THE VAPOR PRESSURES OF MOLYBDENUM OXIDES

AND TUNGSTEN OXIDES

DISSERTATION

Presented in Partial Fulfillment of the Requirements

for the Degree Doctor of Philosophy in the

Graduate School of The Ohio State

University

By

PAUL EDWARD BLACKBURN, B. A.

The Ohio State University 1954

Approved by* 1

ASMQMtfMlgMI

The writer wishes to express his appreciation for the helpful guidance and criticism of Professor Herrick

L. Johnston, under whose supervision this work was oarried out.

He is also most appreciative to Dr, Michael Hoch for their many useful discussions and for his sugges­ tions .

The author is very grateful to Mr. Janes Jones and Mr, L. £. Cox of the laboratory shop for their fine work on the apparatus used in this study. THE VAPOh PRESSURES OF MOLYBDENUM OXIDES

AND TUNGSTEN OXIDES

IflTBQPgCTIQM

Although earlier investigations on the vapor pressures of molybdenum trioxide and tungsten tri- oxide have been reported in the literature (1,2) there is some doubt about the accuracy of the data.

For this reason, and in order to extend the work to

the other oxides, a study of the solid-vapor equili­ brium of the molybdenum-oxygen and tungsten-oxygen systems vas undertaken. The study was carried out by measuring the vapor pressures of the oxides using the Knudsen rate of effusion method. THEORY

Sato of fiffmlga Hitfaad The Xnudaen (3) rate of effusion method consists of measuring the rate at which gas molecules escape through an orifice in the wall of a cell in which the gas la in equilibrium with a solid sr liquid. The equation for the pressure of a gas determined by the rate of effusion method is (4 )

p - » ^ * 5 “ , a ) where m is the rate of effusion in grams per square centimeter per second, R is the gas constant, T is the absolute temperature and M is the moleoular weight.

In order for this equation to hold true, two conditions must be met. The pressures measured must be low enough that the mean free path exceeds the di­ mensions of the orifice (5), and the area of the orifice must be small compared to the area of the solid to maintain conditions close to equilibrium. This equa­ tion also assumes a hole of Infinitely small wall thickness so that molecules passing through the surface pierced by the orifice will not enoounter the walls of the orifice (6). for this reason many investigators use knife edged holes.

If a knife edged orifice is not used, the amount gas passing through the orlflee Is diminished by a

factor K

1 “ 1 ♦ 0.5 1/a ^

in which 1. la equal to the depth of the hole and a

la the radius of the hole (7).

Thermodynamics

According to the Clasius-Clapeyron equation the

logarithm of the pressure of a gas in equilibrium with

a solid or liquid is related to the heat of vapori­

sation (AH) by the following equation,

lnp - ♦ C (3)

where T is the absolute temperature and C is a constant*

The value obtained for AH in this equation is the average

heat of vaporisation for the temperature range covered

by vapor pressure measurements*

It is usually not possible to extrapolate the ex­

perimental curve without error. However if heat capacity data are available, or if values may be reasonably

estimated, the vapor pressure curve may be extended

over the range for which the heat capacity data apply (8,9).

If the difference la heat capacities of the gas and the solid or liquid is expressed in an equation -4- auoh as

ACp - Aa ♦ AbT - ACT”2 U ) than the standard ohangs in free energy (&F°) nay be expressed as follows,

. -Ri,p . £5a - 4. 1.T - * ♦ ffcr-2* I (5) T T 2 2

The equation nay be rearranged to giTe

—RlaP ♦ AalaT + - 4 £ t ~2 , + I (6)

Since the left aide of equation (6) oan be calculated from experinental data, a least squares fit will giTe the best values for 4H 0 and I, the integration con­ stant. APPARATUS

All heating was done by induction, using a

General Electric or a Westlnghouse 20 kilowatt elec­ tronic heater.

Three furnaoea were uaed for the vapor pressure determinations. A furnace for temperatures below the optical range {Figure 1) was uaed for vapor pressure runs on molybdenum trioxide and molybdenum dioxide- molybdenum trioxide mixtures. The vapor pressure runs on molybdenum dioxide, molybdenum-nolybdenum dioxide mixture, tungsten trioxide and tungsten dioxide were made in either a high temperature pyrex glass furnace

(Figure 2) or in a metal furnace (Figure 3).

A Furnace for Temperatures Below 800°C.

This furnace, shown in Figure 1, consisted of a water cooled pyrex glass cell connected to the vacuum system by means of a flat ground glass joint sealed with a dry "0" ring. Stopcocks and a 20 liter Distilla­ tion Products oil diffusion pump were separated from the furnace by a trap refrigerated with liquid nitrogen. -5 The vaouum in the system was of the order of 1 x 10 mllllmetars of meroury or less, as measured with a cold cathode ionisation vaouum gauge placed between the furnace and the trap. -6-

C = D

^ 3 D KN'J DSFN WA 7 ER CUD CELL JACKET radiatio n SHIELD

THERMOCOUPLE

TABLE

TO VACUUM

STUPAKOFF SEALS

AN INDUCTION FURNACE FOR TEMPERATURES BELOW 8 0 0 ° C

FIGURE I -7-

PLANE LENS

RING

TO VACUUM

o r

o r KNUDSEN CELL

o r o

WATER JACKE T

— ii—'

PYREX GLASS EURNACE FIGURE 2 -8-

TT FLAT LENS

u u V S / A

TO VACUUM GAUGE

fla r KNUDSFN CELL LENS

O VACUUM

QUARTZ PI ATE

METAL INDUCt ILN f u r n a c e

FIGURE 3 -9-

Temperatures in this apparatus vara aaasurad with a .010 inch platinum to platinum-10$ rhodium thermo­ couple, lntroduead into the system by maana of Stupa- koff Beals. The eight inches of the thermocouple extending from the hot Junction were insulated with porcelain protection tubes. They were further encased in an Invar tube 3/32 inch in diameter, which was mounted vertically from a brass baae resting on the bottom of the vapor pressure furnace. Tne tip of the thermocouple was wrapped with a small piece of platinum to increase the thermal contact with the Knudeen cell.

A platinum cylinder the same size as the Knudeen cell was placed Just below the position occupied by the cell in order to reduce any losses of heat through the Invar tube.

The thermocouple leads were shielded from the hot junction to the potentiometer. In order to filter out radio frequency currents, a filter consisting of three

0.02 microfarad condensers was connected across the thermocouple leads and from each lead to ground. The cold junction was immersed in a bath of crushed ice and distilled water. A Leeds and Northrop White Double

Potentiometer with a capacity of one hundred thousand mlorovolts was used to measure the voltage of the thermocouple. - 1 0 -

Before installing tbs thermooouple in ths furnaes

it was carefully annealed by suspending it in a strain-

free fashion and passing a ourrent through it. The

temperature of the leads was held at 1A50°C. for one

hour.

It was then calibrated against a platinum to

platinum-10£ rhodium thermocouple which had been stan­

dardised using lead, oopper, aluminum and sine melting

point samples furnished by the Bureau of Standards.

The calibration of the thermooouple used in the vapor

pressure measurements against the standardised thermo­

couple consisted of joining the two couples at their

Junctions and placing them in a quarts tube frosen in

a cylinder of copper in a small resistance furnace.

Temperature was measured with the standardised thermo­

couple, and both the differences between the voltage of

the platinum leads of eaoh thermooouple and the differ­

ences between the platinum—2.0% rhodium leads were found

by means of a switchboard. When the differences ware added and plotted against the tenperature of the stan­

dardised thermocouple a calibration curve was obtained.

As a final check of the aecuracy of the thermocouple, it was calibrated under the conditions which existed during the runs. A tantalum bucket containing a thermo­ couple wall, which was lined with platinum to eliminate - 1 1 - bimetal effects, was filled with aluminum (National

Bureau of Standards sample) and placed In the furnace.

The system was evacuated to 1 x 10~^ millimeters of mercury. The aluminum was melted and an attempt was made to get a cooling curve through the freeaing point.

Due to the small masa of aluminum, however, the rate of cooling was so great that the plateau expected in the curve did not appear. More aluminum was added in the form of small chips and the system was again evacuated.

The temperature was Increased very slowly while the ch lpe were observed. The first indication of melting occurred at 660.0°C. as computed from the calibration curve de­

scribed above, and the whole sample was molten at

661.2°C. This compared very well with the melting point

of 659.7°C. quoted by the Bureau of Standards.

A High Temperature fjcrex Glass furnace

The pyrex glass furnace ahown in Figure 2 consisted

of a water Jacketed tube connected through a trap to a

10 liter Distillation Products vacuum pump backed by a

Velah Duoaeal fore pump. Pressure was measured with a

3C-24 vaouum gauge and was leas than 1 x 10“^ milli­ meters of mercury. The sample rested on a tripod of

•060 inch tungsten rode projecting from a base which was

held in place by a .060 inch molybdenum wire frame. The

top of the furnaoe was sealed to the furnace by a dry - 1 2 -

"0" ring between the flat ground glaaa aaals. Thia arrangement facilitated access to the furnace and per­ mitted easy removal of the top containing the optical flat window for calibration*

Temperatures were measured with a Leeds and Northrop optical pyrometer. The window was calibrated at fre­ quent intervals by plaolng it between the optical pyrometer and the standard lamp calibrated by the Bureau of Standards. The standard lamp was connected through a variable resistance to a direct current generator.

The current was measured using a .001 ohm standard re­ sistor the leads of which were connected to a potentio­ meter. ..ith this arrangement it was possible to keep the standard lamp constant to within about one degree centigrade during the calibration.

A High Temperature Brass furnace

The brass metal furnace (Figure 3} was water cooled. All openings were sealed with dry tt0a rings.

The work coil was Introduced Into the furnace through a 3/8" plate which bolted to a port in the side of the furnace. The work coll was sealed into the plate with

"0* rings separated by one inch glass plugs which served to insulate the coil from the furnaee. A detailed des­ cription of this has previously been published (10). By use of soft solder Joints Inside the furnace, work colls could be readily changed*

Observation ports were placed in both the side and the lid of the furnace so that temperatures could be read from either position. Optical flat windows were sealed to the furnace with dry "0" rings.

The furnaoe was connected to a 50 liter Distilla­ tion Products oil diffusion pump through a liquid air trap constructed of nonel with a capacity of about two liters.

The pressure in the system was of the order of

5 x 10*^ millimeters of mercury or better as measured with a 3 C- 2 4 vacuum gauge*

The sample was supported Inside the work coil on a tripod of *060 inch pointed tungsten rods projecting from a copper base resting on the floor of the furnace.

The windows were calibrated at frequent intervals in the same way as the window on the pyrex glass fur­ nace* knudeen Celia

All Knudsen cells were drawn from 0*010 inch platinum sheet furnished by the Ameriean Platinum

Works. Dimensions of the cells will be given with the vapor pressure data.

As previously mentioned the cell used in the - 1 4 - furnace equipped with thermocouple had a thermocouple

well eunk in tha bottom. The oelle uaed in the other

furnaoea were made with blackbody wells in the top of

the colls.

All Knudaen cells were made with the effusion ori­

fice in the side of the cell. It was found that the

top and bottom of the cells were cooler than the sides.

When effusion holes were placed in the top of the

cell the thermal gradient caused the material inside

the oell to condense around the orifice. Placing the hole on the side also prevented condensation on the

top window.

The orifice was not beveled to e knife edge as is the practice of some investigators. It was found that beveling the hole produced a ragged edge which

Introduced an uncertainty into the measurement of the area of the hole. The dimensions of the holes were measured with a traveling microscope.

The temperature measured through the orifice in the side of the cell were a closer approach to black- body conditions than was the blackbody well at the top.

Hence these temperatures were used to oalibrate the blackbody well. This calibration was completed before the Knudsen cells were used to make vapor pressure runs.

£aoh cell was placed in the brass metal furnace inside - 1 5 - e work coil. This work ooll had widely spaced turns

so the orifice In the side of the cell could be sighted through the side observation port In the furnace.

Headings were aade on the orifice and on the blackbody well. These readings were corrected for the window cali­ bration, and a curve was plotted. - 1 6 - SAMTfcES The molybdenum trioxide was obtalnad from the

J. T. Baker Chemical Company. Am analyaia furniahed by the oompany stated the purity as 99.9%, with the principal impurities of chloride 0.002%) nitrate 0.003%) phosphate 0.0005%) sulfate 0.02%) ammonium 0.001%) and heavy metals 0.005%.

The tungsten trioxide was made by dehydrating tungstlc acid at 900°C. to constant weight. The tungstlc acid lost 0.918 moles of water for each mole of tungsten trioxide. The Coleman S. Bell Company supplied the tungstlc acid which contained the follow­ ing impurities, a-ccording to the analysis furnished by the company: alkalies (as sulfates) 0.30%) arsenic

0.001%) lead 0.005%) molybdenum 0.005% and iron 0.005%.

The tungsten powder and molybdenum powder were obtained from the Calllte Tungsten Corporation. The tungsten was 99.9% pure according to the company. No analysis was available for the molybdenum powder.

Molybdenum dioxide was prepared by thoroughly mixing 0.138 moles of molybdenum trioxide with 0.065 moles of molybdenum powder and heating in an atmosphere of nitrogen at 750°C. for several hours. The excess molybdenum trioxide was removed by sublimation in a vacuum. The purple-brown molybdenum dioxide was - 1 7 - oxidised to constant weight. The sample was found to oontain 97.8 mole per cent molybdenum dioxide. An

X-ray powder pattern gave characteristic molybdenum dioxide lines (11).

Tungsten dioxide was prepared In the same fashion as molybdenum dioxide by heating 0.056 moles of tungs­ ten powder and 0.134 moles of tungsten trioxide at

1000°C. for seven hours. The X-ray pattern for the chocolate brown powder showed only tungsten dioxide (12) and tungsten trioxide lines. The excess trioxide was vaporized during the first few runs. - 1 6 -

rftQCIPTOSg The sample, in the form of a powder, was intro­

duced into a weighed Knudsen cell through the orifice,

and the cell was again weighed. The cell was placed

in the furnace which was then evaouated. The sample

was gradually brought to temperature during the initial

runs in order to degas it. These runs are not included

in the results since the degassing gives an erroneous

pressure. Once the sample was degassed, it was brought

to temperature as rapidly as possible to reduce the

correction for heating. After reaching the temperature

of the run, the sample was held at a temperature con­

stant to within two or three degrees at low tempera­

tures (600°C.) and to within ten degrees at higher

temperatures (1*>00°C.). The temperatures were read at

frequent intervals during the run. The time for each

temperature measurement was recorded using a stopwatch

for shorter runs and a wrlstwatch for longer iuns.

After the sample had cooled to room temperature it was removed from the furnace and weighed. - 1 9 -

CALC PLAT IQBS

The equation (l)

2BRT p * m M vas uaed to calculate the pressures of the gases. In this equation the rate of evaporation, m, was found by dividing the weight loss In grams by the effective time In seconds, and by the effective area in square centimeters. The weight loss of the sample was corrected for the weight loss of the container. In cases where this correction applied, it was found experimentally. The rate of evapo­ ration of the empty container was determined and the curve obtained was extrapolated to the temperature of the run. The effective time is obtained by a graphical integration which may be represented as follows:

(7)

-A T where A is the slope in the Clausius-Clapeyron equation

l o « p . 1 + B (8)

T is the temperature measured against time during the run, and T-t la the average temperature of the run.

This process yields a temperature (T#t) and a - 20- tlme (t eff ) for the run.

A rough calculation of tha pressure was made for each run by using the apparent tine and average tem­ perature of the run in equation (1), With several pressures calculated at different temperatures, a first estimated slope A 1 was obtained from equation (6 ).

The slope A* was then used in equation (7) to calculate the effective time for each run. The process was re­ peated using the effective time in place of the apparent time to get a second estimated slope A". The first estimated slope A 1 usually gave a t eff which was well within the experimental error for a run.

The effective area is the area of the orifice at room temperature multiplied by the Clausing factor K and by a thermal expansion factor.

"The factor K may be regarded as representing the ratio between the rate at which gas leaves the outlet of the tube and that at which gas strikes the inlet*. (13)

The thermal expansion of platinum determined by

Edwards, Speiser and Johnston (14) was used in computing the thermal expansion of the orifice at the temperature of the run

*£* - 7.543xl0" 6 (T-291) ♦ 2.362xl0~ 9 (T-291)2

The temperatures measured with a thermocouple were obtained from the calibration curve mentioned in the description of the apparatus uaed.

Temperatures measured with an optloal pyrometer were corrected for the difference between the blackbody

well and the temperature measured In the orifice.

Corrections were al8° applied for the window absorption

Both corrections were made from experimental curves

plotted from calibration data.

If the gas constant R is in ergs and thi molecular

weight M is in grams with the rate of evaporation in

grams/centimeter2/second the pressure will have units o

dynes/centimeter » The pressure may be obtained in

atmospheres by multiplying by a conversion constant, -7 / ? 9.8697x10 atmosphere s/dyne s/centime ter . - 2 2 -

A.gguxflgy qX tfao Pat* For the temperatures measured with a pyrometer the

following uncertainties apply.

The maximum uncertainty of the standard lamp was

5°C. The standard deviation In calibration of the lamp

was 1°C., while the standard deviation of the tempera­

ture of the run was 1°C. In addition to this, there was

an uncertainty of 2°C. due to calibration of the black—

body well against the orifice. These give a total

maximum uncertainty in the temperature of a run of 6,0°C. o There was an uncertainty of 0.05 C. for tempera­

tures measured with a thermocouple, which, combined

with the accuracy of X°C., gives a maximum uncertainty

of 1°C.

The error in the weight loss was less than 0.1 milligram, and the error in the time was less than a

se c ond.

The error in the area of the orifice was for the smallest orifice used. The error in the Clausing factor K was about 4$, within the precision of the measurements and the estimated accuracy of equation (2).

A combination of these uncertainties gives a total maximum uncertainty in the pressure of 10$ for runs where the temperature was measured with a thermocouple and of 22$ for runs where the temperature was measured with a pyrometer. - 2 3 - EXPERIMEKTAL RESULTS ttplrbflaaua Xrlaxldc The data are tabulated in Table I and Table II*

In Figure L,t the data are plotted with thoae of

Ueno (2) and Fleaer (1) .

All but the last two rune were made in the same cell, which waa 2 . 5 centimeters in diameter and 1 . 6 centimeters high. The first twelve runs were made with an orifice 0.838 millimeters in diameter. Runs

13 through 16 were made with an orifice 2,76 millimeters in diameter. A second Knudaen cell 1 centimeter in diameter and 2 centimeters high with an orifice 1.52 millimeters in diameter was used for the last two runs.

Although the thermocouple had been earefull7 calibrated, it seemed possible that an error might still exist in the temperature due to a thermal gradient. The thermocouple well was located in the center of the cell, which would be the coolest portion. In order to check this, the Knudaen cell with the smaller diameter was used for two runs and the pressures obtained were com­ pared with the other data. Both runs fell within the average deviation from the vapor pressure curve. Hence it may be assumed that the thermal gradient was not large enough to have a significant effect on the pressures measured. T a b U I

Dvaporation of Molybdenum Trioxide

Run * Temp Effec­ ✓it. 1 hernia! Effec­ *.ate of Press, - Log 1 ho. °K tive Loss expan­ tive Kv«p, atm. T ime Orans sion Area g / c m V x 107 o e c. Factor cn»2xio^ sec x 10

16 808 6242 0.002 31 1.0090 53.00 6.982 3.731 6.428 15 836 4284 0.00814 1.0096 53.03 3 5.3 3 19.47 5.711 10 850 9134 0.00168 1.C099 3.334 47.94 26.30 5.580 13 869 2014 0.01919 1.0103 53.07 179.5 99.47 5.002 11 880 7229 0.01374 1.0105 3.336 495.5 a 76.8 4.568 5 890 8725 0.01529 1.0107 3.337 456.7 256.2 4.591 9 898 1399 0.00242 1.0109 3.837 450.8 254.0 4.595 14 904 1009 0.04799 1.0110 53.11 395.5 50h. 2 4.296 7 928 1993 0.02678 1.0115 3.040 3490. 2 000. 3.699 ~ 0<7 6 930 1779 0.02725 1.0015 3.340 3989. 1 * 3.641 3 940 1327 0.03095 1.0118 3.841 o072 . 3501. 3.456 ^ ,» ‘ -3 “5 0 946 2163 u . « ~ J 1.0119 3.841 7262 . 4199. 3.377 2 954 1358 0.05758 1.0121 3.842 11040 6409 3.193 1 953 1187 0.C7817 1.0121 3.342 17140 9975 3.001 0 ^ / r 12 958 1051 ^ t 3 1.C121 3.342 12050 7006. 3.155

* nil runs were made in c cell 1,9 cn high ar.d 2.5 cm in diameter, i.uns 1 through 12 had a Clausing K. factor of 0.688 while the remaining runs had a Clausing K factor of 0 . 8 7 9 * Table II

He bdenum Trioxide

C, Kun Tenp -R log -2.07 log0T -2.95x1a-3 I ^ I O ' 33111 -I iiQ. 0K T

16 808 29.41 13.86 2.38 0 2 ^ *12.83 102.39 90.01

15 836 26.13 13.9 3 *■- * 47 0.36 ^ 9.47 99.40 89.93 10 850 25.53 13.96 2.51 0.2 6 + 8.31 97.32 89.01 13 869 22.89 14.01 2.56 0.24 + 6.07 95.66 89.59 11 380 2 0.90 14.03 2,60 0.24 * 4.04 94.41 90.37 5 890 21.01 14.06 2 L ^ 0.23 + 4.09 93.42 89.33 rv on 9 898 21.03 14.0S 2.65 - * *- J + 4.07 92.59 38.52 H 904 19.66 14.09 2.67 0.23 * 3.68 91.92 89.24 7 928 16.93 1 4 . H 2.74 n r> ■* - 0.16 39.59 89.75 6 930 16.6<_ 14.15 2.74 0.21 - 0.45 89.34 89.79 3 940 15.81 14.17 2.77 0.21 - 1.34 88.43 89.77 n c / c 8 946 ■J- ^ • *+ > 14.18 2.79 0.21 -11.73 87.35 89.53 O A_ 954 14.61 14.20 2.81 0.20 - 2.61 87.10 89.71 1 958 13.73 14.21 2.33 0.20 - 3.31 86.77 90.08 12 958 14 • 44 14.21 2.83 0.2 0 - 2.80 86.77 39.57 I LOG P (ATM.) 1.05 AO PESR F Mo03 OF PRESSURE VAPOR L 10 IUE 4 FIGURE UENO N E U O FEISER • 0 HS RESEARCH THIS

120

- 2 7 -

Calculation of the data yielded an equation for tha vapor pressure and the heat of sublimation for molyb­ denum trioxide. The vapor pressure of molybdenum trioxide may be expressed in logarithmic form by re­ arranging equation (6 ).

- 1 8 1 6 4 log p - J - 1,04 log T - 6.45x10 T

4 .0 2 x1 0 ^ " ---- 5— + 19.585 (1 0 )

The heat of sublimation was obtained by use of equation (6 ) in which the left hand side is designated as Ii

2 « —RlnP ♦ as InT + ^ T - ^ T “ 2 (11)

There was no data available on the heat capacity of the gas,so Kubaschewski and Evans' method for estimating heat capacities of multlatomlc gases was used. The heat capacity was "obtained by adding 6 calories/degree to four times the number of interatomic linkages In the molecule". (15) For molybdenum trioxide the heat capacity was then 18 calories/degree. Cosgrove and

Snyder's (16) heat capacity data were used for the solid,

Op - 20.07 ♦ 5.90xlO~3T - f:.68^ 1 ?.5. (12) - 2 8 -

The heat of sublimation of molybdenum trioxide was

83*111 calories by the sigma plot, equation (6 ) (Table II)

The constant I in this equation was 89*62 * 0,33 calories/degree. The heat of sublimation at 298.16°K. was calculated as 80,970 calories.

The vapor phase of molybdenum trioxide was molyb­ denum trloxide gas. This was established in the follow­ ing manner.

During the runs the vapor condensed as a yellow transparent film on the walls of the pyrex glass fur­ nace. After several hours in air or in a vacuum, the deposit turned deep blue. An X-ray powder pattern showed the deposit to be amorphous. Some of the blue deposit was heated in a vacuum to crystallize it, and another

X-ray pattern made. The lines obtained corresponded to those for molybdenum trloxide.

In order to determine whether dissociation takes place in the temperature range at which the molyb­ denum trloxide was run, a series of runs was made in which two or three grams of molybdenum trloxide were added to a weighed Knudsen cell. The cell was then rapidly heated in vacuum to between 1100°C. and 1200°C. vaporising the molybdenum trioxide. After the cell had cooled it was weighed again. It was assumed that if there were dissociation it would occur either In the solid state or in the gas before the molecules had left the cell. Sinoe, as will be shown, the only gaseous oxides of molybdenum are the trioxide and dioxide mole­ cules below 1500°C,, it follows that dissociation would result in either molybdenum or molybdenum dioxide. Both of these would condense inside the cell below 1200°C.

During these runs the temperature was increasing while the molybdenum trioxide vaporized. A rough esti­ mate of the molybdenum trioxide left in the cell above each temperature interval 1 0 0 ° was obtained by heating the cell containing the trioxide to a given temperature and then turning off the oscillator. The cell was weighed, more trloxide was added, and the cell was heated in a vacuum to the next higher temperature. In this way it was found that, under these conditions, out of two grams of trloxide 150 milligrams vaporized be­ fore the sample reached 1050°C. When 2.217 grams of trloxide were heated to about 1400°C. in 30 seconds, there were 5.82 milligrams of residue in the cell.

This failed to vaporize when the sample was reheated. o After deducting the 150 milligrams lost reaching 1050 C. we find that the 5.82 milligrams assumed to be molybdenum dioxide corresponds to a dissociation of the molybdenum trloxide of 0,32%. This value la at 1050° 1 100°C.

When the trloxide was heqted to temperatures below 1000°C. there va> no residue In the cell, within the

experimental weighing error of less then 0 , 1 milligram*

Measurements at higher temperatures were not possible due to the high rate of evaporation. This method is

very rough but it indicates that there ia little if any

dissociation of molybdenum trioxide at the temperatures

where the vapor pressure measurements were made.

Although the above measurements indicate that dis­

sociation of the trioxide is negligible at the highest

temperature at which vapor pressure measurements were

made (960°K) they also show that dissociation occurs to

a detedtable extent at 1370°K. It is of interest to

compare the extent of dissociation observed with that

calculated using thermodynamic data in the literature.

In order to make this comparison we proceed as follows.

The dissociation pressure of molybdenum trioxide

was computed at 1300°K and 1000°K. The free energy

equation,

g f o - -130,884 - 4.304TlnT ♦ 2.21xlO"3 T2

+ 67,57T (13)

derived by Thompson (17) from thermal and equilibrium

data for the reaction,

Mo ♦ O2 * MoOjj

was used as well as the free energies for the reaction computed from the free energy of formation at 2 9 8 .1 6 °K given by Brewer (18), from the thermal data of Cosgrove and Snyder (1 6 ), from the data for oxygen and molybdenum given by Kelley (19) and from the vapor pressure data determined In this research.

The dissociation pressure of oxygen over molybdenum trloxide according to the reaotion

MoO^(g) - Mo02 (S) + l/202 is atmospheres at 1300°K as calculated from the above data. The pressure of oxygen at 1000°K (40° above the highest temperature at which the vapor pressure of molybdenum trloxide was measured) is cal­ culated as 1 0 - W . l atmospheres. This latter pressure is too low to account for the degree of dissociation at

1370°K. In view of the disagreement between experiment and calculation at 1370°K it seems clear that the above calculation must also be inaccurate at 1000°K. A more accurate estimate of the dissociation pressure at 1000°K can be obtained as follows.

The dissociation pressure of oxygen at 1370°K found in this research for the reaction,

Mo03(g) ■ Mo02 (s) +1 /2 O2 may be computed If the oxygen pressure is assumed to be half the percentage of dissociation multiplied by the extrapolated pressure of molybdenum trloxide. From tble

dlseoeletlon pressure end the estimated heat of dissocla<

tlon the oxygen pressure can be calculated at 1 0 0 0 °K as desoribed below.

The molybdenum trloxide Tapor pressure curve was extrapolated to the melting point (1 0 6 8 .4 °*) (16) and a linear equation was derived, In which the slope was decreased by 12,540 calories for Cosgrove and

Snyder's (1 6 ) heat of fusion. The equation for the

pressure above the melting point Is then

-1A12 3 log p - T ♦ 11.873 (14)

The dissociation pressure of oxygen at 13706K w a s calcu­ lated as 5 .9 x1 0 “^ atmospheres.

The equilibrium constant for the reaction,

Mo03 (g) - Mo02 (S) * l/202 (g) at 1370°X, using the oxygen dissociation pressure and molybdenum trloxide pressure calculated from data ob­ tained in this study, Is

* \ / 2 K - --- -2- - 6.7xl0- 3 (15) *MoO

The equilibrium constant may be expressed as a function of temperature in the usual manner, In which AH is the heat of the reaction given in the

above reaction. This heat nay be evaluated by using

the heats of formation for solid molybdenum dioxide and

trloxide given by Brewer (18) and the heat of sublime—

tlon determined in this researob. The heat of reaction

O A was evaluated at 2 9 8 K rather than at 1 3 7 0 K since the

whole calculation was very rough* The constant C may

be evaluated by using the equilibrium constant in equa­

tion (12) at 1370°K.

Equation (13) then becomes

Log K « - 7.280 (17)

Equation (17) may be expressed as follows using equa­ tion (15) for K,

log K - 1 / 2 log P0 - log p (18) 2 HoU^

When equation (18) la solved for log p^ using equation (17) and (1 4 .) we get,

log POo * ---— — ° ♦ 9.186 (19) * T

The dissociation pressure of oxygen calculated from equation (19), gives 1 0 "1*®, lO* ^ * 1 and 1 0 “^"® atmoa pheres at 1300°K, 1000°K and 800°K respectively. These values may be oompared with and 1 0 " ^ #* atmos­ pheres at 1 3 0 0 °E and 1 0 0 0 °K using data from the litera­ ture and molybdenum trloxide vapor pressure data from this research. If the temperature at which the dissoci­ ation was measured was in error by 1 0 0 °, the dissociation pressure of oxygen would only differ by a factor of 0 2 10 * . It aay then be concluded that the factor of A about 1 0 ° between the oxygen dissociation pressure found in this study and that computed from the literature shows that some of the data in the literature must be in error. Both calculations show the dissociation of molybdenum trloxide gas to be less than X% at the tem­ peratures of the runs.

M°4 °ll

The data are tabulated in Table III and plotted together with the molybdenum trloxide data in Plgure 6 .

The molybdenum dioxlde-molybdenum trloxide mixture was run to determine the existence of oxides intermediate between molybdenum trloxide and molybdenum dioxide.

7 .6 4 x1 0 ~ 3 moles of molybdenum trloxide were mixed in an agate mortar with 2,31x10 ^ moles of molybdenum dioxide.

This was added to a Knudsen cell 1.9 centimeters high and 2.5 centimeters In diameter. An orifice 2.76 millimeters had been drilled in the side of the cell,

A series of runs was made at a constant temperature of

880°K. The results are tabulated in Table IV and plotted in Figure 5* Since the oxides of molybdenum vaporise "able 111

Ivs ;oration of Moi ° n

hun* Temp 6 f f ec- « t . * herne1 n f fec- hate of - Log ii 0 °K t ive ^OE £ a x :a n- t ive Z. VH o. b tm. / o t 1 i me c i on .res xl 0 ec. Pact or cm^xlO' ^ecxlO"

9 846 29484 0.00257 1.0098 W? t £J- W7 w 7 2.274 1.244 5.897 7 869 16575 0 .0046? 1.0103 3.835 7.268 4. 02 9 5.386 7 p « 4 871 2790 0.00080 1.0105 > • J3 s 7.477 4.148 5.374 5 876 8954 0.00340 1.0104 * p T c 9.901 5.509 5.250 6 885 9484 0. 00r 2 ;■ 1.0106 3.836 17.12 9.577 5.009 1 912 3094 0.00781 1.0112 3.839 65.75 37.34 4.423 2 938 2072 0.00949 1.0117 3.841 119.2 68.69 4.153 10 938 2554 0.01243 1.0017 3.841 126. 7 72.97 4.127 3 959 1103 0.01114 1.0121 3.«42 262.9 153.0 3.805

* nuns were made in e cell ce b’gh and 1.5 cm in diameter. The Clausing factor K wes 0,688. LOG P ( A * m ) 3 6 5 4 1.05 VAPOR PRESSURES ABOVE ABOVE PRESSURES VAPOR 1.10 IUE 6 FIGURE - I04 X y- 1.15 • M OYDNM TRIOXIDE MOLYBDENUM 0 M o o 40„ AND M D N A 3 0 120 o 40„ \25

Table IV

Vapor Pressure versus Molybdenum-Oxygen Rnt'o at 880°K,

Run Mo. Atomic Rf f e 0tive ..'eight Pres s. Ratio of Time Lose Atn. Oxygen to sec. Grams xlO6 Molybdenum

3.000 18.33 1 2.764 1558 0.03124 21.07 O 2.759 1493 0.013“5 13.22 A ^ ^ / 1 > 2.753 — ^ 0.941*5 19.60 4 2 , 74O 8130 0.06375 S.139 5 2.726 6312 '1. 05053 7.794 r 701 n ^ ^ 6 w X ^ *■* 0.10564 6.892 7 A / T 7 I* f 8 576 0.04593 7.034 O *- . 0A S--'> a 15167 0.11392 10.30 o✓ ^ rp7 16930 0.105 ~t 8.150 ^ CIO i: 16111 0.07093 5.783 'Ad 2 c *5 c - -X C l li '• * ~r ^ '+ ‘.04571 • * -* - - 12 2 . 4 H 10603 0.03940 4.380 13 2.381 3404 0.01560 *-f/ • an-r c^ 14 2.309 21447 0.07774 4.531 15 2.194 19755 0.03713 2.469

* Value interpolated from molybdenum trioxide vsror pressure curve. A P P (ATM.) o 2.0 LO 0 3 )- o. () - PRESSURE VERSUS MOLYBOENUM - OXYGEN COMPOSITION AT 2.8 1 ATOMICRATIO OXYGEN TO MOLYBDENUM

0 c b

oo 1 i — 1 i FIGURE 5 0 2.6 0

o --- 0 2.4 1 --- o o 1 --- JL o --- 880*K w t* I i as trloxide in this range (the dioxide vapor pressure

is negligible at this temperature) the composition

changed from MoOg^y^g to MoC^*

If there is a stable oxide between Mo02 end MoO 3 which dissociates to molybdenum trioxide gas and molyb­

denum dioxide, the pressure of the molybdenum trloxide

gas over the oxide must be lover than that over molyb­

denum trioxide. If the pressure over the intermediate

oxide were equal to or higher than the vapor pressure

of molybdenum trioxide the intermediate oxide would

disproportionate into the two solid phases, molybdenum

trloxide and molybdenum dioxide. Hence, if any oxides

are stable at this temperature their presence should be

indicated by a deorease in the pressure when the molyb­

denum- oxygen ratio reaches the value corresponding to the compound. The only change in pressure observed occur­ red at the oxygen-molybdenum ratio 2.75. Thus it appears that Mo^Oj^ is a s table molybdenum oxide at 880°K.

The vapor pressure runs on were made in the

Knudsen cell used for vapor pressure determinations of

MoO^* The orifice was 0.838 millimeters in diameter.

1 .5*1 0 ”^ moles of molybdenum dioxide and 2.67x10“^ moles of molybdenum trioxide were ground together in an agate mortar and added to the weighed Knudsen cell.

Several preliminary runs were made for a total time of 14 hours at temperatures from 880°K to 940°K. The composition of the solid phase during the 10

vapor pressure runs changed from Mo0 2 to

This was based on the assumption that the gas phase was

MoO^.

Since no heat capacity data were available for

either the solid or gaseous state the heat of sublima­

tion was determined by a least squares fit using equation

(3). The vapor pressure over is given by the

following equation,

log p - - I5 0 0 -9 + 11.891 (20) T

The heat of sublimation for the reaction

l/3Mo>;01 1 - 1/3 Mo02 +Mo03 (g)

is 68,680 * 5 8 0 calories/mole obtained from the slope

in equation (2 0 ).

An extrapolation of the vapor pressure curve for

**°4 ^ 1 1 tntercepts the MoO^ vapor pressure curve at

763°K, indicating that M o^O^ is unstable below this temperature.

The vapor condensed as yellow film which turned deep blue on standing. An X-ray pattern of the con­

densate, which had been crystallised by heating in a vaouum, gave the molybdenum trloxide lines. Thus the vapor

phase was molybdenum trloxide gas. - 4 1 - Molybdenua Dioxide

The data are tabulatad in Table V and plotted to­

gether with the data for Mo^O^ in Figure 7.

The molybdenum dioxide waa run in two Knudsen

cells. The cell used for runs 1 through 4 was 2.5 centi­

meters in diameter and 1 . 9 centimeters high, while runs

9 through 12 were made in a cell 1.9 centimeters in

diameter and 2.5 centimeters high. There was a thermal

gradient of 43° at 1 8 6 0 °K measured across the top sur­

face of the cell with the greater diameter. The thermal

gradient was 2 3 ° at 1 8 5 0 °K measured across the surface

of the cell of smaller diameter. Since the cell was

partially filled by the sample, the gradient inside the

cell would have been less than that measured on the top

outer surface. Even though the gradient for the larger

cell was nearly twice as great as that for the smaller

cell, the data from both cells were within the experi­

mental error for the same curve. ThuB it appears that

the thermal gradients observed in the cells were not

larp enough to significantly affect the results.

There were no thermal data for molybdenum dioxide available in the literature. The heat of vaporisation was obtained by using equation (3). The vapor pressure

of molybdenum dioxide is given in the following equa­ tion, obtained by a least squares fit Evaporation of I'.olybdenum Dioxide

Run*Temp. Effec- ■

12 1620 18346 0.00873 0.00012 0.00861 1.0284 4468 10.50 8.431 5.074 10 1 7 H 10786 0.02621 0.00056 0.02565 1.0311 44 80 53.11 43.88 4.358 4 1736 5037 0.01921 0.00070 0.01851 1.0318 4483 81*98 68.12 4.167 3 1776 3449 0.01796 0.00106 0.01690 1.0329 4488 109.2 91.77 4.037 11 1780 2650 0.01699 0.00052 0.01647 1.0330 4488 138.5 116.5 3.934 2 1818 3734 0.03376 0.00255 0.03121 1.0341 4493 186.0 158.2 3.801 9 1838 3518 0.03259 0.00203 0.03056 1.0347 4496 193.3 165.3 3.782 1 1860 1653 0.02451 0.00243 0.02208 1.0353 4498 297.0 255.4 3.593

* Kuns one through four were made in a cell 1.9 cn. high and 2.5 cm. in diameter. Kuns nine through twelve were made in a cell 2.5 cm. high and 1.9 cm, in diameter. The Clausing factor K was 0.700 for both cells. i LOG P < ATM ) 4.8 4.4 4.0 6 3 AO PESRS BV MQ AD MogOj AND MoQ2 ABOVE PRESSURES VAPOR 56 4 5 IUE 7 FIGURE * I04 x 4* • MOLYBDENUM DIOXIDE MOLYBDENUM • O MOLYBDENUM DIOXIDE - MOLYBDENUM O OYDNM MIXTURE MOLYBDENUM 58

0 6

64 “ 4 4 * log p - - *7j73 + 6.035 (21)

The heat of sublimation calculated from the Blope in equation (21) is 81780 * 400 calories/mole.

In an effort to establish the nature of the mole­ cular species in the vapor phaee two different procedures were followed. In the first of thene, thermodynamic calculations were made of the dissociation pressures for each of two alternatives: (i) the dissociation products are oxygen and molybdenum metal, and (il) the dissocia­ tion products are molybdenum trioxide and molybdenum metal. Enough data are available to permit two calcu­ lations for each alternative, a total of four thermo­ dynamic calculations. The second procedure involved determining the X-ray diffraction pattern of the con­ densate. This procedure leads to more convincing results but the thermodynamic calculations are also described since they show clearly the need for more accurate thermodynamic data on the molybdenum oxides.

The dissociation pressure of oxygen according to c the reaction

Mo02 (g) * M °(s) + wa e calculated using the free energy of formation of

Wo02(S)* aquation (13), derived by Thompson (17) and the free energy of vaporization given in equation (21).

The equilibrium constant K expressed as a function of temperature wee then - 10530 log K ------♦ 2 . 4 4 1 (2 2 ) T where

K - 1 1 1 (2 3 ) PMo02

The dissooiatlon pressure of the oxygen was found by combining equation (21) and (2 2 ) to give

log p - " 2 8 * 0 0 - 3.60 (24 ) T

At 1600°K the dissociation pressure of oxygen is

10~21*4 atmospheres, and at 1900°K it is 10“^®*^ at­ mospheres. Although the data in the literature are considered unreliable, the calculated pressures are low enough to allow for a large error.

The dissociation pressures can also be calculated in another way by using tthe data obtained in this study.

If equation (17), which was derived from the dissocia­ tion pressure of oxygen measured in this study assuming the reaction

Mo03 (g) - MoG2 (5) + l/202 , is combined with the free energy of formation of the gaseous molybdenum trloxide, using data from the litera­ ture (16,18,19) with the vapor pressure data determined here, an equation for the free energy of formation of molybdenum dioxide la obtained. This equation - 4 6 -

A* • - 1 2 1 5 8 0 ♦ 13.38T (25)

gives a dissociation pressure of oxygen for the reaction

Mo0 2 (g) ■ Ho ♦ O2 of 1 0 *"^" and 1 0 " 1 1 *1 atmospheres at 1600°K and 1900°K*

These values nay be compared with 10“^^*^ and 10"^®*^ at 1600°K and 1900°K calculated from equation (13) derived by Thompson.

Both of the calculated values are so much lower than the measured vapor pressure of molybdenum dioxide that we may conclude, despite Inaccuracy in the thermodynamic data, that there is negligible dissociation to molybdenum and oxygen. This is consistent with the experimental results.

There remains a possibility that molybdenum dioxide may dissociate according to the disproportionatlon re­ action

3/2Mo 02(s j . Mo03(g) + 1/2 Mo(S)

By use of Thompson's equation for the free energy of formation of molybdenum dioxide (17), Brewer's free energy of formation for molybdenum trioxide (18),

Cosgrove and Snyder's thermal data for molybdenum trl­ oxide (16) and the free energy of vaporization of molyb­ denum trloxide from this raaearoh we obtain an equation for the free energy change for the above reaction

& T m 104,470 * 77.23T (26) and for the equilibrium constant

log £ • -2283° + 1 6 . 8 8 (27) T

The equillbrlua preasure for aolybdenua trloxide gas at 1600°K and 1900°K calculated froa equation (27) are 10^'^ atmospheres and 10^*9 atmospheres. The data in the literature give results which are inconsistent with the experimental observations. The pressures cal- Q culated froa the literature are about 10 times as high as the measured pressures and the calculations would lead one to expect molybdenum dioxide to vaporize as molybdenum trioxide, which is contrary to the X-ray diffraction experiment to be described later. The above calculation was therefore repeated utilizing the dissoci­ ation pressure of molybdenum trloxide determined in this study. This calculation is described below.

Equation (17) was derived for the equilibrium con­ stant of the reaction

Mo03 (g) . Ko02 (S) ♦ 1/2 02 (g) using the dlss oc 1 a11 on prossurs of ox jrgen found in this study* If this la combined with Thompson's equation for the free energy of formation of molybdenum dioxide (17), an equation for the reaction

3/2 Mo02 (S) - Mo03 (g) + 1/2 Mo is obtained. This equation

A T • 98180 - 54*84 T (28) log k + 11#99 (29) T

when combined with the free energy of vaporisation of

molybdenum dioxide computed from equation (2 1 ) yields

a value for the pressure of molybdenum trloxide in

the reaction

3/2 Mo02 (g) - Mo03 (g) + 1/2 Mo.

The equation for the pressure of molybdenum trloxide is

-482 70 l°g PM0 O3 * — J + (30)

Pressures calculated from this equation at 1600°K and

1900°K are 10”9*1 atmospheres and 10“A.4 atmospheres

which may be compared to 1 0 ”^*^ atmospheres and 10*"2.4

atmospheres computed from the vapor pressure curve of

this research.

The conclusions from the thermodynamic calculations

on the disproportionstion reaction

3/2 Mo02 (S ) - Mo03 (g) ♦ 1/2 Mo(S)

are as follows. First, the results of the two alterna­

tive calculations are in very poor agreement. Second,

even the thermodynamic calculation which indicates the

least dissociation into trloxide still indicates that

the vapor phase would contain 10£ of MoC>3 at 1600°K.

It seems quite likely that both of the thermodynamic calculations are subject to suoh large arrora that the - 4 9 -

ca 1 n ul n t ion;' cannot establish whether or not molybdenum

trioxide is an i mportant vapor phase constituent.

In order to decisively ffrtablish the constitution

of the vapor phase, X-ray diffraction patterns of the

condensate were obtained. The diffraction lines agreed

with those of molybdenum dioxide. It is therefore

conclude! that -r'lybdenun dioxide is the vaporizing

speeler.

In addition t > an X-ray pattern of the condensate

f r o 1.1 t lie- molybdenum !ioxide vapor, the results of the

measure .cuts over the M 0 -MCO 2 :ai x t ur * describe! in the

next section show that the vapor phase was molybdenum dioxide gas.

riolyfrdcnma- ..esauioxide

The data are tabulated in .able VI and plotted,

together with the data for molybdenum dioxide, in

F i „ >re 7.

The vapor pressure over molybdenum s-squioxide is

given b„ a least squares fit of the data to equation (3).

.he equation for the vapor pressure is

log p - — 9g-9? ♦ 11.769 (3i;

Fromihe slope of equation (31) the heat of vaporization,

133,140 1 349 calories is calculated.

The condensate, which was collected on a pyrex glass plate gave the same X—ray pattern as that Vaporization of Kolybdenum-Molybdenum Dioxide Mixture

Run*Temp.Effec- Wt. Calc.Wt. Effec- Thermal Effec- Kate Press. - Log Ho* °K tive Lons Loss of tive Expan- tive of Ttn Time Gro ms Pt. \lt. cion Area Evap. xl05 Grams Loss Factor cm^xl05 g/cm sec.x1 0 ^

3 1760 9558 0.01020 0.00142 0.00878 1.0325 4.486 2.048 1.713 4.766 6 1 8 1 6 1303 0.00490 0.00054 0.00436 1.0341 4.493 7.447 6.329 4.199 7 1834 3492 0.01543 0.00198 0.01345 1.0346 4.495 8.569 7.318 4.136 2 1854 3324 0 . 0 2 0 0 2 0.00266 0.01736 1.0352 4.498 11.61 9.970 4 . 0 0 1 Prel* 1884 857 0.01204 0.00113 0.01091 1.0360 4.501 28.28 24.48 3.611 1 1904 1628 0.02816 0.00298 0.02518 1.0366 4.504 34* 34 29.88 3.525

* The Clausing factor K was 0,700, - 51- obtained from the solid molybdenum dioxide end from the

condensate for the molybdenum dioxide vapor pressure

runs. Thus the gaseous phase over molybdenum sesqui-

oxide is molybdenum dioxide gas.

The runs for molybdenum seaquloxlde were made in

two series. The first series consisted of six runs

made on a aixture of Mo-MoO^ in the cell of smaller

diameter previously mentioned in the discussion of molyb­ denum dioxide. The Mo-MoO^ mixture was composed of

8.8 x 10 moles of molybdenum dioxide which were ground

in an agate mortar with 2.12 x 1 0 " ^ moles of molybdenum.

2^6 milligrams of this mixture were added to the weighed

Knudsen cell.

Kuna were made until there was no loss in weight above that calculated for the platinum container. The last two runs of the first series are not included in the data since there was insufficient sample to obtain equilibrium. The total weight loss was 122.5 milligrams, of which 2 0 . 9 were calculated to be platinum. The difference of 10 1 . 6 milligrams approximates the weight of molybdenum dioxide, 91.5 milligrams, added to the

Knudsen call. The M 0-M0O2 mixture was assumed to be homogeneous. Thus the percentage of molybdenum dioxide in the portion of the mixture placed in the cell was assumed to be the same as the percentage of molybdenum -52-

dioxide in the total mixture. The cell contained 8 1/-

of the total M0-M0O2 mixture.

In the following table are Hated the weight loss

of sample computed for different gaseous species,

labii ..II

Weight Loss of Sample of M 0-M0Q2 in

which 91*5 Milligrams of MoO^ are Present

Gas Weight Loss in Milllarama

MoO 1 6 0 . 1

M o 2° 3 1 1 4 . 4 M o ° 2 9 1 . 5

MoO^ 6 8.6

From the table, MoO and MoO^ may be eliminated, for they are considerably greater and smaller respectively than the experimental errors. The molybdenum sesquloxide seems very unlikely sines it is almost twice as heavy as the molybdenum dioxide. This leaves molybdenum diox­

ide as the most likely gas.

The 10 milligram difference in weight loss between the total weight loaa and the assumed amount of molyb­ denum dioxide in the M0-M0O2 mixture could be due to an error in extrapolating the curve for the weight loss of the platinum Knudsen cell or to an error in the *^sumption of uniformity In the mixture.

The second series of runs for molybdenum sesquloxlde

were comprised of runs 6 and 7. These runs were made

with a fresh charge of Mo-MoO^ mixture* Thia mixture

was oomposed of 128 milligrams of molybdenum dioxide

and 192 milligrams of molybdenum.

The condensate from these runs gave the same X-ray

pattern as that obtained from the solid molybdenum diox­

ide and from the condensate for the molybdenum dioxide

vapor pressure runs. Thus the gaseous phase over molyb­

denum sesquloxide is molybdenum dioxide gas.

ho direct evidence for the composition of the solid phase was obtained. The large difference in the free energy of vaporisation of molybdenum dioxide and of the

H0-M0O2 mixture indicates the presence of a stable lower oxide. There are two possible oxides, molybdenum ses-

-uioxide and molybdenum monoxide. Molybdenum monoxide seens unlikely since solid monoxide would be expected to vaporize as monoxide gas. «fe have already shown that the gas is the dioxide hence molybdenum sesquloxlde is probably the solid phase.

In his review on the oxides Brewer (20) finds much uncertainty about the existence of oxides below Mo02 .

Some investigators claim to have found Ho^O^ while several others have failed to find thia lower oxide. -$4- Vapor j-rcaaurea In the MQlybd>nai-Qxygcn System

The curves for the vapor pressures of the four molybdenum oxides determined in this research (see

Figures 6 and 7) are plotted with the curve of the molyb­ denum vapor pressure of Edwards, Johnston and Blackburn

(21) in Figure 8 .

Xflttgfltea IgiQ&iai The data are tabulated in Table VII and are plotted together with Ueno's (2 ) data in Figure 9.

The runs were made In a cell 1.9 centimeters in diameter and 2.1 centimeters high. The orifice was

0.688 millimeters in diameter. The thermal gradient o o measured across the top of the cell at 1580 K was 16 .

Jince this i3 a smaller gradient than that measured for the molybdenum dioxide runs, the affect on the results would also be expected to be less.

The preliminary runs on tungsten trioxide yielded pressures which were much higher than the later runs.

This was believed to be due to the presence of water resulting from Incomplete dehydration.

Out of 465 milligrams of tungsten trioxide intro­ duced into the cell, 198 milligrams were vaporized during the five preliminary rune.

The pressure is given by fitting the data to equa­ tion (3 ) by least squares -LOG P (ATM.) 4 AO PRESSURESVAPOR IN THE MOLYBDENUM-OXYGEN SYSTEM FIGURE8 Y x I 8 4 0 126 » 8 w-5 >1 i

-> v n p r ■ tion of Txnrr ten Trio xi do

■> 'Y ^ *r\ r Kun * ^ * • • - :: c. c - 3 r.ornfi1 1. f f e c- . X: t a of . rc r ~o r r - o. °K tive lor s a n- ti ve ~ v" ; - r.tn. lime 11*n r ’ f ri on 4rqn "/cm' ! x1Q6 ec. r> a tor c n*" x 10 recxlO^

r* " A t: / 1314 17636 c .00146 1.02 04 *■- * ^ 3.119 1.675 5.776 13 1340 17794 0.00163 1.0211 2. 1 Q~ O ,656 3.555 —' • f *- — 5.715 7 1383 3136 0.00179 1. 022 3 2.659 33.46 18.46 4.734 9 1464 755 0. 002 53 1.0243 2.664 125.s 71.27 4.147 10 1503 940 0.00676 1.0253 2.667 269.6 154.9 3.810 t 11 VJt 1535 1000 0.02156 1.0261 2.669 8 00. 3 464.5 3.333 0 1 6 1560 398 0.02405 1. 02 67 2.670 3265. 1324. 2.888 12 1569 650 0.03423 1.0270 2.671 1972. 1157. 2.937 1531 C O / 14 - >rfc+ 0.05865 1.0273 2.672 1434. 2.343

-I- • .^ n> — • -LOG P (ATM ) 6 5 4 3 6 4 AO PRESSURE TUNGSTENVAPOR OF TRIOXIDE - 7 5 - FIGURE 9 • O THIS RESEARCH O UENO 72 76 -58- —23591 log p ------♦ 12.070 (32)

The heat of vaporization from the slope of equation (32)

la 107,94.0 * 740 calories.

The gas condensed aa a deep blue film which on

crystallizing in a vacuum, gave an X-ray pattern cor­

responding to tungsten trioxide.

By use of Kubachewski and Evans' (21) equations derived for the free energies of formation of tungsten dioxide and tungsten trioxide, together with the free energy of vaporaization calculated from equation (3 2 ), we obtain an equation for the reaction

W03 (g) - W02 (a) + l / 2 0 2 (g)

The thermal functions obtained for the above reaction are

AF = -44,100 + 33.27T (33) and

l o g K - - 8.364 (34)

By combining equation (32) with equation (34) we get

log p02 = 2 log 4 + 2 log pWq3 (3 5 ) or -27906 log P02 “ t + 7*^12 (36)

The dissociation pressure of oxygen may then be calcu­ lated as 10"“^^*^ atmospheres at 1300°K and atmospheres at 1600°K, both of which are negligible compared to the tungsten trioxide pressure determined

in this ntudy, The vapor phase above tungsten trioxide

is tungsten trioxide gas.

luagaAgja Plaxlda The data are listed in Table VIII, and are plotted,

together with the tungsten trioxide vapor pressure data, in Figure 10.

The runs were made in a cell 1.9 centimeters in diameter and 2,2 centimeters high with an orifice 0,716 millimeters in diameter.

An X—ray pattern of the crystallized condensate gave tungsten trioxide lines, indicating that the vapor fhaee was tungsten trioxide gas.

Out of 303 milligrams of sample introduced into the cell, 47 milligrams were calculated to be tungsten trioxide from the analysis mentioned previously. The first five runs, in which 54 milligrams of the excess tungsten trioxide were vaporized, were not used.

Runs were made until there was no further weight loss, as in the molybdenum dioxide runB. nfter deducting the calculated weight of tungsten trioxide from the weight of sample added, there were 255*8 milligrams of dioxide in the cell. At the completion of all the runs, there were 73.7 milligrams of substance remaining in -vs < : dc

- ^ . - .0 :V -

- J . , . x 1C'

-* , r*- :.:?7 s • ~ s 1 ■ . ■; 32 5.159 " 7 ' 7 273.9 * ' *T '

r- -

■ T * T • - .2.3 »• » ! 1 99B.4 -»• 9 * 1 - 0 3 7 0 1 . 1 *• 0 2233! 1 I LOG P (ATM ) VAPOR PRESSURE TUNGSTEN ABOVE TRIOXIDE AND TUNGSTEN DIOXIDE 60 64 FIGURE10 x I04 6.8 • TUNGSTEN DIOXIDE • O TUNGSTEN TRIOXIDEO 72 7.6 the cell. If tungsten dioxide dlaproportionatee accord­

ing to the reaction

3/2 W02 (e) - W03 (g) ♦ 1/2 tf

the weight of tungsten produced would be 72.6 milligrams

which is in very good agreement with the weight left in

the cell.

In a preliminary experiment a cell with a removable

top was filled with tungsten dioxide. The cell was

heated at 1800°K. ^hen the top was removed a grey metallic powder was found which gave an X-ray pattern corresponding to tungsten.

k least square fit of the data to equation (3 ) gave an expression for the vapor pressure of tungsten tri­ oxide over tungsten dioxide

—19702 log p - j + 9.093 (37)

From the slope of equation (37), the heat of vapori­ zation is 90,150 * 880 calories.

Kubaschewski and Evans' (22) equations for the free energy of formation of tungsten trioxide and tungsten dioxide, combined with the free energy of vaporization of tungsten trioxide from equation (3 2 ), gives

dF - 109730 - 57.30T (38) for the reaction

3/2 W02 (s) - W03 (g) + 1/2 W -63- Equation (36) gives an equation for the pressure of tung-

aten trioxide over tungsten dioxide

-23981 loS *W03 - — t— + 12.523 (39)

Pressures calculated from equation (39) at 1300°K and

1600°K are 1CP5.92 atmospheres and 10”^ * ^ atmospheres

as compared to 1 0 " ^ * ^ and 1 0 ~2 . 2 2 from the measured

vapor pressure curve. The very good agreement between

the calculated and measured pressures is in marked con­

trast to a similar comparison for molybdenum dioxide.

It may be concluded that the data used in this calcu­

lation are essentially correct. The calculated values

are consistent with the observed pressures and vapor

phase composition.

The curve for tungsten dioxide intercepts the tung­

sten trioxide vapor pressure curve at 1306°K. The dioxide should be unstable below this temperature.

However, Magneli (23) prepared tungsten dioxide by heat­ ing tungsten and tungsten trioxide at 1223°K. Therefore,

it appears that there ia an error of at least lj£ in the slope of either the WO^ or WO2 curves. COMPARISON WITH 1AKLIE& DATA

IfcJali 1 Heats of Sublimation and Vapor Pressures

Investigator Compound AH sub. Pressure Temp, Calories Atmos. °K ______U Q ° ______Blackburn M0 Q3 76,300 51.0 900 Ueno M0 Q3 63*600 5.06 900 Fieser M0 O3 66,570 25.2 900 Blackburn V03 107,900 5.30 1360 Ueno WO3 112,600 4 . 79 1360

The heats of sublimation for molybdenum trioxide and tungsten trioilde from the literature and from this study are given in Table X. The values were obtained from the slopes of the vapor pressure curves (see Figures

4 and 9 ), except for the heat of sublimation of molyb­ denum trioxide determined in this study. Thia was found by correcting 4 HQ, found by the sigma plot, for the change in ACp with temperature. The value given in

Table X is the heat of sublimation at 880°K,

The pressures of the gases are given for the tem­ perature listed in the last column of Table X. It may be seen that the pressure of molybdenum trioxide found by Ueno is about one tenth of the pressure found in this study. The discrepancy is several times greater than the expected experimental errors.

Ueno (2) determined the vapor pressures of molyb­ denum trioxide and tungsten trioxide by the effusion -65- method. Hie apparatus consisted of a quarts Knudsen

cell suspended by a platinum wire from the beam of a bal­

ance which was sealed inside a quarts furnace. The

balance which used an optical lever for serolng waa

brought to sero by moving a weight on the end of the

beam opposite the one from which the sample was hanging*

n rider of known weight was placed on the end of the

beam from which the sample was hanging. The electric

tube furnace was brought into position around the quartz

furnace and the time elapsed until the balance returned

to zero was noted. Temperatures were measured with a

platlnum-platinum rhodium thermocouple fused into a glass tube beneath the quarts Knudsen cell. The vacuum in the

system was 10 ^ millimeters of mercury.

This method used by Ueno gives very good internal coneiatency but the accuracy of the method is probably not so great as the method used in this research, Ueno's data for molybdenum trioxide (plotted in Figure 4 ) are over a temperature interval of 4 0 ° compared to 150° for this study. The slope through hjs data would be very sensitive to small errors in his measurements.

Since the furnace surrounded the Knudsen cell, the effusing gaaes had a rather small fraction of the total surface on which to oondenae. Thia could result in soma of the gases returning to the cell, giving a lower apparent pressure.

A least squares fit of Ueno's data to equation (3) gives

+ 10.152 (40) for molybdenum trioxide. The heat of sublimation of molybdenum trioxide from equation (4.0 ) is 6 3 , 6 0 0 calories

Although Ueno's paper failed to say whether the 3 runs on molybdenum trioxide were made in one series or more than one series, it seems reasonable to assume they were all made In the same series. If this were the case then there would be no check on his date as there la in this study. In this research two Knudaen cells and three different orifice dimensions were used. All the data obtained from the two cells were within the experi­ mental error* Ueno's data for molybdenum trioxide may have had a constant experimental error which was not de­ tected due to the possibility that all tthe runs were made in one series,

Ueno's equation for tungsten trioxide is

log p - T ♦ 12.757 (41) which gives a heat of sublimation of 112,600 calories,

Ueno's data for tungsten trioxide (plotted in Fig­ ure 9) are over a 5 0 ° temperature Interval as compared to 173° for this research. Here again the slope should be very sensitive to errors in measurement, although the -67-

author claims lees than 2,5% arror In the vapor pressures.

Ueno's tungsten trioxide data are in good agreement

with the vapor pressures determined here while for molyb­

denum trioxide there are marked discrepancies. There

seems, on the whole, to be more evidence to support the

values found in this study. The temperatures were care­

fully checked, the weight loss was determined by differ­

ences in weight, the areas of the orifice were accurately

determined and the time for the runs was corrected for

temperature variations.

Fieser (1) determined the vapor pressure of molyb­ denum trioxide by rate of evaporation from open trays at

50° intervals from 600°C to 1100°C. He calculated his data three ways. The method based on the rate of vapori­ zation which he preferred was used in plotting his data in Figure 4 , A least squares fit on the data below the melting point gives

log P - ~14|49 ♦ 11.567 U2)

The slope of equation (42) yields a heat of sublimation of 66,570 calories. The vapor pressure equation above the melting point gives

log p * + 5*494 (43)

From the slope of equation (43) ve obtain a heat of vaporization of 35,770 calories. The difference in the heat of sublimation and heat of vaporization gives a heat - 68- of fusion of 30,800 calorics, twica as graat as tha haat of fusion of 12,540 calorics found 6 7 Cosgrove and

Snyder (16). Tha intersection of aquation (42) and

(4 3 ) gives a malting point of 1 1 0 S°K, 4 0 ° higher than tha malting point found bj Cosgrove and Snyder (16).

Although Fleser'a data for molybdenum trioxide are closer to the data found here than are Ueno's, his method is open to the criticism that the accomodation coeffici­ ent is assumed to be unity. In the present research a

Knudsen cell is used and no assumption concerning accomo­ dation coefficient need be made in the calculation of vapor pressures. Moreover, the heat of fusion calculated from Fleser's data does not agree with the more recent calorimetrio determination of Cosgrove and Snyder.

Since no vapor pressure measurements were made on the liquid in the present research no comparison with the work of Cosgrove and Snyder can be made. Despite this fact it is believed that the data of the present research are more accurate than those of Fleser since no assump­ tion concerning accomodation coefficients need be made. -69- CQttCfctfglPlI In the molybdenum-oxygen system there irt four oxides which are stable aboTa 800°K. These are molybdenum trioxide and which vaporise to molybdenum trioxide gas, and molybdenum dioxide and molybdenum sesquioxide, which vaporise to molybdenum dioxide gas.

The two tungsten oxides, tungsten trioxide and tungsten dioxide both vaporise to tungsten trioxide. -70- BIBLIQGRAPflX

(1) Felser, J., "Behavior of Metallic Compounds at

High Temperatures, Molybdenum Trioxide", Metall, u. Era,

(1931) pp. 297-302.

(2) Ueno, K.,"Determinstion of the Vapor Pressure

of Solid Salts. IV. The Vapor Pressures of WQ^, MoO^*

CdO, and TeQg ant* Their Thermodynamic Values", J. Chem.

Soc., Japan, ££ (1941)> pp. 990-4,.

(3) Knud sen, 11., "Experimental Determination of the

Vapor Pressure of Mercury at 0° and at Higher Temperatures,"

Ann. Physik, £ £ (1909) pp. 179-93.

(4) Loeb, L. B., The Kinetic Theory of Gases. hew

York: McGraw-Hill Book Company, Inc., 1934* PP. 103-106.

(5) Ditchburn, E. W., and Gilmour, J. C., "The

Vapor Pressures of Monatomic Vapors," Rev. Modern Phys.,

12 (1941), PP. 310-27.

(6 ) Spleser, R., and Johnston, H. L., "Methods of

Determining Vapor Pressure of Metals," Preprint No. 11,

Thirty-first Annual Convention of the American Society

for Metals, Cleveland, Ohio, October 17-21, 1949.

(7) Dushman, S., Scientific Foundations of Vacuum

Technique. New Torkt John Viley and Sons, Inc., 1949,

p. 99.

(8 ) Brewer, L., and Searcy, A. V., "Utilization of

Equilibrimm Vapor-Pressure Data," J. Chem. Education,

(1949) pp. 548-52. -71- (9) Kelley, K. K., "Contributions to the Data on

Theoretical Metallurgy, III. The Free Energies of Vapori­

sation and Vapor Pressures of Inorganic Substances,"

Bureau of Mines Bulletin 383, (1935), pp. 1-13.

(10) Hoch, M., "New Insulation for Radiofrequency

Leads in High Vacuum Work," Rev. Sci. Instr., £2, (1952) p. 651.

(11) Joint Committee on Chemical Analysis by X-ray

Diffraction Methods. X-ray Diffraction Data Cards.

Philadelphia: The American Society for Testing Materials,

1950, Card No. 3.42 for MoOj.

(1 2 ) Joint Committee on Chemical Analysis by X-Ray

Diffraction Methods. X-ray Diffraction Data Cards.

Philadelphia: The American Society for Testing Materials,

1950, Card No. 3.44 for W02 .

(13) Dushman, S., Scientific Foundations of Vacuum

Technique. New York: John Wiley and Sons, Inc., 1949, p. 96.

(14) Edwards, J. W., Spieser, R., and Johnston, H, L.,

"High Temperature Structure and Thermal Expansion of Some

Metals as Determined by X-ray Diffraction Data. I. Plati­ num, Tantalum, Vlobium and Molybdenum", J. Applied Phys.,

22 (1951), p. 425.

(15) Kubaschevskl, 0., and Evans, E. L., Metallurgi­ cal Thermochemistry. London: Buttervorth-Springer Ltd.,

1951, p. 169. -72-

(16) Cosgrove, L. A., and Snyder, P. E., "High

Temperature Thermodynamic Fropartles of Molybdanum Trl-

oxida,■ J. Am. Chem. Soc ., ^5 (1953), p. 1227.

(17) Thompson, M. da K., The Total and Free Energies

of Formation of the Oxides of Thirty-two Metals. New

Yorkt The Electrochemical Society, Inc., 1942, p. 56.

(13) Brewer, L., "The Thermodynamic Properties of

the Oxides and Their Vaporization Processes," The

Chem. Rev., .52 (1953), p. 9.

(19) Kelley, K. K., "Contributions to the Data on

Theoretical Metallurgy, X. High-Temperature Heat-

Content, Heat Capacity, and Entropy Data for Inorganic

Compounds," Bureau of Mines Bulletin 476, (1949), pp. 120,

131.

Kelley, K. K., "Contributions to the Data on

Theoretical Metallurgy, XI Entropies of Inorganic Sub­

stances," Bureau of Mines Bulletin 477, (1950), pp. 66,72.

(2d Brewer, L ., "The Thermodynamic Properties of

the Oxides and Their Vapo^nBilon Processes," Chem. Rev.,

.52, (1953) p. 28.

(21) Edwards, J. W., Johnston, H. L., and Blackburn,

P. E., "Vapor Pressures of Inorganic Substances. VIII.

Molybdenum between 2151 and 2462°K." J. Am. Chem. Soc.,

14, (1952), pp. 1539-40. (22) KubaBohevskl, 0., and Evans, S. LL., Metallurgi­

cal Thermochemistry. London* Butterworth-SprIngar Ltd*,

1951, p. 309.

(23) Magneli, ▲ "The Crystal Structure of tha

Dioxides of Molybdenum and Tungsten", Arklv. Kemi,

Mineral Geol. 2LL No. 2 (1946) pp. 1-11. 74

ADTOBIOttlAPHY

I, Paul Edvard Blackburn, vaa born in West Branch, Iova, May 24, 1924, My aaeondary education vaa received

In tha public aehoola of Dalavara, Ohio, I entered the Bevy V-12 program at Harvard University in July, 1943* vhere X obtained the first tvo years of my undergraduate training. After being separated from active service In the Navy in July, 194&, I entered Ohio Wesleyan Uni­ versity vhere X received my Bachelor of Arts degree in June, 1948, Xn September, 1948, I vas employed by The Ohio State University as a research ohemlat. In September, 1951, I received a Research Fellovship from The Ohio State University vhich I have held vhile completing the requirements for the degree Doctor of Philosophy,