Design of a fixed bed reactor for the direct synthesis of dimethyldichlorosilane

. ,

\

\.

.,' ; _{.... _ 'L} J.._

;

/-w

.

t ! " ~ l ,1 1

FOR TEE DIRECT SYNTHESIS OF DIHETHYLDICHLOROSILAlrE

Laboratory of Chemical Technology, T. H. Delft,

January,

1969.

c.

Coulaloglou,

232B, Rotterdamsedijk, Schiedam.

-~ -I INTRODUCTION

-

---. / The purpose of this work is the study of a fixed bed reactor for the direct synthesis of dimethyldichlorosilane.

Although the fixed bed reactor has been replaced in the recent years by the fluidized bed in industrial applications, the low

installation-and operation costs of a fixed bed reactor make it worthwhile to reconsider the problem and detect its feasability.

By using rate data based on a) The Prague School investigations, and b) R. Voorhoeve's thesis at the T.H. Delft and carrying out design

calculations by a method which neglects the radial temperature gradients we reach at the follOldng results :

For a yearly production of 5000 t. (CH

3) 2SiC12 we need two reactors vii th 200 tubes, 2.06 inches diameter.

The operating conditions are

a. Feed rate

=

12 lb mOl/Sq ft, hr b. Feed temp.= 300

°c

c. Conversion= 35% C~Cl and 60% Si d. Pressure =

7

atm.

From the discussion which follOviS v1e conclude that because of the poor heat transfer properties and particularly because of the frequent shut-downs the synthesis of (C~)2SiCl in a fixed bed reactor is not economical-ly attractive.

CONTENTS

--

---1. The synthesis of Dimethyldichlorosilane

1 .1 • The direct synthesis

1 .2. The catalyst and the contact mixture 1.3. The composition of the feed

1.4. The temperature

1 .5. The silicon conversion

Page 1 1 1 2 2

3

2. Kinetics 4

2.1. The catalyst particles

4

2.2. Effect of the copper content on the reaction rate 5 2.3. The effect of temperature and pressure on the rate equation 5

3.

General considerations

9

3.1. General considerations of the problem 9

3.2. Proposed method of manufacturing 11

3.3. Reactor tubes 14 3.4. Cooling medium 14

3.5.

Contact mass-catalyst 14

3.6.

Heat of reaction 16

3.7.

Physical properties 16

4.

Process design 17 4.1. Statement of assumptions 17

4.2. Hall heat transfer coefficient 17

4.3. Effective thenlal conductivity 19

4.5.

The design procedure

4.6.

Pro duction

4.7.

Pressure drop in tubes

I I I

4.8. The minimum fluidizing flovl rate

5.

Remarks and conclusions

5.1.

The heat transfer coefficient

502.

Isothermal conditions

5.3.

Activity of the catalyst

5.4.

The contact mixture

6. Literature Page

23

29

31

32

33

33

34 35

35

36

TEE SYNTHESIS OF DIJlTETHYLDICHLOROSlLANE

1.1. The direct synthesis

The most widely used method for the preparation of Dimethyldichlorosi-lane is the direct synthesis. The' principle reaction can be represented as follows (Li t.1)

Si + 2CELCl Si - Cu (CH) S·Cl

5

3

2 1 2

One of the most characteristic of the direct synthesis is the great number of by products resulting from the reaction of Si with CH

3

Cl,which

can be described by a number of hypothetical equations Si - Cu ( ) 2Si + 4CH 3Cl CH3 3SiCl + CH3SiC13 Si - Cu Si + 3C~Cl CH 3SiC13 + C~C~ Si

+

HCl +

C~Cl

SiCu

C~HSiC12

Si + 2C1 2

Naturally a still greater number of products is to be expected with higher organic halides. The reaction rate and the nature of the products depend on a large number of factors some of which have not yet been inden-tified. Among these are the starting material, the catalyst and the

technique of preparing the contact mixture, the temperature, the pressure, the type of the reactor and the degree of conversion of the silicon and the organic halide.

1.2. The cat alyst and the contact mixture

The best catalyst for the synthesis of dimethyldichlorosilane is

copper. The composition of the products depends to a great extent on the amount of copper in the contact mixture. The greater the amount of

copper present, the higher is the chlorine content of the resulting pro-ducts. When the copper content is increased the reactivity of the contact mixture increases, but the conversion of silicon is decreased (Lit.2).

- 2 '

-In industrial applications a copper content of 5 - 10% is normally used. The contact mixture consists of silicon, catalyst and other sub-stances (co-catalysts, promoters) and is prepared in a mechanical, phys-ical or chemphys-ical process. It has been suggested that the reactivity of the silicon copper contact mixture is connected with the formation of an intermetallic n-phase (Cu

3Si), t~e- presence of which is of great impor-tance for the selective synthesis of dimethyldichlorosilane.

1.3. The composition of the feed

The presence of hydrogen chloride can cause significant changes in the reaction rate and the composition of the product mixture, bccause hydrogen chloride reacts with silicon or silicon copper (Lit.2) to form trichlorosilanes and tetrachlorosilanes faster than that between silicon-copper and methyl chloride. Oxygen and oxygen compounds are undesirable. Small amounts of R2S (0,1 - 0,2 mol/100 molC~Cl) can increase the sil-icon conversion up to 90% and the reaction ráte sometimes by 507~.

The heat distribution in the contact mixture can be improved by dilu-ting the CH

3Cl in N2, but because of the difficulties caused in thc

distil-lat ion of the product the use of N

2 is recommended only to moderate a reaction that is going out of control.

1.4p The temperature

One of the most important parameters in the direct synthesis is the reaction temperature, which must be maintained at the required'level on

the surface of the contact mixture, where the reaction proceeds. It must

also be made sure that no hot spots develop in the agglornerates of the solidphase. In fixed bed reactors, in which local overheating may occur, the reactinn rate and the composition of the product mixture are highly temperature dependent. Another factor which has an indirect but strong effect on these two parameters is the flow pattern of the feed (Lit.1). The reaction between methyl chloride and silicon copper in a fixed bed reactor is generally characterized by moderate selectivity (70% at 3000 and only 50% at 3500 (IJit.1)). At higher temperatures the deposition of carbon and the absorption of chlorine on the surface of the contact mix-ture slow do"m the reaction and may even stop it. Other disadvantages of the fixed bed reactor are the uneven consumption of silicon, and the presence of the spent contact mixture throught out the process,

which strongly catalyses the decomposition of the CH)Cl. On the other hand the reaction of Si with CH)Cl in a fluidizedbed is more efficient-ly controlled and this is the strongest argument for replacing the fixed bed reactors first by stirred bed and then by fluidized bed reactors. Nevertheless many experiments are still carried out on the laboratory scale in fixed bed reactors, af ter special precautions have been taken for the effective temperature control. Other special reactors of rare application in the laboratory and the industry include the riser tube

reactor (Lit.4), the horizontal reactor and the slurry reactor (Lit.1).

1.5. The silicon conversion

The direct synthesis of dimethyldichlorsilane is an interesting process

from the chemical engineering viewpoint, because the abrasion of the

catalyst surface, which is highly undesirable in comparable process, has

here a favourable effect by constantly cleaning the surface of the con-tact mixture so that the deactivation of the catalyst from the deposition

of carbon and inorganic halides is slower. The direct synthesis is also

remarkable as a catalytic process in that a gas-solid reaction is

cata-lysed by a

secon

~

solido

It is expected that the conversion of silicon

during the process will influence the reaction rate and the selectivity. The most obvious consequence of the silicon consumption is the increase in the copper content of the contact mixture resulting mostly in an

increase in the halogen content of the product mixture.

The spent contact mixture can no longer be reacted with the organic halide in an economically feasible manner and is discharged from the reactor. The silicon of the spent contact mixture could be reacted with

hydrogen chloride to give trichlorosilane and silicon tetrachloride, but

the marke~-Ior these is limited (Lit.1). S. Nitzche and R. Rielde (Lit.5) have claimed that the spent contact mixture can be regenerated by adding

fresh copper to it and can then be used for the synthesis of phenyl-chlorosilanes, particularly phenyltrichlorosilane; a silicon conversion -of 98% is said to have been reached in this manner.

:;

4

-KINETICS

2.1. The catalyst partieles

Polarization-microscopic investigations of the polished surface of pieces of contact mixture have shown that the silicon particles are part-ly coated with a layer of the ~-phase (app. CU

3Si), which in turn is partly coated with a thin layer of free copper that has not yet been converted into the ~-phase (Lit.1). It appears that ~-phase is formed at the boundary surface between silicon and copper, whereas a relatively copper rieh outer surface remains. Thc latter disappears almost entirely during the first period of the synthesis converted into the rt-phase. It

LV

is significant that only the copper present i$ theform a Si-Cu alloy

,

participates in the conversion of silicon into methylchlorosilanes. In the process of heterogeneous catalysis we visualize five steps occuring in succession:

a. Diffusion of gaseous reactants to the active sites of the catalyst b. Adsorption on the solid surface

c. The actual reaction on the surface d. Desoption of the gaseous products

e. Diffusion of the gaseous products into the gaseous phase.

In the present case of the direct synthesis of dimethyldichlorosilane a sixth step is introduced which is the diffusion of silicon atoms, through the solid phase onto active sites of the catalyst (n-phase of Cu-Si), where they can react with the CB3Cl. This diffusion probably becomes the rate determining step towards the latter part of the process, aft er muoh siliconhas been consumed but not before (Lit.1). Because of this re-striction, in order to increase the silicon conversion, bydecreasing the proportion of the silicon surface that is coated by a layer of the ~-phase,

_we enlarge the silicon surface by dLTIinution of the particle size, a technique vlidely used in industry. This increases the fraction of the reactive Silicon/~-phase interphasial boundary on the surface of the sil-icon particles (Lit.6). p~ additional advantage of this is that the finer partieles ensure a faster ind_tiation of thereaction between the methyl

chloride and the silicon copper mixture. On the other hand the use of

fine silicon partieles is complicated by other factors such as the large

internal surface area which increase the danger of local overheating and

cracking of methyl chloride resul ting in a lovler selectivi ty.

2.2. Effect of the copper content on

the reaction rate

The relationship between the

cop-per content and the reaction rate is

non linear and beyond a certain

opti-mum quantity of copper the rate

be-Gins to fall~~(Fig.1)(Lit.1).

..

;:j o _I 3 -,-j IJ} tu) ~2 r-l o (J) ~ r-l 1 o El ...

2.3.

The effect of temperature and

Eressure on the rate eouation ~,?O I I

O~--~3--~6----9~--1~2~ Cu in contact mass,

%

VI

Af ter systematic investigations the

Prague School arrived at the

follm-J-ing rate equation (1) (Lit.1).

Figs1. The effect of the copper

content, on the rate of the

reaction

r

=

---\l1here

r = The reaction rate, mol C}~Cl/Kg Cu.hr

k

=

The-rate constant, mol CrsC1/Kg Cu.hr PA

=

'Partial pres3ure of CH

3Cl, Atm.

PB ::: Partial pressure of (CIS)2SiC12' Atm.

K

.

=

~ d t - ... t f' CLT C - A ... -1

rl sorp lon cons~an

0

L3~' Ö~~ •

.t~

IC. -13::: A' asorp lon t - cons t 'an t _OI n (C_nrT ) 4,-(")1 ' t -1

3 2 "l\., 2' il. ffi.

For an infinitesimally small conversion of the methyl halide eq.(1)

-I

. I

\..1

t

where r _o = The reaction rate at zero conversion of methyl halide. The kinetic data obtained by,the Prague School are surnmarized in Table (1).

TABLE 1.

KINETIC DATA :B'OR THE REACTIONS OF ALKYL HALIDES WITH Si-Cu. Temp. , k, TT I~ E DH A 1;' (c) hA .>J a 0

RX Si-Cu

_°c

mole RX/kg ( atm -1 ( atm -1 (kcal/ (kcal/ (kcal/

mixture CU/h x 105 ) x 103 ) mole) mole) mole) CISCI Pure Si+ 280 240 ; 760 530

11 .9% Cu 300 600 ; 680 400 25.3 4.7 20.6 320 1170 570 280 300 625 900 \

-

-Techn.Si-t 280 ! 1000 800> 570 r 10.0% Cu 300 2025 ·280 ) 520 20.1 21.0 - 0.9 ; 340 8000 100' , 460

(c) Over-all activation energy of the initial reaction rate ro:

-. ---~/ tJ--'A ~/

~ V'~p- ?

jY ... ~ .

~ == ~ - ])I{

o a A

The experiments \-lere carried out wi th technical silicon and 101'1 feed

rate of the gas. The maximilm selectivity ''las giv8n by 65;; dimethylchloro

si--lane in t~ product mixture and the pressure was found to have a small effect an the selectivity.

R. Voorhoever points out that the system used in the investigations of Prague School did not permit accurate control of the temperature

.throughout the reactor. He used a differential fixed bed reactor (Lit.6) to determine the kinetics of the reaction betVleen methyl chloride and two kinds of contact mixture with an efficient temperature control. In one case the contact mixture was prepared by t he reaction of copper chloride with silicon. Table (2) shows bis results.

7

-TABLE 2

KINETIC DATA OBTAINED WITH A DIFFERENTIAL FIXED-BED REACTORa

Temp. , Hourly _{Crs Cl} Amount

--,

W/F

~

Expt.

°c

yield, conversion of Si(W), roc

No. g/h (x),

%

g

1 ₃₀₃ 1.85 0.43 68.5 0.21 0.021

2 ₃₂₃ 4.00 0.75 67.1 0.16 0.047

3 343 8.40 2.64 63.5 0.25 0.104

Expt. i Composition of the

pr

oduct

mixture

, w t 4 j

No. _{C C}~) 2SiCl2} C~SiCl3 CC~)3SiCl CH

3SiHCl2 CiCl4

1 76.0 13.1 . 4.1 4.1 2.6

2 74.2 15.2 4.3 4.0 2.3

.3 73.0 14.0 4.7 4.0 4.4

a

The contact mixture was prepared from Si and CuCl b

Fis the feed-rate of methyl chloride in g/h fY"".

--~

c _{Fx/W in g of C}~Cl per g of Si per h.} r't" ... ·~'V"" ro =

From table (2) we see that the composition of the product mixture is independent of the temperature in the range 300 - 340oC. The variation

of~e reaction rate with temperature can be calculated by substituting

the appropriate value for the temperature and the value 26~9 ~ 1 kcal/mol

for the activation energy in the Arrhenius rate equation. In a second series of experiments R. Voorhoeve used contact mixture consisted of a

-comminuted silicon-copper alloy exploiting the advantage that copper

is present from thc very beginning in the catalytically active form i.e.

as the n-phase. Some of the results are reproduced in Fig.(2) and Fig. (3). From Fig.(2) we see the agreement of the rate equation with the Arrhenius

law and Fig.(3) shows that in the temperature range investigated the

temperature has hardly any effect on the comllosition of the product mixture and this must be the result of efficient heat transfer of the

reactor. When the reactor has good heat transfer properties, local over-heating is avoided and the composition

o

Temperature C 350 325 300 275

o

r - - r . . . - - - r - - - r - - - - . - - .

fr-,.;:' _ of the product mixture is basically

_-0.4

,J'

independent of the reaction mixture. By applying eq.(1) at tempera~ure

o

300 C R. Voorhoeve shows that at high conversions the reaction rate reaches a maxL~um at low pressures (Lit.1). Thus for methyl chloride conversions of 60% and 50% at 3000C the maximum reaction rates are reached at pres -sure of about 6 and 8 atm. res

pecti-~-0.8

~

-1.2

1.60 1.70 1.80 1000T-1

Fig.2. Arrhenius curves for the reaction between methyl chlorid.e and Si-Cu.

vely. This is because the products are absorbed on the surface of the contact mixture, and thus the number of active sites on it decreases

-at higher pressures and conversions (Lit.1). The adsorption is still stronger at lower temperatures, and the maximum reaction rates coincide wi th stilllovier pressures. Such maxima are responsible for the fact that the methyl chloride conversion in industrial reactors is lirnited to about 50% and the optlinum pressure is about 6 atm. At higher con-version or higher pressure the selectivity of the industrial process is decreased.

---Product composi tion, %1-1

100'r-~--'--'-ïï--.---r--'.-~~-r--~ 80

"

0 lil 8 0 60 ~.le2SiC12 40 20 NeSiC1 3

,

0

•

~ ~

.,

280 290 300 310 320 330 340 350 360 370 380 'l'empcraiure

°

c

Fig.3. T1..,ü ' -o effect of temperature on the com

po-si tion of th8 metr.ylchlorosilane prOduct m ix-ture obtained at 4.6 at:l.

9

-GENERAL CONSIDERATIONS

3.1. General considerations of the problem

Before carrying out the reactor calculations it is necessary to have enough rate data in the form we can use in the design equations. The rate equation according to Baza~t et al is

~~---~--- (1 )

The molal flow rate of CE)Cl entering the reactor is F. At a point \-lhere the conversion is X i t will be NA

=

}1 (1 - x). The molal rate of

(Crs)2 Si(CI)2 is NB = ~ and the total floltl rate Nt = F(1-x) + F ~ =

F (1 - ~).

Assuming that the perfect gas law holds and calling P

t the total pressure, we have NA 1 - x PA - Nt Ft = Ft 1

-

x 2 x PB 2 _x Ft _{= 2-x} x Ft 1--2

Hence the rate equation (1) becomes

From table (1) we At 2800C k=1000 At 3000C k=2025 At 3400C k=8000 kK 1 - x A 1

_-

-

x r -_A- 2

(1

+

J

K 1-x A -_1-2S. 2

have for contact K =: 800.10-5 A -5 K A= 280.10 _t:; K A

=

100.10 .-I Ft

ptf

Ft +

K:s

_{2 -}x _x mixture consisting

-3

K_E

=

570.10

-3

K...,= 520.10 15

K:s=

460.10-

3

(2 ) of Techn.Si+10% Cu

By plotting the reaction rate constant versus temperature in Fig. (4) we get a straight line \'Jhich is expressed by the usual Arrhenius rate

6 ~ ~ 0 r-I 5

,

,

4 3 2

,

_,

,

1.6

1.8

2.0

2.2

1 T· 1000

Fig.

4.

Plot of the reaction rate constant, versus temperature

at

7

atm.

equation.

9 12·16~

k

₌

5570.10 .e

-

T.

(3)

k

=

rate constant in mol _{CIS Cl/Kg Cu., h.}

T = absolute temperature oK.

The change of the adsorption equilibrium constants with the temperature is _shown in fig.(5). We see that

KB

changes slightly with temperature at the range of 280 - 340oC: Rence we can assume that it is constant

, -1

and equal to about 0.500 Atm. •

For P

t

=

7 atm~

k = given by eq.(3) or fig. (4) KA = by fig. (5) •

We calculate the reaction rate for different temperatures and conversions (see tables 3,

4, 5,

6 and 7). The results are plctted in fig. (6).

---'-~---- . ___ ---.l___ _

I

909

\ ~ _\ I • S op <Xl r<'I 0 ~ • ~

..

-

,,_

_

- - - - ---

K:s

-

-

-... I • El op <Xl 300 lî'I 0 ... • ~<Xl 100 280 290

Fig.

5.

Effect of temperature on the adsorption constants KA'

KB

at P =

7

atm.

o 0

We observe three regions of the rate curves in the range 280 - 340 C

(fig. 6). In the first region of the curve the reaction rate drops till 300oC. This is owed to the decrease of the adsorption constant KA and

i

the small increase of the rate constant k. At the _second region

!

~~~~i:n~3~:~)

t:::

::::~:nb::::S~n::e:::: :::::l:/::::~:~:r::

~:Wh:::-a steep increase of tne rate constant and only a small decrease of the

adsorption constant (see fig. 5,6).Further the reaction rate drops once

more because the adsorption constants become very small and cannot be

balanced ,by the increase of the rate constant. In addition to the drop

of the reaction rate above 3300C the selectivity of the reaction is

smaller too and above 3500C CH

3CI is decomposed extensivily. This is

obviously the reason 'n'hy the optimum temperature conditions lie bet1>leen

, 0

280 and 340 C.

3p2. Proposed method of m~rillfacturing

o

Methylchloride \'lill enter the reactor at a temperature of 300 C, a pressure of 7 atm. anci a molal rate of :E' = ~ lb mol/ ft 2 ohr.' ,.;hich will

" .. ' .~ ,.. '.

"(

12 -TABLE

3

TOe KA k _{KA·Pt IVt}

_~

-Atm.-1 mol

_e~el/kg.Cu,h

-1

Atm. 280 80.10-4 1000 560.10-4 0,236 50,0-10-3 290 50.10-4 1370 350.10-4 0,187 500.10-3 300 28.10-4 2025 196.10-4 0,140 500.10-3 310 23,5.10 -4 2950 164,5.10 -4 0,128 50,0-10-3 320 18,9.10 -4 4210 132,3.10 -4 0,115 500.10-3 330 14,4.10 -4 5970 100,8.10 -4 0,100 500.10-3 340 10,0.10 -4 8000 70.10 -4 0,084 500.10-3 TABLE 4 1-x 1-~ 1-x 2-x x

Fr

x 2 _1--x 2-x _1-~ 2 2

°

1 1 1 2

°

1 Q,1 0,9000 0,9500 0,9473 ,1,9000 0,0521 0,9738 0,2 0,8000 0,9000 0,8888 1 ,8000 0,1111 0,9419 0,3 0,7000 0,8500 0,8235 1 ,7000 0,1764 0,9074 0,4 0,6000 0,8000 0,7500 1,6000 0,2500 0,8658 0,5 0,5000 0,7500 0,6666 1,5000 0,3333 0,8164 0,6 0,4000 0,7000 0,5714 1,4000 0,4285 _, 0,7565 ! I

provide the reactant gas vii th the appropriate residence ... time for a con-version of about 30 - 35%. The Si concon-version is chosen to be about

60 - 65~, although a more realistic one would be a silicon conversion of about 50%. The temperature of the reaction must not exceed the level of 3400e because of the low selectivity and the decomposition of e~el above

this temperature. We assume that the selectivity is of the order of 80% in the range280 - 3400e and this is in agreement with the experimental results of R. VoorhoevG (Lit.1), altbough it was claimed that the selec-tivity drops at the e:ld of the process af ter a high silicon conversion. (Bazant et al)

i-x k.KAeP t x x 12" 0 1 0,1 0,947~ 0,2 0,888E 0,3 0,8235 0,4 0,7500 0,5 0,0666 0,6 0,5714

!

M

V

KAPt x I _; - _x ;

1--l

-

2 I I 0 I

I

i 1 0,1 : 0,9738 I : I

°

,

2 I 0,9419 1 1 _{0,3 i 0,9074} 0,4 1 0,8658 0,5 0,8164 0,6 0,7565 TABLE 5 280°C 290°C 300°C 310°C 320°C 330°C 340n '"'C 56,00 47",95 39,69 48~52 55,69 60,17 56,02 : I : 53,00 45,42 37,59 45,96 52,75 56,99 53,12 ; 49,77 42,61 35,27 43,12 49,49 53,47S 49,84 46;11 39,48 32,68 39,95 45,86 49,54 46,13 42,00 35,96 29,76 36,39 41,76 45,12 4290Ó 37,33 31,96 26,45 32,343 37,12 40,10 37,38 I 31,99 27,39 22,67 27,72 31,82 _34,38

I

32, C~-280°C 0,236 0,229 0,222 0,214 0,204 0,192 0,178 1

\

_I i \ l TABLE 6 K P

1.::2s

A t 1_2S. 2 290°C _300°C

I

_310°C

I

0,187 0,140 0,128 0,182 0,136 0,124 0,176 0,131 0,120 0,170 0,127 0,116 0,162 0~121 0,108 0,152 0,114 0,104 0,141 0,105 0,096 I I

i

j 1

--320°C 330°C 340°C. , 0,115

_i

0,100 0,084 0, 111 I 0,097 0,081 0,108 0,094 0,079 0,104 0,090 0,076 0,099 1 0,086 0,072 0,093

I

0,081 O,Oó8 0,086 1 0,075 0,063 ' 1

I

- 14

-TABLE 7

r. _{mol CE)CI/kg Cu, hr or lb mol CILCI.1 0}

3/

_{lb C} hr

j u, TOC

₂₈₀

₂₉₀

₃₀₀

₃₁₀

₃₂₀

₃₃₀

₃₄₀

x

0

3

6,67

34,05 30,5

38,1

4

44,85 49,72

47,78

0,1

33,15

31,13 27,84 34,76

40,83 45,24

43,36

0,2

30,49

28,11

25,03 31,20

36,55 40,48

38,78

0,3

27,20

24,95 22,14 27,57

32,23 35,66

34,10

0,4

2

3

,76

21,70 19,16 23,

9

4

27,8

30,77

2

9

,37

0,5

20,21

18,36 16,12 20,02

23,39 25,76

24,51

/

0,6

16,50

14,80 13,01

16,13

_18,8

_I

20,7

19,65

3.3.

Reactor tubes

The fact that it is necessary not to exceed the mentioned temperature

l~vel in the catalyst bed and that .the reaction is considerably exothermic, leads us to the choice of small diameter tubes with better heat-transfer properties. It also means that the radial temperature variation will be less.

2.06

inches internal diameter

40/

₂₁

NichelChromium alloy steel

tubes were employed for the reaction duty because this material has superior

strength properties at elevated temperatures and good corrosion resistance to organic halides.

3.40

Cooling medium

'I:Y' 0

.~)'"' .çI'" Stearn of about

8.5

atm at a temperature of

175

C wi11 be used as cooling

,

~

medium.

3.5. Contact mass-catalyst

The contact mass consists of

0.50

mm diameter particles containing

10

%

copper. \'Ie assume that the catalyst is present from the beginning in·the form of Si-Cu alloy (C~Si). Af ter about

60 - 65

%

Si conversion the contact

..

(5 ~ "-r<'I o ..-c r-l o r<'I ... >-'-< o rl o S ~ , f 50

1

/

/

~

.

_x=o 46

//

/

~

x=0,1 , 42 I

(

Ij

/

/

~

x=0,2 _/ 38

'~

L_- •.

-

/

,//

/

/

-34

/t

~

~

~

11/

_1/

/

X7

30 ~, I / "

""'"-.

~

~

l/

_/

_{/ '}

_;5

/x=O,4

r""

~

/

/ " .;'/ ,,/ 26

/

~

""

_>/~

/

~/

/ " / ...

~

. /

p~

VP ""'" ,// -22

-

-l><

V

~

, t> 18 _, 14 280 290 300 310 320 :530 340 350 360 Temperature

oe

Fig.; 6. Plot of reaction . rate against temperature at 7 Atm.

I

I

!

.. j -

-

--- ---

- -

-16

mass will be discharged and replaced by a new one. The porosity of catalyst was estimated to be equal to 0.5.

3.6.

Heat of reaction _,

The complexity of the reaction and the side reactions occuring during the process make it difficult to ~stimate with great accuracy the heat of reaction. For the present conditions the heat of reaction is estimated to be about -92.000 cal/mol Si' which is equal to -46.000 cal/mol CH

3 Cl (Lit.1).

3.70

Physical properties

The physical properties of the reactant and products were estimated at

o

\

- - - -

17

-PROCESS DESIGN

4.1. Stat ement of assumptions

The following assumptions are made

a. Smooth variation of the properties b. Eddy diffusion dominant

c. Axial diffusive process neglected

d. Insignificant temperature varations in the radial direction across the diameter of the reactor.

e. Heat transfer coefficient at the wall and the bed.

4.2.

Wall heat transfer eoiffieient

I

Leva's correlation may be used for this purpose (Lit.7)

cooling

, -

--

'

where h .d t .W k g 0.7 d

.e-4,6.~

h =

w wall transfer coefficient, Btu/ sq ft.hr. 0 F

d t k _g d P G

tL

For -= = = -= tube diameter, ft. CH 3Cl thermal conduetivity, part iele diameter, ft

superficial mass velocity, CH 3Cl viseosity, lb/ft• hr d t

=

0,172 ft d

=

0,0016 ft P ~

=

0,050 _{lb/ft •hr} The Reynolds Re

=

~

=

12 x 50,5 • 0,0016

=

P

0,050 19,4

I

I

I

- 1$ -and d

p/

_d

=

0,0093 t / Hence eq (4) becomes 3,5. k (19,4) 0,7 " 3,5.902.8 hw

=

d t .e-

f.

6•O'0093

=

0,172.1,043

=

3,15 hw

=

3,15 _Btu/s ft.hr.oF

---g---I

The wall transfer coefficient is a critical quantity in the design method

and small changes in it can cause large difference in the conversion.

The use of Leva's correlation for this small Reynolds number is a

question-able procedure and only experimental resul ts ,dIl provide us wi th an

accurate wall transfer coefficient. However it is expected that the wall

transfer coefficient will have a low value not only because of the small Reynold number, but also because of the very smal 1 partiele size which decreases the size of the eddies ànd the distance over which each mixing

process occurs. Also, there is a large number of more or less stagnant

films between the fluid and the solid partieles, vlhich the heat must cross

in reaching the wall. Colburn~ (Lit.8) claimed that.the ratio of heat

transfer coefficients varied with the ratio d

p/d

value of about d

I

=

0,15 (TabIe 8). t

p d t TABLE 8 I·~ :, i reaching a miximum

J

~

:>

Ratio of heat transfer coefficient in packed- (h )and empty (hi)tube (8).

P I d I P/d 0,05 0,10 0,15 0,20 0,25

_!

0,30 t h P/h. . 5,5 7,0 7,8 7,5 7,0 6,6 l

4.3 Effective thermal conductivity K

e

The effective thermal conductivity is estimated by the analytical method proposed by Vl. Argo and

J.

Smith (Lit.9).

a. Radiation contribution K'

r "

e Ta3

K'

= 4·---·d .0,173

.r 2-e p 1004

T

=

mean temperature of the bed a

e = emissivity

d = part iele dia~eter

P K'

₌

_4-

ih2.

_{.0,0016.0,173·} ,1070₈ 3 r 1 ,5 10 b. Point eonductivity K' p

=

=

= = K logK'

=

-1,76 + 0,0129·-ä p e (5) 1070 oR 0,5 0,0016 ft 0,004 (6)

Ks

=

particle conductivity == 48,4 Btu/sg.ft.hr.oF/ft

e

=

void fraction=0,31. For d

1

=

P d 0,0093 e = 0, 31 ( Li t . 1 0 ) t ~ r 6 logK_{;= -1,76+0,0129. 0}

,

3

1

=

-1,70+0,0129.15 ,1

=

0,254 K' = 1,8

-12 _____ _

c. Coefficient h between pellet and gas

c For Re <350

.

2/

3

.

(d . G) -0. 51

(~)

(?;)

= 1.95 ..lL- (7)

~

I '1

20

-i

=

0,33.Btu/lb~F.

c == mean specific heat of reaction mixture

p

i

(c ; == specific heat of C~Cl _{at constant pressure== 0,38 Btu/1boF}

P1 )

c _{= specific heat of (CEJ)2 SiC12 at constant pressure} =

f2

0,20 Btu/1b0:E').

-..

G =:superticial mass velocity

=:CEJC1 viscosity

== 606.1b/ _sq. ft _• hr

= 0,050. 1b/ft • hr

kg

=

CH)Cl conductivity _{== 0,020 BtU/Sq ft.hr OF/ft}

By replacing in eq.(7) we get

I

~

)

2/

3

"

~~h~c~~ _ 0,33.0,05 == 1 ,95 (19,4) -0,51 0,33 x 606 0,02 and d. Coefficient hand h r p 0,33.606.1,95 h = =94 c 0,86.19,4.0,51 h _c == 94

The calculation of the radiation and point-to-point contact coefficient

from eq (8) and (9) is a trial- and error procedure, since both

expres-sions involve the summation coefficient h.

h == K' (2K' T hd ) r r s p d .K P s d .K P s

KS = particle conductivity

=

_{48,4 Btu/sq} ft.:b..2.~. oP/ft

(8)

!'

We assume that h = 2445, then eq (8) becomes:

0,004(2.48,4 + 2445.0,0016)

and from eq.(9)

h _r = ---

=

5,2 h = P 0,0016.48,4 h _r = 5,2 1,8(96,8 + 0,0016.2445) 0,0016.48,4 = 2.346,4

Since the sum of h _c + h _r + h _p = _{_} 94 + 5,2 ~ 2346,4 = 2445,6 , the original assumption of 2445 for h is satisfactory.

e. Knovling the coefficient h at the surface of the partiele, the series contribution to K can be evaluated from eq.(10).

e

h.K .d K'series

=

s p

=

2K + hd s l? 2445.48,4.0,0016 188,3 =2.48,4+2445.0,0016

=

100,7

=

1,87 (10)

f. Xurbulent d1ffusion contribution Ktd

The turbulent diffusion contribution is evaluated from eq (11)

-

.~ ~. d .c .G

K.

td

=

(~e

m)

e.

d .u

n

Pem

=

Average Peclet Number = ~

De

=

Effective diffusivity

;

- ,

~- - - _ .

-I

- '

22-d .G

From R.A. Bernard and R.H. Wilhelm (Lit.11) for ~ = 19,7

and d / _{P d = 0,0093/} t / Pem

- - - =

9

d 2 1 + 19,4 (

p/

_{dt )} Pem

=

9(1

+

19,4.0,00932) = 9.1,0016

=

9,014

Rence eq.(11) becomes

d.c G 0,0016.606.0,33

Kt~

=

~em:d

=

9,014.0,31

=

0,107

g. Molecular conductivity contribution K '

c

Xhe thermal conductivity of CE3C1.at 315

°c

is about

h. Effective thermal conductivity k~.

K = eek' + Kt'd + K' ) + (1 - e) K'series --e g r =.0,31 (0,020 + 0,107 + 0,004) + 0,69.1,87

=

0,041 + 1,3

=

1,34 Etu/ ft h 0F/ sq • r. Ft K = 1,34 Btu/ f t h 0F/ e sq • r Ft

~'

T

.,

--

---j,...~-I

_I ,\ , , \ j ' L

I

I

/

23

-404. The over all heat transfer coefficient

The heat transfer capability of a fixed bed ofregularly shaped small size solids may be expressed in the form of an over all transfer coefficient from eq.(12) (Lit.12).

I

I

(12 )

,

I

i

U = over all coefficient Btul 0

sq ft.hr, F

h = heat transfer coefficient for cooling fluid outside of wall

o

r

=

radius of bed solid, ft .

Btul _sq f+-_IJ. h _r, oF

?

We assume that the outside of the wall Rence from /éq. (12)

heat

~~e

is negligible.

.:::.

!

t

0,086

U 3,15 + 4 x 1,34

=

0,326

U = 3,1 Btu/c _{oq .} ft _• hr _, OF

4.5.

Thc design procedure

At this point we have enough data to proceed to the evaluation oI' the conversion, the temperature profile along the reactor and the size of the_. reactor.

The rela.tionships required are

a. The basic design equation r.dVI = F.dx

or (13)

S r () Q' Z = :E'. dx

CO ·lBo

r

=

rate

.

Jj,

J

(-

24

-I

equation,lb Cu, per hour W = mass of the solid catalyst, lb F = feed rate, lb mol/

hr x = conversion

z

=

longitudinal distance in reactor, ft Sc= cross sectional area of

t~be,

ft2

eB= Bulk density of solid catalyst, lb/ _{cu. ft.} b. The rate equation

In fig. (6) the reaction rate is shown as a function of temperature and conversion.

c. The energy balance taking into account the heat transfer to the reactor wall

F·dx· (-DI-I) - ·U.S - (tD-tv,)-dz = Sm.c .dtm (14)

R l Pi

DH

= heat of reaction, Btu/_{lb mol oF}

2

S = interna_{l surface of tube per ft, ft /ft} R

hl = wall temperature, oF

tm = bulk mean temperature of reaction mixture, oF mi - weight rate of flOiv of component i, lb/ hr

c = specific heat of component i at constant pressure, Btu/1b*OF

Pi

The bulk densi ty of the catalyst

e

B

is found as follOi-Js Thc.density of the contact mixture particles

e~

= Ö.(0,1 eCu + 0,9 eSi)

ó

= porosity of particles

=

0,5

~....;--eCu= density of copper = 555,6_1b/cu ft

.eSi= density of silicon

=

151,71b_{/ cu} ft

e~ =0,5-(0,1°555,6+151,7-0,9)= 96 _{lb/cu ft} I 1 J I j f ,\ " I,

________________

---_______________

~I

---The void fraction of the bed e = 0,31

Rence the bulk density of the contact mixture partieles ec is

ec = e~·(1-e) = 96.0,69 = _{66,3 lb/cu ft}

and the bulk density of the catalyst (10% copper)

The heat of reaction is DH=-46.000 cal/

g

mol

CB)CI 0C'

which in British units is equal to

-46.000 Btu/1b _mo. 1 CH-C _{"5 1,} oF

The wall transfer area/per unit length of tube SR = 0,541 sq ft/ft ';: il.I ...

The cross sectional area of the tube

2

Sc

n~d

= 3

_à

14 = 0,1782

=0,023~

sq.ft

The molal rate/per tube

F = 12·S

c

=

12-0,0232

=

0,278 lb mol/hr

The heat capacity of the reaction mixture Sm. cdenends on the tempera-_{1. p. _ ..} ture and conversion. Por an average temperature l of 31SoC and conversion x

+ F • liill • ~. c

2 2 P2

rill

1, lffi2 = Holecular weight of C~Cl and (CI~)2()S1 C12 respectively

By substi tuting these values in the desibJTI equation (13) we get

,

or

dz = 1 80·d.x _{, r}

26

-(A)

The energy balance equation (14) may be written

,t

0,278-46.000-dx-1 ,8-3, 1-0,541 ~ (tm-tw) ·dz·1 ,8=0,278- (19,2-7,7') ·à.tm-1,8 (The

ture

conversion factor 1,80F/ OC have been introduced so that the tempera-may be expressed in degrees centigrade)

or

For conversion x = 0 the energy balance eq~ation becomes

.12,788·d.x-1,68-(tm-tw)-dz = ·5,34 (-lh .. , for x = 0,05 12.7SS·d.x-1,6S-(tm-tw)·dz = 5,23 for x = 0,10 12.78S·dx-1 ,68· (tm-bl) -dz = 5,13-dtm for x = 0,15 12.788.d.x-1,6S-(tm-tw)-dz = 5,02-dtm

Proceeding in this way we can get the form of the energy balance equa.tion for every conversion.

Equations (A) (B) and fig (6) or eq (2) are the design energy and rate relationships necessary to solve the problems. A numeri cal approach is indicated ivhich is carried out in the following llay

1.

An

incremental value of conversion dx = 0,05 is chosen 2. A temperature at the end of the increment is assu~ed

3. The rates are obtained at the beginning and end of the increment from fig_(6)

4. The increment dz is computed fr om eq (A) 5. The assumed temperature is checked in eq (B)

,.I f .1 /. ..

f

.

-a...J~)

·

I -.

:5

':.,.J ... 01

I

27

-The design equation (A) may be written in a simplified form considering the average reaction rate in the increment dx = O,OS

dz

=

1,s-0,oS·{l +

1) ·

-21 =

\r

1

r

2

dz = 0, 04S -(

1

_r +

1 )

_{(A1 )}

1 r 2

r

1, r2

=

the rates at the beginning and end of an increment dx=O,OS Illustrating these calculations we start with an increment dx = O,OS

First increment It is assumed that ;tm 1 = 303 oe Hence dtm = tm l - 300° = 3 0 e

From fig_ (6) the rate at the beginning of the increment, zero conversion and t = 300 oe is 30,4-10-3 and at the end of it, t = 303 oe is 31,3-10-3 , then in eq (Al) dz = 0,045- 1 + 30,410-3 1 - - ' - - - = 2,91 ft 31,3-10-3 In eq.(B

1) we check the assumed value of tm1

/

"

-

;.'"

.:

'1

,

,-

.

0"",,.,.,... :J, 'ol 12.788-0,05 - 1,6S-(300-17So)-2,91=5,34e dtm= 119,7 - 0,32-12S-2,91 = tm 1 - 300 - -tm1 - 300 = 3,1 tI!l1 z: 303,1

(versus the assumed value of 303°C)

Second increment

I

,

t'

~,34-(tm1-300)

At' the end of the second increment we assume Tm~

=

306,S

-Then from fig.(6)

r

1 = 31,3-10-3

r

2 = 32,2-10-3 and eq.

A,

gi yes

+ __ 1.:-_ _

=

2,83 32,2-10-3

Checking the assumed value of tm

2 in eq, (B2), 5,23.dtm = 12_788 - 0,05 - 1,68-(303-11'5)-2,83 dtm = 122,2 - 0,322·128-2,83

=

3,50C

tm2 = 303 + 3,5

= 306,5

(Versus the assumed value of 306,5)

Continuing the computation give the results summarized in Table (9) and shovln in fig. (7)

TABLE 9

Convers ion x TeI:1perature

_°c

Catalyst bed depth_ :ft

°

300

°

~ '\ 0,05 303 - 2,91

-

0,10 306,5 2,83 ~ 0,15 316,5 2,62 Ó,20 ~, 331,5 2,'35 '. I 0,25 336,5 2,34

)i

i I 0,30 _, 328,5 2,48 /~ 0,35 290 3,20 • '7 ~

\

0,37 (265 ) (4,8) \ " )

.

- . _ - 29 -. / 340~ ________ ~ __________________________________________ -, 330 320 310 300 290 2 ₄ 6 8 10 12 14 16 18

Catalyst bed depth-ft.

Fig_ 7. Longitudinal ternperature profile of the bed. 4.6 Production

i } ' It is obvious from fig_ (7) and' table (9) that the reaction slows dovm

/' for CH

3Cl conversions greater than 0,35. The length of the tubes for this

~ c:> conversion is equal to 18,73 ft _ A reactor consis ting of 2,06 inches tubes,

.--ft high operating 260 days/year (about 1 day in operation and 9

r-

18,73

~'Á-hours shut-down) the yearly production is estiluated as follo\vs

The mass of contact mixture per tube is

Mc= Vte c

=

0,434-66,3=28,78 lb/tube and

MSi=the mass of silicon in a tube=28,7S-0,9=25,9 lb/+ b

For 60% silicon conversion the mass of silicon converteN- rs 1'1si = 25,9. 0,6 . ) = 15,54 lb

or

Msi

=

1,11 lb mol

30

-The amount of C~Cl to convert 15,54 lb·Si is

"IK M .-2 h_{CLLC};: Slo} j 0,35 · 1,11·2 = 0,35

=

6,34 lb mol ~.{,Z ;r {L~ ,~v 'I

For a feed rate 0,278-lb mol CB)Cl/ hr

b2i

'

it will take about h

=

0,278

=

23 hr to convert 60% of Si.

For a selectivity of 80% the production per tube per hour will be

p' = 0,278 -0,35-0,8 = r 2 or 0,0389-115 = 4.473 lb/ 1 lr, tube or P ' _r = 4,473-₁₀₀₀ 453,5 _{= 2,03 kg} / _{hr tube} .

,

And the yearly production per tube

P _r

=

260-24·2,03 = 12.670 _. kgf _year, tube

'"

.

A

reactor with 100 tubes will produce P =·100·12.670

=

1.267.000 kgf .

r year

and with 200 tubes

P

=

200·12.670

=

2.534.000 kgf

r year

In a reactor of L

=

15,53 ft the C~Cl conversion will be 30% Then

h ·= 22 hr

p'-

_{r - '}

1 74 k / _{g hr tube}

,

A reactor with 100 tubes will produce P _r

=

1.086.000 kgf _year

31 -and with 200 tubes

P _r = 2.172.000 kg! _year

Because the synthesis of dimethyldichlorosilane is a semibatch process it is necessary to use at least ~wo reactors, even it is expected that the use of three reactors will secure smoother operation of the whole

unit_

_/

4.7- Pressure drop in the tubes

Using Kozeny's equation for laminar flow we evaluate the pressure drop in the tubes (Lit.13)

2 -1 DP _ ~G .a •

L

L - 2 a·u g"e·e~",) l ~'~~J..~tr for

SL

.4

00

a.~

'\

~)

(15)

r-

-e

---

<

v

> \)

P <e. v., D~ . . - - - " ' : .

1.

G (1_ ~ ) b'1. !:( \ -.

J

.::. ~ G - 606 lb! _sq fot' _.nr

p

= oO,05 -lb!ft.hr _~e'

=

densi ty of C~CI =

.

=

0,31 g = 4,18.108

a

₌

specific surface of the bed .sq _{ft! cu ft}

=

§JJ-eJ

D a

₌

Hence

6

{1-0 221 }

₌

2587 0,0016 606 G a~ 2587-0,05 P sq _{ft/cu ft.} =

4,7

L - 8 2 •

=

175 psf! ft

~p

5-6062.2.587

(606-1

4,18-10 .0,45·0,31 2587·0,05

DP

178 ~

=

144'= 1,2 _psi/ft DE

=

1,2-18,73 = 23,4 psi

rfJ~

rwJ

vr~lr

T:'AS

çJp

c>

P "/

re,

cU-

,L.cJ,2';'/~

~),

(>_1)1-

"l,,\;e)

y',~

)"'7'

I

4_8 The minimum fluidizing flow rate

-The CB)Cl will pass upward through the bed. In order to check that the gas velocity is less than the minimum fluidizing velocity C we must

m

estimate it for the conditions under which the reactor

operaf~s.

Using equations (16) proposed by Hiller and Logwinuk (Lit.13) we have

= 5,23°105·e1,1·(ep-e)0,9.dn2 ~

e

= density of CH

3Cl=0,45 lb/cu ft

e

p

=

density of particles ~~lb/cu ft

D = 0,0016 ft p ~ = 0,050 lb/ ft.hr Rence 5,23.105.0,451,1.(95,55)0,9.256.10-5 0,050 (16 )

=

5,23-2,56.0,415-60,5 0,05 = 336,18 0,05 = 6.723 lb/ sq ft h , r.

33

-'

-REMARKS ANTI CONCLUSIONS

From the previous discussion we dravl the following conclusions:

5.1. The heat transfer coefficient

This critical parameter for the design of a fixed bed reactor was

found to be low at the present operating conditions (hw=3,15 Btu/Sq ft,hro:. Although Leva's correlation for the wall transfer coefficient is a

questionable procedure for not provide us with enough

(d

I

= 0,0093)

p d

low Reynolds numbers, the litterature data for so small d and d

I

ratios

P P d

t

does

t

In an attempt to improve the heat transfer properties of the reactor by using higher gas velocities the results we got were not satisÎactory,

(For F= 24 lb moll _sq fi: _{_,} h the h = 5,1 _{r w} Btul _{s q ,} ft h) probably, because _r

of the very small particle size of the contact mixture. On the other hand the conversion was low (about 10-15%) and the pressure drop very high. The length of the reactor necessary for 0,05 conversion vlas of the order dz = 6ft . .

Another mean to irnprove the heat transfer properties was the use of smaller tube diameter. Here follm"l the resul ts we got

For

F = 10 lb mol CELCI/ _{--5 sq ft, hr} d

t = 1,61 inches Feed temp.= 2950C

Cooling me~ium temperature 2000C The total length of the reactor

L = 15,85 ft

The conversion = 30% CB)CI and 60% Si The production p'= 5.500 kgf _tube

r year,

and for a yearly production of 1.5000000 kgf we needed a reactor

year

vli th 273 tubes, a considerably expensi ve construction.

I

/

./

o

0,1 0,2 0,3 0,4 0,5 0,6

Conversion x Fig. 80 Graphical integration of the design equation at

T = 3300C and P

=

7 atm.

--It seems that we could improve the heat transfer properties of the reactor by introducing an inert gas (N

2) in the feed stream. We have

already seen (Par. 1 .3) that the use of N

2 \vould cause difficulties in

the distillation of the product and the acceptance of this method cannot be predicted without taking into account the separation problems.

5.2 Isother!1".al conditions

If we"could operate the reactor isothermally, at the temperature of higher reaction rate T = 330oC, something which is qui tOe unrealistic for

the direct synthesis of (CH

3)2 SiC12, the length of the reactor for

COI1-version x= 35% C~Cl eQual to that of the examined problem is found as

follows:

35

-7 W

_jO,35

_d.x \'.,/ F -

- -

_r S _~ ~/ (17) JvV 0 . À YJ'

/ "

'it ~...r ('w,.JU-'\ S _{t ,25 '33,7 ;-}

~~2

hrl lb Cu lb -mol C~Cl and ~ . " \ W = P-8,42 = 0,278·8,42

₌

2,340 lb Cu

The bulk density or Cu

eB

=

6,63

lb/

cu ft

...

\

Hence the volume

6tf

the tube is

~ \

V_t = 6,63 = 0,35~U ft

and the length ~Of the' reactor _ ~ _ 0,353

L - S - 0 0232 = 15,2 ft ~

c '

503

Activity of the catalyst

It is expected that the conversions will be lower than that predictod by eq. (2), because of the uneven consumption of· silicon and the decrease of the activity and selectivity of' the catalyst vlith time.V.S---Fikhtengol' ts and A .L. Klebanskii claimed that in the fixed bed process the relative

amount 'Of dimethyldichlorosilane drops from 60 to 30% in the course of

20 hours (Lit.14).

5.4. The contact Dixture

This which makes the fixed bed reactor quite unattractive for the direct synthesis of (C~) 2SiC12 is the frequent shut-dm'm for removal of the contact mixture. This becomes the overriding cons~deration for rejecti LG

LITERATURE

==========

1. R.J.H. Voorhoeve, "Organohalosilanes Precursors to Silicones", Elsevier Publishing Company, Amsterdam, 1967.

R.J.H. Voorhoeve, Thesis, Delft, ,1964

2. E.G. Rochow, J. Am. Chem. Soc. 67p 963 (1945)

3. P.G. Dudani and H.G. Plust, Nature 194, 85 (1962) 4. C.E. Reed and J.T. Coe, Chem.Abstr. 40, 1536 (1946)

5.

S. Nitzche and R. Rielde, Chem. Abstr. 57, 13.803 (1962)

6. R.J.H. Voorhoeve, B.J.H. Geertsema and J.C. Vlugter, J. Catalysis 4, 43

7.

Leva,Che~. Eng. (Aug. 1957)

8. J.M. Smith "Chemical Engineering Kinetics" IvIcGra'\'l - Hili Book Company, Inc., New York, 1956

9. W.B. Argo and J.H. Smith, Chemical Engineering Progress, 49 : 443 (1953)

10. J.H. Perry "Chemical Engineers' Handbook" 4th edition,

r.TcGraw-Hill Book Company, Inc., New York, 5-51, (1963)

11. R.A.< Bernard and R.Il. Wilhelm, Chem. Eng. Progr.,

46 : 233 (1950)

12. J. Beek and E. Singer, A procedure for sealing-up a catalytic-Reactor, Chem. Eng. Progr. 47 : 534 (1951)

13. S.H. Walas, "Reaction Kinetics for Che~ical Engineers",

HcGravl-Hill Book Company, Inc, Nevl York, (1959)

14. V.S. Fikhtengal'ts and A.L. Klebanskii, J. Gen. Chem. U.S.S.R. 27, 2535 (1957).

I.

APPENDIX I

1

1 ProceS8 design for zero partiele porosity

For

G

=

superficial maS8 velocity

=

_{1212 lb/sq} ft.hr

tw

=

wall temperature

=

145°C

and

S

= poroai ty of catalyst particles = 0

We follow the same design procedure.

Wall transfer coefficient

__ G.d 12

The Reynolds number Re

r-

-

=

From equation (4): = 3,5.0,02.(38,8)°,7

=

0,172.1,043 hw

=

5,02 _{Btu/sq ft.hr.op} 1212.0,0016 0,050 5,02

13

Effeetive thermal conductivity Ke

a. Radiation contribution K' r

From

equation (5) K' lC 0,004 r

..

-

....

_--

...

_--_

..

b. Point eonductivity K~ Prom equation (6) K' =

_{..,E--... ... ___}

1,8

_.

_{__}

c. Coeffieient h _{. c} between pellet and gas

From equation (7) he

=

0,33.1212.1,95

=

146 0,86.6,23 hc == 146 -.-.

...

---.---=

38,8

-""---" ~... ..'

r-

-

'

-

-II. d. Coefficient h_r and hp

We

aesume

h

=

2.499.

From equation (8)

and from equation (9)

h

_{_12}

_{... __}

=

_{... ____}

2.347,5

_{_}

_

----

-.

-

~ . - ::.~,

_". h'-':'~

Since the eum hc+ h_r+ hp

=

146 + 5,2 + 2347,5 = 2.498,7 the original assumption of 2499 for h is satisfactory. e. Series contribution K'series

From equation (10)

f.

Turbulent diffusion contribution Ktd For Re :: 38,8 dp/d t

=

0,0093 = 9,2 Pem : 9,22 andfrom equation (11) Ktd = 0,225

---

...

----g. Molecular conductivity contribution k~

" -'.' ... ~.

k~=O,020 BtU/ft2~hr.(OF/ft)

...---_

...

---h. Effective thermal conductivity Ke

Ke

=

e (K~ +

Kid

+ K;) + (1-e) K~eries

~ 0,31 (0,020+0,225+0,004) + 0,69.1,9 f: 1,39 Ke= 1,39 BtU/eq.ft.hr.op/ft

l I l .

14

The over

all

heat transfer coefficient

From equation (12)

1 1 0,086

'U ~ ~ +

4

.

1,34

=

0,214

u

=

4,67 _{Btu/sq ft.hr.oF} ~---~--~

15

The design procedure

The density of the contact mixture partieles f~ (See page 24).

For 10% Copper

r

~ = _{0,1.555,6+151,7.0,9 = 192 lb/cu•ft}

The void fraction of the bed e

=

0,31. Rence the bulk density of the contact mixture particles

Fe

is

re

=r~(1-e)=1920,69=132,5 _lb/cu ft

and the bulk density of the catalyst (10% copper)

The molal rate/per tube

, , ,

F ~ _, 24.S _c = 24.0,0232 = 0,556

.:.;/

~ ~.

The heat capacity of the reaction mixture Sm_i c Pi

Smic = _{F.MB1.{1-x).c +F.MB2·~·c}

Pi P1 P2

By

substituting in equation (13) we get O,0232.13,25.r.dz = 0,556.dx or

or

dz = 1 80 dx _{, r} dz

=

1,80.dX(~

+ '

);~

1 r ₂ (A)

The energy balance equation (14) may be written

25.576 .dx-2-.53. (tm-tw) .dz=O, 556. (19,2-7, 7x) .d tm (B) For x=O and an increment of dx=0,025

1 1 )

dz

= 0,0225 (-- +

--r,

r

₂

59,8-0,237.(tm-tw).dz=dtm

For a feed temperature of 270°C it is assumed that the temperature tm1 at the end of the first 1ncrement is 2900

c.

From fig. (6) the rate

at

the of the increment, zero conversion and t

=

270° C is 34,3.10-3 , and at the end of it, t = 290° C is 33,2.10-3 •

Then in equa.tion CA1)

1

,

dz=0,0225 + =1,3

.... :---

....

34,3.10 -3 33,2.'0 -3 In eq.(B,) we check the assumed value of tm,

59,8-0,23~(290~).1,3=dtm=tm1-270

59,8-39,7

=

tm,-270 20,1

=

tm,-270

tm,= 270+20,1=290,10C (versus the assumed value of 290°C)

ft.

Cont1nu1ng the computation give the results summarized in Table (10) and fig. (9).

560 ~20 SOO ~80 !60 (.) o E-4 4 Fig. (9) Conversion x

0

0,025 0,050 0,075 0,100 0,125 0,150 . 0,200 0,250 0,300 0,350 0,400 ' ; . 6 .8 . _.i,.

...

..

.

. . .. ,.~,.. .. ; 10 12 14 16 18 20 22

Catalyst bed depth, f t

Longitudinal temperature profile of the bed.

TABLE 10.

Temperature

°c

f' . Catalyst bed depth, ft.

270,0 0 290,0 1,30 300,0 1,45 308,0 1,43 322,5 1,19 338,0 1,06 346,5 1,12 348,0 2,47 343,0 2,58 339,5 2,60 334,0 2,71 291,5 3,50 ./

I --'

.

_

... -. _,', ~

..

I . . -, i ,'? VI. 16 Production

A reactor coneisting of 2,06 inches tubes, 15,20 ft high

..

----operating 260 days/year (about 1 day in operation and 9 hours shut down) the yearly production for eonversion x=0,30 ie estimated as follows:

V

_t

=

_L.Se

=

15,20.0,0232 = 0,353 eu ft The mass of contact mixture per tube is

Mc= 't·re= 0,353.132,5 = _{46,77 lb/tube}

M

_{Si = the} mass _{of silicon in a tube=46,77.0,9=42,1 lb/tube}

For 60% silicon eonversion the mass of silicon converted is

M

_Si = 42,1.0,6 = 25,26 lb = 1,8 lb.mol The amount of CH 3Cl to convert 1,8 lb.mol Si is

M

_{Si ·2} 1,8.2 M_{CH Cl=}

_0,30

= _0,30

=

12 lb mol CH₃Cl 3

For a feed rate 0,556 lb mol CH

3Cl/hr it will take 12

about h = 0,556 = 21,7

hr.

to convert 60% of

S1 •

Assuming a mean select1vity of 65% during the course of 21,7 hours, the production per tube will be

;;=O,~56.0,30.0,65=O,0542

lb mol(CH3)2SiC12! hr, tube

or

0,0542.115

=

6,233 lb/_{hr ,} tube

or

P; = _{2,82 kg/hr, tube}

And the yearly production w111 be

I •

•

A reactor

with

100

tubes will produce

Fr= 100.17.597

=

1.759.700

_kg/year

and

w1th

150

tubes

P

_r

=

150.17.597

=

2.639.550 kg/year

17 lressure drop in the tubes

From equation

(15)

~

=

359

pSf/ft

=

2,4

_psi/ft

and

dP =

2,4.L

=

2,4.15,20

=

36,5

psi

Ia

The minimum fluidizing flow rate

\" , ) ,

~l

L.

l ,

---From equation

(16)

we see that the minimum fluidizing

flow

rate is

about

5,5

times higher than the

ueed velocity of CH

,

.

VIII.

REMARKS ANTI CONCLUSIONS

We see from the previous discussion that considering

the particles without porosity the _{production/year} is

improved but the main disadvantagesof the fixed bed reactor i.e. the poor heat transfer properties and the

frequent shut-down remain.

For a production or 5000 _tUbes/year we need:

Two reactors with 150 tubes, 2,06 inches diameter and

15,20 ft length.

The operating conditions are

a. Feed rate = 24 lb mOl/Bq ft, hr b. Feed temperature 2700 C c. Conversion

30%

CH

3

Cl and 60% S1 d. Pressure

=

7

atm. e. Wall temperature 1450 C \