V O L U M E 27, NO. 10, O C T O B E R 1 9 5 5 a total sulfur method and the o-clichlorol,rnzeiie solution procedure (S), appears to be required for determining organic sulfur reliably. .iCKNOWLEDGMENT
The authors wish t o thank John hlandel of the Testing and Specifications Section for his statistical analysip of the dats. LlTERATURE CITED
1609 ( 2 ) British Standards Institution, Bull. 903, Method 3 . 1 4 , p. 27, 1950. (3) Federal Testing Method Standard. S o . 601, .\lethods 1G231 and 16241. (4)
Hale, C. H., and SZuehlberg, W.F., I s n . ESG.CHEM.,AXAL.ED., 8, 317 ( 1 9 3 6 ) .
( 5 ) Linnig, F. J.,Milliken, L. T., and Cohen, R. I., J . Research .\-atZ. B u r . Standards, 47, 135 ( 1 9 6 1 ) . ( 6 ) Reconstruction Finance Corp., Office of Production, Synthetic Rubber Div., "Specifications for Government Synthetic Rubbers," Washington, D. C., rev. ed., (October, 1952). FVall, L. d., J . Research, .Vatl. B u r . Standards, 41, 315 ( 1 9 4 8 ) .
"' I"\
(1) Ani. SOC.Testing Materials Philadelphia, Pa., Part ti, Designa-
tion D-297-50T, pp. 63-4, 1952.
RECEIVED for re\.iew . i p r i l 5 , 19.55.
Accepted May 26, 1966.
Densities and Refractive Indexes for Ethylene Glycol-Water Solutions EDWARD T. FOGGI, A. NORMAN HIXSON, and A. RALPH THOMPSON2 University o f Pennsylvania, Philadelphia, Pa.
The lack of agreement aniong previouslj pui)lished data indicated the need for precise anal) tical data for mixtures of ethj lene glycol (1,2-ethanediol) and water. Investigation of gl?col purification by vacuum distillation defined the necessit? for extreme care in prekenting contamination with trace amounts of oxidation products. Densities at 23" C. and refractive indexes at 20" C. w-eredetermined for niiutures of highly purified ethylene gl?col and water. ?'he density data provide a basis for analysis to within +0.11$%by weight, whereas refractive index determinations yield analyses accurate to within f0.03 weight %. Refractive index is not a linear function of weight per cent compoSition. as is commonl? asqumpd for this sJsleni.
I
N TIIE course of a recent distillation study irivolving t'hc
et,hylenc glycol-n-ater syst,em a precise analytical method was of critical importance. Variations in published valiies for the density and refractive index of purified ethylene glycol suggested the presence of trace impurities, since the range of values v.xs considerably in excess of that which would be expected from a series of independent determinations on identical mittcrixls. Therefore, purification by vacuum distillation was studied froni the standpoint of identification and subsequent elimination of trace cont,aminauts. Once a satisfactory distillation procedure was found, densities and refractivc indexes were determined for the aqueous glycol system over t.he eutirr composition rmigc. PURIFI(:.ZIION
OF >lATERI.&LS
Tecliriical grade. 99.570, ethylene glycol was fractionated a t 7 to 10 mm. of mercury absolute pressure and with a reflux ratio of 10 to 1 in a nitrogen-blanketed, adiabatically operated, packed column (1 inch in diameter and packed to a depth of 36 inches with glass helices "16 inch in diameter). Only the middle third of the const,aiit boiling point distillate was collected. Iieproducibility of the product was adequate as evidenced by the constancy of the physical proprl,ties. Of elwen lots thus purified. thv densities a t 25' C. varied from 1.1098 to 1,1099 grams per nil. with the average k i n g 1.10982 grams per ml., as compared to previously published values which range from 1.1097 (8) t o 1.1110 ( 4 ) with the most probable value. according t o Curme and Johnston ( S ) , being 1.10986 gr:tms pcr ml. Despite all precaiitions taken to prevent, water contamination, t8he purified glyc.01 analyzed (by Karl Fischer reagent) 0.07 f O . O l ~ o(weight') water. Correcting for the presence of water b y extrapolation rtwilted in an :tverage density of 1.10988 grams per ml. for the 1 Present address, E. I. d u F o n t de Kcinours & Co., Inc., T i l i n i n g t o n , Del. 2 Present address, D e p a r t m e n t of Chemical Engineering, University of Rhode Island, Kingston, R . I.
purified ethylene glycol. Similarly, the refractive index, %go, varied from'1.13179 t o 1.43182 with the higher value being the most common. Extrapolating to zero water content led to a value of 1.43188 for purified glycol as compared t o previously published values ranging from 1.4304 ( 1 ) to 1.43192 (IO).
In general, there are two primary sources of glycol contamination, both of which tend t o lo\wr density and refractive index values, and which could well account for the wide range of reportwl values for density and refractive index. The most obvious is t.he :Ilniost unavoidable presence of small quantities of water; in much of the early literature no water determinations were reported and prohably this accounts for many of the discrepancies. The second class of contaminant,s consists of oxidation pmducts: aldehydes and acids. It was shown during the first attempts a t purification that the oxidation reaction was platinum catalyzed; however, t,race amounts of aldehydes were detected (Schiff hasej even when all-glass equipment was used. Repeated purging of the distillation equipment xith oxygen-free nit,rogen prior to st,art-up eliminated the formation of analyzable quantities of 3 Idehydes. Contamination hy this mechanism would hc a diptinct possildity in much of t>hepreviously published work and very probable in those cases n-here a,n air bleed was introduced into the still pot to promo Water used in the preparation of the glycol solutions used in this work rras freshly distilled and had a specific conductivity of the order of 10-6 ohm-' em.? As an added precaution, t.he n-at.er was boiled to remove dissolved gases and then cooled 11-ithout apit,ation immediately prior to use. PREPARATIOW OF SOLUTIOIVS
Solutions of known composition were prepared by injecting approximat? amounts of purified glycol (0.07% water) and water into dried, stoppered 60-ml. vaccine bottles. The exact compositions were determined by n-eighing to 0.1 mg. Hypodermic syringes used to transfer the glycol mere dried a t 110' C. and then allowed to cool in a desiccator until used. Since the smallest amount of either component was 2.7 grams. the compositions n-ere known to a t least 1 part in 27,000 or 0.004 veight 7' D E S S I T l AND REFR.ZCTIVE INDEX DETEKRZINATIOYS
Density measurements were made in 10-ml. Weld-type. capped, specific gravity bottles which had been calibrated with boiled, distilled water: Temperature control was obtained by partially submerging the bottles in a 5-gallon constant ten>0.03" C. The maximuin perature bath controlled a t 25.00' error resulting from uncertainties in the volume calibration of the specific gravity bottles T o d d be =!=0.00009 gram per ml. However, strict adherence to a standardized procedure resulted in duplicate measuremente having a maximum deviat,ion of .000011 gram per ml.
+
1610
ANALYTICAL CHEMISTRY
All weighings were reduced to values in vacuo and the relative densities a t 25" C. were calculated in grams per milliliter. Expressed in these units, the density is numerically equal to the specific gravity a t 25" C. relative t o water a t 3.98" C. Refractive index measurements were made with a Bausch and Lomb Precision oil refractometer, with water controlled at 20.00" f 0.02' C. circulating through the prism blocks. The temperature coefficient for the refractive index of water is 0.0001 unit per degree Centigrade; t h w , in order t o obtain readings accurate t o the limit of the instrument (0.00003 unit), temperature fluctuation could be m much as 0.3' C. Similarly for glycol, the temperature coefficient is 0.00026 unit per degree C p t i grade (S), necessitating temperature control to only 0.1 C. Duplicate analyses never differed by more than 0.00003 unit. provided the sample was injected between the closed prisms by means of the hypodermic syringe. Earlier work on pure glycol showed that the refractive index decreased as rapidly as 0.00012 unit per minute if exposed to room air a t approximatel) 35% relative humidity.
Coiiiposition, Weight % Ethylene Glycol
Refractive Index, n go
n
1.33300 1,33810 1,34760 1.35724 1,36925 1.38107 1,38990 1.39990 1.40939 1.41407 1.41835 1.42184 1.42668 1.43182
5.44 15.14 24.85 36.58 48.05 56.65 66.53 76.13 80.97 86.50 89.23 94.45 114.93
Table 11. Composition, Keight r6 Ethylene Glycol
n
1 .'do38
1.0163 1.0294 1.0449 1 ,0597 1.0701 1.0813 1.0910 1.0954 1.0993 1.1023 1.1062 l.lOW3
Smoothed Data
Refractive Index, n 5' 1.33300 1.34242 1.35238 1,36253 1.37276 1,38313 1,39336 1.40340 1.41316 1.42262 1 43188
10 20 30 40 50 60 70 80 90 100
Absolute Density, Grams/MI. a t 25O C .
Absolute Density, Grarns/Ml. at 25O C. 0.99707 ( 5 ) 1.0096 1.0229 1.0364 1.0495 1.0620 1.0740 1.0850 1.0945 1.1029 1,1099
WEIGHT % ETHYLENE GLYCOL Figure 1. Densit>--compositiondata Ethylene glycol-water solutions
Experimental data are listed in Table I. Large scale plots were drawn for both density and refractive index (Figures 1 and 2), from which the smoothed values at even composit,ion increments given in Table I1 were obtstined. LIMITATIONS OF DATA
Analyses based on density determinations accurate to within 10.0001 gram per ml. will he good only to 1 0 . 1 weight %. Further, the comparatively involved manipulations required increase the chance of experimental error considerably over that likely with refractive index determinations which are inherently more accurate (+Q.03%). Then, too, refractive index determinations have the obvious advantages of requiring smaller samples and much less time per sample. The primary requisite for precise :tnalysis by refractive index is t h a t of preventing sample contact with moist air; the use of a hypodermic syringe is strongly recommended. The experimental density data agree closely n-it'h t'hose of Spangler and Davies (11)except over the composition range from 50 to 70 weight % of glycol. A large scale plot of their data indicates slight breaks a t 54 and 69% glycol in an otherwise smooth curve. DISCUSSION
The density-composition curve determined in this work is very slightly concave upward over the range 0 to 20 weight % of glycol, and concave downward over the remainder of the composition range. This double curvature appears to be characteristic of aqueous glycol systems in that it is indicated in the work of Spangler and Davies (11) on the ethylene glycol-water system a t 2,;" C., Cragoe ( 2 ) on ethylene glycol and m-ater a t 20" and
WEIGHT % ETHYLENE GLYCOL Figure 2.
Refractive index-composition data Ethylene glycol-water solution
15.6" C., LlacBeth and Thompson ( 7 ) on aqueous propylene glycol a t 25" C., and the same authors ( 8 ) on aqueous diethylene glycol at, 2.5' C. T h a t a volume change occurs upon mixing ethylene glycol and water is indicated by the lack of a linear relation beh-een density and volume per cent composition. The refractive index-composition curve is also S-shaped, being ('oncave upward from 0 to 40 weight % of glycol. This is contrary t,o the relatively common assumption (3) that refractive index varies linearly Kith weight per cent composition for t,his p:irticular system. Analyses based on a st,raight line het'ween the refractive indexes of water and purified glycol would result in a maximum error of 1.6 weight % a t about 75% of glycol. As Lvith the density data, the characteristic shape of the refractive index curve is indicated for the ethylene glycol--mter system in the data of Spangler and Davies (11) and Romstatt (9) and for aqueous propylene and diethylene glycols by the work of MacBeth and Thompson ( 7 , 8 ) . ACKNOWLEDGMENT
This work, representing a portion of E. T . Fogg's doctoral research, was supported by a grant awarded to A. R. Thompson
V O L U M E 27, NO. 1 0 , O C T O B E R 1 9 5 5 by the Committee on the A4dvanceiiientof Research of t h r University of Pennsylvania. LITERATURE CITED
(1) Conrad, F. H., Hill, E. F.. and Ballman, E. A , . I n d . E I L UC.’ / , e m . ,
32, 542 (1940).
(2) Cragoe, C.
S;,
“Properties of Ethylene Glycol and Its Akqueous Solutions, Cooperative Fuel Research Committee of the Cooperative Research Council, Kew York, S . 1’.(Declawified 1946). (3) Curme, G. O., and Johnston, F.,“Glycols.” d.C.6. .\lonograph 114, Reinhold, S e w York, 1952. (4) Dunstan. A. E.. Z. p h g s i k . C‘hem., 51, 733 (190;).
1611 (5) Lange, N. A., ”Handbook of Chemistry,” 6th ed., Handbook Publishers, Sandusky, Ohio, 1946. ( 0 ) Lawrie, J. W., “Glycerol and Glycols. Production, Properties, and Analyses,” A.C.S. Monograph 44, Chemical Catalog Co., New York, 1928. (7) hIacBeth, G., and Thoinpsnii, - 4 . I ? . , ANAL.CHEY..23, 618 f1951).
(8) Ib;d.:24, 1066 (1952). (9) Romstatt, G., Industrie chinziptie, 22, 648 (1935). (10) Schierholtz, 0. J.. and Staples. 11. L.. J . Am. Cheni. Soc.. 57, 2709 (1935). (11) Spangler, J. A,, and Davies, E . C. H . , 1x11. EM+.CHEM.,.\SAL. ED.. 15, 96 (1943). RECEIVEDfor rrriew Frhriiary 8. 1955. .4cwpted .lune 3. 1955
Determination of Halogen in Organic Compounds Potentiometric Microtitration with Silver-Amalgamated Silver Electrode System EVERETT C. COGBILL and J. JACK KIRKLAND’ D e p a r t m e n t o f Chemistry, University o f Virginia, Charlottesville, V a .
‘I‘he p o t e n t i o n i c t r i c titration of halide w i t h silver n i t r a t e , employ i n g the silver-anialgamated silver electrode system of Clark, has been adapted to the niicrod e t e r n i i n a t i o n of chlorine, bromine, and iodine in org a n i c compounds. This electrode sj stem lends itself w i t h a d v a n t a g e to micro work, because it is easj to cons t r u c t , c a n be made very small in size, and gives a sensit l \ e indication of the end point even in t i t r a t i o n s w i t h v e r y d i l u t e solutions. Procedures are given w h i c h provide a c o n v e n i e n t and rapid analysis of o r g a n i c conip o u n d s on a micro scale, w h e r e 1 to 4 mg. of s a m p l e are abailable. If the substance contains ionizable halogen, it may be titrated directly in alcoholic solution; compounds w i t h nonionic halogen are decomposed most conveniently prior to the titration b y the c a t a l y t i c d r y combustion method. The procedures descrihctl have been applied to the determination of halogen i n standard o r g a n i c substances w i t h a precision of 3 parts per thousand I
T
HE direct determination of halogen in organic conipouride
invariably involves the measurement of halide ion, usualll. after preliminary decomposition of the organic subst’ance. Thr, determination of halide on a micro scale by gravimetric means ie a tedious and lengthy procedure, and hence t’itrimet,ricmethods are preferred for routine analysis. Of the volumetric nicthodq for the determination of halide, those employing visual indicators s u f f ~ from r a lack of end point sharpness, because of the high dilution of the sample and titrant which must necessarily be The potentiometric titration of chloride, bromide, and i with silver nitrate is, however, applicable to very di1ut.e solution^ and lends itself well to the determinations required in micronnalyticnal work. The potentiometric titration of halide with silver nitrate is comnionly carried out with a metallic silver indicat,or electrotit. and a reference electrode of the conventional type, such as silversilver chloride or calomel, which is connected to the titration cel! by a halogen-free salt bridge. For the titration of halide on :I niicroanalytjical scale, where approximately 0.5 mg. of the ion is to be measured, the volume of the solution must be kept small in order t80 minimize end-point errors due to solubility of the precipitated silver halide. Consequently, the electrodes u ~ c d must also be kept small in size. Moreover, the insertion of n 1 Present address, Experimental Station. E. I. du P o n t de Nemoiirs & Co., Inc., Wilmington, Del.
s:ilt, bridge int,o the titratioti V P I i p often iiiconvenierit rtnd iritrodures a poteutial so~v(:c’ of error in contaminating the solution with traces of halide or loss of the ion by diffusion into it,. In 1926 Clark ( 1 ) suggest,ed a bimetallic type of electrode system for the titration of chloride iyith silver nitrate, which has none of the disadvantages which makr c,lectrode pairs of the conventional kind awkward for micro work. This electrode pair consists of a plain silver elec-trodo and one oi amalgamated silver. This system was later adapted by Cunningham, Kirli, and 131ooks ( 2 , 3) to the ultramicrotitration of chloride in biological fluids. The latter authors found that in drop-scale potentiometricn t’itrntions, using a pair of tiny silver and amalgamated silvcr electrodea, microgram quantities of chloride could be deteriliitied Ivith an accuracy of about 0.5%. This silver and amalgamated silver electrode system deserves R wider application in the determination of small quantitirs of halide than it appears to have enjoyed. It is extremely sinil)lc t o construct and can be made to fit any requirements of size. It gives a sensitive indication of the end point in the titration of bromide and iodide, as well as chloride. The authors have n.dapted :I modification of the procedure of Kirk arid his coworkers to the routine elementary analysis of organic