenient and stable source of oxygen-free water for such purposes may he obtained by making up two bottles of reagent in the asual manner, but omitting the dye from one of them. T h e color of the reagent bottle then may be used as an indicator of the degree of reduction within the water bottle. I t is preferable to reduce such bottles alowly a t room temperature, as alkaline glucose solutions turn hi-own on being heated to high temperature. Because reduction in the size of the sample changes the slope of the calibration ourve, a now calibration factor must be determined for every differmit size of sample used.
Figure 1.
Final assembly of apparatus
As the improved method is calibrated directly against Winklerdetermined standard., it is flee from errors induced by changes in the blank absorbance of reduced and oxidized samples. As with the Wmkler method, however, compounds capable of oxidiaing or reducing indigo carmine, such as nitrates, chlorates, s nitrites, iron salts, or sulfites, will interfere d e ~ previously removed by appropriate methods (1 ). LITERATURE CITED
(1) Am Public Health Asioc , New York, "Standard Methods for Examination of Water and Sewage," 10th ed , 1955 (2) Loomis, W F.,ANAL CHEM26,402 (1954). (3) Rawson, D S , Limnological Soc A m , Umver-ity of Ahhisan, Ann Arbor, h h h , Spec Pub 15,1944 (4) Ricker. W E ,Ecology 15,348 (1934) (5) Welch. P S , "J,~mnologicalIllethods," p 366. Blaknton, Philadelphia,. 1948 (6) Wmkler. L W , Bcr 21,2843 (1888) RECEl