Determination of Thiols in Hydrocarbon Gases

THOMAS BARKER1. Cities Service Oil Co. {Delaware), East Chicago, Ind. A convenient dependable method for determination of thiols in hydrocarbon gases,...
6 downloads 0 Views 3MB Size
Determination of Thiols in Hydrocarbon Gases EDGAR W. ELLIS AND THOM.IAS BARKER' Cities Service Oil Co. (Delaware), East Chicago, I n d .

A convenient dependable method for determination of thiols in hydrocarbon gases, which may be performed without complicated or specially fabricated apparatus, is considered desirable for control of stench in commercial propane or other hydrocarbon fuels when specifications require the presence of malodorous material. A method has been developed which meets these requirements and makes use of reagents that require very infrequent restandardization if at aH. The method involves the oxidation of thiols in a measured volume of gas with excess cupric acetate-acetic acid reagent and back-titration of the

U

SI.; of liquefied petroleum gas and other gaseous furls has

greatly increased during the past 15 years and many states have enacted legislation making it mandatory that all domestic fuel gases contain a minimum concentrat ion of some malodorant in order that leaks may be readily detected and repaired. In order t o meet this requirement most producers of fuel gases add measured amounts of volatile thiols (mercaptans) directly to the product before it is marketed. Many natural and practically all refinery gases (butane and lighter) contain varying, though frequently small, quantities of thiols of low molecular weight. In order to control the final concentration of thiols in the product, a dependable method is needpd for determining the concentration of thiols in fuel gases. IIakewill and Rueck (4)have presented n method similar to that of Shan ( 6 ) , in which the gas sample is passed through a fritted-glass absorber containing 10% cadmium chloride solution and 1.0 AVsodium carbonate. The thiols (and hydrogen sulfide. if present) absorbed by the rc~igentsare liberated with hydrochloric acid, treated with an excess of standard iodine solution, and back-titrated with sodium thiosulfate. The method appears to give good results, but, requires the use of standard iodine solution, which is relatively unstable, and a special glass vessel as an absorber. Furthermore, thiols may be lost after addition of hydrochloric acid before the absorber can be closed. The Railroad Commission of Texas ( I ) has developed a method in which fuel gas is passed through 0.025 N silver nitrate solution. An excesfi of 0.025 A' sodium chloride solution is then added and the excess is titrated with standard silver nitrate. Recently, some instrumental methods have been evolved. One of these ( 7 ) employs a photoelectric colorimeter t o analyze copper butyl phthalate after the latter has absorbed thiols and subsequently been exposed to ultraviolet light. Another method ( 3 )makes use of the Titrilog for continuous titration of sulfur compounds in gas st reams. The method developed in this study has proved satisfactory in this laboratory for more t h m a yrar. The method is relatively simple and apparatus required is readily available in most refinery laboratories and laboratory supply houses. The procedure has been used primarily for determining thiols in propanepropylene mixtures, but is applicable to all butane and lighter fractions if proper arrangements are made to vaporize the sample completely before it comes in contact with the reagents. The precision of the method is good and the accuracy is satisfactory for control of odorant concentrations in commercial fuel gases. However, the method does not distinguish between thiol@and 1

Present address. American Mineral Spirits Co., Atlanta, Ga.

excess cupric ion with sodium thiosulfate in the presence of potassium iodide and starch. Reproducibility is well within 0.10 and accuracy within 0.2 grain of thiol, calculated as sulfur, per thousand cubic feet of gas. The cupric acetate method is applicable to natural gas, liquefied petroleum gas, and refinery gas streams containing only butane and lighter hydrocarbons. It will serve as a check on thiol concentrations when stench is added to commercial fuel gases or when effectiveness of sodium hydroxide scrubbers for removing thiols from refinery gas streams is being investigated.

hydrogen sulfide. The latter must be removed if present in significant quantities before thiols are determined. Hydrogen sulfide is generally absent in commercial fuel gases or in refinery gases which have pawed through caustic soda or other suitable ~~rutiticrs. OUTLINE OF ,METHOD

The chemical reactions involved in the proposed method may be represented by the following equations:

+ 4RSH 2CuRS + 4HC2H302 + R2Sz 2Cu(CzH302)2 + 4KI Cui12 + 4KC2H302 + 1 2 2CuRS + I? = C U ? I + ~ R2S2

2C'u(CzH302)2

=

(1)

(2) (3)

hdding and simplifying, 2('U((?~H302)2

+ 2RSH + 2KI = C U ~+ I ~2HC2HaOz + 2KC2H302 + RzSz

(4)

A sample of the hydrogen sulfide-free gas or liquefied petroleum gas is taken under full plant pressure in a suitable pressure vessel. The sample is released from the vessel sloaly through a small U-tube immersed in hot water to vaporize all liquid constituents in the sample. The gas is then passed through a frittedglass absorber containing standard cupric acetate reagent which includes an excess of acetic acid. The thiols are oxidized quantitatively to disulfides and cuprous mercaptides. Potassium iodide solution is added to the absorber after completing passage of the sample. Part of the iodine liberated by the residual cupric ion oxidizes yellow mercaptides formed in the absorber to the corresponding disulfides with simultaneous formation of insoluble cuprous iodide. The iodine remaining, which is equivalent to the residual cupric ion after the thiols are completel~oxidized, is titrated with standard sodium thiosulfate. e difference between the original and residual cupric ion in the absorber reagent represents the equivalent of the thiols present in the measured sample. EXPERIMENTAL

I n developing the method many determinations of thiols in various samples were made in duplicate to establish the reproducibility of the method. Duplicate determinations can normally be expected toagree withinO.l grain of sulfur (as thiols) per 100 cubic feet of gas, as is shown in Table I. The accuracy of the method was evaluated by analyzing several samples of propancbpropylene mixtures to which known amounts of ethane- or butamthiol had been added (Table 11). A stream containing a mixture of propane and propylene was used for obtaining samples for these experiments because many refinery gas streams, espe1777

1778

ANALYTICAL CHEMISTRY

cially those used for making liquefied petroleum gas, contain up t o 50% or even more unsaturated campounda, and it wa8 desirable to determine whether possible side reactions with olefins might affect results. The preparation of propane-propylene samples with accurately known concentrations of thiols proved to be difficult, as liquid propane-propylene mmples must be mmipulated in bombs under pressures equal to their respective vapor pressures a t essentially mom t,emperatures, and extremely small quantities of volatile thiol were required to produce a normal concentration in samples even as large as 1 gallon. By "normal" concentration is meant t h a t which would meet most is, of the order of 1 rcquiremonts for a stenched product-that to 5 pounds of sulfur, in the farm of thiols, per 10,000 gallons of liquefied petroleum gas. The thiol concentration in the prepared samples was subject t o an additional error caused by smsll ;Lmounts of impurities in the ethane- and butsnethiols ae purchased from the supplier. The materials were of the highest quality available cammercinlly and were used aithout further purification.

Tahle 11. Analyses 01 Prepared Samples (Total thiols a8 maim oi sulfur per 100 cubic fcet oi gas s a t u m t ~ dxi 00' I?. snd 30 inohes of mercury) PIesent I'oond Difference 1.09 0.08 -0.13 1.90 4.29 2.29 2.29

1.84

4.51 3.64 1.03

1.89

4.33 2.30 2.14 1.70 4.40

3.57 0.78

-0.21

+0.06 +0.01

-0.15 -0.14 -0.05 -0.07 -0.23

Cupric acetate reagent. I n about 600 ml. of water, 19.9670 grams of oupric acetate monohydrate are dissolved, 100 ml. of glacial acetic acid are added, and the solution is diluted to I liter with water. T h e reaeent should be eauivalent. volume for volume, to the thiosulfite. Potassium iodide, 33% solution. St,ilrchsolution '

PROCEDURE

If the sample to be analyzed is undor sufficient pressure to IIC partially in the liquid phase, set up t,he apparatus as illustrated in Figure 1. T h e U-tube below the sample bomb is largely immersed in wat,er a t 120" t,o 150" F. to effect complete vaparisation of any liquid present. If the sample is totally in the gaseous phase, the snurce of the gas may be connected directly to the absorber. Pipet 25 ml. of cupric neetilte reagent into the ahsorher, record the meter reading, start the gas flow cautiously, and regulate the rate a t about 2.0 cubic feet per hour. Under thefie conditions the reagent. will normally froth about halfhay to the top of the absorber. Pass a total of 5 cubic feet of the gas through the absorher. Note t.hc svectge temperature as phou-n by the meter thermometer, and the barometric pvessure. -0-m * 1 0 "

LLLLW BAROMETER INCHES HC

25

24

Figure 1.

Sample Bomb, Vaporizer, 4hsorhar. a n d Meter

23

Reomdueihilit 22 2.30 2.07 1.77 1.67 I. 6 7 1.48 1.08 0.4Y

0.48 0.46

0.20 ~

2.36 2.27

1.77 1.07 1.67 1.48 1.18

n

ii

0 40

C.40 0.20

0.00 c.20 0.00 0.00 0.00

21

0.00 0.10

0.c2 0.06 0.08

0.00

.-

20 ANALYTICAL PROCEDURE

Apparatus (see Figure 1). Wet test meter, stainless steel, 0.1 ouhic foot per revolution. Sample containers, stainless steel, of sufficient size to hold the equivalent of 6 to 10 cubic feet o l gas a t atmospheric conditions, and hydrostatically tested a t 400 pounds per square inch gage pressure. U-tube of '/,inch copper tubing between bomb and absorber for vaporizing liquid products by means of h a t water. Cylindrical jar, Pyrex (Corning No. 6920 or 6940). Absorber with fritted-glass disk and spray trap as used in the ASTM method (a). Reagents. Sodium thiosulfate, 0.100 N sodium thiosulfate pcnfahydrate in water.

Nomographic Chart for Estimation of Thiol Concentration

Figure 2.

If the sample contains unsaturated hydrocsrbona purge the absorber, a t the end of the gas passage, Kith 0.2 cubic foot of nitrogen or carbon dioxide to remiwe i,he unsaturates which react withladine. Disconnect the alrsorber, add 10 ml. of 33% potassium iodide solution or 4 to 5 grams of potassium iodide all cwst.a,ls. " ~nnd hrinn ~ ~ the ~inner ~surfaces , of the absorber and spray trap in con& with the solution to dissolve all traces of vellow cu~)rousmercaptide formed during the absorption. Finally, rinse thespray t& int,o theabsorberwith water. ~Tit.ra.ttethe esccsr iodine in the absorber \\-it,h0.1 A'sodium thiosulfate, d h ~~

~

~~~

1779

V O L U M E 23, N O . 12, D E C E M B E R 1 9 5 1 ncltlition o f starch near the end, until the pale brown color suddenly changes to milky white. The mixture may be agitated during t,he titration t>>- alternately applying suction and pressure with the mouth through a rutiher tube attached to the open end of t h e spray trap while the latter is fitt,ed t'o the alisoher. C A XU LATION S

T h e thiol content may l)e c:ilculated in trrnir o f sulfur by o f the follo\ving equatioAi:

niwriy

s

=

0.05609(25 - T')(460 ~

+ t)

H - P

\ ~ I I I ~ ISY ~= thiol sulfur rontent, gr:iina of sulfur 1x1' 100 culiic: feet of gas saturated Jvith n-atcr v:ipi>r a t 60" F and 30.0 iric.hw crf niei'cwry 1- = volunie of 0.1 .\ thiosulf:itt, i.equirri1, nil. t = tcmprratui~c< , f outlet gas, F. H = t)aroniet rio 1)rc.w11'e,inrhes of niercw p = v : i l i o i ' ~ I I ~ I W ~ofI ~ n-ater < ~ at t o F., iricmh O

.I iioniographir chart for cnlculating the thiol sulfur (,ontent \\-lii~i all st:indard rcxgents are exactly 0.1 S i s given in Figure 2. Tlic. (ahart is Iiased on the v:ilue 1.032 for p , which is t h r vapor ti.iiPion of water a t 80" F. 1-dues of thiol content o1)tainc.d from the chart :it other ga.; teniper:itur(Js normally encount isred art: sufficic~ntlyaccuratii for most purpose^, as variations i l l p :it room temperaturos (mwt- rc~lativel:- small errors in thiol i m t m t . 1 hy 1:iying :I straight-edge betn-een the, oliwrvid LIW lint1 gas temperature>,noting \\-here it i~rossc~s thr refeiwiri~line, and rstrnding the edge across the point on the ri~ftwnceline arid the volume of thiosulfate uwd in the, titi,:ition t o thc thiol contrsnt ; ~ x i i . DISCUSSION

I h p l j c a t r a n a l ~ fxr o~m the sanic sample bonih provitfrd with tl\-o [--tubes in parallel, onr of copper :ind one of glaaa, l c d i n g

t o individual absorbers and nietcra, produced identical rcwlta, showing that the samples are not partially "sweetened" (cxonverpion of thiols t o disulfides) by cxontact with the coppt'r tul)r. It, was found t h a t accuracy could not be improved by u*ing nioi'e dilute than 0.1 S cupric acetate reagent. Reaction of tlir, cupiic ion with the thiols !vas incomplete if less than 25 nil. of rc.:igent ivere usi~din the ukisorlier, prohably liecause of inuficmieiit w n tart tinir, :is only a rcJlatively p m : i l l fraction of this i(.:ig(,iit is consunird by the. thiols. I-sr of hydi,ochloric and Fuliut ic. :ic,itis in the ahsorhei. rwgrnt initrad of acrtic proved t o Iw uiixitisfactor!.. The c'xact r:ingc of pH value rrquired for gc~tr(lt i w l t s \vas not deterniincd, but indications are that it should I K :illout 3.0 or 3.5 2nd not higher than 4.0. The cupric ac,i.t:itij-:ic.c,tic a s drscribed consistentlj- producrd t l q i i ~ n i l acid huffer prc~parc~l nlilt, rrsults aiid pH rc3quirenients were not explored fu1.tlirr - I ( ~ t y l r n eand its homologs aril normally not prewiit i t 1 c.itlier natural or refinrry gases. This highly u n s a t u r a t d g:i- w:+rts \vith cuprous ions only in aninioniacal solution. Furt I ~ I ~ I ~ ~ I I I ~ I ~ , ;ic~c~tylwieis lilirratrd from acrtylides by ncid:: ( 5 :. lio~ice, n if prcwmt, n.oultl not i.v:+c.twith the. : i c . i t l c~)lil)er iwigcwt i n thv a l i s o r h r . LITERATURE CITED (1) Aut. ( A . s .J., 162, T o . 0 (194.5). '21 .\in. Soc. Testiiig lIatei.iaIs, "=\ST11 S t a n d a r d > . " 3Iethod D 90-4iT. r:l) . - \ i d t i , I t . K . , I-'et,cy. L. I:.,ant1 1,:srhet.. E. E.. fAf.,. 26, S o . 5, 47- 5:3 (1950). (4) Hakewill. H.. aiid Rueck. E . 31.. P,.or.. - 4 ~ G. r t a -4 (1946). 1.5, liii-hter. Victoi. m r i , "(kgaiiic ('heinistly,'' 31d et1 , 1.01, 1, .\nici~ii;aiiP h o t o Offset Reyl.int. l i p 110-11, S e n - Tior 1;. \-III rle~ i i a i iI'iihlishiiig r o . , 1944. \I.. En., 12, 668 11940l. \ t i ) Shaw, .J. 1.. I s n . Esc;. ('HEM., , G u s , 25, No. G , 38-9 (1049). ( 7 ) IVhite, D . L., aiid Reichnr.dt. F

RECEIVED .4pril 19,1951.

Determination of Small Amounts of Dimethylamine in Technical Methylamines Simple Method f o r Separation of Dimethylamine f r o m Monomethylamine EDWARD L. STiNLEY, I I i R R Y BiU\.I, 4 Y D JESSIE L. GOVE Rohm a n d Huus Co., Bridesburg, Philadelphia, P a .

T

HI< catalytic vapor phase reaction of methanol arid amnionia produces a complex mixt,urr of water, monomethylamine, dimethylaminr, trimethylamine, ammonia, arid methanol. The monomethylamine and trimethylamine separated from this mixture usually contain less than 1 dimethylamine, but m:iy contain as much as 3%. A precise, accurate, and rapid method for the determination of dimethylamine in the anhydrous amines and in their aqueous solutions was desired. In the procedures used generally for the determination of dimethylamine in methylamine mistures, the monoinethylamiiie is tlecomposd with nitrous acid and the resulting dimethylnitrosoamine and unrracted trimethylamine are steam-distilled into staiidard acid ( I O ) . The excess of the acid is titrated with alkali and the nitrosoamine is then reduced with zinc. Suhsequent titration of the alkaline distillate of the reaction mixture yields the total secondary and tertiary base and, from this, the diniethylamine is calculated by difference. It is necessary to apply Fmpirical corrections and the procedure is unsatisfactory for small amounts of dimethylamine because of the accuniulatioil of c'rrors on the diniethyl:iniine.

One method with which some success lias been ol)tained in this laboratory has heen reported (9). However, this procedure does not determine dimethylamine directly and some interfri,cnce in thr prc'sence of nioriomethylainiiie has been noted. Srvwal procedurc~sfot, the direct determination of tliniethyl:imine have been described ( 2 , 4,7 , 8). The polarogr:iphic. procedure of Smales and JVilson was investigated, hut did not give satisfactory results in this laboratory. English has ingeniously overconic m n i ~of the shortconiings of thr Smales and \\.ikon p i ~ o c ~ d u r ebut , reports that his method is not suitalilr for the determination of sinall amounts of dimethylamine. Iiatcher and \Toroshiloir:i titmted the dimethyldithiocnrbaiiiate salt of the amine with a cupric sulfate solution. This niethod TI-ouldbe applicahlr to the determination of large amounts of tlimcthyl:iniinr in the ahsence of moiiomethylariiine; triniet h:,lainine would presumably not interfere. The colorimetric procedure of Dowlen, originally d loped for the determination of small amounts of dimethylamine in biological fluids, appearrtl to offer thr greatest promise and was intensivrly investigated. T h e Doivden procedure is b a d on the forniatioti of the