FISCHER-TROPSCH SYNTHESIS

FISCHER-TROPSCH SYNTHESIS Differential Reaction Rate Studies with Cobalt Catalyst ROBERT B. .IhDERSON, ilBR.4HAlI KRIEG', AND R. A. FRIEI)El, Central Experiment Station.

L.

r-. S. Bureau

of

Mines, Bruceton, Pa.

s. \L4SOl

1 niversity of Pittshureh, Pittsburgh. P a .

T

L)itferrntial reaction rate data drv presented for the this wax rather completelj HL rate of synthesis of Fischer-Tropsch s>nthesis with cobalt ratalysts at atmoscobalt Fischer-Tropsch filled the pores of the catapheric pressure. The synthesis rate was high in the inivttalysts is dependent upon lyst, but a t higher pressures tial part of the bed, lower and nearly constant throughtemperature, operating presthe catalytic activity rew p , composition, and flow out a large portion of the catal>stbed, until the reactants mained constant even though of synthesis gas, a s well as were fairly completely consumed. Over the range of relathe pores of the catalyst The chemical and physical were filled Kith hydrocarbon tively constant rate of synthesis, the gas composition raried considerably. Methane and carbon dioxide were I t has been postulated that chature of the catalyst ( 1 7 ) \Teller (20)and Anderson (6) formed by primary reactions; methane was also formed the rate of synthesis depends Yhourd that the rate of the b) secondary hydrocracking of higher-molecular-weight upon some process a t the +rnthesis n-as strongly dehydrocarbons and carbon dioxide by the water gas reaccatalyst surface and not upon tion. These reactions occurred throughout the catalyst pendent on temperature, overrates of diffusion or adsorpill activation energies being bed. and their extent depended upon the composition of tion of reactants ( 4 ) . the gas in contact with the catalyst. The postulates of Craxford ( 7 ) postulated thout 25 kg.-cal. per mole. Craxford involving the synthesib occurring on cobalt Fi5cher and Pichler (10) and that cobalt carbide is an inI erniediate in the synthesib carbide and the secondary reactions on cobalt atoms do K e l l ~ r(20) found t h a t the r a w of synthesis a t subatmosnot appear to be an adequate explanation of author's data. and t h a t its presence is depheric pressures was desirable in the normal synpendent upon the operating thesis. Work a t the Bureau # ~ fA\lirie5 ( 4 , 21) showed that bulk-phase cobalt carbide was pressure t o a power less than 1 . .It atmospheric pressure and riot formed in the synthesis and that its presence in large tibove, the pressure coefficient of the synthesis rate was found to he about 0; however, the catalyst remained active considerably :ample. I n every case the rate of rrithdrawal of gas was less , i t atmospheric pressure the rate of synthesis was retarded whm than 2% of the gas flox-ing in that part of the bed. The gas samples were analyzed in a mass spectrometer (Consolidated Present address, General American Transportation Corporation, East Enpinwring Company). I l h i c a e o . Tnd. ~~

2189

INDUSTRIAL AND ENGINEERING CHEMISTRY

2190

1

lnsolute cerne nt

‘

,Thermocouple

’

The volume of gas a t any point, z, is Vo(1 - C z ) . Thus, the volume of component B at point z is given by

flexible copper tubing coil

h

P

\,

17

Vol. 41, No. 10

lrBi/170

= (1 - Cz)fBz

(4)

The partial pressure of component B is equal to VBig,lVtota~,: where I r t o t a ~ , is the sum of the volumes of all coniponents at position I. The voluml- of Rater vapor was not determined directly, but obtained from an oxygen balance, VHgOz

‘ d

1

?/Thermocouple

~

6 8

k2 T-O ‘

WGX

= J’COO

- J’COz - 2 v C O z z

(5)

This is permissible since nearly all of the oxygen in the products occuIs as water and carbon dioxide. The volume of hyclrocarhons other than methane \vas assumed to equal one tenth of the volume of water vapor present. This assumes an average molecular weight of these hydrocarbons of Cg t o Cs; since the volume of hydrocarbon vapor was usually less than 3% of the total volume, these assumptions introduce no significant error. The total conversion from the beginning of the bed to any point is expreswd as liters of hydrogen plus carbon monoxide consumed per hour. The accuracy of the experimental method is difficult to drtermine. The relative errors of the gas analyses were less thari lY0 except for percentages less than 2%, \There the absolute accuracy was about 0 . 0 5 ~ o . The apparent contraction over the entire bed as computed from the change in eoncentration of argon v a s usually about 3 to 5% lower than the apparent Contraction computed from the in- and outflows. The greater part of this difference \vas due to the fact that most of the gaseous hydrocarbons were removed by the charcoal scrubber placed before the outmeter. The accuracy of the sampling method probably \vas satisfactory, since the usage ratios of hydrogen to carbon monoxide computed between alternate ports were nearly equal t o 2 in the parts of the bed where the normal synthesis might be caspected to occur. There was no indication of local overheating within the catalyst bed. The catalyst tube was surrounded by a bath of boiling

h

/ I -

trap

Figure 1. Sampling DeFice in Converter

The gases (synthesis gas, hydrogen, and argon of high puritj ) were premixed in cylinders. Argon was chosen as inert refeierice gas because of the accuracy and simplicity of Its analysis by the mass spectrometer. From the increase in concentration of argon, the apparent contraction and the volume of any component a t that point in the bed pel unit volume of synthesis gas entering the converter could be computed. The following equations were used in the computations. Because the argon does not react, SioT’c

= f4.Vz

(11

where f A 0 ,

fi. = fractions of argon in gas entering converter and at position z in catalyst bed, respectively V o ,V s = volumes of gaseous components entering bed and at position z

Since the gases were sampled at room temperature and analyzed on a water-vapor-free basis, water vapor and heavy hydrocarbons are not included in these volumes. The apparent contraction C is defined as =

-

volume out, per unit time volume in, per unit time

(2 1

with the volumes measured a t room temperature, or using Equations l and 2,

c= =

1

- fAo/fA,

where C, = apparent contraction to point z

(3)

Figure 2. Data for Test 8 2Hg to 1 C O g a s at 1 9 2 O C. and space velocity of 1Cn per hour; inflow, 0.134 liter (S.T.P.) per gram unreduced catalyst per hour

INDUSTRIAL A N D ENGINEERING CHEMISTRY

October 1949

2191

Tetralin, the temperature beT.*BLE I. G.4s AXALYSISDITA, IS I-oLuirr: PER CEST, FOR TEST10 ing controlled by regulating the pressure a t which the R HI t o I C 0 a t 102' C . ) Tetralin vias boiling ( 1 4 , 1 7 ) . Gas sample -4 H? CO CHI C? C>--a Cz C3-CA C4-Ca CI-COz Unfortunately the effect of the hydrostatic pressure of the boiling liquid on its boiling point was not anticipated, and this resulted in a 4 " C. or less temperature gradient along the catalyst bed, the temperature increasing from the top to the bottom of the reactor. -4bout the same temperature gradient, was obFerved viith no catalyst in the reactor. All of the temperatures reported in this were planned to be made in the range of relatively constant catapaper were measured a t the middle of the length of the boiling lytic activity, as just described. Prior to the special experiments liquid bath by a thermocouple in the n-ell in the hoiler, a s Figure under consideration, the c:italyat was operated for 30 days with 1 shows. 2H2 to 1CO gas at atmospheric pressure with reactivations mTth I n these experiments 150 cc. of pelleted catalyst, Reighing hydrogen a t 200 C. evrry sixth day. Before the special experi114.4 grams, were used. T h e catalyst was reduced with hydrogen ment \\-as started, the catalyst was reactivated with hydrogen a t in the reactor a t a space velocity of 3000 for 2 hours a t 360" C. 200" C. and operated with 2H2-1CO for 3 days to ensure con(space velocity taken as volumes of feed gas, a t standard temstant activit'y. Then the catalyst was operated at the desired perature and pressure, per bulk volume of catalyst per hour). temperature and with the special gas mixture for 16 hours before .A11 of the tests were made a t atmospheric pressure. the gas samples a t various bed lengths were taken. The experiments with a given gas composition were performed on successive C-ITA LY ST ACTIVITY days, and the catalyst was readtivated with hydrogen before a Previous studies ( 1 5 ) of the synthesis on cobalt catalysts a t series with another gas composition was st,arted. The series of esatmospheric pressure shon-ed that the activity of a freshly reduced periments m r e performed in the following order: 3H2to 1CO gas, catalyst decreased considerably during the first 12 hours and then 2H2 to 1CO gas, and 1H2to 1CO gas. remained relatively constant for 10 or more days. After hydroThe decrease in activity in a given series of experiments was gen reactivation, the act,ivity was restored t o its initial value, and usually small, and it is believed that these data may be compared the cycle n-as repeated. The average activity remained the same without uncertainty. However, comparison of activities in difover many such cycles ( 5 ) . I n the present work the synthesis was ferent series of experiments with diff went gas compositions may studied with gases of hydrogen-carbon monoside ratios of apbe uncertain because the nature of the catalyst surface or the hyproximately 1, 2, and 3. The esperiments with each gas mixture drocarbons adsorbed upon it may change. The data are inter-

.6

b ' I

I

I

I

W

w (L

%.5 0

c

z

a

F

W 0

W

2

2.4

cc 80

I

z-

0

cc wl W [r a

c V

a 60 E

z 0

-J

5

b.3

a 4

0

c

z W

40

2

.2

a a 4

20

BED L E N G T H ,

.I

CENTIMETERS

Figure 3. Data for Test 10


2Hg to 1CO gas at 192' C. and space velocity of 49.5 per hour; inflow, 0.065 liter (S.T.P.) per gram unreduced catalyst per hour

P H z to 1CO gas a t 206' C. and space velocity of 102 per hour; inflow, 0.144 liter (S.T.P.) per gram unreduced catalyst per hour


2192

Vol. 41, No. 10

a

Methone

+

3.0

*,

-

.

Carbon dioxide

3

'0

20

30 40 50 60 70 BED LENGTH, CENTIMETERS

80

90



J.5Hz to I C 0 gas at 186' C. and space velocity of 99.5 per hour; inHow. 0.130 liter (S.T.P.) per gram unreduced rataly-t per hour

d.SH2 to I C 0 gas a t 187' C. and space velocit, of 49.3 per hour; inflnw. 0.0646 liter (S.T.P.) per gram unrrdiired catalFst per hnur

preted t o indicate that the activit) in terms of volunic;. 01 hydrogen plus carbon monoxide converted per hour a t a given flow were about equal for hydrogen-carbon monoxide ratios of 3 and 2, but that the activity with 1to 1gas was considerably lower Table I gives gas analysis data for a typical experiment. The percentages of CZand higher hydrocarbons are too small to permit a detailed treatment of their rates of formation; however, i t was possible t o compare the total amount of satura ted and ansaturated hydrocarbons present a t any point in the bed fnr C', to C I hydrocarbons

atlube tnose ot rigures s and 3, and the partial prrssures 01 methane and carbon dioxide were greater. In a test with 3.5H2 t o 1CO gas a t a space velocity of 99.5 and 186' C. (Figure 5 ) , the curves had the same general shape as in Figures 2, 3, and 4, except that the part'ial pressure of carbon monoxide decreased more rapidly than that of hydrogen and the partial pressure of methane was greater. I n tests 5 and 7 (Figures 6 and 7) the carbon monoxide was completely consumed by t,hP middle of the catalyst bed. The partial pressures of methane were high and continued t o increase even after t,he carbon monoxide was consumed. The partial pressure of carbon dioxide decreased after the carbon monoxide was completely consumed. Increase? in flow and temperature increased the space-timeyield. The activity of the catalyst had decreased considerably befort tests 15 and 16 (Figures 8 and 9) were made. This was due t'o several days of operation in 1H1 t o 1CO gas in which tests 12 snd 13 were made. The curves had the same general shape as in the experiments with 2Hs to 1CO and 3Hz to 1CO gas, except that the activities and space-time-yields tvere lower and the partial pressure of carbon monoxide remained constant over most of the catalyst bed. For 1Hz to 1CO gas the partial pressure of carbon monoxide should remain constant as long as the hydrogen a n d carbon monoxide are consumed in the ratio of 2 to 1 LIDQ

I

EFFECT OF SPACE V E L O C I l l

k'igurc< 2, 3, and 4 present data for experimenij nitil SH2 ~ c 1CO gas; Figures 5, 6, and 7, data for 3.5H2 to 1CO gas, and Figures 8 and 9, for lH, t o 1CO gas. I n Figure 2, a t 192' (3.and 102 spaee velocity, the partial pressukes of hydrogen and carbon monoxide decreased and the partial pressure of water vapor increased continuously throughout the bed. The slope of the curve for synthesis gas converted from the beginning of the catalyst bed to any port was large in the first part of the bed, then became nearly constant, and finally decreased near the outlet end of the bed. The partial pressure of carbon dioxide was small and increased throughout the bed, and the curve of the partial pressure of methane was concave upward. Decreasing the space velocit) bo 49.5 (Figure 3) increased the degree of conversion (apparent contraction) but decreased the space-time-yield as shorn n by the liters-per-hour curve. These ourves and those of the partial presewes had the same general shape as in Figure 2. The partial pressures of methane and carbon dioxide were greater t h a n in the previous figure. I n Figure 4, at 206" C. and a space velocity of 102, the space-time-yield (liters per hour) Curve was consider-

1

COILSUhXIYTlOh OF HYDROGL\

4\L)

CARBO\ .I1ONOXIDL

Table I1 gives the ratios of usage of hydrogen to carboil mouoxide along the catalyst bed for the data of tests 8, 5 , and 14 The ratios were computed for the gas consumed in the catalysl bed between alternate ports-Le., the first t o third, second t o fourth, etc. This tends to smooth the data and eliminates unimportant fluctuationy The usage ratio w3s somewhat greater thaD


October 1949

2193

c

z W

loop w

2-

2

BC

$ c

2 (0 3

c

2

60

a

Q

a 4c


.._- ..-_

___

1Ht to 1CO gas at 197' C. and space velocity of 24.1 per hour; i ~ f l o w , n n w c I:+-- la T P e a . .d,.nn ..-+.,I..-+ .i L

Figure 7. 3.5%

-.._

Data for Test 7

to 1CO gas at 200" C. and space velocity of 102 per hour; inflow. 0.134 liter (S.T.P.) per gram unreduced catalyst pcr hour 0

2 a t the brgiiiiiirig of the bed in all tests With 2Hz to 1( ( 1 g . i ~the ratio reniaiiied \omen hat above 2 throughout the catall s t bed except for the last port. K i t h 3Ht to 1CO gas the usage idtio remained -omenhat above 2 in the first half of the bed ind increased to very high values when nearly all of the carbon inoiioxtde had heen consumed. With 1Ht t o 1CO gas the ratio rtmiainpd about 2 . d e c r e a h g considerably near the bottom of the h14

5 4.

u 3.4 "?

J1 "J

a a

-

5.5 (r

1 U

TABLE

11. US I G E RATIOS OF HYDROGES TO AS A

Port a h c

d

e I) 6,

CARBON \IOSOXIDE-

FUNCTION OF BEDLESGTH --Usage Ratio, Ha: CO- Test 8 Test 5 Test 14 2Hz: 1CO 3.5H9:lCO lH?:lCO 2.39 2.40 2.18 2.17 2.20 1.89 2.20 2.20 1.77 1.98 2.27 2.00 1.90 2.63 2.05 2.12 7.62 2.17 2.13 8.50 1.74 1.85 m 1 . 2:1

2

I

B E 0 LENGTH, CENTIMETERS

In Figures 10, 11, and 12 the Sield of methane formed in each portion of the catalyst bed per cubic meter of hydrogen plus carbon monoxide consumed is plotted against bed length for tests 6-ith 2 H t t o 1C0, 3.5H2 t o 1C0, and 1Ht to 1C0, respectively. The data werp qmoothed by calculating the methane formed per cubic meter o f gas consumed in an interval of tN0 holes. With 2Ht t o 1CO gas the methane -bed length curves passed through a minimum in the middle of the catallst bed. With3.5Ht to 1CO gas the methane curves increased from the beginning of the bed, becoming very high in tests 5 and 7 above the point in the bed where most of the carbon monoxide had been consumed. These

F i g u r e 9. Data for Test 16 IH1 to 1CO gas at 207' C. and space velocity of 95.5 per hour; inflow. 0.125 liter (S.T.P.) per mam unreduced cataljst per hour

iiirge yields of riit:thaiir kwr cubic nietwr of gas converted ortn be explained only by the assuwpt ioii of hydrocracking of hiyhermolecular-weight hydrocarbons. IT-ith 1H? t o 1CO gas (Figure L2j, the yield of methane either remained constaiit or decreased throughout the catalyst bed. In all ewes the yield? of mc~l,hnnr were greater a t higher temperat,ures. T h e perwntager nf light hvdrocnrbons other than met,hane in

'

2194

INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY 100

100 90 1

0

70

n

w

I

1

1

I

I

I

5

25

m z

3,

40

n

U

w

IO 7.5

- W

1-r

30

u mo

W W

I

5.0 25

u v

5

1

50

a >

rar

t

I

+

60

2 ,"

50

W

I-

cn

I

75

ao

E

Vol. 41, No. 10

20

I

U

0 8

192'C

10

I0

I

2 0 30 40 5 0 60 7 0 BED LENGTH, CENTIMETERS

80 90

Figure 10. Differential Formation of .\IethaneBed Length Curves for Tests 8, 10, and 11 with 211a to 1CO Gas Methane yields (in grams per cubic meter of 15: plus CO consumed) were computed between alternate ports

w

250

100 75

50 I

25

'

0

II IO

0' 4 dl 5 e,7 I

I 186OC. 187'C 200oc. I

l

I

H P :'CO=3.5 l

I I

I

I

40 50 6 0 70 BED LENGTH, CENTIMETERS

20

I

I

0 I I 206OC

IO

-0 16 207°C

30

I

80

1 I

90

Figure 11. Differential Formation of .\lethane-Bed Length Curves for Tests 4 , 5 , and 7 with 3.5142 to 1CO Gas Methane riclds (in grams per cubic meter of H: plus CO rqnsumed) were computed between alternate ports; arrows indicate position within bed where pressure of CO had decreased below 0.01 arm

the gas samples Twre too sinall to permit a detailed calcu1:itioii of t h e amounts formed per cubic iiieter of gas reacted, hut it 173s possible to prepare informative plots of ratio< of snturatrtl to U I I saturated hydrocarbons in the gas against bed lt>iigtli,i i q diowi for C1hydrocarbons in Figure 13. K i t h 2Ha to 1CO ga2 this ratio reiiiaiiied about coiistant in the first half of the bed arid inti in t h e last half. The fractioii of saturated hydriicarl)oiis iricreased although the ratio of hydrogen to carbon niiiiioside decreased. K i t h 3.5H: to 1CO the Ernctioii uf aaturntr, increased throughout th(2 bed, arid with lHr to 1CO gas it decren;eii. FORMATIOS OF C 4 R B O S D I O X l n E

Figures 14 to 16 illustrate the formatiori of carbon tiioxide as a function of bed length. Again the data are smoothed bv cornputing the carbon dioxide .formed over intervals of t x o ports. Since t.he total number of moles of hydrogen and carbon monoside is unchanged if carbon dioside is formed by the x t t e r gas shift .reaction, the formation of carbon dioxide was computed as grams

of c u b o n dioxide formed per cubic meter of carbon monoxide consumed. For 2H2 to 1CO gas (Figure 1-1)the differential amount of carbon dioside formed increased with bed length, the amount of carbon dioxide being greater for the erperimeiits in which the fractioiid conx-rrsiori \vas the greatest. K i t h 3.5H2 to 1CO the amount of carboii dioxide also increased r i t h increasing percentage of conversion uiitil the carbon monoside vias practically COIIsumcd (esperimrwts 5 and 7 ) . Beyond this point the carbon dioxide i'oriiied iii prcvious portions of the bed n-as consuinfd cither I t i o n to methane or by the reversal of the water gas itli 2EI2 to 1CO gas the aniouiit of carbon tlioside forilled \vas greater than with 3H2to 1CO gas for similar conver11, t o 1CO gas (Figure 16) the yields of c:trlmn did xitli increasing esterits of coilversion, arid for of conversion the carbon dioxide was greater than for t,itlier 2H? to 1CO or 3.SH. to I C 0 gas. The :miouiit of carboii dioxide formed i m s slionm to iiicre:ise with the extent of coiiversioii a11d thus with the partial pressure of water vapor iii the gas stream, arid also increased with the partictl pressure of carbon monoside. Thu., the data strongly sugshift rextiori as the source of carbon tlioside, d o n i i r i a n t osygeri-containing product of the 1)riiii:uy m i c i i o i i . The amourit of carbon dioside usually increoaed vitli temperature, but its formation vas inore strongly a fuiict ioii (Jf 1xirti;tI pressure (Jf T\-ater vapor than of tempcrature. The c1ai:i of Figures 2 to 9 are summarized in Figure 17, where of incoming hydrogen plus carbon iiionoside consumed nguiiisr the reciprocal of t,he space velocity for tests n-ith iiyrllogc~nto c:irtxin inoiioxide ratios Gf 3.5 to 1,2 to 1, rand 1 t o 1. Tliwc. pliit s h o w tile same characteristic iiientioiicd pre:tiit1 t,lie points obtained a t the same temperature but ere111iiiitinl flows f d l about the same curve. The middle portio11 of the conversioll curves is almost linear, the slope of this parr ilecre:isiiig onl>-slightly with decreasing space velocity. I n the initid portions of the bed the slope of the conversion curve is greater tllaii that of the "linear'' portion, and when the reactants are about 70y0coiisuiixxl the slope is lon.er. The rate curve for 1H2to 1CO gas flattens a t a lower fractional conversion than those for the oilier gas nlixtures. IVith 2 H a to 1CO gas the over-all activation energies computed from the rates of reaction at equal estents of conversion were about 29.0 l;g.-cal. per mole at high extents of conversion and were somen-liat larger a t low corlversions. With 3Hn to 1CO gas larger activation energies than 29.0 kg.-cal. per mole )\-ere observed, and with 1H2to 1CO gas, lower activation energies.

2195


October 1949 10.0

I

d

0.0 6.0 5.0

Y

/

*/(

i' I

1

40 .

1rZ-j

8

300

a IW

y

200

m 0

u IO0 W

a

Ln

Test 14 I HI! to I C 0

2

197°C.

0

.

70 50 40

ON

.2I

0 6

~

0

IO

20

30

40

50

60

70

00

90 Hg G O =

BED LENGTH, CENTIMETERS

IO

Figure 13. Ratio of Saturated to Lnsaturated CI Hydrocarbons CS. Bed Length for Tests 5,lO. and 14 Ratios of Ci h*drocarbons were computed from total amounts of these components present at any given position in bed and are not differential values

20

30

40 50

60

70

I ! 80 90

COz 3ields were computed between alternate ports

I

200 0 w t

+ H9O

the usage ratio of H2 t o CO being about 2 over a wide range of gas composition. The rate of the synthesis was large at the beginning of the bed, then decreased to a nearly constant value in the middle section, and finally decreased near the end of the bed or a t a point where the reactants were strongly depleted. In the middle port'ion the rate was nearly constant although the partial pressures of reactants and products changed considerably. I t is difficult t o estimate to Khat extent the integral rate curve was influenced by the temperature gradient along the bed; however, the extent of divergence of the data in Figure 1 i indicates that, the temperature differences did not influence the data greatly. The fact t h a t the extent of conversion a t a given temperature can be correlated with t,he space velocity (Figure 17) suggests that the rate is a function of gas composition. The rate !vas found to be practically independent of total pressure above atmospheric (4). This and the data of Figure 17 indicate that the dependence on partial pressure of the various components !vas slight. The data suggest that the rate-controlling step is some process at the catalyst surface and not the rate of adsorption of reactants which would be strongly dependent upon the part'iai pressures of the reactant?. A mechanism in which desorption of products is the r:ite-determining step predicts qualitatively niaiiy of the characteristics of the synthesis. The application of siiiiplified rate equations, such as derived by Hougen and Watson ( 1 4 ) v5ll be discussed in a later paper. The rather high values of the apparent activation energies probably exclude diffusional processes as ratecontrolling steps. The reaction is very complicated, because the catalyst is heavily covered n i t h the hydrocarbon products, and differences in the nature of the catalyst surface may be a function of the composition of the gas in cont,act viith catalyst. Prob-

0 I I 206%;

Figure 14. Differential Formation of Carbon Dioxide-Bed Length Curves for Tests 8, 10, and 11 with 2H2 to 1CO Gas

The principal primary reaction of the synthesis on cobalt catamay he represented by the equation,

l/n(CHZ)n

IO

192'C:

I O 19Z0C,

BED LENGTH. CENTIMETERS

l?--t.q

+ CO

2

I 0

DISCUSSIOX OF RESULTS

2H2

6

5>

5

100

80 60

g'

50

28 &a 8E

30

9

20

w

40

9

0

IO

0

10

20 30 40 5 0 60 70 BED LENGTH. CENTIMETERS

80 90

Figure 15. Differential Formation of Carbon Dioxide-Bed Length Curves for Tests 4, 5, and i with 3.5R2 to 1CO Gas CO? yields were computed between alternate porfs. Arrows indicate position within bed where pressure of CO bad decreased below 0.01 atm.; beyond these points in the bed Cog was consumed

ably no simple rate equation should be expected to espress the observed rate. The rate of synthesis was about equal for 2H2 to 1CO and 3.SH2 to 1CO gases, but n-as smaller for 1EIz to 1CO gas. With 1 to 1 gas the activity decreased more rapidly with time, which is in agreement with data reported previously (6). This may result from a reversible poisoning by the high carbon monoxide concentration siniilar to. but much less rapid than, the poisoning of iron catalysts in the ainnionia synthesis by water vapor ( 1 , 9 ) . This poisoning effect may be related to formation of g r e d e r

2196


500 400 300

200

0 m V 3

100

0

10

20

30

40

50

60

70

80 9 C

BED L E N G T H , CENTIMETERS

Figure 16. Differential Formation of Carbon Dioxide-Bed Length Curves for Tests 14, 15. and 16 with 1H2 to I C 0 Gas

Vol. 41,

No. 10

gralrter than the equilibrium constant. ' h i s indicates that carbon dioxide was consumed by direct reduction t o methane and not b y reversal of the water gas reaction. However, this interpretation a m m e 3 that the gases reacting at the catalyst surface have the m n e compositioii as those in thr main gas stream. The rcaults on the consumption of cubon diosicle agree wit!, the worL of Fischer and Pichler ( 2 1 ! 26) in which carbon diosidr wax uot consumed i11 tlie synthesis until slniost a11 of the carbon monoxide had reacted. Methane ivas the only h y i r o c a r h I . , f o r n d t)) tile liydrogenation of carbon dioxide. Tliud, the processes that form niethaiie appear LO bc fuiic;ionz ( i f the p:irtisl pressures of liydrogrn and hydrocarbons, : m l proce m s forming carbon dioxide appear to be fuiirt ions of t he partial pressum of water vapor and carhon monoxide. T h e f o r m d o n of hoth rnrtlinne and carbon dioxide ir~cri~:isrd n it#liirrc,rr:i*ing tern-

COz yields were wmputed between alternate ports

quailtities of surface carbide by the gas v i t h high cdJoIl illoiloxide content. The rate appears to be ouly slightly "inhibitd" high partial pressures of carbon innnoside. Some conclusions may be dranii as t o t h t r i a ~ i i r eui tiit: rewLions forming methane, saturated ::rid uiisat urstrd hydrocariions, and carbon dioxide. W t l i 2Hz t o 1(10 g:ts, methane forni:iiiori was large in the first part of the tlrd, passed ihrough :I minimuni, and then increased in the lattcr part, of thc bed. The high >-icxld of riiethane in the first part of thc bed may lixvr been due t.o eithei priiiiary or secondary reactions. ivhile t,he iricreasiiig aniouiits of methane formed in the latter parts of the bed may have b w i i produccd by hydrocraclcing of higher hydrocarbons. With 3.5112 iij IC0 gas the yields of methanc i u u r e a d throughout tlie catalyst hed, : ~ n din tliese experimenb the ratio of hydrogen to c : d w i i nioiioside increased continu:illy tlon.11 tlie e:it:ily$t l)cd Irr experimmts 5 and 7 (Figure 11) there is defnitr c~vitlenceio7 t11c hydrocracking reaction since thi, yields of nirthaiic escecdcil 179 grams per cubic meter of hytlrogeii p l w carl~onniorioxiilp converted, corresponding to tlie reaction 3% CU --+ CH. q- ILO (179 grams per cubic meter is also the mariniuni yield for tlit. rcaction 4H2 $. COn + CH, i- 2Hz0). K i t h 1Hr to 1CO gas [Figur(' 12j the yiVld8 of methane either remained constant 01' :I+ creased dowi the catalyst bed. In tliese esperiments the rAtio oi hydrogen to carbon monoxide decreaaed dorm the catalysi hcd TLir yields of methaiie were higher at higher temperature. Thus, for ench of the three compositions of synthesis gas, the reaction> producing riicthaiie were favored by high ratios of hydrogcii to wlbon Iiioriouide, high partial pre,ssures of hydrocarbon prodiict. in the gas sireani, m i d high temperatures. The ratio of saturated t o unsaturated hydroearbum prrsiii\ ;tl any poini in the cat,al>-stbed varied with the ratio of hydrogeii to carbon monoside in tlie gas a t that part of the bed as slioivn for the C, hydrocarbons in Figure 13. This indicates that olefins ma>-lw :i primary product of the reaction, and that these olefins may then be liydrogeiiated or undergo further reaction to foriii a liy.. droc:irhoii of higher molecular weight at thc poiut of formatinn OI in later parts of t,be c:ttalyst bed. Carhon dioxide formation was chiefly a function of the part id prer;sure of n-atvr vapor in the g nd to soiiie estent of ihr ratios of hydrogen to carbon monoside. M o d of the carbon dioxiile appears to be formed by a eecondiry water-gas reoct,ioii. I2vc.r: with the hydrogen-rich gas, carbon dioside MBS forrnecl i n sid:tt)le quailtities until allnost 311 of the carbon inonoxide had bcwi coilduiiied. K h e n the partial predsure of earboii monoside deci,e:tseti to less than 0.01 atniosphere, the carhon dioxide forint?tl in the reaction was cunsurned with thc forniatiou of methane. .\t t h r point in the bed where carbon dioxide began to be coiisuiiied, tiic quotient of tlie product of the partial presures of carbon ino~ioside and water vapor divided by the product of the partid pressures of carbon dioxide and hydrogen \vas fifty to a huridretl t init+

+

,5!

. .

.-

.

..

.I.

..

.. .

~

.. . .

...

501

Figure 17. Variation of Extent of Conversion of Ht ply6 CO Gas with Reciprocal of Space Velocity at Atmospheric Pressure with Synthesis Gas Having Ratios of Hz to CO of 3.5, 2, and 1

October 1949

2197

INDUSTRIAL A N D ENGINEERING CHEMISTRY

perature. There reactions occur t o at' least some extent over the entire catalyst bed. T h e data in Figures 10 to 16 were extrapolated t o zero bed length, as shown by the broken extensions of the curves. Alt,hough the validity of this extrapolation is rather uncertain, the extended curve.< indichte that a t least part of the met,hane, saturate,d mid unsaturared hydrocarhons, and csi,hon iliosidr are Formed by primary proceasr3. Craxford (8) rt9yvrted data on the formation of inethane and aarbmi dioxide which are similar to those reported in this paper, and attriluted the formation of these components to secondary hydrocracliirig of hydrocnrbons and the water-gas reaction, respectirely. Hon-ever, Craxford (8) postulated that the nwniai synrh& occurs on parts of the surface converted t o cobalt carbide whereas the secondary reactions occur on parts of the surface covered by cobalt atoms that are not actively engagrd in t h e of hydrocarbons. The authors' data indicate thnt thest reactions occurred throughout moat of the catalyst bed, Jvhere the $ y n r h r Gof hydrocarbons was proceeding in a normal m:innrr: I n thrir experiments carbon dioxide was formed in greater amount+ 6-ith 1Hz to 1CO gas than with 2H2 t o 1CO gas. The gas with rhe highw carbon monoxide concentration should favor forrnatiorc of carbide on the surface and, according t o Craxiord's poptulates, should produce less rather than more carbon dioxide. I t has never bccn established that a surface carbide is the active catalyst in the synthesis, and it has been shown that cobalt cstnlysts rather completely converted to carbide were inactive in the aynthesis is! ?I ). Thus, the postulates of Craxford are not a simple nsplaiiii tjon of methane and carbon dioxide formation. The authors' data indicate that these products are formed coiicurrently n ith the normal synthesis by secondary and to wme 9st>eiitby primary reactions, and that their FormRtiorr i q a fiirrntinn ' i f t h e gar? composition a n d trmpw:i,tiire

\(:KYOW- LF;D(:MEhT

I t 16 a ple,ibure tu :ickriowledpe t8heassistance of several perJons in this Kork: H. H. Storch for helpful criticisms of the final manuscript j J. Lecky, A , Dudash, and the crew of operators for assistance in obtaining and computing the data; and A. Sharkey. R. Borgnian. and the niws apect,rnmeter group for the gas nnnlv?is rht;j. LITERATURE CITED

Allrnqui,t and Hlark, J . Am. Chem. Soc., 48, 2814 (1926,.

Anderson, Hall, Hewlett, and Beligman, Ibid..69, 3114 (1947) Anderson, Hall, and I-Iofer, Ibid., 70, 2465 (1948). Anderson, Hall, Krieg, and Beligman, Ibid., 71, 183 (1949). \nderaon, Krieg, Seligman, and O'Neill, Im. E m . C H E M . .39 1848 (1947). C raxford, Fuel, 26, 119 (1947); J . SOC. Chem Jnd. 66. $117 (1847). Crexford, Trans. Faraday Soc., 35, 946 (1939). Ibid.. 42, 576 (1946). Emnwtt and Brunauer, J . Am. Chenz. Soe., 52, 2682 (1930, Fischer and Pichler, Brennstof-Chem., 12, 365 (1931). I b i d . , 14, 306 (1933). I b i d . , 20, 41 (1939). Hall and Smith, J . SOC.Chem. I d . , 65, 128 (1946). Hougen and Watson, IND.ENG.C n a x , 35, 529 (19433. Jerosejev, Runtso, and Volkova, Acta Phusicochim. U.R.S.S.. 13 111 (1940). Pichler, Brennslof-Chem., 2 4 , 3 9 (1943). Storch et al., Bur. Mines, Tech. Paper 709 (1948). Tsuneoka and Fujimura. J . S n c , Chem. I n d . J a p a n . 37 w L ) r ) l binding, p . 363 (1934). [bid.,p. 704. Weller, J . A m . Chem. ~ ' o c . ,69, 2432 (1947), KPllrr, IIofer iinrl \ i i c l ~ ~ : n n ,T b f d , , 70, 799 jl9.&8, R h c h I ' m J u l i 30, 1 B l b . From a. thesis presented by Abraham Krieg t u thr Gniversity of Pittihurgh in partial fulfillment of t h e requirements for t h e iegree of master of science, 1948. Previous articles of this series have appeared in ISD. Eso. CHEM.,40, 2347 (1918); 39, 1548 (1947): and J. A rn phpm. P n r 71. 188 '1949): '70. 2408 (1948): 69, 3114 (1947)

Viscose Processing of Cellulose CHANGES IN BASIC VROP~RTTZES R. L. MITCHELL Kayonirr f n t * o r p o m t e d ,Shdtori. W a s h . Changeb in degree of polymerization, a-cellulose, carboxyl, and pentosan content which occur during the steeping, aging, xanthation, and spinning stages of viscose processing are shown for several pulps of different initial degrees of purity. Special consideration is &\en to depolymerization reactions since the permanent drop in degree of polymerization has perhaps the greatest effect on the rither hasir ana1)tiral propertie9 o f the relliilnse.

I

3ome elfects of time, temperature, and oxygen concentration on the extent and type of alkaline degradation are shown. The presence or absence of oxygen may be the factor determining which of two types of alkaline degradation will occur. A special recovery technique i+ described for retaining low degree of poll mrrization gamma r a n s t i t u e n t s a - the water-in~nluhle n i t t a t i . rlerivatites.

K THE rnanufarture of rayon or cellophane by the viscose

process, certain operations command positions of prima,ry importance---namely : ( 1) steeping, which under suitable conditions serves the double funrtioii of purifying 2nd mercerizing the cellulose starting mat,erial; ( 2 ) aging, which serves in large rneasurc to effect the desired degree of depolyrnerizatioii: (3, santhntion, which t ' o n V w t , * the ot:llulose to the aoluhlv forin: ttnd (4j spinning, which regenerates the cellulose iu :i ilesired physical form more usable than the original. I n t'hese steps, not only is the native cellulose changed to hydrate cellulose of a different physical structure, but many of t,lw hasic analytical properties of the cellulose are modified as well.

Thebe alteratioua in analytical properties result mainly

ri u i r l

.!ravage of the w l h i l n v chain due to alkaline oxidation which m u r - at certain .tLtges of proresbiug, notably in alkali cellu1o.r iging m d i n uitrtrhation. A typical wood pulp- -for example,,

having an initirtl dtgree nf po1yrneriL:ttrun of 1150-is converted by preeeni i;omnieicial prarticr into J, rayon having a degree of pol~merisatioriof 360. hnalj tiral properties of the celldoze itre dfiected LLIV by the ieniu\,tl i n the hteeping operdtion of varying amounts of hemicellulose and of noncellulosic impurities. Cei Gain of the properties are further affected by the physical form and molecular order of the structure laid down in the spinning I

)perntioii

FISCHER-TROPSCH SYNTHESIS

Recommend Documents