PARTIAL OXIDATION OF LIGHT HYDROCARBON GASES AT ATMOSPHERIC PRESSURE by LAWRENCE BERT HEIN A THESIS Submitted to the Graduate School of Michigan State College of Agriculture and Applied Science in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemical and Metallurgical Engineering 1946 ProQuest Number: 10008475 All rights reserved INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted. In the unlikely event that the author did not send a complete m anuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion. uest ProQuest 10008475 Published by ProQuest LLC (2016). Copyright of the Dissertation is held by the Author. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code Microform Edition © ProQuest LLC. ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, Ml 4 8 1 0 6 - 1346 ACKNOWLEDGMENT The writer wishes to express his sincere appreciation to Dr. C. C. DeWitt for his aid and guidance throughout this investigation; to the Engineering Experiment Station, Project No. 89, for equipment and supplies; and to the General Education Board for the Fellowship which made this work possible. CONTENTS Page Introduction...................................... 1 Illustration............... 15 Equipment......................................... 17 Procedure................. 25 Data and Results........................... 36 Discussion........................................ 52 Summary...................................... 60 Literature Cited.................................. 61 1 INTRODUCTION For the last thirty years the idea of obtaining useful chemicals from the abundant naturally occurring hydrocarbons by partial oxidation has occupied a high place in the research laboratories of the world• At the present time it is still impossible to effectively stop the oxidation of all of a hydrocarbon at a stable intermediate stage. This is because the partially oxidized carbon atom is more easily oxidized to comple­ tion than is the carbon atom, which is bonded only to hydrogen and carbon. The work to date includes reports on the flow rate of the gases, the gas composition, methods of heat removal, temperature at which the gases can be made to react, pressures of the reacting gases, introduction of diluents and gaseous catalysts, and the effect of solid catalysts. Most of the effort has been directed along the lines of catalyst improvement. It is at least conceivable that some material would selectively adsorb hydrocarbon and oxygen and at the same time repel the aldehyde formed, thereby freeing this useful product to the gas stream flowing over the solid surface. Such a catalyst giving a high reaction rate would be ideal. In spite of the apparent fed.lure of research to provide high yields a number of commer­ cial partial oxidation units are in operation t o d a y 3 The previous research on this subject can be z divided into two large groups: one was carried out in the neighborhood of one atmosphere total pressure; the other at elevated pressures up to several hundred atmospheres. Another classification might be made on the basis of static or dynamic experimental conditions. The bulk of the material published is in reference to new catalysts, and has been done on gases which have not been too carefully purified. Newitt and Haffner^have shown the following effects of increasing the pressure in a static system of methane and oxygen. 1. Increase the rate of reaction or lower the temperature required for a certain rate. 3. Increase the quantity of methanol and formalde­ hyde in the product. 5. Increase the ratio of alcohol to aldehyde in the product. 5 Pichler and Reder found that in the dynamic system at 100 atmospheres, higher temperatures were needed than in the static system, but better yields of useful product were obtained. By lowering the incoming oxygen to 0.6 per cent of the gases, a 60 a per cent yield is reported. Newitt and Schmidt found that higher pressures favored the survival of 7 the higher alcohols. Newitt attributes to pressure the following influences on reaction: 5 1* Increases collision rate. (Permits use of lower temperature) 2. Permits survival of intermediate compounds. 3. In chain reactions it deactivates carriers, thus terminating the chain. 4. May cause alteration in the electrostatic field, so as to induce an increased polarity. 5. Le Chatlier*s Principle. Since the production of formaldehyde is the only case in which no contraction of volume occurs one would expect at least a tendency for low pressures to give the highest ratio of formaldehyde to other products. This is verified by a number of investigators.4 ’7 *1 6 *22 7 Newitt further shows that with methane the diluents nitrogen, water vapor, and carbon dioxide are detri­ mental to the survival of both methyl alcohol and form­ aldehyde at 50 atmospheres pressure. He has also shown that pressure favors the oxidation of the middle carbon atom of propane. This is attributed to a change of polarity of the hydrocarbon molecule, and an alter­ ation of the vibration period of the C-C or the C-H linkages. Pease Q has shown that with propane and oxygen each at 300 mm. pressure and 280°C. the addition of 20 mm. acetaldehyde caused explosion. Without the addition of the acetaldehyde an induction period of 39 minutes was required followed by a 5 minute reaction 4 time. Aldehyde did not eliminate the induction period except in those cases where explosion occurred* Coating the wall surface with potassium chloride slowed down all reactions* Pidgeon and Egerton Q agree with many others that these are chain reactions, that the products of the later stages influence the initial stages, and that the chains are initiated and termi­ nated on the surface, but propagated in the gas stream* Bone and Hiir**° report the explosion of ethane-oxygen mixtures when 1 per cent acetaldehyde is added. Bone and G a r d n e r ^ show that induction periods can be shortened with the aid of 1 to £ per cent alcohol, aldehyde vapors, or nitrogen dioxide. They also pro­ longed the induction period and slowed down the reaction rate by increasing the surface-volume ratio. IS Norrish believes that the induction period of methane-oxygen mixtures represents the interval during which an equilibrium quantity of formaldehyde is built up at the surface. Aldehyde is always essential for the occurrence of chain reactions. 13 Harris and EgertonA have shown that coating the walls with potassium chloride, sodium chloride and ferric oxide slow down the reactions and increase the quantity of carbon dioxide formed. This is overcome to some extent in static methods by the constantly changing composition of the gas. Formaldehyde increased the induction period, acetaldehyde and propylene 5 decreased it, and methyl alcohol had no effect, Egloff, Nordman and Van Arsdell^ report that heat­ ing the reaction mixture slowly gives oxidation at lower temperatures. They also indicate that methyl alcohol, ethyl alcohol, formaldehyde, acetaldehyde, and water vapor shortened the induction period; while iodine, hromine, and lead tetraethyl inhibit the reac­ tion. Boomer and T h o m a s ^ used small electrolytic copper blocks as a catalyst with dry natural gas containing 3j per cent ethane. Using dry air as the oxidizing agent, yields of 1§ to 2 per cent of the natural gas introduced were obtained at a pressure of 185 atmospheres. used. Temperatures of 525 to 350°C. were Pressure and reaction time did not have much effect on yields in the pressure range of 147 to 199 5 atmospheres. Pichler and Reder noted a sudden pressure increase from 160 atmospheres to 550 atmos­ pheres when methane-air mixtures were heated in a bomb to 350°C. Newitt and Thornes were able to convert to useful products, 13 to 18 per cent of the propane burned. These results were obtained at 274°C. and 400°G. at reduced pressures. They explain that the cause of the cool flame inflammation is a critical high concentration of higher aldehydes; during the passage of the flame the aldehydes are further oxi­ dized to aldehyde peroxides and finally to formalde­ hyde . 6 Another explanation is that the cool flame is formed when peroxides and aldehydes reach a critical high partial pressure* Newitt and Schmidt were able to obtain somewhat better yields of useful products when propane was burned in cool flame combustion than when burned by slow combustion. The following comparison is also made for the first three members of the hydro­ carbon series. Hydrocarbon Reaction Temperature CH^ C2H 6 °3H8 380°C. 305°C. 275°C. 40 64 Total condensible products, as per cent of the carbon of the hydrocarbon burned 7 Wiezevich and Frolieh^® have found air just as effective as oxygen on natural gas if the pressures are increased to correct for the diluting effect of the nitrogen. These reactions were carried out under a pressure of 135 atmospheres. The flow rate affected reaction only to the extent of ability to heat the gases sufficiently. They obtained 70 cc. of useful product from one cubic meter of gas. Their study included the gases up to and including heptane. 17 Smith and Milner, using only nitrogen dioxide as the oxidizing agent, have obtained 3.4 per cent of the methane decomposed, of formaldehyde. materials. (3£.5 per cent) in the form An analysis was not made for other Z to 13 seconds contact time, and tempera­ tures in the range of 440 to 680°C. were used. 7 Bibb, 18 using natural gas containing 16.55 per cent ethane, was able to obtain 19.8 pounds of 40 per cent formaldehyde solution from 1,000 cubic feet of natural gas. A chrome-nickel steel tube was used. The temper­ ature outside the tube is given as 735°C., but gas 19 temperatures were not taken. Matsui and Yasuda found that sodium chloride, potassium fluoride, phosphoric acid, and boric acid, adhering to the sides of a pyrex, quartz, porcelain, or copper tube, decreased the effectiveness of nitrogen dioxide as a catalyst with air. Uranium and beryllium oxides were only slightly better. Bromine was a good catalyst, but hydrogen chloride and sulfur dioxide were poor. Lewis 90 has shown that powdered glass and pumice increase the rate of peroxide formation while charcoal 21 catalytically destroys peroxides. Bone believes that the initial step in oxidation occurs by the hydroxylation process. One reason for this is that approximately 30 times more alcohol than aldehyde was formed in the work of Newitt and Haffnerf He took this to mean that only the first step in a hydroxylation process had taken place. Higher aldehyde forma­ tions are explained by a non-stop run to dihydroxy and aldehyde. Newitt and Bloch trace of peroxide. op were unable to find any In their copper-lined reaction vessel any oxygen concentration above 11.8 per cent resulted in very difficult temperature control. The 8 author had this same experience, but found that various catalysts changed the value of this maximum oxygen conps centration over a large range. Beal and Reniger&° indicate that Kaolin or Floridan are suitable cata­ lysts, but that they require frequent revivification either by superheated steam or recalcination. Fuller* s earth is also an excellent short-life catalyst which has a longer life when used with superheated steam. Egloff, Nordman and Van Arsdell2^1 report the following catalysts useful to various degrees. cupric chloride uranium oxide cobaltous oxide platinum wire manganese dioxide silver vanadium pentoxide copper ferric oxide lead eerie oxide selenium dioxide The extent of their value is not given, except an indication as to which products predominate. This will depend a great deal upon the history of the cata­ lyst and the conditions under which it is used. Lewis2** shows that the temperature of explosion for paraffin hydrocarbon-oxygen mixtures is raised by the metals platinum, copper, and lead, but not affected by tin, zinc, and aluminum. Iron gave the most violent explosion of any of the tests. Octane was used. A tube of small diameter required higher ignition 26 27 2R temperatures. James, * * in working above the 9 explosive limit at temperatures of 530 to 700°C., and contact times of J to 4 seconds was able to obtain 6 3/4 per cent of the natural gas treated going to formaldehyde. His catalysts were mixed oxides of molybdenum, chromium, tungsten, uranium, cobalt, and vanadium in various combinations. In any partial oxidation process carried out above 360°C. some cracking always occurs with the fragments formed combining with themselves and/or oxygen to give an oxidized product or polymer. Other portions of the hydrocarbon most probably combines through hydroxylation and peroxide formation. Thomas, Egloff, and 29 Morrell have arranged the metals in the various groups of the periodic table according to their cata­ lytic behavior. The catalytic actions of these metals have been divided into three reaction types: dehydrogenation, cracking, and complete decomposition. The results obtained from the catalytic activity of glass, porcelain or quartz was used as the arbitrary zero for evaluating the activities of the metallic catalysts. Taylor believes a catalyst surface to consist of unsaturated atoms, more or less loosely held to the underlying lattice structure with the loose atoms distributed haphazardly over the surface. These surface atoms have free valences which depend upon their relative positions to other metal atoms. These 10 are called active centers. One of the great difficulties of evaluating the action of a catalytic substance is attributed to minute quantities of impurities, which may serve as promoters. A promoter is a substance, the addition of which to a catalyst increases its activity. Alone the promoter does not exhibit catalytic activity. C.P. chemicals are no exception. Modern methods of spectroscopic analysis detect the presence of these impurities, but no known method of purification can remove them all. This factor may explain the dis­ crepancy of results obtained by investigators apparently using the same catalyst?,4:»'*-3 >'^ Balandin3'*'*52 and Burk33 agree that catalysis decreases the heat of activation by disrupting the molecules due to a strain action, caused by two or more parts of the molecule striving to reach two different surface atoms. When this strain is great enough, multiple adsorption occurs and the bond between the atoms is broken. A bond might be formed in the opposite case where two molecules occupy the same activated surface atom. The degree of fit between the interatomic distances of the molecules and those of the catalyst lattice determines whether or not positive activation will occur. Some of the major mechanisms of oxidation are as follows; 11 1. Formation of intermediate compounds of higher oxidizing power than the oxidizing substance itself, such as peroxides or ozone. High pressure liquid films may also be formed, E. Activation of the oxygen by exchange of the oxygen from the gas stream for the activated form, or simply liberation of oxygen from the carrier until it reaches a stable state of oxidation, 3. The formation of chain reactions. 4. Dehydrogenation. 5. Autoxidation. In all the above cases the catalyst is useful in that it lowers the temperature of oxidation and provides for increased selectivity. In many cases products are obtained which it is impossible to gain without the presence of the catalyst. 34 Ipatieff believes that alcohols generated by the reaction are further oxidized to aldehydes in a metal filled tube due to reoxidation of the metal. This can either be brought about by oxygen introduced or by the more active form generated by the decompo­ sition of the water produced. In the latter case the cycle consists of oxidizing the alcohol to aldehyde by the oxygen carrier, decomposition of the water formed in the first step, and finally regeneration of the metal oxide. The equations are as follows: 12 Metal + HgO = MeO +- Hg G H 3CHeOH + MeO = CHgCHO + HgO + Me The greatest danger in working with this type of reaction lies in the possibility of obtaining explosive mixtures. The exact values of the explosive range have been determined very carefully. These values are not affected materially by flow rate nor type of surface present. Coward and Jones report the following limits of inflammability. Hydrocarbon Limits in air, fo Limits in oxygen, fo ______________ Lower Higher Lower______ Higher methane 5.3 14 5.4* 59* ethane 3.2 12.5 4.1* 50* propane 2.4 9.5 - - -- The lower limit means that if the gas mixture contains less than this percentage of hydrocarbon explosion will not occur, likewise when the percentage compo­ sition of the hydrocarbon is above the upper limit it is safe regardless of temperature. For instance, the danger range for methane in air is in the interval of composition between 5.3 per cent and 14 per cent methane. Yeau reports the following explosive limits in air. *Values obtained in a closed apparatus 13 Limits in air, fo Lower Higher Hydrocarbonb methane 4.9 15.4 ethane 2.5 15.0 propane 2.2 7.3 Some discrepancy appears, however, this is not serious, because of the safety factor allowed in the work under discussion* The literature is incomplete on the slow com­ bustion of relatively pure hydrocarbons* As has been cited, natural gas, having a wide range of composition, is the favorite subject. In no case has a comparison been made as to reacting ability of the three lightest saturated hydrocarbons when partially oxidized under identical conditions. The number of variables which control yields are quite numerous. In the present work no attempt has been made to obtain the effects of gas flow rates, tube sizes, concentration of catalyst and various rates of heat removal. In the case of propane, the effect of increasing the reaction temperature from 470.to 650°C. is shown. Since the higher temperature gave the best results for the molybdenum oxide catalyst this was used on the lighter hydrocarbons, because they are more resistant to chemical reaction. Likewise, the effect of oxygen concentration in the propane-oxygen system was investi­ gated, as was the concentration of air in the natural gas-air mixtures* The upper temperature limit of 14 650°C. was observed because of equipment limitations. Although copper oxide, vanadium oxide, and molyb­ denum oxide have been used as catalysts for this process the idea of supporting them in very low concen­ trations on silica gel is new. The catalysts reported are normally pasted over the carrier to the exclusion of the latter*s surface. greatly This not only decreases the surface available, but also fails to make use of the carrier surface, which in this case was found to have some activity of its own. Most investigators have preferred to work with the cheap and easily obtained hydrocarbons found in the mixture of natural gas. "When this was done, large quantities of water vapor were introduced, which may have helped the reaction in some cases and retarded it in others. No reference was found in which powdered metals were used, probably chiefly because of the difficulty in holding the powder in place. was thought this type of catalyst, with its large surface, was at least of sufficient interest to warrant a trial. It PARTIAL OXIDATION UNIT 15 cc UJ O x cc U J Q CL < z cc x x o — UJ I LU £ < X UJ C OU J CD z g O cc U JC C OO o: ui z CO 2 L_ O O 00 z> o < n >o -c viro ^ in c o s c o o z UJ cc C D CC 1- ZD h- Ho LU o LU c c LU CO O x CD z LU z z CD o < z z z □ o X o h- o UJ X < cc >- CD z> m O CD o x UJ o X CO X X CC o h < < < LU o z z z o z o 2 o LU LU o UJ o LU o O CD X CD cc e? O X z >- Q X Q >* o >< X X X > X >- X 2 o X o X o X o _ CM fO if) (D S 00 OJ LINE OJ d cr> CD 3 00 [L _J *----- Isf o> a ID CD 1 ro UNIT LU CO Z g _ X fLxJ LU o z L U UJ h 1o o CL CO n OXIDATION Ixl >o r~ PARTIAL lu OF o DIAGRAM o 17 EQUIPMENT The equipment necessary to determine yields,under various conditions,of the partial oxidation products of hydrocarbons should measure gas flow, provide for a thorough mixing of the gases, control temperature, regulate pressure, and isolate the products for analysis. Part of the temperature control problem involves the rapid removal of heat from high-rate, exothermal reactions. This heat removal can be pro­ vided by conduction through the reaction tube wall, or by the use of high flow rates which result in convection of heat to the condenser. Pilm resistance decreases the value of the first possibility. The necessity of a proper balance among induction period, reaction rate and gas flow rate increases the diffi­ culty of the second. The convection method was used in this investigation and proved satisfactory at low oxygen concentrations for most of the catalysts. However, with a nickel catalyst, or with the higher oxygen concentrations, the flow rates used were too slow. Many attempts to solve this problem, when the reactions are fast and tend to go to completion, are reported in the literature, but the effort has not been successful. It was decided that rotameters would serve best for the measurement of gas flow rates. The oxygen rotameter (5) had a flow range of 75 to 5,000 cc. per 18 minute, the air rotameter (5), 75 to 1,300 cc. per minute, and the larger hydrocarbon meter (6) measured rates accurately from 500 to 6,000 cc* per minute* Tha Fischer Porter Company considers these instruments accurate to within 4 per cent* These flow meters were calibrated, at 70°F. and 739 mm., by the dis­ placement of water. Since the calibrations were made during warm weather it was necessary to place the two gas cylinders, and two surge tanks into a large container, which was held at 70°F. by means of circulating water. This same system was also used to control gas temperature for the earlier runs, during which time room temperature was well over 70°F. It seemed advisable to control gas tempera­ tures at the rotameters rather than rely on alinement charts to correct flow rates for temperature deviation. The accuracy of the latter method is questionable. Use of the manometer (9) and air pump (16) permitted close control of gas pressures behind the manometers to 739 mm. mercury. One of the first difficulties encountered was the fluctuation of gas pressure from both the hydro­ carbon and oxygen cylinders. These pressure vari­ ations made it impossible to hold the rotameter floats steady, and thereby made gas flow rate control inadequate. This was remedied by the insertion of the surge tanks (3) and (4). These tanks were identical. 19 They were made of two pieces of 6 inch pipe 4 feet long, welded shut on each end, drilled and tapped on the sides for quarter inch fittings. The diaphragm valves on the gas cylinders reduced the pressure in the surge tanks to 5 pounds per square inch gauge. Perfectly smooth flow was obtained through the rotameters. Hoke valves were placed at both inlet and outlet of the surge tanks. This allowed removal of the surge tanks from the system for evacuation to an absolute pressure of 7 mm. before the introduction of a different gas. An evacuation to this degree gave contamination of 7/739 or, roughly, 1 per cent. Since gas was passed through the tank rapidly for several minutes before use, the impurity was decreased to an inappreciable amount. Rather than rely on the gases mixing while flowing through the tubes, a mixing chamber (8 ) was placed in the line. This consisted of a 1 inch cylindrical glass tube 8 inches long in which the gases were made to reverse direction. This reversal of direction mixing is quite successful in commercial installations. From the mixing chamber these gases passed through 3 feet of inch copper tube before entering the reaction chamber. Further homogeneity was assured by the mixing in the packed reaction chamber. Reaction tubes made of two different materials 20 were used, The first copper tube burned through, so one made of nickel was inserted. The nickel was found to catalyze the reaction to completion; it was discarded. Another copper tube replaced it. Spot thermocouples were placed between the heater and tube wall near the mid point of the copper tube, and near the exit end. These thermocouples were used to detect excessive temperature, which might again cause the tube to melt. The reaction tube size was 1.050 inch outside diameter, 0.112 inch wall, and 28 inches long. 4 inches on each end of the tube was left bare; an electrical heater surrounded the remaining portion. Two thermocouples were placed within the tube. The iron-constantan couple, placed at the mid point of the tube and near its center, controlled the Brown Potentiometer Controller (12). Temperatures of the chrome 1-alum.el couple, placed 6 inches from the gas exit end of the tube and in its center, were read on the Foxboro Potentiometer (14). The two chromel- alumel couples outside the tube and the one inside were placed in the circuit of the Foxboro Potenti­ ometer by means of the switch station (15). Thermocouple leads were brought inside the tube through flanged copper fittings which were packed with asbestos and drawn tight to prevent leakage. The wires were protected from the copper tube and fittings with ceramic insulators. Originally, a 21 reactor tube designed to permit rapid cooling was placed between the heater (1 0 ) and condenser (1 1 ), Enough radiant energy was lost by this reactor to decrease the gas temperature well below that obtained in the heater. This may have proven satisfactory for conditions under which a sizeable induction period occurred, but for each condition a determination of the extent of reaction in the heater should be made. This reactor was discarded, and high gas velocities were depended upon to dissipate the heat. The heater caused more delay in the earlier work than any other item. After building one 500 watt unit it was found that a 1,000 watt element was needed. This element was made up of 45.2 feet of Number 16 Chromel A wire. The wire was arbored on a 3/16 inch welding rod, and wrapped over a length of 20 inches. After being set in alundum cement the cylinder was slipped over the copper reaction tube. The copper tube was wrapped with sheet asbestos to prevent a short circuit. Ends of the heater were supported on the copper tube by means of rings of fire clay and alundum cement. Outside the heater was placed suc­ cessively £ inch of ground asbestos, 3/4 inch 85 per cent magnesia pip© insulation, and J inch of ground asbestos. The ends of the unit were insulated by ground asbestos only. The whole unit was supported by two iron rings around the 85 per cent magnesia covering. 22 This assembly could be moved about with little danger of short circuit; it was sturdy enough to permit placing in a vertical position when catalyst was added or removed. The condenser (11) consisted of 18 feet of J inch copper tubing turned into a coil of 8 rounds. This copper tube was coiled on a lathe, and fastened into a firm unit of constant downward slope with the aid of 3 pairs of iron strips. The coil turns were held in place and spaced by bolts passing through the pairs of vertical iron strips on either side of the coil. The assembled coil was placed in the cooling tank; ice and water provided the refrigeration medium. At the bottom of the condenser a 100 cc. bottle served as a trap (15). From the trap, which held the condensate formed by chilling, the gases traveled to the absorbers. Two absorbers (16) of the fritted disc type, and one of the Milligan type, were used. The fritted disc absorbers are the more flexible and served exceedingly well at the higher gas flow rates. The Milligan absorber is more efficient at lower flow rates, but has the disadvantage of creating a large pressure drop. Additional absorbers created such a pressure drop that 20 or more inches mercury vacuum had to be maintained. This would have defeated the purpose of separation by the removal of aldehyde vapors. 23 Since the data was to be obtained at the one pressure, 739 mm, mercury, it was necessary to compen­ sate for changes in atmospheric pressure as well as back pressure occurring in the line behind the rotameters. An air pump (17) was placed behind the absorbers to provide the vacuum. Control was obtained by placing a T connection between the last absorber and the air pump. A valve was placed on the atmos­ pheric leg of this T to close off or admit air from the room. Opening the valve decreased the vacuum while closing increased it. Exact pressure at which reaction occurred was but slightly different from the pressure indicated by the manometer. It decreased uniformly over the entire length of the 28 inch reactor. There would be some difference in the absolute pressure in the empty tube and the packed tube. For the tube packed with silica gel this variance between ends is approximately 10 millimeters. Temperature control, for all the data taken at a temperature of 470°C. or below, was made with a Brown Potentiometer Controller. An iron-constantan thermocouple, and thermocouple extension leads furnished by the manufacturer were used. of this instrument was 0 to 500°C. The range For temperatures above 470°C. a Hoskins Resistance Type Temperature Regulator was used, which had a range of 0 to 800°C. The circuits were connected according to recommen- 24 dations of the manufacturers?^ In working with gases of this nature the possi­ bility of explosion always exists. To decrease the probability of explosion a J inch check valve (7) was placed just behind the oxygen rotameter. Although the chances of hydrocarbon backing into the oxygen surge tank was shown by calculation to be small, it did exist at very low flow rates. Both the mixing chamber and the condensate trap were designed to blow off and release pressure in case of explosion. Copper lines were used throughout to eliminate the danger of flying glass. Liberal use was made of Hoke valves to prevent gases from leaking through to some section of the equipment where they were not wanted. Two fuse boxes were placed in the electrical circuit to prevent damage to control instruments by a short circuit. 25 PROCEDURE Most of the runs were of one hour duration. In the earlier work, some were shortened to 30 minutes in order to determine whether o r not the catalyst was losing its activity. Since this was not the case, the time was increased successively to four hours. It was decided that one hour gave enough product for accurate determination and would permit a broader scope of investigation. One diffi­ culty encountered was passing the proper mixture of gases through the reactor without risking some deactivation of catalyst before the run was begun. The procedure followed was to have the heater on with the temperature coming up, then start the hydro­ carbon into the system, followed as quickly as possible by the oxygen. When the reaction temperature desired was 470°C., this was done at 425°C., while for a run of 650°C. the gases were introduced at 600°C. A maximum time of 10 minutes was then required to obtain the desired temperature. When this higher temperature was reached the condenser and absorber were placed Into the stream and the pressure regulated. In all cases where natural gas was used the oxygen flow was momentarily stopped until the vacuum from the air pump had been roughly set. This was necessary because of the low pressure at the natural gas source. On those runs made at 470°C. or below, temperature control ± 5°C. was possible, while at the higher 26 temperatures the deviations amounted t o l t l 0 oC. Since the thermocouples were not shielded from infra red radiation, the actual reaction temperature deviations were probably somewhat higher than those indicated by the potentiometers. There were a few runs in which the temperature control indicated above was impossible because of the highly exothermic nature of the reaction. This was true when nickel or copper oxide on silica gel was used as a catalyst and practically zero product was obtained; also when molybdenum oxide catalyst was used in mixtures containing a high oxygen content. In some of these cases just enough heat was given off to hold the temperature fairly constant above the controller temperature, while in others a very rapid temperature increase was observed. In the latter cases it was necessary to stop oxygen flow and allow the unit to cool. The following order was followed in stopping the run: 1. Turn off valve controlling oxygen flow. 2. Disconnect absorbers from condenser and each other. 3. Stop air pump. 4. Turn off hydrocarbon control valves. 5. Introduce nitrogen. After a 10 to 15 minute purge with nitrogen, oxygen 27 was introduced to burn off carbon deposits. Since all of this work was done above the explosive range, constant checks were made to be certain that air did not leak into the system, particularly in front of the reaction chamber. Precautions were also taken to keep rotameters vertical, prevent surges in the gas flow system and always to introduce hydrocarbon into a nitrogen filled reaction chamber. Surges in gas flow from the storage tank to the surge tank occurred with both methane and ethane unless heat was supplied to the diaphragm valve. The cooling effect of these gases was sufficient to cause the diaphragm to stick. Heat was supplied by a stream of air flowing from a small compressor. At the end of each run the condenser was washed with 4 or 5 25 cc. portions of distilled water. In cases where condensate was obtained it was weighed separately and combined with the washings of the con­ denser, unless it occurred in sufficient quantity to be analyzed alone. The solution from each absorber was analyzed separately in order to be certain that excessive quantities of product were not lost. In all cases where metal tubes and turnings were used activation was made in the same manner by alter­ nate oxidations and reductions. Three such cycles consisting of 10 minute oxidation by oxygen, 15 minute purge with nitrogen, and 10 minute reduction 28 toy hydrogen were carried out before the initial run. Following this first use of the particular catalyst a long oxidation of one hour or more burned off the carbon deposit. This was followed by two cycles as described above. Inspection of the catalyst showed the desired thoroughly roughened surface. This was simply an attempt to obtain the type of surface which B a l a n d i n ^ *^ believes to be active. Prolonged use embrittled slender turnings to such a degree that they crumbled; therefore even if a catalyst of this type proved useful it would probably need to be made up of larger pieces. The silica gel used was made by Davison Chemical Company of Baltimore, Maryland as a commercial catalyst support. It was 8-14 mesh. When used alone this gel was quickly transferred from its container to the catalyst tube in order to prevent it taking on exces­ sive water. Any water taken on, however, was driven off in the tube at 450 to 600°C. into a stream of nitrogen. the gel nor any of the Neither catalytic substances supported on it were reduced by hydrogen. Ordinary drying of silica gel (100-110°C.) reduces its moisture content to 5 to 8 per cent. Higher temperatures were used to remove the chemically held water almost to zero. 38 According to Ruff and Mautner, one gram of silica gel contains about 450 meters of surface, and the pore volume is square 40 to 50 per 29 cent of "tli© total volume. The 120 grams of catalyst placed in the tube filled it completely. Glass wool was used at the entrance and exit ends of the tube to prevent carryover. Carryover of catalyst in this equipment was highly improbable because of the con­ struction, nor were solids observed in the condensate. The condensate resulting from high temperature runs using molybdenum oxides on the gel did give a straw color solution, which was believed to be due to polymer. It was not a solid unless of colloidal dimensions. In order to compare the catalytic activity of copper turnings to copper oxide supported on silica gel, 40 grams of copper nitrate was deposited on 129 grams of the dry gel. A solution of the copper nitrate was made and added to the desiccant. This substance, after being heated to 800°:F. for 2 hours, appeared black, indicating that the nitrate had broken down to the oxide. 120 grams of this material was charged to the reaction tube. The catalyst caused the reactions to go all the way to carbon dioxide and water as was indicated by the quantity of heat evolved. Temperature control was impossible even though the rate of oxygen flow was reduced from 560 cc. per minute to 140 cc. per minute. Some useful product was detected as shown in the following section under runs 54 and 55. 30 The vanadium oxide catalyst was prepared by adding 170 grams of silica gel to a solution obtained from E grams of ammonium metavanadate (NH^VOg) dissolved in 150 grams of water. The catalyst was heated to 1,300°F. in a muffle furnace for several hours. This higher temperature was used, rather than the 750°F. recommended by Jaeger and Bertschf9 to increase the rate of decomposition. Since the catalyst was subsequently to be used at temperatures up to 1 ,S 00° F ., any detrimental effects of high temperature would be exhibited throughout its use. In the preparation a stream of oxygen was not passed over the salt to form only vanadium pentoxide. This higher oxide would surely be reduced to the various states of oxidation after only short use. The catalyst after use did appear to be a mixture of the black oxide "VgOg, the blue oxide V g0 and reddish-yellow V£°5* The molybdenum oxide catalyst was prepared in the same way as the vanadium catalyst, only the starting material was now molybdic acid, A thin slurry was made of this insoluble material and the desiccant added while rapidly stirring the mixture. Water did not appear in the bottom of the beaker in which the mixing was carried out, nor was there any evidence of powdered chemicals which had not been adsorbed. New molybdenum oxide catalysts, which were made up in as 31 nearly the same way as possible, were charged just before runs number 6 6 , 83, and 93. The only evidence of a loss of activity on the part of the catalyst was the failure to obtain checks on one high yield, which was obtained in the first of a series of runs under identical conditions. This may have been caused by carbon deposit even though long periods of oxidation were used. After use the molybdenum appeared to be in both the form of the colorless trioxide MoOg, and the blue oxide MoOg. One notable difference between the low tempera­ ture reactions and those carried out at higher temperature was the larger quantity of carbon deposited in the second case. Since all of the high temperature runs except numbers 106 and 107 were made with silica gel in the tube, the temperature condition may not have been entirely to blame. caused pyrolysis. Silica gel may have This condition was especially noticeable in those runs using propane. Oxidation for burning off the carbon was carried out for 2 to 3 hours at 550°C. in these cases. Because of the limitations of the copper tube, higher temperature oxidations could not be carried o u t , and some carbon always appeared on the discharged catalyst. A great deal of cracking occurred, especially with propane, and light oils were observed on the glass around the watergas interface of the absorbers. These same polymers 32 may have caused the straw color In the condensate as mentioned previously. The substances used as catalysts contained the following impurities as reported by the supplier. Molybdic Acid Cupric Nitrate C.P. Bakers .Analyzed Component C.P. Bakers Analyzed ^ Component ^ Cl 0.001 Insol. Matter 0.005 30( 0.01 Cl 0.001 PO 0.0005 SO 0.002 Insol. in NH40H 0.01 Alkalies as SO 4 0.02 H. Metals (as Pb) 0.001 Fe 0.0015 NH3 0.0005 NO, 0.003 Ammonium Sulfide Metals other than Fe (as N i ) 0.005 C.P. ammonium metavanadate obtained from Fisher Scientific Company was used. No analysis was given by the supplier. The sintered bronze catalyst tube used in runs 106 and 107 was obtained from the Moraine Products Division of General Motors Corporation. The tube was filled with bronze shot of mesh size minus 40, plus 60. The analysis was 7.95 per cent tin and 92.07 per cent copper. The propane obtained from Phillips Petroleum Company was of their Pure Grade, and was listed as 99 Mol per cent minimum purity. The methane and ethane of 99 and 95 per cent purity respectively were 53 obtained from the Matheson Company at East Rutherford, New Jersey. Consumers Power Company’s analysis showed the natural gas to be of the following composition. Natural Gas Component Methane ^ 82.3 Ethane 7.2 Propane and higher 0 Nitrogen 10.5 Walker^® suggests the following method for the estimation of formaldehyde in the presence of alkali sensitive products, and impurities such as eresol, phenol, and resinous materials. The method is that of Brochet and C a m b i e r ^ The basis of reaction is the liberation of hydrochloric acid when hydroxylamine hydrochloride reacts with formaldehyde to form formaldoxime as represented by: CHgO (aq. ) + NHgOH . HC1 = CHg : NOH + HgO 4- HC1 The procedure consisted of treating a 50 gram sample (20 grams when sample was condensate) with 10 milli­ liters of 10 per cent hydroxylamine hydrochloride solution. At the same time another 10 milliliters of hydroxylamine hydrochloride was pipetted out to serve as a blank. After 15 to 20 minutes the two were titrated with standardized tenth normal potassium hydroxide using bromophenol blue as an indicator. The end point is marked by a color change from yellow 34 to light purple* The difference between the milli­ liters of standardized base required for the sample and for the blank indicates to what extent hydrogen chloride has been liberated in the above reaction* This method is specific for formaldehyde determina­ tion, but part of the ketones formed might show up here* Ketone has not been detected by investigators at atmospheric pressure, and only to the extent of 5 to 5 per cent of the total product at elevated pressures? *^ Walker 42 recommends titration with carbonate free sodium hydroxide for the determination of formic acid. Bromothymol Blue is used as an indicator* This indi­ cator changes from yellow to blue at a pH of 6 to 7.5 so that acetic acid as well as formic would be neutral­ ized. The results were reported as formic acid. 20 milliliter portions of sample were titrated with a standardized tenth normal alkaline solution. The alkali was made carbonate free by using barium chloride after which atmospheric air was excluded. The determination of aldehydes above formaldehyde 43 was made according to Cumming, Hopper and Wheeler. Use was made of the addition reaction between aldehydes and sodium bisulphite. Since the quantity of formalde­ hyde was determined independently, the higher aldehydes could be obtained by difference. ?ftien higher aldehydes are absent this is a good method for the determination 35 of formaldehyde. The method consists of contacting 20 milliliters of aldehyde solution with an excess of sodium bisulphite, containing 12 grams of the salt per liter* The difference between the iodine titre of the sample, and that of the blank indicates the quantity of sodium bisulphite that has formed the addition product with aldehyde. The bisulphite solution should not be too strong because in more concentrated solutions the hydrogen iodide liberated reduces the sulfuric acid formed to give the reverse reaction* Tenth normal iodine and sodium thiosulphate solutions were used. Detailed procedure for the determination of alcohol is available in Walkerf4 The original work was done by Blank and Finkenbeiner. The method con­ sists of complete oxidation of alcohol, aldehyde and acid to carbon dioxide and water* about by an excess of chromic acid. This is brought The quantity of chromic acid used in the oxidation can be determined with the aid of a blank. The only time calculations are possible is in those cases where exact quantities of compounds are known. For example one quantity of chromic acid is required to oxidize a molecule of acetaldehyde, but a different amount is needed for a molecule of propanal. 36 DATA AJSTD RESULTS Table I is the result of 107 partial oxidation tests made under various conditions. analyses were made. Over 1,600 The average yield of formaldehyde as obtained by two different methods of analysis is given. The first 31 runs are not listed, because only traces of the desired product were obtained. The nickel tube used in the first 21 runs catalyzed the oxidation reaction to completion. The next 9 runs were necessary to locate the approximate conditions Tinder which detectable yields could be obtained. Table II shows the effects of catalyst and experimental conditions on the ratio of grams of formaldehyde to grams of formic acid surviving. Table III compares the yields of the various catalysts toward propaneoxygen mixtures treated in identical ways. 57 Table I Run No, Duration of Run, min. Hydrocarbon Plow Rate, cc./min. Table I Oxygen Flow Rate, cc./min, Reaction Temperature, °c. Propane Formaldehyde Surviving, mg,_____ - Oxygen Formic Acid Surviving, mg. Total Quantity of Useful Products Surviving, Per Cent of Oxygen Surviving as: Formaldehyde Formic Acid Total Useful Products ■ .. *ng.«________ Copper Turnings 52 60 2290 560 470 56 6 62 0.07 0.01 0.08 55 90 5000 715 470 40 11 51 0.05 0.02 0.05 56 90 5000 715 470 47 16 65 0.05 0.05 0.06 57 180 2290 560 470 56 15 71 0.04 0.02 0.06 58 90 2290 560 470 44 28 72 0.05 0.05 0.08 59 90 2290 560 470 54 14 48 0.05 0.02 0.05 40 50 2290 560 470 21 18 59 0.07 0.05 0.10 41 60 2290 560 470 58 24 82 0.07 0.04 0.11 Propane - Oxygen ;y Copper Tube 42 60 2290 560 470 51 10 61 0.06 0.02 0.08 45 60 2290 560 470 49 20 69 0.06 0.05 0.09 44 60 2290 560 470 48 18 66 0.06 0.05 0.09 45 60 2290 560 470 17 16 55 0.01 0.05 0.04 46 60 2290 560 470 18 5 25 0.02 0.01 0.05 Propane - Oxygen Lea Gel 47 60 2290 560 470 212 25 257 0.26 0.04 0.30 48 60 2290 560 470 156 5 139 0.17 0.00 0.17 49 60 2290 560 470 187 6 195 0.24 0.01 0.25 50a 60 2290 560 470 228 5 253 0.28 0.01 0.29 51a 60 2290 560 470 516 5 321 0.59 0.01 0.40 52a 60 2290 560 470 244 6 250 0.50 0.01 0.31 55 60 2290 560 470 245 19 262 0.51 0.05 0.34 a— Gases bubbled through water took on about 10 grams per one hour ruiu 38 Table I -Run No. Duration of Run, min. Hydrocarbon Oxygen Flow Flow Rate, Rate, cc./min. cc./min, Reaction Temperature, °c. Propane Formaldehyde Surviving, continued Formic Acid Surviving, m ? ______ - Oxygen Total Quantity of Useful Products Surviving, Per Cent of Oxygen Surviving as: Formaldehyde Formic Acid Total Useful Products P fc_________ Copper Oxide on Silica Gel 134 170 304 0.16 0.27 0.43 (470) 51 77 128 0.14 0.12 0.26 75 375 22 9 31 0.13 0.02 0.15 240 300 7 15 22 0.00 0.03 0.03 54 60 2290 560 470 55b 40 2290 560 56 60 3000 57 60 3000 Propane - Oxygen Vanadium Oxides on Silica Gel 58 60 2290 560 470 329 26 355 0.40 0.04 0.44 59 60 2290 560 470 285 28 313 0.36 0.05 0.41 60 60 2290 560 470 308 16 324 0.38 0.05 0.41 61 60 2290 560 470 303 16 319 0.37 0.03 0.40 62 60 2290 560 470 306 31 337 0.37 0.05 0.42 63° 60 2290 560 470 306 25 331 0.34 0.03 0.37 64° 60 2290 560 470 259 23 282 0.30 0.04 0.34 65° 60 2290 560 470 259 18 277 0.32 0.03 0.35 Propane - Oxygen Molybdenum Oxides on Silica Gel 66 60 2290 560 470 967 43 1010 1.19 0.07 1.26 67 60 2290 560 470 1182 48 1230 1.44 0.08 1.52 68 60 2290 560 470 952 49 1001 1.18 0.08 1.26 69 60 2290 560 470 951 43 994 1.18 0.07 1.25 U— Temperature control impossible due to highly exothermic nature of reaction. c— Gases bubbled through water took on about 10 grams per one hour run. 59 Table I -Run No, Duration of Run, min. Hydrocarbon Flow Rate, cc./min. Oxygen Flow Rate, cc./min. Reaction Temperature, Formaldehyde Surviving, °C.________ Ethane - conti nued Formic Acid Surviving, mg.__ Oxygen Total Quantity of Useful Products Surviving, Per Cent of Oxygen Surviving as: Formaldehyde Formic Acid Total Useful Products __________ Molybdenum Oxides on Silica del 70 60 2290 560 470 535 19 554 0.67 0.03 0.70 71 60 2290 560 470 466 IE 478 0.57 0.02 0.59 72 60 2290 560 570 1301 39 1340 1.60 0.06 1.66 73 60 2290 560 570 1280 30 1310 1.58 0.05 1.63 74 60 2290 560 650 1813 43 1856 2.25 0.07 2.32 75 60 2290 560 650 2098 39 2137 2.59 0.06 2.65 76 60 2290 560 650 2121 46 2167 2.65 0.08 2.73 Methane - Oxygen Molybdenum Oxides on Silica Gel 77 60 2290 560 650 605 13 618 0.75 0.02 0.77 78 60 2290 560 650 751 19 770 0.94 0.03 0.97 79 60 2290 560 650 597 17 614 0.65 0.03 0.68 Propane - Oxygen Molybdenum Oxides on Silica Gel 80 60 2290 560 650 2058 62 2120 2.55 0.10 2.63 81 60 2290 560 650 1285 42 1327 1.58 0.07 1.65 82 60 2290 560 650 1159 48 1207 1.43 0.08 1.51 83 60 2290 560 470 1319 50 1369 1.64 0.02 1.66 84 60 2290 560 650 1879 46 1925 2.30 0.08 2.38 85 60 2290 560 470 955 39 994 1.18 0.06 1.24 40 Table : Run No. Duration of Run, min. Hydrocarbon Flow Rate, cc./min. Oxygen Flow Rate, cc ./min Reaction Temperature, Formaldi Surviv: °C. Natural Gas m - . continued Formic Acid Surviving, fflg.*_____ Total Quantity of Useful Products Surviving, -P&* - Per Cent of Oxygen Surviving as: Formaldehyde Formic Acid Total Useful Products _________________________ Oxygen 86 60 2290 560 650 640 25 665 0.80 0.04 0.84 87 60 2290 560 650 646 25 671 0.80 0.04 0.84 88 60 2290 280 650 660 25 685 1.63 0.09 1.72 89 60 2290 280 650 675 22 697 1.65 0.07 1.72 90 60 2290 140 650 588 10 598 2.89 0.07 2.96 91 60 2290 140 650 474 14 488 2.34 0.10 2.44 92 60 2290 140 650 517 16 533 2.57 0.11 2.68 93 60 2290 140 650 470 22 492 2.32 0.15 2.47 94 60 2290 75 650 227 8 235 2.09 0.10 2.19 95 60 2290 75 650 218 9 227 2.00 0.11 2.11 96d 60 2290 840 650 603 10 613 0.49 0.01 0.50 Air rbdenum Oxides on Silica Gel Natural Gas - 97 60 2290 500 650 311 6 317 2.06 0.05 2.11 98 60 2290 500 650 218 6 224 1.43 0.05 1.48 99 60 2290 800 650 259 6 265 1.07 0.03 1.10 100 60 2290 800 650 351 7 358 1.41 0.04 1.45 101 30 2290 1000 650 181 5 186 1.23 0.02 1.25 102 60 2290 1000 650 448 15 463 1.48 0.06 1.54 103 60 2290 1000 650 446 9 455 1.47 0.04 1.51 104 60 2290 800 650 368 6 374 1.54 0.03 1.57 105 60 2290 500 650 v 293 5 298 1.99 0.04 2.03 d— Temperature control impossible due to highly exothermic nature of reaction. Highest temperature well over 800 0. 41 Run No. Duration of Run, min. Hydrocarbon Oxygen Flow Flow Rate, Rate, cc./min.cc./min. Reaction Temperature, Table I -- continued Formaldehyde Surviving, mg._____ Formic Acid Surviving, mg. Air Sintered Bronze Tube C. Natural Gas - - Total Quantity of Useful Products Surviving, m&.________ Per Cent of Oxygen Surviving as: Formaldehyde Formic Acid Total Useful Products 106 60 2290 500 650 69 5 74 0.48 0.04 0.52 107 60 2290 500 650 75 6 79 0.48 0.05 0.53 42 Table II Ratio of Formaldehyde to Formic Acid A B Effect of Catalyst Effect of Hydrocarbon Propane-Oxygen-470°C. Catalyst Oxygen-Molybdenum Oxides-650°C. Ratio Hydrocarbon Ratio Empty tube 5 Natural Gas 24 Copper oxide on Silica G-el Methane 38 1 Ethane 45 Propane 28 Vanadium oxides on Silica Gel 13 Molybdenum oxides on Silica Gel 21 Silica Gel alone 43 C E ] Effect of Temperature Effect of Oxygen Flow Rate Effect of Temperature Ethane-OxygenMolybdenum Oxides Temperature Natural Gas-OxygenMolybdenum 0xides-650oC. Ratio Flow Rate Ratio Propane-OxygenMolybdenum Oxides Temperature Ratio 470°C. 31 840 26 470°C. 21 570 36 560 24 650 28 650 45 280 27 140 35 75 26 43 Table III Average Percentage Yields for Various Catalysts a,390 cc«/fliin, Prouane-560 cc./min. 0xygen-470°C. Catalyst Yield, % Empty tube 0.07 Copper turnings 0.07 Silica Cel 0.39 Copper Oxides on Silica Gel 0.55 Vanadium Oxides on Silica Gel 0.39 Molybdenum Oxides on Silica Gel 1.17 700 600 o 3 O o cc o. 500 UJ CO 3 u_ o 400 a -I UJ >300 o I— OXYGEN AIR AS 0 XYGEN CONTE NT 200 0 200 OXYGEN FIG. 2 EFFECT OF 400 600 800 FLOW RATE (CC./M IN.) OXYGEN FLOW GRAM YIELD NATURAL GAS-MOLYBDENUM RATE ON OXIDE - 6 5 0 ° C. 3.0 o Z ^ > > EC h° 3 5 O 3 CC CO CL­ OXYGEN AIR AS 0 XYGEN CONTE NT 2.0 UJ CD > UJ X CO o 3 u. CO O o