THE DETERMINATION OF SULFUR IN PETROLEUM OILS BY GEORGE ROCKWELL BARNETT B. S. Monmouth College, 1918 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN CHEMISTRY IN THE GRADUATE SCHOOL OF THE UNIVERSITY OF ILLINOIS, 1922 URBANA, ILLINOIS Digitized by the Internet Archive in 2015 https://archive.org/details/determinationofsOObarn \ UNIVERSITY OF ILLINOIS THE GRADUATE SCHOOL MAY 31 1 .192 2 i HEREBY RECOMMEND THAT THE THESIS PREPARED UNDER MY SUPERVISION BY GEORGE RQCEKELL BARRETT ENTITLED THE DETSRMIRATIOR OP SULPUR IR PETROLEUM OILS— BE ACCEPTED AS FULFILLING THIS PART OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OP SGIEROE ceS In Charge of Thesis art. tTK cr Head of Department Recommendation concurred in* Committee on Final Examination' ‘Required for doctor’s degree but not for master’s (I) TABLE OF CONTENTS ACKNOWLEDGMENT I INTRODUCTION II FORMS OF SULFUR III METHODS FOR SULFUR ANALYSIS IV REVIEW OF METHODS V OEJECT OF INVESTIGATION VI EXPERIMENTAL VII MISCELLANEOUS POINTS TAELES VIII SUMMARY AND CONCLUSIONS IX BIBLIOGRAPHY page 2 FF* 2-4 FP. 4-5 pp. 5-9 pp. 9-17 p . 17 pp. 17-38 U* 28-40 p p . 4 1-44 pp. 45“ 46 ( 2 ) ACKNOWLEDGMENT The writer wishes to express hi s sincere apprec- iation to Prof. S. W. Parr. Much of the present work would have been discouraging had it not been for the inspiration of his real interest and wise guidance, which were ever available to the worker. The writer is also grateful to Dr. T. E. Layng, whose problem of sulfur removal was the starting point for the present work. Acknowledgment is likewise due the Standard Oil Company of Whiting, Indiana, and the Texas Company of Houston, Texas, for complimentary samples of Louisiana and Mexican crude oils. ( 3 ) THE DETERMINATION OF SULFUR IN PETROLEUM OILS, I INTRODUCTION The problem ox determining the total sulfur content of petroleum oils confronted the writer in connection with an investigation of the efficiency of various metallic oxides in removing the sulfur. A lar^e number of sulfur determina- tions were to be made, and for this reason a method was sougnt which would d 0 a i mp 1 a 00 c ^ u 1.0 x ana r a p x. u , a c c i 1 as accurate, ana a x t n e same t i m a ap 1 i a a u r e to ^ ^ li c u.^-. 0 x w * a a 1 y v a x*y i ng percents of sulfur. A search through the literature would convince one, at first, that it only remains for the chemist to pick at random any one of a large number Oi methous waich nava, seemingly , been t r i e a. out ■< 1 i:i success j v ax x 0 us workers, , - - - u 1 s e each is seen to nave his preference, but it is apparantly only a problem 0 a m a k i ng a 1 i a l Ox t a e m e t xx 0 ds wn 1 c n * xx e app- aratus available to the particular laboratory will permit, and making the individual selection according to the temperment of the individual worker, 1 1 xx a a 0 e a ti w Xi e exw^-axxenae ox 0 h e w x x 1 0 x , xx o w e v e x , L xx a x most metnods prove to be disappointing. Althougn with practice one oan 0 b a x n good o xxe oka wr ^ xx c* pxooeuuxe, 0 xx e oonxidence t e mp or ax xij plaoeu x n txxe p x 0 c e a a i a uiaai gated •. xx a n o xx e x c a u 1 1 s w a s xx one m etxxod a x a 00 m p a x a a i on x e s u 1 1 s on 0 n a a a m e o x i ( 4 ) obtained by other methods. Although all me o ho u s were not ti i & u f tne ^ e n o x* ax experience uas been the same. First comes the straggle to master the re- tails oi the metnod (in general involving more or less teuxous procedure) to make eacu step in tne process exactly tne same in all cases in order to obtain good checks. Only to find in the end that the pursuit ol check results nas indeed been but the pursuit ox a mirage, wne n it is seen tha ♦. another metnoo which gives ecpuali^ ^ood ciiecks does not give the same results. It follows, then, that check* results mean little; that the same oil will analyze d i x X e r e n 1 1 y n 1 u a dixici'Siii/ m e t n o d x ; a n u. that tiie more me^non*** u s e x i t *. determining i iiC p e r ■ a ^ o x wUli'ui in a certain oil, tne great . me's doubt as to now sulfur it contains. In \ 1 & it Ox t n e s e c o n x i a e x e t x o n x , xt > » cl s v i 16 d t o x o. ^ t n x main p r o b i e m ox tne removal Ox suliux t e mg o r a x i 1 y upon tne x n e 1 x , until the matter of a satisfactory method for the determination of sulfur could be v/orked out. II FORMS OF SULFUR PRESENT. Unfortunately xOi t n e analyst one s u 1 x u r in gecioxeum uoes not occur in one xox*m alone, x u t in a vai'xety Ox f o r m s . In x p i x e of the well-known difficulty of isolating pure substances from sucii complex mixtures as petroleum and its fractions, a large- number have been identified. V.'hen the latter are classified it is seen that the sulfur in petroleum is present not only in various types of organic molecules, but as free sulfur 1 , and . ( 5 ) sometimes as inorganic sulfates if the oil contains ground water emulsoids or inorganic solids in suspension 2 . For analytical considerations, it is well to keep in minu the forms in which the sulfur is present. For purposes of tabulation, the following are named: (a) free sulfur 1 ; (b) hydrogen sulfide 2 ; (c) thiophenes 4 and thiophanes*; (d) alkyl sulfides 6 ; (e) mercaptans 7 ; (f) carbon bisulfide 8 ; (g) sulfonic acids 9 ; (h) sulfonates' 0 ; and (i) alkyl sulfates 1 '. Inorganic sulfates mentioned in the previous paragraph are not usually considered in sulfui analysis, because water and solids are supposedly separated previous to the sampling. Ill METHODS FOR SULFUR ANALYSIS. It is seen that it is quite a problem to choose a combination of procedure and reagents that snail be efficient in converting all of the sulfur in its various forms to one certain form, and separate it from the other substances present without loss of sulfur at any point in the process. The fact that results by different methods disagree among themselves makes it self-evident that all of them are not efficient, as they should be, at least in the way they are applies. To the present there seems to be no one method which is applicable in all cases. Some of them seem altogether inapplicable. At the most, it seems that every method proposed is subject to special and often serious limitations. The problem in sulfur analysis is, then, to find out which of the proposed methods give trustworthy results; and, if , ( 6 ) possible, to find a new method or improve one of the old so that it shall be more generally applicable. For future consideration it might be well, at this point, to review the possible methods which have been proposed for the determination of sulfur in petroleum oils or similar substances. They divide themselves into two main classes: [1] Reduction to Hydrogen Sulfide. The possibility of reducing all the sulfur to H 2 S is interesting because tne latter is easily and quickly estimated. The first attempt in this direction, as far as known, was that of Goetz!' 2 , in which he tried to fAUge tne a... cunt of sulfur by the amount of H 2 S which came off wuen tne oil was uis tilled. Although it is known that oils containing no H 2 S originally will evolve the gas when distilled 13 , no usable relation between the amount' of sulfur present and the amount of H 2 3 evolved could be found. Next Waters 14 attempted to hydrogenate an engine oil with gaseous H 2 in the presence of a catalyzer, oxidize the H 2 S in an absorption solution of hypobromite, and precipitate the sulfur as BaSCU. Catalyzers tried were metallic nickel and nickel oxide, ana platinum foil. Waters recovered only a very small fraction of the total sulfur in this way. The most promising attempt yet made seems to be that of Heulen 15 , by which he heats the substance in a combustion tube in a current of H 2 , passes the gases formed through a platinized asbestos catalyzer, and absorbs the H 2 S in the effluents either in alkali and determines it iodome tricaily, or absorbs it in ( 7 ) sodium plumbite solution, and determines it colorimetrically. Meulen gives a table of comparative results on seven different petroleum fractions and two crudes, ranging from .024^-5. 100^ S, in which he compares the percents of sulfur obtained gravimet- rically (method not stated) and obtained by his method. The results in each case check remarkably well in every case, being in some instances lower and in others higher than the gravimetric, but on the whole slightly higher. It would be interesting to know what gravimetric method he used for comparison. Meulen 1 s method is extreme Ly interesting, because he applies it equally well on any petroleum or fraction oi high or low sul- fur content (and on other organic substances, such as coal) . Put in the opinion of the writer it is not, at present, to be viewed with undue optimism from the practical analytical standpoint. His samples (50 mg.) are too small to give good checks on any but the most sensitive balances, and with larger samples it would be ex- pected that trouble would be had with rapid poisoning of the cat- alyzer, which was probably Waters trouble. In any event it is too early to state just what possibilities may lie in this method until it is further tried in other laboratories. No work has been done on this method in the present investigation. [2] Oxidation to Sulfate. To the second general class belong ail of the so-called standard anu rapid methods* Ail of them depend upon oxidizing the sulfur to sulfate, eliminating the residual oil (usually by complete oxidation), and determining the SC 4 ”. For convenience, — • > ' ■ - - ■ ( 8 ) the methods have been outlined as follows: 1 - Digestion With Wet Oxidizing Agents. (a) Strong KMn0 4 - Waters 16 . (b) Nitric or fuming nitric acid- Waters' 7 ; Andrews' 8 ; Francis and Crawford' 9 ; Calvert 20 . (c) Nitric or fuming nitric acid, and oxidizing agents- Waters 21 ; Gilpin and Schneeberger 2 2 ; Francis and Crawford' 9 ; Gill and Grindley 23 . (d) Hydrogen peroxide- Lecocq and Vandervoort 24 . (e) Eoil NaCH + Zn + * + Pb + + to form PbS- Schulz 25 . 2 - Burning Methods. (a) Lamp- Heusier 26 ; Engler 27 ; Friedlander 28 ; Magruder 29 ; Ellerton 30 ; Conradson 31 ; Lomax 32 ; Esling 33 ; Bordas 34 ; Bowman 85 ; U. S. Bureau of Mines 36 . (b) In enclosed atmosphere of oxygen- Hempel 37 ; Grafe 38 ; Marcusson and Doscher 39 ; Hauser 40 . (c) In combustion tube- Mabery 41 ; Barlow and Tolleus 42 ; Bay 43 ; Sauer 44 ; Dammer 44 ; Dennstedt 46 ; Brunck 46 ; Vita 47 ; Lant and Lant-Ekl 48 . 3- Open Fusion Methods. (a) With alkalis- Eschka 49 ; Hundeshagen~° ; Handy 51 ; Heath 82 ; Hertig 53 ; Sadtler~ 4 ; Garrett and Lomax 55 ; Parr 66 . (b) With mixtures of alkalis and oxidizing agents- Aufrecht 57 ; Lidow 58 ; Peckham° 9 ; Langmuir 60 ; Dubois 61 ; Schillbach 62 ; Blair 63 ; Chari tschkoff 64 ; Koch and Upson 65 ; Schreiber 66 ; Smith 67 ; Falciola 68 . ( 9 ) 4 - Bomb Methods. (a) Oxidation with fuming nitric acid in a sealed tube under pressure- Carius 69 70 ; Anelli 71 ; n oland 72 , (b) Oxygen Bomb- Filiti 73 ; Allen 74 ; Lord 7 **; Lohmann 76 ; Philip 77 ; Falciola 78 ; U. S. Bureau of Mines 79 ; Christie and Bisson 80 ; Parr 8 '. (c) Sodium peroxide- Osborne 82 ; Konek 83 ; Parr, Wheeler, and Eerolzheimer 84 ; St. Warunis 85 ; Lidow-Holde 8 6 ; Francis and Crawford 87 ; Franks 88 ; Parr 89 , 5 - Combination Methods. (a) Wet digestion followed by dry fusion with alkalis- Goetzl 90 ; Sanders 91 ; Whittum 92 ; Rothe 93 94 95 96 ; Waters 97 ; Otsubo 98 . IV R'EVIEW OF METHODS. In reviewing the oxidation methods under [2] , a strict historical presentation will not be attempted. The modifications will be taken up in the same order as the classification given in the outline. In the wet digestion methods. Waters obtained low results when he attempted to oxidize the sulfur with strong KMn0 4 solution. He also obtained low results with the nitric acid digestion, even when adding such oxidizing agents as KCIC 3 and KMn0 4 . Francis and Crawford came to the same conclusion regarding nitric acid diges- tion, with or without oxidizing agents, even though the digestion is carried out under a reflux condenser to prevent loss of sulfur. Andrews, however, claims good results when the reflux is used. ( 10 ) Gilpin and Schneeberger claim this method checks very well with the method of Carius, which will be mentioned later. Gill and Grindley vary this method by using KN 0 3 as the auxiliary oxidiz- ing agent. Lecocq and Vandervoort attempt to oxidize with H2O2. Still another variation of the wet method is that of Schulz, in which he boils the sample in a strongly alkaline solution of Zn(C 2 H 3 02)2 and Pb(C 2 H 3 C2)2» to change the sulfur to PbS, filters, and fuses the residue with sodium and potassium carbonates. Inas- much as the workers who have tried these methods are not agreed among themselves as to whether the results are to be trusted, it is thought doubtful if much confidence is to be placed in such style of attack. Of the burning methods, by far the greatest amount of work has been done on the lamp method, and its modifications. This method adapts itself most readily to light burning fractions containing a very small percent of sulfur, on account of the fact that a much larger sample can be used than can be taken by other methods. The method simply consists in burning the oil itself (or mixed with a suitr' i •: , . h ;,m 11 , solvent) in a lamp, catching the gases of combustion and passing them into an absorbing alkali. The apparatus needed can be improvised easily, but the method is objected to because it is long and tedious, and subject to error because of the retention of sulfur in the wick. However, if the lamp method is to be used, the latter objection can be overcome by running a sulfur analysis on the wick, according to Conradson. Or as suggested by Bordas, a bundle of capillaries can be used instead of a wick. On the whole, the lamp method is to be avoided • (M) on account of the labor involved, except in cases of burning oils which have so little sulfur as to require a larger sample than can be taken in the oxygen bomb. For these fractions, the lamp method is considered the standard by the U. S. Bureau of Mines, and others, and is frequently used as a basis for reference. Burning the sample in an enclosed atmosphere of oxygen is used by some workers. It consists of suspending the sample in a small container suspended in a large bottle (!C-I£ liters capac- ity) filled with oxygen. After the combustion an absorbing liqu- id is introduced. The metnod seems to have little to recommend it in the way of convenience of manipulation, and in addition incom- plete combustions are troublesome. Burning the sample in a com- bustion tube in a stream of oxygen likewise has nothing in the way of simplicity of operation to recommend it. The method re- quires the usual combustion Lrain, a platinized asoestos cat&j.^ - aer to insure complete oxidation, and the reaction has to be kept down by a careful dilution of the entering oxygen fc;> means of carbon dioxide. When properly controlled in every det ail t n e method is accurate, but as a practical analytical procedure it is to be studiously avoided. From the standpoint of simplicity of operation, fusing the sample in an open dish with alkalis, or a mixture of alkalis and oxidizing agents, would seem to be the choice procedure. The object is to catch the sulfur from the decomposing oil by means of the alkali present in intimate contact, completely oxidize the residual organic matter by open ignition, and precipitate the sulfur from a solution of the residue as BaSC 4 . The method • t . ( 12 ) was first worked out by Eschka, and although more often applied to coal, has been used by some workers as a basis of reference for petroleum oils. The reagents most commonly used are MgC and Na 2 C0 s (2: I), known as Eschka mixture. This is mixed with the oil, sometimes leaving a layer of the Eschka mixture covering the sur- face during the burning. Or a double crucible arrangement can be used, whereby the escaping gases are forced to travel through 2-6 cm. of Eschka material before escaping. Very often, also, oxidiz- ing agents, such as KCIC3, KNO3, NH4NO3, and I{ 2 Cr 2 C 7 have been used in conjunction with the Eschka mixture, or in special comb- inations without the latter. For example, Falciola mixes the sub- stance before ignition with KC1C 2 , KNC S , and NH 4 N0 £ (4 : 1:1). Smith uses precipitated SiC 2 , N a 2 0 2 , and KClOo. Schillbach uses Ba0 2 , treats with i T Cl, filters all the soluble matter, and weighs the remainder as BaS0 4 . A great deal of work has been done on these dry fusion methods, and the Eschka type of procedure has enjoyed a widespread popular- ity. A great many writers seem to regard it as a standard, although of late there is a tendency to question the accuracy of the results with the Eschka method. Cn the one hand a worker 68 is congratulat- ing himself because his method checks so well with Eschka s, al- though giving slightly higher results; while on the other hand two in conjunction 46 condemn the Eschka method, claiming that on a high sulfur petroleum it gives 2.6% of total sulfur less than the actual sulfur content. On the whole the dry fusion method, with or without oxidizing agents, cannot be regarded with undue confi- dence at present. Part of the present investigation has been in . * (13) this direction, and will be covered further on. Of the bomb methods, that of Carius was the first orthodox procedure, and was intended for use on any type of organic sub- stance containing sulfur. The procedure is to seal the substance with nitric acid in a hard glass tube and heat until oxidized. The SC 4 ~~ is then precipitated with Ea**, Roland includes KC1C 4 . Anelli claims more satisfactory results by introducing Ba(N0s) 2 with the nitric acid. The Carius method doubtless gives reliable results on oils not too low in sulfur, but only small samples can be taken, and if the percent of sulfur is small the BaS0 4 obtained is so little as to be difficult of satisfactory manipulation. The method is also objected to on account of the danger of the tubes exploding. Oxidation of the sample in a bomb under 3C-40 atmospheres of oxygen is at present generally regarded as the most reliable method. About I gram of the oil is placed in a small metal cru- cible inside the bomb; a few cc’ s of water or alkaline absorbent placed in the bomb; the oxygen introduced through a one-way valve; and the sample ignited electrically. The sulfur is oxidized to S0 3 and unites with the water forming S0 4 “' . Careful analysis of the residual gas has failed to detect the smallest trace of incom- pletely oxidized sulfur" l0 °, when the proper oxygen pressure has been provided. For oils not too low in sulfur, this method is the standard in this country, as shown by the fact that it is so regarded by the U.. S. Bureau of Mines and the U. S. Bureau of Standards, in references quoted. It is highly recommended by the best writers, and the main objection to it is that the apparatus (14) required is more costly than the ordinary laboratory can afford. Cutting down the size of the bomb and eliminating the other special apparatus required with the oxygen bomb, the sodium per- oxide bomb has come into favor. It is simple to operate, and with Na 2 0 2 of standard purity the results are entirely satisfactory. However, the allowable sample is only half that of the oxygen bomb, so this method is even less adapted to low sulfur oils than the former. The procedure consists in placing the sample in the bomb together with about 14 grams of Na 2 C 2 , and sometimes an ac- celerating mixture of benzoic acid and potassium chlorate. The ignition is started by heating the lower part of the bomb, after which the residue is dissolved in water, the solution acidified and boiled to decompose the H 2 0 2 > the iron precipitated by ammonia and filtered, and the acidified filtrate treated with BaCl 2 to precipitate the S0 4 ~**. The article by Franks 88 is a good example of the careful application of this method in routine analysis on various oil fractions. Another variation in the method of attack is a combination of wet digestion, followed b> dry fusion with alkalis. Sanders, Whittum, Waters, and Otsubo treat the oil at first with nitric or fuming nitric acid, alone, or with an auxiliary oxidizing agent such as KBr or Br 2 . The mixture is heated and gradually brought to small volume, when an alkali is added and the organic matter is driven off by ignition. In the Rothe method the oil is placed in a flask with MgO, and then a large excess of fuming nitric acid is added. It is subjected to gentle neating for a time, and later ignited to decompose all the organic matter. In ' ’ (15) all these cases the residue is treated with dilute HC1 and fil- tered, and the SO*”” in the filtrate precipitated with EaCl 2 . Such methods are usually applied to low sulfur oils, because from 5 to 10 grams of sample can be taken, as compared with .5-1.0 gram in the bomb methods. More will be said later about this type of procedure, in connection with work done in the present investiga- tion. It is seen from the above that the selection of a method for determining sulfur is not always a simple one. The analyst must take into consideration, in the first place, the type of oil to be analyzed. Next, he must consider the accuracy necessary to the work at hand, the number of determinations to be made, the labor- atory equipment available, and so on. The type of oil to be analyzed has to be considered because there is no method which is perfectly adapted to all oils. The oxygen bomb method can be relied on to give dependable results with the widest range of oils of any other one method (Allen and Robertson 10 ! ) , but cannot be used on fractions of very low sulfur- content because the sample is limited to one gram. The peroxide bomb method is very comparable, but the allowable sample is only half that of the oxygen bomb. The lamp method is adaptable to rather large samples, and when the proper precautions are taken gives dependable results, but even when inflammable solvents are used it can hardly be regarded as successful with heavy fractions. The Carius method is very seldom used because it is subject to the same limitations as are the bomb methods, and in addition has objectionable features of its own. Of the other methods or modifi- ■ ( 16 ) cations cited, they are either held to be inaccurate, or accepted with reservations. Examples of the latter class are the Waters method and the Rothe method. Both are combinations of the initial wet oxidation with concentration of the sample, together with dry ignition at the end. Any type of oil can be used (although Vandaveer 96 applies the Rothe method to low sulfur oils only) and relatively large samples can be taken. If they give accurate re- sults, the methods would seem to recommend themselves, from the standpoint of flexibility at least. Waters claims good results with his method, and gives comparative results with his method and others on the same oils. Good results are also claimed for the Rothe method 96 . Investigation of the Rothe method formed part of the present work. In considering accuracy, e e d which is known to he in- accurate will find very little application. Since goou qualita- tive methods are known 19 , in which a rough estimate of the amount of sulfur can be made, it is questionable whether an inaccurate quantitative method has any use at ail. The length of time required to make an individual determina- tion is an important point to be considered, even when only an occasional analysis is run, and for routine work the tediousness of some methods is an absolute veto as far as the} are concerned. Of the more dependable methods the bomb methods are about the speediest and the lamp method the slowest, with the Waters and Rothe methods in between. As to the apparatus required, the oxygen bomb method invol- ves the greatest outlay, including the tank of compressed oxygen . (17) and fittings besides the bomb itself. The peroxide bomb is less expensive, but cannot be improvised. Apparatus for the other methods is of the ordinary laboratory type, or can be improvised. V OBJECT OF THE PRESENT INVESTIGATION. Such was the situation at the beginning of the present in- vestigation. The object in view has been to try out various phases of sulfur analysis by several different methods; to see what methods seem to show the greatest possibilities for usefulness; and how they might be modified or improved. All in the hope that a method might be determined upon which would be reliable, simple to carry out, and applicable to a greater variety of oils than the metnous now usee. ^ cannot s e c 1 e i m c, u u n a t s u c n a nc^.e <■ uo realized in the present work., But useful information was gained concerning the application of tsc metaods tried. VI EXPERIMENTAL. The work done can be grouped roughly under four heads: (A) Application of the peroxide bomb; (B) Work on op^n fusion methods; (C) Work on the Ro the and Tiiittum methods; and (D) Determination of 30 4 _ ~ in solution. (A) The Parr peroxide bomb 89 is a simple and easily manipulated piece of apparatus, and its use in the complete oxidation of oil samples for sulfur determinations is becoming more and more pop- ular. Some laboratories use this method almost exclusively, A . . (IS) case in point is the work of Franks 8 '-, in which he analyses shale oils and a complete sex 1 s o oi shale oil ^.xacuions o j t h i s m e t n o d . Ihe method, as applied by Franks, is as follows; One me us ure of sodium peroxide [uuout 14 g.j, 1 g. of powdered po t as s i um . chlor a te and 0.0 g. ox sen- zoic acid are placed in a Psrr borne and well mixed by shaking. About 0.5 g. of the uniform s aisle is weighed into this mixture by means of a medicine dropper and a smell weighing bottle. Ihe who © mass is then thoroughly mixed over a piece oi gla- red paper with a thin glass rod, any solids ad- hering to the rod or falling on the paper oeing subsequently returned to the bomb. After ignition in the usual manner the fusion is carefully dis- solved in about 50 cc. of water, the bomb ohox-ou— ghly washed with water, the inside being finally rinsed with about 1 cc. of concentrated i* 01 and a little more water, and the solution made acid with concentrated HC1. About 5 cc. of s at y rated bromine water is then added to oxidise ail the sulphur to sulphuric acid and the iron to the # ferric condition f i f not already in that condi- tion] and the solution boiled to decompose the KaQa and expel the Br a . Ammonia is added until the liquid is alkaline and the solution brought to a.vigorous boil to coagulate the ferric hy- droxide and expel the excess of ammonia [sso 2, under kISCELL ANEOUS PGIhTSj. The former is then filtered off, a small waa of absorbent cotton being placed in the bottom of the fil- ter to hasten the operation and facilitate washing. Jour thorough w a s h i n g s with hot water are sufficient. The solution is acidified with 1 cc. of concentrated PCI t using methyl orange indicator, adding 1 cc. in excess 102 ], made up to 22a cc., brought to a boil and 5 cc. of 10 per cent barium chloride solution is added slowly from a pipette. The boiling is continued until the precipitate is 'well fox'med, which sometimes requires as much as twenty minutes. Ab out 200 co. of liquid will remain. After standing over night the precipitate is filtered througn a close 7 cm. filter paper (Munkt el 1 s ho. 00 is preferred; and is carefully washed free from chlorides . In general the method quoted is a good one, and could be questioned in few particulars. Ihe writer tried varying proportions of KCIO3 and CqH 5 C00H, and found those given are about the best. Very few incomplete fusions were encountered wuen tne accelerator was weighed out with fair accuracy, but failures were frequent if the right amounts were not used. It might be questioned if four washings of the voluminous Fe(0H)s precipitate are enough to entire- ly free it from SC 4 _ ~. Efforts were made to demonstrate that slight- ly higher results would be obtained by dissolving tae precipitate, diluting the solution and then precipitating the Fe(CH ) 2 again. . . ' ( 19) It was found in practice, however, that the added step in the procedure did little more than make it difficult to obtain good checks. In addition it increased the concentration of ammonium salts in solution, which is undesirable 102 . The amount of Barium chloride used can be criticised in this as well as most standard methods, but this point will be touched upon more fully under (D) . Using this method Franks obtains checks which are within one percent of error from their mean with oils containing .5 % or more of sulfur y and successfully analyses fractions containing as low as .3% of sulfur. The latter figure is perhaps about the minimum for satisfactory results with this method. This for the reason tnat wi tn sucu small quantities o - s a d s L a nc e , a small loss of material or error in weighing would cause a relatively large error- in the result. For example, with a .5 g. sample of oil containing .3 a S, or . CO 1 5 g. S, corresponding to about II mg. of BaSCU, an error of .2 mg. in one weighing would cause a 2% error if' all the other operations in the analysis were perfect. The peroxide bomb method was first tried on a sample of Louisiana crude oil, using the procedure given by Franks. Ten individual samples were analyzed to see how tiie checks were runn- ing. It was found that slight errors in weighing came out a s large errors in the final results, but with careful work good results could be obtained and repeated. The last pair of samples came out • 122 % S and ,732% S respectively. Thereafter the standard of agree- ment between checks was taken as .02% of total sulfur. It might be added, however, that such would be altogether unsatisfactory for oils of very low sulfur content. • . ( 20 ) The next oil tried by this me thou was a gasoline engine cylinder oil. The runs on this oil were of value only in demon- strating that this method could not be applied on such a low sulfur oil. Results obtained ranged between .03$ S and . 17$ S. A comparison between this method and the oxygen bomb method was made using a sample of California crude oil,. which was the oil used in all later experiments on methods.. Two samples ana- lyzed 1.005$ and .990$, giving a mean of .998$ S, as compared with 1.003$ S, the average of 4 determinations with the oxygen bomb. In general the peroxide bomb method was found to check well with the oxygen bomb, without correction for sulfur in the reagents. Care was always exercised that the reagents were the best obtain- able (the Na 2 0 2 prepared especially for the purpose), and that they were not contaminated witn laboratory vapors. (E) The simple technique of a dry fusion method is rather entic- ing to the analyst. If one could mix the sample with a small quan- tity of powdered reagent, burn it off, dissolve it and determine the sulfate from the solution, the question of routine sulfur an- alysis would be solved. A large number of samples could be weighed out and burned off simultaneously, nothing in the way of special apparatus would be required, and the operations would be easy. An attempt was made to utilize the oxidizing principle of the peroxide bomb, by using Ka 2 C 2 mixed with various other substances to be ignited in an open dish. The advantage sought was that of using larger samples of oil than can be used in the peroxide bomb. r • ; -v ( 21 ) Mixtures of Na 2 0 2 , MgO, and Na 2 C 03 (using 5 g. samples of Califor- nia crude) were tried. The reagents were thoroughly mixed with the oil sample in a porcelain dish, then the mass was ignited by heat- ing the dish cautiously. The object was to dilute the Na 2 0 2 to such an extent that the fusion would travel quietly throughout the mass. It was found, however, that the oxidation was in ail cases so vigorous, even when the oil and Ma 2 0 2 were greatly diluted, that particles of the reacting mass were always carried off in the flame. This held true even when the Na 2 C 2 was entirely omit- ted from the mixture, except when only Na 2 C0 3 was mixed with the oil. In the latter case the residue consisted of a caked material with carbonaceous matter imbedded in the mass. The latter could not be successfully burned to an ash, for it assumed a pasty con- sistency when heated. Calcium peroxide was the next oxidising agent tried. It was first mixed with Na 2 C0 3 , but the activity seemed to be slignt. Then the Ca0 2 was mixed with the oil alone and heated. Oxidizing action was evident, the mixture giving off dense fumes during the action. The most encouraging feature with the Ca0 2 was that the mixture after ignition was light and powdery, containing very little carbon, and that easily burned off. The method was tried quantitatively by mixing 10 g. Ca0 2 with about 3 g. of oil. This a Sample from Bausch and Lomb Optical Co., known to have been in stock over IS years. Not more than a trace of sulfur in it. Another sample, commercial grade, was tried but it was loaded with sulfate. For the last trials a new s amp le of CaO s was obtained, which was fresh and much more reactive than the previous samples. It contained a small amount of sulfate, but blanks were run in every case. ( 22 ) mixture was ignited by carefully heating one edge of the dish, when the fusion would start spontaneously and creep throughout the mass, white fumes being given off meanwhile. After the reac- tion was completed the organic matter was completely burned off, the ash dissolved in HC1 and filtered, and the S0 4 — in the fil- trate precipitated with BaCl 2 . Average of the two samples was .383% S on the California crude containing 1.0% S. When a large amount of the Ca0 2 was taken (20 g. with 1.5 g. oil), the result was .579% S, after correcting for blanks on the reagents, which correction was made in every case in this phase of the work. The next oxidisin fe agent tried was Ba0 2 . It was hoped that the method proposed by Schillbach 62 could be used. It is the most direct method that has ever been proposed. It consists in mixing the sample with BaC 2 , heating the mixture to start the reaction, burn off any residual carbon, dissolve the ash in HC1 and weigh the insoluble matter as EaS0 4 » Various proportions of Ba0 2 and oil were tried, but the reaction always ended in a melt which was hard to free from residual carbon and formed a hard, horny mass when cooled. It was found that this horny residue would become more friable upon standing in moist air, but the method was aban- doned because of the difficulty of igniting the residue to a sat- isfactory ash. A desirable ash could be obtained, it was found, by using a mixture of BaC 2 and CaC 2 , and some runs were made with this mix- ture. After correcting for a large tare in the blanks, the result- ant sulfur was only .412%. This was in using CaC 2 of doubtful strength, however, so a new sample was obtained. The reaction ( 23 ) with the new Ca0 2 in the mixture was much more vigorous than with the old, but the net result was about the same. In fact it was found that all the runs made with the Ba0 2 or mixtures thereof had such a high sulfur content in the blanks as to make the results erratic. When so much sulfur was present in the reagents, and the latter were weighed out on rough balances, little could be hoped for in the way of accuracy. Qualitative tests on the reagents individually showed not more than a trace of sulfur in the new sample of Ca0 2 , but the BaC 2 which had been used all along, and which was supposedly pure, was found to be loaded with sulfate. Attempts to use the fresh CaC 2 alone were unsuccessful be- cause the reaction was so vigorous that considerable of the react- ing mixture was thrown out of the dish. It was tried in a beaker covered with a watch glass and also a lightly covered flask, with the same condition obtaining. A mixture of Ca0 2 and Ba0 2 had been tried in a small steel bomb and the bomb exploded, so the CaC 2 alone was not tried in a bomb. Sodium carbonate was tried as a diluent with the Ca0 2 , but the percent sulfur given by this method was .650$, or roughly two-thirds of the actual sulfur content. The proportions used were, 10 g. Ca0 2 , S g. Ua 2 C0 s , and 1.5 g.. of oil. Using half the amount of reagents in the same proportion, with about the same amount of oil, resulted even lower, .54!$ S. In concluding the trials with the open fusion method, it may be said that the results are disappointing. Although an accurate open fusion method may be developed, possibly, none of the methods tried to date are accurate. If it is a choice between the Eschka method, and the Ca0 2 -Na 2 C0 3 method last tried, the writer would (24) choose the latter. It is easy to manipulate, but even with the oxidizing agent present the results are low. Open dry fusion methods proposed are all subject to loss of sulfur. There is no doubt that a considerable amount of the sulfur content of the oil is lost through volatilization of light frac- tions in the initial stages of the reaction, and although the presence of oxidizing agents may help to conserve the sulfur at this point, the loss is still appreciable. Where the reaction is at all vigorous there is also the tendency of mechanical loss. On the other hand,- if the reaction is not vigorous enough, and much carbonaceous matter is left in the end, the sulfur already caught in the alkali may be lost again through reduction to decomposable sulfites by the incandescent carbon or CC present in the burning y off. Loss in this A was later prc with experiments on the Ho the and 7/hit turn methods. (C) Whittum 92 developed a combination method at the Universi ty of Illinois, and some further data concerning it was obtained by Otsubo 98 . The method was first proposed as a substitute for the lamp method and intended for light burning oils. An effort was made to prove its reliability, and also to extend its use to a wider variety of fractions. The procedure is as follows: Y.'eigh £— 1C g . cf the oil into a porcelain disa. Cuts I . h— . t g. of finely powdered KBr over the surface. Add 4 — E cc. of fuming nitric acid, drop by drop, with constant stirring. After the reaction has ceased, place the dish on a steam bath or asbestos pad and evaporate the liquid until it becomes viscous. Acid about h A . of Eschka mixture and mix thoroughly with the oil,, being careful to take up all ohe oil that may adhere to the sides of the dish. Place trie dish over an alcohol flame, or ( 25 ) Eur.een protected with an asbestos pad, and allow the mixture to dry slowly. The heat should be increased gradually until the mixture catches fire and burns tc a white powder, breaking up the large particles occasionally with a glass rod. After cool- ing add water and ECl to dissolve the resi- due and filter the solution. The SC 4 — in the filtrate is precipitated in the usual manner. Leaving aside the question of whether the method retains all the sulfur in lamp, oil, its behavior with crude oil was studied. Certain difficulties were encountered using this method with crude oil, and considerable work was done to try to eliminate them; and, when the procedure could be modified to suit a heavier oil, to see what results could be obtained by it. The first difficulty encountered was in the addition of the fuming nitric acid. Even adding it very slowly, drop by drop, the reaction was always so violent that some of the oil would spatter- out. Benzene ’was tried as a diluent to cut down the violence of the reaction somewhat. The spattering was diminished by the benzene, but not entirely eliminated. The results of duplicate samples us- ing the benzene were a trifle higher that those obtained by the straight Whittum method, so it was seen that the diluent in no way effected loss of sulfur. It was then resolved to try inert solvents. Chloroform, tetraclilor ethane, and carbon tetrachloride were tried, and of these the latter seemed the best. Quantitative runs using different amounts of CCi 4 failed to show any diminution in the amount of sulfur obtained, and tiie results were in some cases higher. It was concluded that the addition of enough CC1 4 to cut down the violence of the reaction had no undesirable effect, and was to be recommended. The method is to mix the CC1 4 , say about 15-20 cc., with the oil and then to add the fuming nitric . ( 26 ) acid slowly. The next difficulty met in the Whittum method was that, if the sample were heated for a time to concentrate the residue after the addition of the fuming nitric acid and before adding the Eschka mixture, the mass would solidify when it was attempted to stir in the latter, and a uniform mixture was impossible to obtain. It was found that a good mixture could be obtained if the Eschka reagent were mixed with the oil directly, or very shortly after the reaction with the fuming nitric acid had subsided. Another mechanical difficulty tnat was encountered was the fact that when igniting the Eschka mixture at the end to complete- ly burn out the residual carbon, the mass would assume a pasty consistency and be hard to nandle. This could be obviated by in- creasing the proportion of MgO in the mixture. As the chief mechanical difficulties seemed to have been met, quantitative runs were made on the California crude oil, of 1.0$ sulfur content. The Whittum method modified only by adding the Eschka mixture directly after the reaction with fuming nitric acid had ceased gave .684$ S. The same, except that an inert solvent was added before the fuming nitric acid, gave .727$ S. Varying the amounts of fuming nitric acid and of MgO and Na 2 C0 2 , results were obtained: .758$, .768$, .734$, .813$, and .749$. It was found that a larger percent of sulfur could be obtained by increasing the amount of fuming nitric acid to about 7 cc. A larger percent of Fa 2 C 02 in the absorbing mixture had the same effect, but the limit for the method was the highest figure given, .813$ S, and this was obtained without correcting for sulfur in the reagents. Also (27) the larger amount of Na 2 C 03 used made the residue difficult to handle, and it was ignited for several hours in the effort to obtain a white ash. It is probable that in this time some sulfur was absorbed from the laboratory fumes. It was thought that perhaps an absorbent present at the time the fuming nitric acid were added would conserve the sulfur. An- hydrous MgCl 2 was first tried, for the reason that it would in no wise take up the energy of the oxidizing acid as would MgO or Na 2 C0 E , and could at the same time unite with the less volatile S0 4 ““, HC1 escaping. But only .328% of sulfur was obtained in this attempt. Another modification was tried, by using anhydrous SeOCl 2 as the oxidizing agent. Some very interesting reactions ensued, chief among which was the escape of reddish fumes of selenium. Some of the sulfur was oxidized to S0 4 ~~, and the residual Se was easily gotten rid of in the end by volatilization during the subsequent ignition of the residue. But the oxidation was either incomplete or selective, for the percents of sulfur obtained by the SeOCl 2 were .278% and ,290% in two runs. The last trial of the Whit turn method was made about, as the first, using the larger amount of fuming nitric acid with smaller samples (2. 5-3.0 g.), and then during the ignition with the Eschka mixture, the dishes were piled full of MgO to force the escaping gases to travel through a thick layer of absorbent before they could escape. The fact that I. 108% of sulfur was obtained in this run was not taken as proving that all the sulfur could be obtained by this method, but rather as an indication that other sources of * ( 28 ) error were being multiplied, e.g., high concentration of Mg salts causing occlusion in the EaS0 4 precipitate; and sulfur in the re- agents, for which no correction was made. In later experiments it was proven that in this general type of procedure the sulfur is lost in two ways: by volatilization during the addition of the fuming nitric acid, and by reduction to S0 2 and consequent volatil- ization or otherwise in the burning off of the residue. In view of these facts it will be seen that even if all the sulfur could be retained in the burning of the residue, by passing the escaping gases through a large quantity of absorbing alkali, there would still have been the loss encountered at the initial stage in the analysis. The Whit turn (and Rothe) type of sulfur analysis was held to be inherently at fault in these respects, at least as applied to crude oils. Reasoning by analogy it might be questioned whether such methods are reliable on low sulfur oils. Of course the total error in sulfur with a low sulfur oil would be small, in propor- tion with the small amount of total sulfur, but the results by the Whittum type of procedure might still be questioned. The method of Waters is essentially the same as the Whittum method, except that Waters uses fuming nitric acid which is sat- urated with bromine. This method was not tried in the present in- vestigation. The Rotiie method has come into some favor, as shown by the fact that it has been tentatively adopted by the A.S.T.M.; and the recent work of Vandaveer 96 would seem to recommend it. The principles of the method are the same as in the Whittum method, variations in procedure being that the operation is carried out ( 29 ) in a flask, that the absorbing material (MgO) is added before the addition of fuming nitric acid, and that a large excess of the latter is used. The procedure recommended by Vandaveer: Weigh from 5 to 10 grams of oil into a liter pvrex flask containing 2 grams of magnesium qxi de . Immediately add carefully 25 cc. of fuming nitric acid. Allow the mixture to stand until the reac- tion-ceases, Then heat gently on a sand bath for about three-quarters of an hour. At the end of that time gradually raise the temperature until all the nitrates are decomposed. Continue the heating over a full Bunsen flame for several minutes. Blow in air or oxygen and heat the flask with a free flame to drive off the carbon. Cool the flask, moisten the white residue with water, add concentrated hydrochloric . a \ ( r . I . ; ■ ;■ ■: i » (34) a justifiable outlay. The method is to make up the solution to be tested to 250 cos., taking 50 cc. of this solution, diluting to !00 cc. and adding an excess of sized BaCl 2 crystals. This results in a finely divided, non-settling precipitate. The tur- bidity of the solution is a direct factor of the amount of sul- fur, which can be obtained by direct translation of the photometer reading by means of a curve. The method is surprisingly accurate, when the eye of the operator has once been standardized by com- parison with solutions of known sulfur content. It has another advantage, also, that the relative error is no greater with a low percent of sulfur than with a high percent. The method using a BaCr0 4 suspension as the precipitant is interesting; but it is thought doubtful if the method proposed by Bucherer would be found satisfactory. The recognised standard procedure, however, is the gravimet- ric estimation of the BaS0 4 precipitate. It is also the most prac- tical when only occasional analyses are to be run. As there are points which are too oftei) lost sight of, some of the variable factors of this precipitation were tried out from a practical standpoint. Since it is a well-known lact 103 that BaCl 2 is apt to be ad- sorbed or occluded in a precipitate of BaS0 4 wnen it is forming, one would expect to find that standard methods for sulfur analysis would lay down a careful procedure to prevent such occlusion. There seems to be a laxity in this respect. Holde, the U. S.. Bureau of HI. Mines, the U. S.. Bureau of Standards, and Hamor and Padgett (in references cited) simply say that the precipitation of BaS0 4 is , .4 ' (35) made with BaCl 2 in the usual manner, except that in the Mines Bulletin No. 5, the directions are to add EaCl 2 solution drop by drop until there is an excess present. Parr precipitates by means of the addition of 10 cc. of 10% BaCl 2 «2H 2 0, regardless of whether the oxygen bomb, Eschka, or peroxide bomb methods are used. Franks adds 5 cc. of 10% BaCl 2 , using the peroxide bomb method. On the whole, it seems to be customary in oil analysis to add 10 cc. of 10% barium chloride solution. Since too much BaCl 2 in solution may introduce an error in the results, the question arises as to just how much to add, and in what manner, so that the error caused by occlusion of BaCi 2 in the PaS0 4 will be reduced to a minimum; and at the same time have enough Ba + + present to completely precipitate the SC 4 “~. From the standpoint of common ion effect, it would seem desirable to have a certain excess of Ba + + in solution, to decrease the sol- ubility of the B aSC 4 to be recovered . Consulting a table of solubility data ,M , the solubility of BaS0 4 in water at room temperature is 2.5*!0~ 3 g. per liter. Say precipitation was made in 400 cc. of solution, from the fusion of .5 g. of oil (the amount taken by the peroxide bomb method), this would mean that the sulfur which would go undetected, with no ex- cess of Ba* + , could be only .027%, due to the solubility of BaS0 4 in water. Treadwell-Hall ,0 3 recommends the addition of 10 cc. of N. BaCl 2 for each gram of BaS0 4 precipitate to be obtained. With this standard, let us assume an analysis of a ,5 g. sample of oil con- taining 1% of sulfur, and see what percent of sulfur could remain . A , . * ■ . ' . ' (36) undected in solution. The solubility product of EaS0 4 is !.2l*IO” 10 , or . 000000000 1 2 L In the sample is .005 g. of sulfur, which is equivalent to .03635 g. of PaS0 4 . According to our stan- dard, we shall therefore add .3635 cc, of N, BaCl 2 . Amount of BaCl 2 added, in grams , 03795 Theoretical amount necessary . ...... ,03365 Amt. BaCl 2 in excess of theoretical .... .00435 This amount of EaCl 2 in excess of the amount necessary to precipi- tate all the sulfur from 400 cc. of solution is equivalent to .0000376 gram ion concentrat ion of Ba + + , BaCl 2 being 72% ionized. The Pa** concentration due to the BaSC 4 in solution is .000011, giving a total Ba + * concentration of .0000436 g. ion. (Ba) ( S0 4 ) = .0000 11 x .00001! = .000000000 12! (S0 4 ) = .000000000121 = ^000000000121 = .00000249. “(Pa) .0000486“ PaSC 4 in solution is the same or .00000249 moles per liter. This is equivalent to .00023 g. of BaS0 4 in 400 cc. of solution, and on the .5 g. sample is equivalent to . 0063 % of sulfur. In the same manner, the amount of sulfur which could go un- detected, owing to the solubility of PaS0 4 in water, using the amount of BaCl 2 solution called for by Treadwell-Hali, in a .5 g, sample containing 2% of sulfur, would be only .0036%. Applying the same solubility data to an actual analysis of a sample of Mexican crude oil: sample,' .5232 g., analyzing 4.8 14 % S, the amount of sulfur which went undetected in this way was only .0015%. Of course the solubility of PaSC 4 is somewhat affected by presence of various salts in solution, as well as free acid, but such effects are usually slight except for the presence of acid 1 ( 37 ) ( -arid the trivalent metals such as iron, which are removed hy ammonia). The acidity can be regulated by adding I cc. of HC1 to the neutral solution before addition of the PaCl 2 . It will be seen that in all these cases the amount of sulfur which could go undetected owing to the water solubility of BaS0 4 in the presence of only a small excess of Fa** is well within the experimental error, and therefore to be disregarded. Then why add a larger excess? There is the practical reason, of course, and that is the uncertainty of the amount of EaSC 4 pre- cipitate to be obtained. Even from this standpoint it would hardly seem desirable to add so much PaCl 2 as 10 cc, of 10$ in all cases, regardless of the method or the size of the sample. When using the peroxide bomb method Franks has cut the usual amou'nt of PaCl 2 in half, and yet he is adding enough for the complete precipitation of 13.2$ of sulfur, considering the necessary excess and the size of the sample taken. With the usual amount in the peroxide bomb method, enough BaCl 2 is added to precipitate 26.4$ of sulfur. From a practical standpoint, does this unnecessary excess of BaCl 2 usually obtaining actually effect the results appreciably? The following results answer the question: Analyses By I he Parr Peroxide Bomb. $ S obtained by adding 5 cc. of 10 % BaClg Eolru f, S obtained by adding amt, £ a C 1 » called for by Tread- wel 1— Hall Increase in app — arant % of S due to addition of too much B aCl a Louisiana Crude 0.835 $ 0. 727 $ 0. 108 $ Mex. Crude No.. 1 4.940 4.8 14 0. 126 Mex. Crude No. 2 4.983 4.819 0. 164 In every case the filtrate was tested with a drop of H 2 S0 4 to . ( 38 ) make sure that BaCl 2 was in excess. In view of these considerations, it would seem advisable to take into account the size of the sample and the probable maximum percent of sulfur present when adding PaCl 2 , and not ado. an un- necessary excess. In the opinion of the writer, the me thou given in one refers . 0 of adding the : aOl 2 solution crop by urop is no c a practical one, for with small amoijuts of - ui fur no precipitate of EaSC'4 is visible at first, and only after boiling does it appear. Then it is seen as a turbidity, and it is a practical impossibility to gauge the amount of tiie precipitant which should be add-. . The latter condition also holds true for larger amounts 01 &uixui . The manner of adding the BaCl 2 also has somethin! tc ith whether it is occluded in the BaSC 4 precipitate. The bos . pi ; is to dilute the necessary amount of BaCl 2 to ICC . and add io lot to the boiling hot solution to be precipitate J • i -• c^itio, is made slowly, with stirring. In ordinary routine work, however, it should be satisfactory in most cases to add the EaCl 2 in a 10# or a 5 a solution, being careful to add it slowly and with constant stirring to the boiling hot solution. VII MISCELLANEOUS POINTS IN OIL SULFUR ANALYSIS. 1 -General . The mo s t convenient method for weigiiing oil samples was found to be a weighing bottle which has a medicine dropper ground into a tight joint at the neck. This was easy to handle, and pre- vented volatilization of the oil. - ( 39 ) Reagents should be the best obtainable, and laboratory fumes should be carefully excluded. Blanks should always be run on each new lot of reagents, and more often if necessary, as im- purities crop up where least expected. A case in point during the present investigation was a fresh bottle of Merkels ’’pure' BaC 2 , which was found to contain a high percent of S0 4 ~“> It is advisable to run blanks on every fresh packet of fil- ters, by selecting about four from different parts of the pack. It was found in practice that the weight of the incinerated fil- ter which had been treated exactly as the sample was always higher than that given on the label. For routine work it is sometimes convenient to use Gooch crucibles, but they have to be carefully filled with digested and washed asbestos fiber. It is easy to go wrong with Gooch crucibles in ignition work. In igniting the precipitates a blast should not be used, for BaS0 4 decomposes slowly at the temperature of the blast lamp. There is also danger of blowing particles of PaSC 4 out of the crucible. R-Peroxide Bomb. Attempts to guess out the amounts of accelerator reagents to be used result in frequent incomplete fusions. These should be weighed with fair regard for accuracy. For speed of manipulation it was found convenient to weigh out the accelerator in doses, folding the doses in small papers. These papers of doses can be bunched into packets, and are then in convenient form to add imm- ediately witiiout loss of time during the runs. . •'31 (40) In precipitating out the iron, a good excess of NH 4 0H should always be added, or else some of the sulfate may be held in com- bination in the precipitate as basic ferric sulfate. For the same reason, the solution should only be heated enough to coagulate the F e ( 0 H ) 3 sufficiently for good filtration. See Treadwell-Hall 10 2 The Fe (OH ) 3 precipitate should be well washed with hot water, for gelatinous precipitates are notorious for retaining soluble salts. When it is considered that it takes from 10 to 20 washings to entirely rid the PaS0 4 precipitate from chlorides, it does not seem reasonable to suppose that 4 washings of the Fe(0E) s precip- itate are sufficient, as stated by Franks. ' , ( 41 ) TABLE I Comparative Results By Different Methods- Otsubo 98 Results by the methods given: Oil Oxygen Bomb Peroxide Bomb Whi t tum-Otsubo Mex. crude 3.79 % S 3.97 % S 3.76 % S Light oil 0.587 0.580 0.528 TABLE II Comparative Results By Different Me thods-Vandaveer 9 6 Results by the methods given: Oil Peroxide Bomb Whi t, turn Ro the-Vandaveer Fuel oil 0.367 0.33 1 Fuel oil a 0.340 0.360 0.408 • Kerosene 3 0.037 C.040 Color oil 0, 295 0 . 2! 1 Crude 0.430 0. 200 Crude 0. 200 0. 187 The only instances given in frhich results by the wet and dry combination methods were not lower than in the bomb methods . MOTE: Results given on this page are averaged results from the references cited. TABLE III (42) Test Funs on California Crude Oil Method Percent S. Peroxide bomb method; samples Average of _ - s amp 1 e s . about .5 g.; Duplicates shown 0.998 % S checked within .015% sulfur 2 Oxygen bomb method; samples about .9 g.; individuals, 1.003 .988%, 1.028%, .993% S r? u AMOUNT OF SULFUR IN CALIFORNIA CRUDE taken to TABLE IV be 1.000 per cent. Experimental runs on California crude. Method. Percent Samples Whittum, anhydrous MgCl 2 in dish with oil 0.328% 2 Whittum, but Se0Cl 2 the oxidizing agent 0.278 O Whittum, but SeCCl 2 the oxidizing agent 0 . 290 2 Whitt um, Eschka directly after fuming nitric 0.684 2 Whittum, oil diluted with benzene 0. 727 2 Whittum + 10 cc. CC1 4 0.758 2 Whittum + 15 cc. CC1 4 0. 768 2 4 ^ CO ) TABLE IV (continued) Method Percent S Samp les Whittum, CCi 4 and 7 cc. fuming nitric acid 0. 734 2 Whittum, CC1 4 + 7 cc. fuming nitric + 2 g. MgO + 4 g, Na 2 C0 3 0.8 13 2 Same as previous, except 12 cc. fum. nitric 0. 749 2 Whittum, large amt. Eschka heaped over dish 1 . 108 ! Average by the selenium oxychloride 0.284 4 Average most representative Whittums 0. 734 10 TABLE V California crude by the Ro the method. Method Percent S Samples Without modification^ 0.685 1 T. enty cc. CC1 4 added, before the fuming nitric 0.782 1 Same as previous 0.75 ! 1 Same as previous 0.786 ! Same as previous C. 686 1 Representative average C. 773 3 a Five hundred cc. flasks were used in all exp er iments with the Rothe method. . ( 44 ) TABLE VI Open fusion experiments with the California crude. Method Percent S Samples 10 g. Ca0 2 + 3 g. oil 0.383 2 20 g. CaO 2 + 1.5 g. oil 0.579 2 lOg. Ca0 2 +6g. Ba0 2 +2g. oil 0.412 2 10 g. CaC 2 ( fresh) + 3 g. Ma 2 CC 2 + 1.5 g. oil 0.650 2 5 g. CaO 2 (fresh) + 4 g. Pa 2 CC £ + 1.5 g. oil C.54! 2 TAPLE VII Comparative runs on Mexican crude. Method Percent S Samples Peroxide bomb 4.940 2 Eothe 1.670 2 ( 45 ) VTIJ SUMMARY AND CONCLUSIONS. 1. A review has been made of the various methods which have been proposed for the analysis of sulfur in petroleum oils or similar substances. 2. Methods where the oil is fused in open dishes with alkalis or mixtures of alkalis and oxidizing agents are held to be inacc- urate, due to loss of sulfur. Efforts to modify the general pro- cedure, together with the trial of new oxidizing agents to pre- vent such loss of sulfur, have not met with success. 3. The Whittum and Rothe methods, whereby the oil is treated with liquid oxidizing agents, followed by dry fusion of the resi- due, were also tried out with crude oils. Loss of sulfur was found in every case, and efforts to modify the methods were unsuccess- ful in preventing such loss. 4. It was demonstrated that loss of sulfur, as shown under 2 and 3, is occasioned at two points: (a) loss by volatilization of the oil in the preliminary stage; and (b) loss by reduction or otherwise in the burning off of the residue. 5. Addition of an inert solvent to the oil before treatment (by wet oxidizing agents) does not diminish the efficiency of the oxidation, and has the advantage of preventing loss of sample occasioned by too violent a reaction. 6. Good results have been claimed for the Whittum, Waters, and Rothe methods when used on low sulfur fractions, but in view of the inaccuracies found in the Whittum and Rothe methods as applied to crude oils, there is also reason to doubt the accuracy ( 46 ) of such procedures with low sulfur oils. 7. As far as the present work would show, the only practical methods of sulfur analysis which can be viewed with entire con- fidence are the oxygen bomb and the sodium peroxide bomb methods, which can be applied to medium or high sulfur oils; and the lamp method, which is resorted to with very low sulfur oils. The Carius method and the combustion tube method, although accurate, are not held to be generally applicable. The Meulen reduction method is subject to the same objection. 8. It is shown that too'much barium chloride solution is fre- quently added, to obtain the best results. 9. Miscellaneous points are given in connection with oil sul- fur analysis. 10. Tables of data are given showing analyses of oils by diff- erent methods. IX BIBLIOGRAPHY ( 47 ) 1. Peckham, Proc. Am. Phil. Soc. £6 108 (1897). Petroleum and Its Products i 296. Kissling, Chem. Zeit., 2 £ 499 (1902). Richardson and V, 'all ace, Eng. Min. J . , 78 359 (1902). Rogers, Trans. Am. Inst, Min. Eng., 52 969 (1917). Schwartz and Nevitt, Petroleum 7 No s, 2 , 23, 96, 98, 100, 102 (1919). 2. Clarke, Data of Geochemistry 727 (1920). 3. Mabery, Proc. Am. Acad. Arts, Sc., 31 17, 43 (1694). Richardson and Wallace, J. Soc. Chem. Ind., 20 690 (1901). Waters, Sulphur in Petroleum Oils, U. S. Bureau of Standards Technologic Paper No. 177 , page 5 (1920). 4. Girard, Petroleum 2 No. 3 (1906). Scheibler, D er, 46 1815- 26 (1915); c. A., 1C 840 (1916); *' ters. Standards Paper- 177 P . E (1920) . 5. Mabery, J. Soc. Chem. Ind., 19 ECS (1900); Mabery and Cuayle, Proc. Am. Acad, Arts Sc., 41 89 (1905). 6. Second reference under 5; Ellerton, J, Soc. Cehm. Ind., 31 10-12 (1912); Waters, Standards Paper 177 p . 6. 7 . Waters, Standards Paper 177 p. 6. 8. Chem. Zeit., 21 203 (1897); Waters, Standards Paper 177 p. 6. S. Veith, Dingl . pol. J. , 277 £67 (1-890); Waters, Standards Paper 177 p. 6. 10. J. Ind. Eng. Chem., 9 479 (1917); Waters, Standards Paper 177 p - 6. 11. Waters, Standards Paper 177 p. 6. 12. Z. angew. Chem., ie 1529 (1905). ( 48 ) 13. Waters, Standards Paper 177 p . 5 , 14. Waters, Standards Paper 177 p. ie. 15. Pecueil, 41 112-120 (1922). 16. Waters, Standards Paper 177 p. 17. 17. Waters, Standards Paper 177 p. 10. 18. Chem. Analyst gg 21-2 (1917); C. A. 12 761 (1916). 19. J. Ind. Eng. Chen.., 9 479-61 (1917). 30. Chem. News £4 76 (1871) . 21. Waters, Standards Paper 177 p. 11 . 32. Am, Chem. J. 50 65-6 (1913) • 23. J. A. C. S. 31 62 (1909). 24. J. Soc. Chem. Ind. 21 neo ( 1902 ). 25. Allen and Robertson, Methods of Determining the Sulphur con- tent of Fuels, U. S. Bureau orf Mines Tech. Paper £6 p. 6 (1912). 26. Z. angew. Chem., p. 2&s (1696). 27. Chem. Zeit. 20 197 (1896). 28. J. Soc. Chem. Ind., is 950 (1899). 29. Proc. Va. Chemists Club 1 53 ; C. A. 3 116, 2627 (1909). 30. J. Soc. Chem. Ind. si 10-12 (1912); C. A. 6 1846 (1912). 31. J. Ind. Eng. Chem. 4 £42 (1912). 32. J. Inst. Pet. Tech.. 4 6 (1917). Chem. Age (London) s 684-5 (1920); n 0 « A . 1 5. 812 ( Ann. fals. 13 539-43 (1020); C. A. 15 £717 (1921) 35. J. Inst. Pet. Tech. 7 334-338 (1921); J. Soc. Chem. Ind. 40 878 A (1921) . 26. U. S. Pureau of Mines Bulletin No. 5 , Report of the Committee on Standardization of Petroleum Specifications, page 14 (1920). ( 49 ) 37. Zeit. angew. Chem. is 393 (less)* 38. Zeit. angew. Chem. 17 616-19 ( 1004 ); J. Soc. Chem. Ind. 23 563 (1904) . 39. Chem. Zeit, 34 417 (1910) . 40. Anales soc. espan. fis quim. is 175-91 (1921); C. A. 15 3957 (1921). 41. Am. Chem. J. 16 544-51 (1894). 42. J. A. C. S. 26 341-67 (1904). 43. Compt. Pend. 146 333 (1908); C. A. 2 1543 (1908). 44. Zeit. angew. Chem. 2£ 4^0 (isos). 45. Anleitung zur vereinf achten Elementaranalyse, p. 62 (1906). 46. Zeit. angew. Chem. 22 436-49; 493-97; 1361 (1909); C. A. 3 2281 (1909) . 47. Stahl u. Eisen 40 933-6 (1920); 0. A. 15 483 (1921). 48. Brennstoff Chem. 2 33 O -2 (1921) ; C. A. ie 613 (1922). 49. Oesterr. Z. f. Perg. u. Huttenwesen 22 ill (1891); Mitteilungen 9 107 (1691). 50. Chem. Zeit , J 6 1070 (1682); J. An. and Ap . Chem. e £85 (1692)* 5 I, J. An, . and Ap. Chem. 6 611 (1892) • 5 2. J. A. n 0 Vi/ « KJ * 20 630— S 7 ( 1896) . c 0 tj ♦ Chem. Zeit 768-9 (1699). 54. J. A. C. S • kl 1188-92 (1905). O' O • J. Soc. Chem. Ind. 24 1212-13 (1905). 56. Parr, The Analy sis of Fuels, Gas, Tar,- Water, and Lubricants page 172 (1922). 57. J. Chem. Soc. 72 595 (1697). 58. Chem* Centralbl* 70 493 (1899); J. Soc, Chem. Ind. 1056(1899) ■ ' (50) 59. J. A. C. S. gg 772-76 (1809). 60. J. A. C. S, 22 ©9-102 (1900). 6 ! . J. Soc. Chem. Ind. go 1241 ( 1001 ) . 62. Z. angew. Chem. 16 1080 (1908); J* Soc. Chem. Ind. 22 1309 (1903) . 63. The Chemical Analysis of Iron page 290 (1906). 64 . Petroleum Zeit. g 714 (1907). 65. J. A. C. S. 31 135 5—64 (1909). 66 . J. A. C. S. 32 977 (1910). 67. Gas World 66 40 ( 1917 ); C. A. 11 1GS4 (1917). • CO QO Giorn. chin, applicata 1 38-45 (1920); C. A. 14 25 45 (1920). 69. Gatterman, Practical Methods of Organic Chemistry, 3rd Eng- lish edition, page 66 (1914). 70. Blount, Analyst 43 £9 (1918); C. A. 12 1120 (1916) • 7 1. Hass. min.. 34 1-4 (1911); C. A.. 5 is7& (1911). 72. Chem. Zeit. 7 99, 130 (1893); J. Soc. Chem. Ind. 2 2 671 ( 189 S') * 73. Bull. Soc, Chim. Paris 21 336-41 (1899). 74. Min. and Sc. Press 81 569 (1900). 75. Ur S.. Geol .. Survey Prof. Paper 48 p. 174 (1906). 76. Chem. Zeit. 35 1119 (l©ll); Petroleum 7 237 (1911) ; Holde, Examination of Hydrocarbon Oils, page 41 ( 1015 ). 77. Analyst 44 95-6 (1919); C. A. 13 1386 (1919)* 73. Ann., chim. applicata e 77-62 (1917); C. A. 12 417 (1918) . 79. Ur S. Bureau of Mines Bulletin Ho. 5 p. 25 ( 1920 ). 00 o • J. Ind. Eng. Chem. 12 171-2 (1920). 8!. Parr, Analysis of F. G. T. W. L., page 171 (1922). 00 J. A. C. S. £4 140 ( 1902 ) and 26 ill ( 1004 ). ( 5 !) 82. J. Chem. Soc. 84 572 (lOOS) . 84. J. Ind. Eng. Chem. i 689 (1909). 85. Chem. Zeit. £4 1285-6 (1910); C. A. 5 1244 (1911). 86. Holde, Examination of Hydrocarbon Oils, page 87 (1915). 87. J. Ind. Eng. Chem. 9 479-81 (1917). 88. Chem. and Met. 2£ 49-53 (1921). 89. Parr, Analysis of F. C. T. W. L. , page 173 (1922). 90. J. Chem. Soc, 88 761 ( 1905 ); J. Soc. Chem. Ind. 24 1 O 86 (1905). 91. Trans. Chem. Soc. 101 ( 1912 ). 92. University of Illinois, unpublished. 98. Z. angew. Qnim. 1528-31 ( 1905 ). 94. Holde, Examination of Hydrocarbon Oils, page 4 c (1916). 95. A. S. _ . Ji: . 20 40 8. 96. Vandaveer, University of Illinois, unpublished. 97. Waters, Standards Paper 177 p p . ie-26 (1920). 98. University of Illinois, unpublished. 99. Fieldner, in Pureau of Mines Tech. Paper 26 , p. 10 (1912). ! 00 - Regester, J, Ind. Eng. Chem. e 812 ( 1914 ). 101. Methods of Determining the Sulphur Content of Fuels, Espec- ially Petroleum, U. S.. Bureau of Mines Technical Paper 26 (1912). 102. Treadwell-Hall, Analytical Chemistry, v. II p. 467 (1919). 103. Ibid pp. 464-470. 104. Allen and Johnson, J. A. C. S. 32 588-617 (1910). 105. Johnson and Adams, J. A. C. S. 829-45 (1911). 106. Parr, Analysis of F. G.. T. W. L. pp. 174-178 (1922). . - (52) 107. Christie and Bisson, J. Ind. Eng. Chem. 22 171-2 (l©20). 108. Pezzi, Giorn. Chim. ind. applicata 3 10-11 ( 1021 ); C. A. 15 8054 (1921). 109. Kolthoff, Eec. trav. Chim. 686-9 (1921); C. A. ^ e • 1056 (1922). 110. Bucherer, Z. anal. Chim. £9 297-802 ( 1920 ); C. A. 15 483 (1921). 111. Treadwell-Hall, Analytical Chemistry, v. I p. 21 ( 1916 ) 112. Hamor and Padgett, The Examination of Petroleum ( 1920 ).