Manganous Hydroxide Method of Test for Oxygen in Butadiene

Manganous Hydroxide Method of Test for Oxygen in Butadiene Vapors ... of low concentrations of oxygen in gases by the manganous hydroxide method...
0 downloads 0 Views 358KB Size
Manganous Hydroxide Method of Test for Oxygen in Butadiene Vapors GRANT W. TAYLOR AND D. S. ALEXANDER Polymer Corp., Ltd., Sarnia, Ontario, Canada A comparison of results obtained on determination of oxygen in butadiene vapor by the ASTM manganous hydroxide method of test and by mass spectrometer indicated that the ASTM method gave misleadingly low values. The variables in the test were investigated to determine the cause of the low recoveries. A s a result of this investigation the procedure was modified by increasing the volume of 1% manganous chloride reagent, increasing the amount of shaking of the reaction flask, and incorporating a calibration curve to convert experimental results to actual oxygen values. The modified method gives reliable results for oxygen concentrations up to 4.0 volume %.

T

HE manganous hydroxide method for oxygen in butadiene vapors was developed by the Butadiene Producers’ Committee on Specifications and Methods of Analysis of the Office of Rubber Reserve and was accepted as the standard test in 1944 ( 3 ) . In this test a vapor phase sample of butadiene is shaken with a freshly prepared aqueous suspension of manganous hydroxide and the resulting manganic hydroxide is titrated iodometrically. A modified procedure using a different type of apparatus was issued by the Office of Rubber Reserve in 1946 (4)and an adaptation of this method appeared as a tentative ASTM method of test in 1949 (1). This present investigation of the manganous hydroxide procedure was initiated by a discrepancy observed between results by the ASTM method and by mass spectrometer analyses. After completion of this investigation, information was received that a typographical error had been made in stating the concentration of the manganous chloride reagent in the ASTM procedure ( 2 ) . However, it is shown that the use of the more dilute reagent (lye i n s t e a d of 107G)was responsible for only part of this discrepancy.

furic acid (1 to 1) are then added and the flask is gently shaken to dissolve the precipitate. The funnel attachment is removed, the contents of the flask are diluted with 150 ml. of oxygenfree distilled TTater, and the liberated iodine is titrated with 0.1 IT sodium thiosulfate to a starch end point. A blank is run on the reagents omitting the sample.

Table I.

Comparison of Oxygen Analyses by ASTM Method and by Mass Spectrometer ASTM method

Oxygen, Volume % Mass spectrometer 0.01 0.15 0.25 0.50 2.8 5.5

0.01 0.08 0.12 0.18 0.70 2.0

Table 11.

Accuracy of Oxygen Analyses by Mass Speotrometer

(Synthetic blends of oxygen i n butadiene vapor) Theoretical Oxygen, Vol.

r0

0.34 0.52 0.74 1.04 1.20 2.26

Exptl. Oxygen, VOl. % 0.32 0.50 0.74 1.01 1.20 2.27

COMPARISON OF METHODS

In the ASTM procedure, the 500-ml. reaction flask as shown in Figure 1 is evacuated t o a pressure of 1 mm., and the sample is passed through it for 5 minutes by way of glass tubing and a one-holed rubber stopper fitted into the addition funnel. The stopcocks are closed, the flask is allowed to come to room temperature, and any excess p r e s s u r e is r e l e a s e d b y momentarily opening the stopcock. T w e n t y -f i v e milliliters of m a n g a n o u s chloride solution (15 g r a m of the hydrate per liter) f o l l o w e d b y I 5 m l . of sodium hydroxide solution (100 grams per liter) are a d d e d from t h e f u n n e l attachment and the flask is shaken for 10 minutes. Ten milliliters of p o t a s s i u m iodide solution (100 g r a m per liter) and 25 ml. of sul-

A comparison of typical results obtained by the ASTM method and by mass spectrometer analysis is given in Table I. For the mass spectrometer analyses, the samples were taken in mercuryfilled, 250-ml. transfer bulbs and the oxygen was determined directly by mass 32. From considerations of possible interference from other components and of changes in instrument calibration, the accuracy of the mass spectrometer result8 should be within 5% of the actual oxygen content for concentrations up to 5 volume %. This accuracy was confirmed by analysis of synthetic blends of oxygen in butadiene vapors as shown in Table 11. Accepting the mass spectrometer results as being essentially correct, it was apparent that the ASThI procedure gave misleadingly low values for oxygen concentrations above 0.1 volume %. VARIABLES IN MANGANOUS HYDROXIDE METHOD

In an attempt t o increase the values obtained by t h e ASTM method, a series of eqeriments was conducted to determine which variable, or variables, had the greatest effect upon the results.

Figure 1. Apparatus

The samples used throughout this work were synthetics prepared from polymerization grade butadiene and cylinder oxygen. The butadiene was taken from the vapor phase of a bomb sample and passed into a 3.5-gallon bottle filled with a saturated solution 1.083

ANALYTICAL CHEMISTRY

1084 of sodium chloride acidified with hydrochloiic acid. It was brought t o atmospheric pressure by venting and the volume was determined from the amount of brine displaced. A known volume of oxygen was pressured in over mercury from a gas buret and the actual oxygen content of the blend was checked by mass spectrometer analysis. Amount of Shaking of Reaction Flask. The first variable investigated was the amount of shaking of the reaction flask. A Burrell wrist-action shaker set a t maximum agitation was used; the results obtained with intervals of 5 to 60 minutes are given in Table 111. The ASTM procedure does not specify the degree of shaking, but simply states, "shake the flask continuously for 10 minutes." It was found that the Burrell shaker produced as much agitation as that normally attained manually by routine laboratory workers during a 10-minute interval. From the results in Table I11 it is evident that the amount of shaking is an important variable and one that must be carefully standardized

bination of increasing the actual amount of manganous chloride and increasing the liquid volume. Amount of Caustic. The amount of sodium hydroxide eolution (100 grams per liter) was varied from 5 to 25 ml. tvithout any change in the expeiimental results. An excess of caustic was present in all cases, holyever, as only 1.5 ml. would be required to react with 25 ml. of the manganous chloride solution. Amount of Acid. The amount of sulfuric acid (1 to 1) was varied froin 5 t o 50 ml. xithout any appreciable effert upon the recovery. TIODIFIED ASTM PROCEDURE

Table 111. Effect of Amount of Shaking of Reaction Flask

As a result of the investigation of the variables in the ASTM procedure, the method was modified by increasing the volume of manganous chloride (15 grams of the hydrate per liter) from 25 t o i 5 ml. and b:. niechanically shaking the reaction flask for 15 minutes. Three additional modifications R ere also incorporated into the procedure. In preparing the apparatus for taking a sample it was dried prior to evacuation This prevented water vapor from entering

Burrell wrist-action shaker set at niaxiinuni agitation Remainder of procedure as described in ( 1 ) Theoretical oxygen content = 0.52 vol. % Recovery, shaking Time. Exptl. Oxygen, 3Iinutes Vel. 70

Table V. Effect of Concentration of Manganous Chloride Solution 26 ml. of 31nCL solution Flasks shaken for 10 minutes on Burrell shaker Theoretical oxveen content. 0.61 vol. % " _ ." Concn. of Exptl. 3InClz Solution. MnClt in Flask, Oxygen, Recovery, G./L. a3 3lnClz.4HKJ Gram Vol. % '7c 5 0.08 0 17 33 15 0 22 0.24 25 0.40 0.28 40 0.64 0 31 61 60 0.35 0.96 09 150 2.38 0 34 6i

2

Volume of Manganous Chloride Reagent (15 grains of hydrate per liter). The volume of manganous chloride reagent n-as varied from 5 to 100 ml. and the results obtained on a 0.5% oxygen blend are shown in Table IV. I t appears that the amount of manganous chloride solution has a definite effect upon t'he experimental values, as the recovery increases with increasing volume of reagent. Theoretically, there is sufficitnt manganous hydroxide formed with even 10 ml. of the manganous chloride reagent to give complete recovery on an 0.8% oxygen sample. Concentration of Manganous Chloride Solution. The concentration of the manganous chloride reagent was varied from 5 to 150 grams per liter (as the hydrated salt) and 25 ml. were used in each case. The results shown in Table Sr indicate that the recovery is related t o the manganous chloride concent'ration, with approximat,ely 60 grams per liter shoning maximum recovery. The effect of increasing the concentration, however, is not as great as that produced by increasing the volume of the 15-grain-pt~-litersolution (Table IV).

GRAPH

I

0.4

ACTUAL OXYGEN-V0L.X 0 0.4

0.8

12

1.6

2.0

2.4

2.8

32

3.6

4.0

4.4

GRAPH I1

Table IV.

Effect of Volume of Manganous Chloride Solution

Flasks shaken for 10 minutes on Burrell shaker Concentration of hlnClp solution, 15 grams per liter as MnC1z.4HzO Theoretical oxygen content, 0.53 vol. 75 Volume of Exptl. hZnClz Solution, hInCli in Flask, Oxygen, Recovery, 111. Gram Vel. o/G 70 10 25

50 100

Volume of Liquid in Flask during Shaking. Using 25 ml. of manganous chloride reagent (15 grams of the hydrate per liter) the total volume of solution in the flask was increased by the addition of oxygen-free distilled water. As sh0R-n in Table VI, this resulted in increased recovery. h comparison of these results with those obtained in Tables I V and V shows that the improved recovery observed on increasing the volume of manganous chloride solution (15 grams of the hydrate per liter) is a com-

I

Figure 2. Correction Graphs for Manganous Hydroxide Oxygen Determination

1085

V O L U M E 2 4 , N O . 7, J U L Y 1 9 5 2

at least as aatisfactoy!. I ~ ~ ~ S U I as I ~ St.hose obtained by thc morc tedious, yet less reproducible manual shaking-for example, 011 a 0.43% oxygen blend, bhe Rurrell shaker gave an average rccovery of 58V0and manual shaking gave 567,. On comparing the Rubber Reserve results (Tahlc 1-111)Tvith t,hose obtained by the modified ASTM method (Table VII), it : ~ p pears that the modified procedure is more satisfactor:, even without applying a correction graph. With a 10% solution of mangaIn 0.21 -10 nous chloride, the suspension of manganous hydroside produced li .i n 26 . 0i is much heavier than at t,he lower concentrat,ion. This prob!IO 0 31 til) 1 1.-, 0 :i7 71 ably results in less eficieiit, contact with the gaseous sample and consequent. low recoveries. Winslow and Iiehhafskp ( 5 ) found that the reactivity of manganous hydroxide decrtar;es as its surthe vacuum pump ;trill tc~iitlr~i t o give s o n i ( ~ ~ v l iiii~i ot w rc~protluc,it>l(~ face becomes oxidized and one would expect thi. effect to lw more resul ts. pronounced thc heavier the suspension. R-hen sampling, thv apparatug \+-asfllcd thivugh t,hc citli) :irrii. This was more convcnient t,han filling through the :tcitiitioii t'unnrl and adequate flushing enwred the removirl ot' O i r small :imoutit of air trapped in thv *ide ami. 'l'al3le L-111. iccuracl- of Kithber Reserve Procediire In determining the hlanii or1 t h r reagc~iitst h c : :rppar:ttns \viis 2 2 ml. of MnCh soluti,on. i n n grams per liter as MnClz.iHz0 rvacuated, flushed wit,h nitrogen, and again c ~ c w t t e dto r e n i o v ~ ~ I3 nil. of S a O H solution the last, trace of air. It was t h r n sn-irled after the addition of I'la:ka ;haken for 10 r n i i i r i t ~ con R11rr~11 -baker each reagrnt, rather than shnkw, sinw it \vas difficvlt t o p i m ~ r i t .. . +v., air leakage when shaking 1 hi, eviccuatett :ipp:tratuy. Deviation

Tahle VII. Accuracy of hlodified iST\1 Procedure 7 5 1111. of bInClz solution, 15 prams per liter as .\InC12.4HzO 1 6 1111 of N a O H solution l'lask> .shaken for 1.5 minutes o n R w r ~ l .baker I

Experimental Oxygen, Vol. $;

Theor. Oxygen. Vol. %

n n

003 03 0 12 n 32 0.49 0 5G

So

c)f

.iv

Nuns

iincorrerted

.\lean Nil 0.03 0.13 0.32 n 47

Corrected . Mini.\laximum niiiiii

n

32 0.43 0 4fi

n .io .A \-, 1)eriation froin AIean Valuit. Tal. 7A

0 74 1 20 2 2i

fi

0 0-1

8

I) 2 1

4 1

2.5 0.38 0.33 0 36

1 8 8 8

n

n

n

AB 88

0.03 0.14 I). 18 0 31 031 0 LY

n

40

0 61

0.04

0.32 0.41 0.43 0.35 n 51 n 82 125

10.01 3=0.04 Io.08

*0.01 1t0.02

+n

06

=n

25

1 0 09

-

0.58 0 74

I

0 . on3

01

1 38 2,il :I 73

r ,

l h e avoragc ixwilts oI)taineci on a series of 12 I)l(~ndsof 0 to 3.7Yc oxygen are given in column 3 of Tahlc: \TI ancl indicate that this modified procedure is a drfinit? improvement ovcr t,hc original. However, t'he slightl2- high rerults obtained beloir 0.50; oxygen and the IOU- results abovr, t'his concentration indicate tho necessit,y of a correction graph such as shown in Figures 2. Graphs I and I1 cover thn concentration ranges 0 to 47; ani1 0 to ly0 osygen, respectively. The result,s obtained on t>heahovc series ~vhenusing thcse graphs are given in columiir; 4,5 , and 6, of Tahle VII. 4CCURAC:Y O F KIJRRF;R R

The main differencc bet~vcrnthr .iST\I procctiurc, ar pullishcd up t o 1952, and the Rubber Reserve inet'hoti is in t h e concent.ration of the manganous chloride reagent. The former proccdure specifies a lYc solution ( 15 grams of the hytiratv per liter), whereas t,hc lat,trr employs a concrntration of 10%. Although Table V indicated that 150 grams per liter did not givc c:oniplete recovery for 0.5% oxygen, a series oE 8 hleiids vias analyzed using nianganous chloride reagent of this st,rength to ohtain an indication of t.he accuracy of thc Ruhher Reserve method for osygen eoncentrations up t o 245. The results, as givpn in Tahl(1 L-III, show that the Rubber R e ~ c r v emethod gives Ion. rwult s, partioularllfor the higher oxygen conwntrations. The analyses \Yere carried out in accordance, Lvith th(J Rubber Reserve procedure, esccpt that niechanical, i n ])lac(: of rnanual, shaking was ernplo>.ed. 1IowPvr:r. t h r Hurrc~ll s1iakc.r pi~otluc~ctl

The, nioclified ASThl procedure is more satisfactory tlian either .iST?\l niet,hod D 1021-49T, as now published, or the Rubber Reserve procedure. I t is cnpnhle of giving reliable results up t.o 4 volume % osygen. Two of the shortcomings of method D 1021-49T are production of insufficient, manganous hydroxide and t,he presence of inaufficient liquid in t,he react,ion flask during shaking. These h a w heen overcome in the modified procedure b y increasing the rolume of 19; manganous chloride reagent from 25 to 75 nil. The continued UPC of a 1% manganous chloride solution prevents t,he production of the heavy- suspension of manganous hydroxide, which is one of the major objections t o the Rubber Reserve method. The change to niechaniml shaking gives more reproducible values, particularly between different operators, and t h r increase in shaking time from 10 to 15 minutes results in a more caomplete reaction of the os>-genin the sample. The use of a correction graph enables th(1 c~nipirioalresults to be converted to actual osygen values. .ICKNOW L EIXMENT

The authors desire to espress t,heir appreciation to Ruth I3ron.n for her assist,ance in carrying out the experimental work, t o K. E. Elston for supervising the mass spectrometer analyses, and to the Polymer Corp., Ltd., for permission t o publish this report. LITER 4'1'1:HE: CITED

(1) Am. Poc. T w t i i i g hIatet.ixls. Designation D 1021-49T. ( 2 ) Dunbrook. R. F.. Chai~nian,Section I, Technical Conitxiittee H, A.S.T.M. Comniittet D-2, private communication to author, February 195%. (H) Office of Rubber Kcsen-e, "Riitadieue Laboratory Manual," 1944, Designation L.M. 2. I .8. (4)I b i d . , 1946, Designation L.11. 2.1.8.1. 15) Winslow-.E. H., and Liebhafsky, H. A . , IND.ENG.ChiF:hr., .h-.4r2. ED.,18,565 (1946). R E C ~ : Ib \ . f~u r rryicw 1 1 a r c i i l o . I!>>?. Awepted M a y 10, 1959