Nov., 1946
CYCLING PROPERTIES AND OXYGEN PRODUCTION OF S O M E [CONTRIBUTION FROM THE
DEPARTMENT OF CHEMISTRY,
CHELATES
2263
UNIVERSITY OF CALIFORNIA]
The Oxygen-carrying Synthetic Chelate Compounds. 111. Cycling Properties and Oxygen Production' BY W. K. WILMARTH,'~ S. ARANOFF'~ AND M. CALVIN The previous papers of this series2 have discussed the formation of stable peroxides by the reaction of gaseous oxygen and certain metallo-organic chelates. These peroxides, upon heating or evacuation, decompose reversibly to form the original reactants. This reversible peroxide formation and decomposition can be carried out many times, in the more favorable cases, without apprcciable loss of oxygen-carrying capacity. However, upon continued cycling, one finds that the oxygen desorbed on each cycle very slowly decreases in amount, indicating a slow concurrent irreversible decomposition of the original chelate. The following study is concerned with the rate of, and the products formed by, this irreversible decomposition. The loss of oxygen-carrying capacity of a chelate was studied by the alternate absorption and desorption of oxygen on a single sample of the chelate. The rate a t which each of these two stages of the cycle proceeds depends upon the temperature and the partial pressure of the oxygen gas. The choice of these variables and the duration of each stage of the cycle was governed largely by two practical considerations. Firstly, we were concerned primarily with the balance of experimental conditions that would produce a maximum of pure desorbed oxygen in a given time with a minimurn of irreversible decomposition. Secondly, the testing apparatus was designed to give results that could be duplicated on a much larger scale with commercially available apparatus. Cycling studies, consisting of alternate oxygenation and deoxygenation with simultaneous measurement of oxygen capacity, were made on two active chelates, CoSaEn . and 3F-CoSaEn. The chelate, CoSaEn, was first studied because it was the first example of successful use of such a chelate peroxide for oxygen production. Later the chelate SF-CoSaEn was developed. Preliminary studies showed that it was less subject to irreversible oxidation than CoSaEn so it was also tested by continuous cycling. Several factors undoubtedly contribute to the deterioration of a chelate, However, the most important appears t o be an irreversible oxidation by molecular oxygen. A preliminary study by chemical analysis of the deteriorated CoSaEn was undertaken here and those data are also presented. (1) Done under a contract (OEhIsr-279) between the University of California and the National Defense Research Committee. ( l a ) Present address: Department of Chemistry, University of Southern Californi:i, Lus Angeles, California. ( I b ) Present address: Donner Laboratory, University of California, Berkeley, Calif. (2) Papws 1 and 11, Trrrs J O U R N A L , 68, 22,iJ, 2137 (1946).
Experimental Deterioration of CoSaEn.-Thc active form of CoSaEii can be prepared directly in water solution or can be formed by vacuum heating of a pyriditiate. The inaterial used in the particular run here reported was of tlie first type. I t was prepared 0 1 1 a relatively large scale, had a lower initial capacity and a greater ratc of irreversible decomposition than the active form prepared from the pyridinate on a laboratory scale. However, it nas d e sirable t o have the data ;is to cyclitig stability oii the conimercial product as this grade of niafcrial would have h ~ i i used at that time iii large scale units. The tests were made iii a doublc pas> heat iiiterchntigcr in which 860 g. of the pelletcd chclate \ r a y c o i ~ t ; i i i i~l l ~ ~ l 30 one-half inch copper tubes twciity iiiehcs i i i Iciigtli. The chelate, as received, would absorb oiily S3.55,of thc theoretical oxygen. The theoretical absorptioii for this peroxide is 4.92% by w i g h t of osygcn. Thc cycling t e s t s were performed in the follo\ving way: A stream of air (dew-poiiit, -40') :it 90 gage pressure was passed through the tubes coli pelleted chelate a t a rate of 65 standard cu. ft water was circulated around the tubes tluritig this pr to dissipate the heat of oxygenation. This first stage of the cycle continued for thirteen and one-third iiiitiutc~ and in this time a 2.5-foltl excess of osygrii had passe(; through the pelletrtl chc-latc,. The desorption stage of t h c cycle is effected by stopping tlic flow of high pressure air through the interchaugcr, release of the air prensurc i n tlic interchanger, and the passage of steam arouiicl the tul)cs containing the chelatc. The steam is acimittc.tl a t 1 1 1 ~t c ~ p of the vertical interchanger and thc oxygcii t1~~sorl)c.d b y the chelate sweeps the remaining air ahead of it a:itl out of tlic interchaiiger. This self-purgitig of the air rciiiaiiii~i~: i i i the interchanger is sufficiciit to produce ovcr 90',: uf t h e d c sorbed oxygen with a purity of 999;. If fiirtlier purity is required the original air can be rcinovcd by cvaeuatioii I,cfore the steaiii is admitted t o the iiiterchaiipcr. This tlcsorption of oxygen is eontiiiued a t atinosphcric prcsirirc for six and two-third miiiutcs, thus coinplctiiig the cyclLh. The times chosen abovc do not pcrniit complete osygenation or deoxygeiiatioi~t o take ])lace. Thc reactioii.; are interrupted in this fashioii because the completioii of either the oxygenation or deoxygenation is elfected very slowly. The unit, operating in this way, can produce more oxygen per unit time by more frequent cycliiig than could be produced by carrying the reactions t o conipletioii. This slowness experienced in the last 10-15% of the reaction is partially illherent in the chemical rcactiori and partially a problem of heat transfer iii tlie cycling unit. The unit operated with solenoid valves which were automatically coiitrollcd by an eltictrical timiiig unit.3 The experimental rcsults 3re showii in Fig. 1. The upper dotted line represents the total initial reversible oxygen capacity in pcr cciit. of tlie theoretical activity This value of 83.5% was obtaiiicd by measuriiig thc osygcii absorption of the original saiiiple \rhen it is allowcd to come t o equilibrium at these temperatures and 1)ressures of air. The 16.5';> of inactive inaterial is niade up of preparative impurities mid CoSaEli in the inactive crystal form. The reversible dcsorption of oxygen under the cycling cotiditions is given by the poiiits which are plotted against the nuniber of cyclcs. This is a iiieasurc of the oxygen produced per cycle as outlined abovc and is soinewhat less thaii the saturation value. I t was measured di-~
(3) This timing unit was conslructed fur us by the Arthur D. Little Co.
2264
W. K. WILMARTH, S. ARANOFFAND M. CALVIN
Vol. 68
rectly in an integrating flowmeter. This oxygen production fell to approximately 50% of theoretical after 300 cycles and the test was discontinued. However, the saturation value was then measured again and is given by the short dotted line a t 61% of theoretical as against the initial value of 83.5%.
procedure, that the rate would be directly comparable except that the oxygenation with air might be slightly faster because of the cooling effect of the continuous stream of air. The time cycle chosen was three and one-half minutes for absorption and six minutes for desorption. The air flow rate was 0.68 standard cu. ft./lb./min. (a 3.3-fold 1 excess). These times were used in testing both preparar: I tions for comparative purposes although this resulted in M 82% of saturation oxygenation and 92% of complete de9 4 oxygenation in Preparation I as compared with 94 and 90%, respectively, for Preparation 11. The sample being studied was placed in the small interchanger and allowed t o cycle continuously. It was again not convenient to measure the desorption volumes directly because of the small sample size. Instead the cycling was interrupted a t intkrvals of 100-200 cycles and the rate of oxygenation and total capacity was measured with pure oxygen as previously described. However, immediately before this absorption measurement the interchanger was attached t o a high vacuum line and pumped a t 100' for one hour. This drying was done because we had hoped to follow the weight changes during 0 100 200 300 the cycling. However, this vacuum heating could also Sumber of cycles. reactivate any loss of capacity due t o formation of the Fig. 1.-Deterioration of the parent compound: - - -, hydrated crystal form. Other tests lead us t o believe that initial saturation, 90 lb. air; -0-, 20 min. cycle, 13I/2 no appreciable hydration of this chelate would take place under testing conditions. min. absorption; final saturation. The data are given in Fig. 2. The per cent. of original Deterioration of JF-CoSaEn.-This chelate can be pre- activity is plotted against the number of cycles. It will be pared' as a hydrate, as a hydrated peroxide, as a piperi- noted that the points now represent saturation values and dinate, and as a non-solvated form. All of these except not, as in the tests on CoSaEn, the absorption during the the lion-solvated form can be activated by vacuum heat- actual test. The rate of oxygenation remained constant ing. However, the various solvated crystals produce throughout the tests so the actual absorption during cyactive chelate samples which have a somewhat different cling will be less than that indicated by approximately the total capacity and rate of Oxygenation. X-Ray studies percentages given above. Preparation I was cycled only indicate that all samples of the activated chelate possess with air with a dew-point of -50' but Preparation I1 was the same.crysta1 form. These differences in capacity pre- also tested with air with a dew-point of -10". I t is probsumably depend upon impurities or formation of some of able that the variations observed are not due to this the inactive 3F-CoSaEn during the activation process. variation in experimental conditions but rather the initial The difference in rates of oxygenation5 is probably caused purity of the sample. by the differences in the shapes of the active crystals. Preparation 1.-The active material in this preparation was obtained in the following way. The red crystals of the hydrated 3F-chelate arc moistened with ether in contact with air. A black peroxide is formed which upon vacuum heating loses weight corresponding to one molecule of water and one atom of oxygen per molecule of chelate. The resulting compound has a high rate of oxygenation and a capacity of 3.85% by weight (4.4% theoretical). Preparation 11.-This material is' prepared by vacuum heating of the piperidinate. Each molecule of chelate loses one molecule of piperidine. The resulting compound has the highest rate of oxygenation of any of the active chelates. The two preparations used had an initial capacity of 4.04 and 4.25% by weight. The apparatus consists of a 0.5-inch copper tube heat 500 1000 1500 interchanger containing approximately 10 g. of the pelCycles. leted chelate. Air flow, steam and water are controlled by solenoid valvcs. The cycle is again autoniatically reguFig. 2.-Deterioration of the 3-F compound: 3 , 3F lated by the t.iming a p p a r a t ~ s . ~The chelate is oxy- active from piperidinate, air dewpoint - 10"; 0, 3F active genated with air a t 20 lb./sq. in. gage pressure. Cooling, during this stage, is effected by circulating tap water from oxy-hydrate, air dew-point -50'; X, 3F active froin through the interchanger. The desorption stage is carricd piperidinate, air dew-point -50". out a t 100' and one atmosphcrct pressure. The time cyck Analysis of Deteriorated CoSaEn.-A sample of deteriowas determined by measuring the rate of absorption and desorption of the material in the interchanger under the rated material for chemical analysis was prepared by placexperimental conditiotis stated above. In these initial ing some active CoSaEn in an oven, exposed to the air, rate capacity tests the oxygenation is carried out with pure and maintained at a temperature of 110'. The reversible oxygen a t a pressure equal to its partial pressure in air a t oxygen capacity, which was originally 4.83% by weight, 20 Ib./sq. in. gage. This was done because of the greater was measured a t weekly intervals and fell steadily. The final value, after six weeks a t l l O o , was 0.38% by weight. ease of measurement with these rather small samples of chelates. We had previously shown, before adopting this The sample was also weighed a t weekly intervals and it was found to be continuously losing weight. The total weight loss over the six-week period amounted to 4.77%. (4) See Ref. 2, paper I. The details of the preparative work will Elementary Analysis.-The deteriorated material was appear later in il paper with I
