Open Collections

• t i f f ft% THE DETERMINATION OF THE RATES OF ATTACK  OF THE HYDROXIDES OF FIRST GROUP  METALS ON 790 VYOOR GLASS by John R. Colclough Submitted in partial fulfillment of the requirements for the degree of Master of Arts. Department of Chemistry , University of British Columbia. April 1947. TABLE OF CONTENTS Pag© Acknowledgment . . . . i A b s t r a c t . . . . . i i P r e v i o u s Work . . . . . . . . . 1 A p p a r a t u s and M a t e r i a l s . . . . 5 T e m p e r a t u r e B a t h 5 R e a c t i o n l u b e s . . . . . 5 7 9 0 V y c o r G l a s s S a m p l e s 4 fydroxide s o l u t i o n s 5 O b s e r v a t i o n s 7 T r e a t m e n t o f R e s u l t s 22 S u g g e s t i o n s f o r F u t u r e Work 2 4 B i b l i o g r a p h y 27 ACKNOWLEDGEMENT Only through the kind and helpful suggestions offered by Dr. J . Q. Hpoley was this thesis completed. I wish to express my sincere thanks for his generous assistance. ABSTRACT 1. The effect of concentration of alkali on the i n i t i a l rate of attack of vitreous si l ica was measured at 80 C for LiOH, NaOH, KOH, RbOH and CsOH. Maximum rates of attaokwere observed at concentrations of 5 N for NaOH and ION for KOH. 2. The effect of temperature on rate of attack was measured in the range 60°0 to 90°0 for NaOH and from 70°0 to 90°0 for KOH. 5. Por a given concentration, the order of a decreasing attack is NaOH, KOH, LiOH, RbOH and CsOH. 4. With 9?% ethanol and 100$ methanol as solvent for NaOH and KOH, the rate is n i l . 5. With crystalline s i l ica in aqueous alkali , the rate of attack is about 1/8 of vitreous s i l i c a . ooOQOoo t * 1 PREVIOUS WORK Comparatively l i t t l e researching has been done in regard to the determination of rates of attack of hydroxides of f i r s t group metals on s i l i c a . Whenever a vit%Jous s i l ica sample was immersed in the hot hydroxide solution, crazing of the sample made results anomalous. Weighings after the immersion could not be regarded as the true loss of s i l ica resulting from chemical action, since this chipping and crack-ing phenomena allowed small particles of the s i l ida to drop out. Also, two samples whose surfaces had apparently been prepared in the same aay would craze to a greater or lesser extent, for no apparent reason. These phenomena were investigated by Hooley and. Wainwright ® with the following results: Their attempts to eliminate the crazing of the vitSfejous si l ica surface failed* but suggestions as to the reason why the glass crazed led to a satisfactory solution, of this problem. They noted that a freshly formed surface resulting from a fracture, did not craze . They also noted that a blown surface did not craze as much as a ground and polished sample. From this they postulated that the crazing was due to minute cracks or other weak areas in the surface, which acted as centres from which crazing spread SurFace over the CSBBOES of the glass. Containers of different materials were used to hold the hydroxide Into which the glass samples were immersed. Crazing varied consistently with the container used, which led them to further postulate that an electrical phenomena also played an important role in crazing, i t being felt that a cell was set up between the glass sample and the container. Measurements on the rate of attack with time indicated that i t was :. t lineafc for periods up to fifteen hours. 2 GENERAL PROCEDURE. The problem of crazing had to be solved in order that reproducible data could be obtained. In view of the work done by G„ Hooley and J . Wainwright i t was decided to study the effect of insulation of the glass sample from the container and also . . to try by some means to eliminate surface scratches on the s i l i c a . The satisfying solution to this problem led to the determination of the effect of concentratation and temperature on the rate of attack. 790 Vycor glass was used for the experiments and was ground as flat squares of approximately 40 mm. square and 6 mm. thick. It was decided to standardize the procedure with Sodium hydroxide* then to include Potassium, Lithium, Cesium and Rubidium hydroxide measurements. The effect of different solvents for the Sodium and Potassium hydroxides was investigated as well as the influence of products resulting from the reaction between the hydroxide and the s i l i c a . 5 APPARATUS AND MATERIALS. A constant temperature bath was made from an enamelled iron tube. It was insulated with three inches of gyprotf wool and set on a sheet of transite which rested on insulating bricks. The top was covered with a sheet of lead with eight holes of 2 " diameter to allow reaction tubes to f i t i n , and with six smaller holes through which a: stirrer, a ther^mometer, a thermostat and three heating elements passed. The level in the bath was maintained by means of an overflow lake which was connected to the bath by a siphon. The f irst thermoswitch was a b i -metallic strip but since this controlled to only — 1°C i t was set aside and a mercury switch with a relay was used instead. This gave constant temperatures with a t 0 , 1 ° 0 aocuracy and proved to be very satisfactory. The current controlled was"constant" and was fed to one element. The other two elements had;- a "variable" current by means, of a sliding re-sistance coi l , which allowed temperatures ranging from 60° C to 9 5 ° c to be obtained. Reaction tubes were made from pyrex tubing, pulled apart and blown, copper tubing with silver soldered base, sheet silver rolled and silver soldered, and iron boiler tubes with welded bases. They each stood about 1 2 " high and had a diameter of i f - 2 ° . They were supported by means of lugs attached to the base of the tubes approximately 2" from the bottom of the bath. The f u l l capacity of the tubes was 5 0 0 r . mis. which allowed the JOO mis. of solution used to be completely under the level of the water in the bath. The solution was maintained at 5 0 0 mis. by means of air condensers set into corks. Eventually rubber stoppers with the air condensers were used. Evaporation^ was negligible and could not be detected by means of the 5 0 0 mis. graduate used to measure the solutions. / 4 The glass samples used were cut from a large 790 Vycor plate, obtained from Corning Glass works. They measured approximately 40 mm. square and 6 mm. thick. The 8 urfaeeof the sample was standardized by grinding in a fine emery powder in a solution of glycerol and water. disc A round flat,cast-ironAwas spun and the grinding compound applied. The glass sample was held against the wheel until the surface became uniform •b and any chips on the edges were removed. To facil^ate the elimination of chips on the edges they were slightly bevelled. For the f irst couple of runs this completed the sample, but later i t was found necessary to immerse the sample in a heated lead bath (50°C) of $0% H.F. (60%) and $0% HgSO^ . (cone. ). A copper wire holder was made to support sixteen samples, so that this number could be etched at one time. This turned the surface from a seHlmVtransparent white sheen to a more transparent oily-l ike surface. Only about 15 minutes in the. bath was required, to do this. Thirty-two samples were ground and i t was possible to etch sixteen samples gt the same time. The small numbers scratched on were done so before the acid treatment. The samples after treatment were dried at room temperature- . and placed in a des|icjptor f i l l e d with Calcium chloride. They were left in the des i^c^ator for at least 2k hours before they were weighed. Heating test samples in an. oven pat a temperature of 150°C for 2-5 hours did not make any significant change in the weight of the sample, hence this procedure was neglected. The glass samples were suspended, by means of holders which were fashioned from the same material as the tube. The samples were allowed to hang freely in the tubes midway in the solution. They were placed in the solution after the solution had a chance to come to equilibrium and the time noted on the immersion of the f irst sample. The final time was noted on the removal of the f irst sample. They were usually allowed to remain in the tubes for 15 hours. The samples were, washed free of the alkali and weighed. They were next treated with the acid bath reedy for the new run. The area was calculated from time to time to allow for the s i l i c a removed by the acid treatment. The Sodium hydroxide and Potassium hydroxide, solutions were made up in volumes of 5000 mis. and stored at room temperature in 5000 mis. Florence flasks. They were, stoppered with corks. Eaoh solution was standardized against HC1 using phenolpthalein as indicator. A fresh solution was used on each run except where It is specified differently. Water used was disti l led and no attempt was made to remove any carbonates or other impurities that were introduced by the NAOH and KOH pellets. The very concentrated solutions used were standardized by pouring when hot into graduate, then allowing to cool and noting volume, diluting with known volumes and thus dissolving the crystallized NaOH. The accuracy of these was checked by treating a known solution by same procedure. 6 The Cesium hydroxide was made from CsCl^I. It was heated until white, then concentrated HgSO^ added and again heated until fuming ' stopped. When dissolved i t gave a test for Cl hence i t was necessary to evaporate the Cs^ SO t^o dryness along with HNOj and to heat until no more brown fumes came off. Then concentrated H S^O^ was again added and the solution evaporated to dryness. This CsgSO^ was dissolved and diluted. Solid BaCOH S^H^O was £j^U : added until no further precipitation [and the solution filtered. It was evaporated in a silver tube, the solution dripping into the tube at about the rate of evaporation. Air, passed through soda lime, and bubbled through three normal NaOH and saturated Ba(OH^solutio$ in order to remove the COgwas kept over the evaporating solution. The final con-centration was &.8Q6 N and the volume approximately 100 mis. This CsOH solution was filtered through a Pt. Munro crucible using suction. The normality was determined by pippetting 1 ml. from the hot (80°0) solution and standardized with HOI, After a run a new normality was obtained by adding a known volume of water to the solution. However a small known volume was removed and titrated, as a check on the normalities. No attempt was made to remove the products of reaction from the previous run. The Rubidium hydroxide was made from a RbNO^ , RbCl mixture. It WSB heated to a white salt with concentrated H^SQ^, This was dissolved in nper and filtered. The clear solution was evaporated to dryness. Its solution gave no test for the halides Cl, Br, or I. The Rb^ SO^ was dissolved, diluted and heated. To i t was added hot Ba(OH) solution Z — _ JiJLt until no more precipitation was noted. ' The solution was decanted ^ through a Pt. Munro crucible using suction. The RbOH solution was • evaporated in like manner as the CsOH. Normalities were also calculated similarly. tubing 6 A of Solution (<I,/Pr*) Silver tithe flppdnxfus usee/ /or RbQH and (sOH sol'ns. tick cr It OBSERVATIONS. Since preparation and procedure for each run varied, these topics shall be discussed under each run heading. The conclusions conclusions arising from the experiment shall also be included. Run No.l (NaOH) T= 7 9 . 5 ° C Time- 1 5 hours N= 1 . 2 0 2 PURPOSEt To discover a means by which the irregular weight losses due to chipping of the glass sample, experienced in previous experiments could be overcome by use of insulation of the glass sample from the container. The glass samples were ground with fine emery powder. Rubber insulators were made for 2 silver, 2 iron and 2 glass tubes, so fashioned that the glass sample did not touch the container. One silver and one iron tube was used as a contfol, that i s , contact made with the glass sample through the holder to the container. All the samples were cracked and jwere chipped. There was no significant differences in weight losses between the insulated and non-insulated glass samples. Inspection with a high-powered microscope failed to bring forth any means of distinguishing the samples . Intention of the insulation was to eliminate any possible electri&al potential which might arise due to the cel l - l ike nature of the glass sample and tube. This supposed potential might be responsible for the crackingand chipping which occured at confined areas; the remaining glass surface remaintRg. semi-transparent as i t had been ground. However the insulation failed to produce the desired results and whereas i t did not by any means prove conclusively that no potential existed , this line of attack was abandoned in the light of the next run. 8 Run No. 2, (NaOH) T = 80°C Time = 15 hours N = 0.6084 PURPOSE: To discover the effect of etching the glass samples with Hydrofluoric acid before they atfe immersed. The samples were ground and etched in a mlxture of 50% RF(60%) and 50% H2S04 (cone) by weight. The reaction took place at room temperature for five hours. Insulation was similar to t r i a l No. 1 with 2 tubes u8ed as controls. The samples were not weighed. There was no cracking or any change by which a sample could be identified as having been treated with sodium hydroxide. As before, the sample afterward had an oily-like semi-transparent surface. Samples are apparently left with a uniform surface which resists chipping and cracking. Any strains in the glass surface put there through grinding would be eliminated by the action of the H 4 F 4 acid and since chipping ceased i t would appear that an area of strain^as responsible for the cracking,, <A a. Jfc^ Run No. 5 (Na6H) T t 79?8 0 Time s 15 hours N • .8808 PURPOSE: To check the results of previous run. Samples the same as run No. 2 w e r e placed in the acid bath for one hour at room temperature • Insulation was contained as above. Glass samples were weighed• There was neither any change in appearance of samples, nor in the weight loss between insulated and non-insulated samples. Agreement in weight losses for tubes of same construction was good. Iron gave the highest weight losses, (av. .0^4 mgs./ hr/ cm* ) with silver container next ( .045 mgs /hr/ cm2) and glass the lowest ( .Q29 mgs. /hr/ cm1). 9 It is interesting to note that previously, glass samples gave the highest values, when the samples were not treated with HF and the clipping observed. Differences in weight losses between the tube containers was felt to be due to interference of products arising from reaction between the alkali and the tubes themselves* However, neither the solutions from silver nor ifcon gave tests for those metals. Reason for low results in pyrex tubes was attributed to the amount of Silicates coming from the tube itself , and hence interfering with the silicates from the glass, sample* The effect of temperature as shown by later experiments gave rise to the fact that radiation losses from the tubes, (about 2/5 of tube above surfac© line of bath) may have effectually re-duced temperature within tube and thus affected loss^of s i l i c a . Run No. 4 , (NaOH) T • 82°0 Time « 15 hours N a . 8 8 7 0 PURPOSEt To determine rate of attack on crystalline quartz and to note effect of container on reaction rate. The glass samples were treated with acid for tfc hours at room temperature. They ware not ground since previous run. A clear crystalline quartz sample was. ground and was not treated with HF. The f insulation was discontinued and the glass tubes substituted with coper tubes. For Iron an average loss of . 0 5 6 gms /hr/ cm?" was noted, for silver a loss of ,052mgs / hr/ cm*" and for copper a loss of . 0 5 8 5 mgs/ hr/ cm2". The quartz sample gave a loss of . 0 0 7 mgs/ hr / cm 1. The copper tube gave a blue coloration to the alkali solution. \ 10 It appears that copper is unsatisfactory as a container* Loss is quite low but this is greater than would be expected even from the bluish coloration. some other factor must have been responsible for this low weight. Purecrystalline quartz has a very high resistance to alkali and is not effected at points of strain., since no chipping was observed. This however may take place i f a longer time than 15 hours had been chosen. Run No. 5 (NaOH) Temp 79.5 Time 15 hours N a .9568 ibihty FURgPOSE: To check reproduc*4*4*y. Samples were reground and placed in acid treatment for 4 hours. Two samples of thin Vycor ware were cut from the side of a Vycor dish and treated as above. Loss in weight in milligrams per hour per square centimeter for iron was . 042; for Ag was .057, and for copper was .0581. The blue coloration was s t i l l , present and no explanation.for the rise in copper values can be given. The Vycor dish losses compared well with the Vycor glass samples. This run shows that reproducibility is not as yet certain as far as different tubes is concerned. Run No. 6 (NaOH) T • 80°0 Time 15 hours N » 1.001 PURPOSEi To determine the resistance of acid treated samples to crazing. The samples were not given any acid treatment but used directly after Run No. 5* The same tubes as before were used. There was no chipping or checking. Very good agreement with the silver tubes which gave highest values. The copper tubes dropped slightly below the iron tubes. 11 Run No. 7* (NaOH) F = 80 .0*0 Time 15 hours N = 1 . 0 2 5 PURPOSE: To determine the resistance of acid treated samples to crazing. Same samples as Run No. 6 with no acid treatment. Same time, temperature and approximately the same normality of NaOH as Run No. 6. The results within experimental error were the same as Run No. 6. Run No. 8# (NaOH) T •• 80°0 Time 42.5 hours N * 1 . 0 2 5 PURPOSE; Same as No. 7 . Same conditions as Run No. 7 except time extended to 4 2 . 5 hours. The Sodium hydroxide solution was not changed from Run No. 7 but left in the reaction tubes. Samples were badly chipped and cracked except the three which were in the iron tubes. On weighing,these samples gave weight losses inagreement with weight losses for only 15 hours. The significance of this is somewhat lost in the light of later work done in regard to temperature control'-., hence i t is well not to congider this result toe closely. It does indicate however that interference of product during the 15 hour runs does not appreciably slow the r a t e of attack. The samples in handling were evidently scratched so that chipping i occured. It is interesting however to note that only the iron tubes did not chip the. glass samples. No explanation is apparent. In any case results indicate that the samples should be treated at least after every second exposure to the NaOH. Iron tubes appear to be the best of a l l tubes used. 12 Run No.9> (NaOH) T * 80 C Time - 16 N » 1.427 PURPOSE: To determine the effect o^uaing glass containers with acid treated sample. New pieces of Vycor were ground and polished with the fine white powder. They were treated with HF and H S^O^ , being left in the bath over-night. The samples after, this treatment were not uniform. Some had a roughness and a cloudiness, whereas others had the character-i s t i c §mooth>oil-like, semi-transparent surface. The normality was increased. The copper tubes were replaced with the former pyrex glass tubes. Iron again gave the highest values. Glass gave approximately half of the loss as shown by the iron tubes. Agreement was poor. It appears -the that f tglass surface may have an effect on the rate of attack. However no agreement could be made between apparent roughness of sample and the loss in weight that i t experienced. Run No. 10, (NaOH) T » 80.0°C Time » 15.5 * N = 1.582 PURPOSE: To determine the effect of f ire polishing the glass samples and to note the. effect of allowing a stream of Sodium hydroxide to pass Over samples. The rough samples from Run No. 9 were reground and giv$n acid treatment. Four of the eight samples were then fire-polished with a oxy-coal gas flame. A reservoir of NaOH was set up in order that a continuous flow (7.5 oc/min.) of solution would pass the glass sample. This was v'done to one iron and one glass tube, the solution coming from reservoir into iron tube, then passing into glass tube and so into the waste reservoir. 15 The sample loss ln glass tube was slightly greater than sample loss in iron tube. No cracking or chipping took place with the f i r e -polished samples, and no difference could be detected in their apparent weight losses. The weight loss was smaller in the iron tube through which the NaOH flowed than in the oontrol. The silver tubes, ( 2 f ire-polished and one acid sample) agreed quite well. Apparently the glass tube interferes with the sample loss because of i t being attacked by the NaOH. This would introduce by-products which would interfere, and also reduce normality of the solution , thus reducing the activity. As above experiment indicates, when solution is renewed continuously, the above factors cannot influence the rate of attack of the glass sample. Fire-polishing is effective as well as the H.j30^ and HF •'. treatment in the preventionvof chipping and cracking. Similar weight losses would indicate the roughness of the surface has l i t t l e effect. This would be explained by the fact that a layer of saturated film exists about the sample, in ; w h i c h the effective area is not that of the glass but is that of the bounory of the saturated f i lm. The temperature drop caused by the continuous addition of cold NaOH solution would explain the drop in the weight loss of the sample in the iron tube(flowing) from that of the control. Runs No. 11 and 1 2 . These runs were similar to run No. 10 and served only to check results. PURPOSE; To find the interference of added Sodium silicate to rate of attack. Runs No. 1 5 , 14, 1 5 . Samples were prepared as befpre. A solution of Sodium silicate was made up to a known concentration. Amounts of the ttluhor* were addecL'So that the tubes contained 5 lOmg and 5 0 mg of 14 Results showed a slight deorease ln the loss weight per sample, but agreement was poor. Temperature control was difficult and was not a constant factor. The manner of adding the Sodium silicate was not correct. The assumption had been made that the product arising from the reaction of NaOH on the Vycor glass would be satisfactorily duplicated by the straight addition of Sodium silicate. This is not necessarily correct. Runs 16, 1?, 18, 19 and 20. PURPOSEt To determine the effect of concentration on rate of attack at constant temperature. New samples were cut from a Vycor flat dish, thirty-two In a l l . They were prepared as described under apparatus and materials. The runs were.standardized on iron tubes and the concentrations on the NaOH varied from 0.1 N to 10 N. e Runs No. 16 and 17 were made at a temperature of 80 C. However, o the temperature dropped to a value approximately equal to 75 °» during the 15 hour run. Run No. 18 was continued for 45 hours and Run No. 19 at 15 hours in order to check effect of product on the rate. However, the b i -ai metallic thermo-switch f a i l e d to mintain a/steady current. Hence results were treated qualitatively only. An important outcome of these experiments was the increasing of the current through both the constant and the variable heaters, which ensured the maintainence of temperature. Also a mercury thermo-switch with i. 1 o relay was installed, which gave a constant current to _ T0~~ 0. 15 In the past fair agreement was obtained in the individual runs. However, agreement was poor from one run to another. Factors which may influence this arei Size of Samples, Size was not standard, a small sample in same amount of solution as a large sample would have less weight loss. That i s , there would be less product to interfere with further reaction and also a relative higher activity. However, this could not have been appreciable as borne out by experiments run at long time intervals. Surface Treatment of Glass. The glass samples were ground and bevelled. Corners were hard to bevel and sometimes chipping would occur. This would facilitate further chipping which may take place during handling. Since cracking does not occur on a chipped surface this would pass unnoticed and account for anomalous, weight losses. The surface as a whole was treated with HF and HjSO ,^ but this procedure was not standardized. Some samples became cloudy and rough to the touch while others retained their translucent surface. Temperature. The bi-metallic thermoswitch would not control, better than t 1°0 . Considerable trouble was experienced in main-taining the same temperature for each run. This would introduce a serious error and would satisfactorily account for previous differences. As previously mentioned the temperature for future rtims was controlled to - 0.1°C. The size of samples very closely equals one another and the surface treatment has been standardized. 16 With the procedure standardized the following runs were made.  Weight losseB are expressed in milligrams per square centimeter per hour. Normality _ " Hun NaOH _2I_ 22 _25_ 24 _25_ 26 _2Z_ 0.0901 .0165 .0217 .0084 .0074 •0059 .0042 .0002 <*89 0.1767 .0265 .0564 .0156 .0159 .0097 .0070 .005 0.477 .0444 .0654 .0256 .0265 .0170 .0119 .005 0.975 .0585 .0902 .0526 .0558 .0220 .0146 .QD62 1.415 .0706 .0104 .0595 .0597 .0254 .0185 .0075 1.949 .0714 .110 .0457 .0415 .0280 .0204 .0082 4.695 .0942 .162 .0598 .0574 .0407 .0271 .0115 9.280 .0897 0.157 .0557 .0517 .0568 .0255 .0085 Temp(!c) 85.5 90.8 79.7 79.7 75.1 70.1 60.0 Time (hr) 15 15 15 42 15 15 15. Note: Run 24 was left for 42 hours using the same solution asprevious runs and the same samples,(i.e, no acid treatment between run 25 and 24.) 17 Normality KOH .4864 1.286 5.018 4.265 5.7174 7.055 9.055 10.05 Temp (0) Time(hr) Normality NaOH 10.4 15.41 20.6 25.4 KOH 9.25 15.15 17.45 18.20 Temp fe) Time (hr) _2Z_ .0140 .0502 .0525 .0457 .0445 .0505 .0450 .0570 80.2 15 40 .0277 .0217 .0166 .0075 .0259 .0707 .0127 .0119 70.7 15 .0524 .0625 .0727 .0877 .102 .1185 .1160 .1590 90.4 15 41 .0614 .0492 .0411 .0182 .0548 .0414 0,0275 .0258 79.8 15 -5SL .0072 {mg*l<*%/br.) .0116 .0145 .0175 .0195 .0226 .0256 .0242 70.1 15 42 .1465 (mg*/un7h<:) .1215 .1082 .052 .1260 .0940 .0656 .0626 90.5 15. 18 Runs No. 45, 44 and 45, were made using Methanol and Ethanol the solvent, for the NaOH and KOH. However weight losses were n i l . Normality LiOH Temp. Loss. Vol. Run 5.096 80.2 .0516 500 46 2.o5o 80.5 .0276 500 48 2.050 80.0 .0258 500 49 1.015 80.0 .0182 500 50 0.555 80.0 .0145 500 51 Time five hours. Normality• NaOH Temp Loss Vol. Run 15.45 80.2 .0585 100 46 7.15 80.5 .0774 150 48 4.86 80.0 .069 185 49 5.009 80.0 .0510 500 50 0.900 80.0 .029 500 51 Time five hours RbOH Temp. Loss Vol. Run 6.079 80.2 .0204 100 46 5.21 80.5 .0210 122 48 5.089 80.0 .0175 212 49 2.048 80.0 .0147 500 50 0.976 80.0 .0087 500 51 Time five hours 19 Normality CsOH Temp, Loss. Vol. Run. 9.806 80.2 .0068 100 46 7.930 80.5 .0075 i4o 48 80.0 .0069 222 49 5.157 80.0 .00616 500 50 1.179 80.0 .OO65 500 51 Time five hours Viscosities in gentistokes were for NaOH and KOH. The correct viscometer was not available at time of these measurements and in its place a viscometer of larger capillary tubing was used. This allowed.a shorter time for the liquid to pass through the capillary tubing*, hence a larger experimnntal error. Normality NaOH 16.10 10,42 5.29 5.65 Temp (0) Time (sec) Vigoosity 69.2 67.7 5.790 80.0 51.8 2.905 91.8 40.2 2.255 70.1 57.7 2.112 80.7 30.5 1.709 91.8 25.5 1.417 68.7 19.1 1.070 79.9 16.9 0.949 92.2 15.1 0.848 1 70.4 15.8 O.887 80.6 14.5 O.805 91.4 15.1 0.755 20 Normality KOH Temp.( C) Time (sec) Viscosity (centistokes) 14.8 69.8 58.2 2.155 80.0 51.8 1.780 91.5 27.45 1.557 12.5 70.5 26.5 1.485 80.7 22.6 1.521 89.0 21.4 1.199 8.55 69.5 18.76 1.052 80.0 16.9 0.948 89.0 15.8 0.886 5.14 69.5 14.8 O.851 80.7 15.4 0.752 90.5 12.6 0.707 The following data was taken from Chemioal Society of Japan-Journal. 52 1951 p.646. Molality -6 KOH Molality J_ 27.06 50.96 20.02 — 67.5 24.105 27.15 18.94 48.8 20.75 19.21 16.25 28.7 18.95 18.02 12.996 14.1 16.95 15.66 10.04 6.7 14.56 8.87 7.056 5.0 11.14 4.57 4.057 1.57 8.099 2.07 2.05 0.875 5.996 1.51 1.022 0.76 4.08 0.917 0.502 0.757 2.05 0.727 0.099 0.792 1.025 0.680 0.552 0.681 0.252 0.726 Temperature of concentration celllfrcm which activity coefficient was calculated was 25 C 21 GRAPH 1 Data from Runs, 21, 22, 2?, 2 5 , 26, 2 7 , 4 0 , 41 and 4 2 . GRAPH 2 " d (i ^ 5 9 < 4 0 > 4 L a n d 4 2 . GRAPH 5 n n n 2 5 , 5 7 , 46, 48, 49, 5 0 and 5 1 . Graph 4 " n n 25 and 2 4 . GRAPH 5 n n n From graphs 1 and 2 . GRAPH 6 • n " From Rune 25 and 5 7 . Normality has been changed to molality and multiplied by the activity-coefficient as obtained from page 2 0 . GRAPH 7 Data from visco B ity measurements page 19 and 2 0 . o 22 Treatment of Results. When the Initial rate of attack is plotted against the concentrate-ion of the hydroxide, the general form of the curve rises rapidly to a maximum, then fal ls off slower in the case of Sodium hydroxide and much faster in the case of Potassium hydroxide. These maxima occur at about 5 and 10 normal respectively. The mechanism of a heterogeneous reaction of this typejis the diffusion of the reacting agent to the surface of the reactant, then the subsequent diffusion of the product or products away from the surface. It wi l l be chemically controlled i f this, is the slower process and diffusion controlled i f this in turn happens to be the slower prooess. Another important aspect of this heterogeneous reaction is the fact that surrounding the s i l i c a sample there will exist a layer or film of solution, which is saturated with the products of reaction. It is this film that controls the reaction since the solution must diffuse through i t to the surface of the glass and the products diffuse back out into the solution. That i s , the effective activity of the hydroxide will be i ts activity in this layer, ^actors which will influence the activity of the hydroxide in this layer are, (l) Temperature, (2) Viscosity of solution, (?) Viscosity of layer, (4) Solubility of the product or products, (5) Thickness of film, (6) Concentration of solution, (7) Amount of product in solution. The samples were immersed for 15 hours which would eliminate the need to consider the amount of product i n solution, since runs No. 25 and 24 indicate l i t t l e influence on the rate in this regard. Viscosity measurements on Sodium and Potassium hydroxide give a ft* k curve with an increasing slope. Since nrj = the effect of the viscosity of the solution on the rate vs. cone, would be to give a curve 25 with a decreasing slope, assuming that rate is directly proportional to concentration. Thus the maximum due to a falling off of rate with higher concentration is not explained by the viscosity of the solution. As yet, an unknown factor is the viscosity of the saturated layer. The viscosity of this layer wil l change with the viscosity of the solution, depending on the concentration of the hydroxide as well as the concentration of the product. However, this film will have a thickness which will also vary with the viscosity of the solution and the viscosity of the layer, becom-ing greater with increasing viscosity of solution and layer. The time taken for an OH ion will thus depend on its diffusion through the solution into the saturated layer and its travelling the thickness of the film to the surface of the sample. It wil l also be controlled by the speed with which the resulting product can escape. An equation which would correctly interpret the values of con-centration vs rate of attack must take the above facts into consideration. Since a l l the required information is not available at present i t is felt that any derived equation would be stiff with assumptions and limited in i ts application. The negative results obtained by the use of alcohol aa a solvent indicate two possibilities. One is that the hydroxyl ion is prevented from reacting with si l ica because of the alcohol and (two) is that the product formed is insoluble in alcohol, thereby putting a thin protective film around the Vycor which prevents further reaction. The effect of the positive ion is indicated by the set of curves obtained by plotting rate of attack vs concentration of a l l five metals. The hydroxides are stronger for the heavier metals but this does not seem to influence the order of the curves. 24 Sodium is highest with potassium next, then comes lithium which appears out of place. The rubidium and eeslum f a l l in line, with cesium giving a very flat and non-characteristic curve. No viscosities were taken on lithium, rubidium or cesium, but indications point to an ex-ceptionally high viscosity for cesium. Since the s i z e of the positive ion increases with, the heavier metal this would make its diffusion to f a l l and thus would correspondingly slow up the reaction. „ The lithium may f a l l into i ts place because of a hydrated positive ion , thereby i n -creasing its size and decreasing its ability to diffuse. It may also be notedtt that lithium in many respects is similar to a group II metal. This may have some heaping onlts displaced position . SUGGESTIONS FOR FUTURE WORK. Whereas the phenomena of crazing was overcome, the reason for i t was not determined. Indications point to a catalytic effect which may be caused by the distortion of surface molecules due to strain. (3) Many catalysts are believed taact as such because of their abnormal and misplaced structure lattice, the secondary forces thus created acting to speed up or hinder the reaction of the material i t is catalysing. The effect of the container cannot be neglected. For untreated glass samples, the pyrex container speeded up crazing. This would i n -dicate that the products of reaction speed up the effect of crazing. Just in what way this i s done remains to be solved. Also for long runs on treated glass, (several handlings had likely inflicted areas of strlee) the iron tube did not cause crazing as did the other tubes used. As a check however asid treated glass should be left in the various containers 25 for a long time in order to positively establish the fact that acid treatment will permanently prevent crazing. In order to check the existence of a saturated layer or film surrounding the glass sample, samples should be so ground that varying degrees of roughness are obtained before final treatment with acid. The rate of attack should be effectively the same since the. r e l a t i v e gize of the sample is that created by the film. This would probably mask any roughness. Another method would be to rotate sample ( a round cylindric sample would be required) at varying speeds - the loss in s i l i c a depending upon the speed of rotation. This same effect may be ob-tained by allowing solution to stream past sample fit a fairly rapid rate. If i t could be put under pressure this would help to remove products of reaction and prevent their forming a saturated layer about the sample. The last two suggestions<are given on the assumption that the rate of a attack is controlled by the saturated Ljrer and its thickness. A method for the sweeping of the sample with a brush or rubber wiper should be considered since i t would be more efficient in removing any products of reaction which may cling to the sample. Further work on the effect of product on the samples should be done. I£ was assumed that the product did not appreciably hinder the re-action for a period of 15 hours. Runs have shown that the addition of Sodium silicate does interfere. However these results could not be counted too seriously because of temperature controlling difficulties and also be-cause the simple addition of sodium silicate does not necessarily give like products as are formed by sodium hydroxide on the s i l i c a . A large volume of standard NaOH should be made up and half of this allowed to stand with the glass samples at the desired temperature necessary to react with a desired weight of s i l i c a . 26 The samples weighed before and after would give the exact amount of St Ojthat had gone into solution. With these solutions and original solutions a series of runs could be made which would determine the inter-ference of product. This procedure would ensure identical temperature conditions which in the past have been largely responsible for anomalous values. Gentle stirring of the solutions should be done since this would keep normality of solution immediately surrounding sample more con-stant. The diffusion of product through the solution should fee investi-gated. This may be done by allowing sample to rest on bottom of tube and to draw off small amounts of solution from the top and determine the amount of s i l i c a that has diffused up to the top surface of the glass. A more exact method than this may be thought necessary however. Determination of the value of the solubility of product in sodium hydroxide solution should be done. This may be accomplished by powdering Vycor and allowing i t to react with hydroxide solutions, of varying C o n e , for a time longenough to ensure a saturated condition. Then an analysis on the amount of si l ica could be made on the filtered solutions.Viscosities of these solutions could then be made which would aid in the derivations of a satisfactory equation. Some experiments may be done using a mixture of methanol and water, or ethanol and water. The influence of the alcohol may give some indication as to the medbanlsm of the reaction. Also a qualitative test in alcohol, voul^ indicate "the to find the solubility of sodium silicate A failure of alcoholic solutions of sodium and potassium hydroxide to react. The effect of group two metal hydroxides on si l ica may help to explain the misplaced lithium curve. BIBLIOGRAPHY (1) The determination of the rates of attack of Sodium and w Potassium hydroxides on 790 Vycor glass and Vitreous s i l i c a . J . fr. Hooley and J . W. Wainwright, University of British Columbia. M*>j B.fl. tU«U.. (2) A Treatise on Physical Chemistry, (vol. 2, page H.S.Taylor. (5) Text-book of Physical Chemistry. (page 1121) S. Glasstone.