0009-2509179/0901-1123/$0200/0

SYNTHESIS OF PROPYLENE OXIDE

FROM PROPYLENE CHLOROHYDRINS-I

KINETIC ASPECTS OF THE PROCESS

S Cm, E SANTACESAlUA,* M MORBIDELLI and L CAVALLI

Ishtuto da Chlmlca F~s~ca del Pohtecmco dr Mtlano. bCentro Nazlonale Tecnologla della Prop&lone ed
Energetrca, Socleta Euteco, Mdano

(Recewed 30 October 1978, accepted I6 Jun~ory 1979)

Abstrac-The reactlons mvolved IIIthe mdustnal production of propylene oxide from chlorohydnns have been
kmetlcally mvestlgated The kmetlcs of dehydrochlonnatlon of propylene chlorohydnns have been studied, both In
the presence of slaked hme and of caustic soda, by potentlometnc techmques The kmetic parameters of the
reactions have been determined The propylene oxide hydrolysis has been studled by measurmg the decrease of the
reagent dunng the time usmg NMR analysis

INTXODUCTION propylene oxide IS flashed out with stream to shorten the

Propylene oxide IS still mamly prepared m industry from contact tune and to prevent hydrolysls[l, 2]
the chlorohydrmatlon of propylene, followed by dehy- Designing the dlstdlatlon column and choosmg the
drochlonnatlon of propylene chlorohydrms, according to optimal operating condltlons requires the determination
the followmg scheme [ 1,2] of kmetic parameters for the reactions Involved in the
process and the evaluation of vapor hqmd equdlbrmm
CH,-CH-CH, data
In the present paper the kmetlcs of dehydroch-
CH, = CH-CH, + HClO dH dl
lormatlon of propylene chlorohydnns and the propylene
oxide hydrolysis have been studled
CH,-CH---CHz A subsequent paper wdl study the vapor-hquld equd-

L1 AH lbrla By combmmg such mformatlon with kmetlc data,

the stmulatlon of a continuous mlcropdot unit for propy-
lene oxide production wdl be made The results of the
CH3YH-YH2 calculations wdl be compared with the experimental data
dH d obtamed by a mlcropdot plant
-CH2-_CH--CH, + Cl- + Hz0
CH,+H-_CHz +OH-
\/ EXPERIMENTAL

41 AH 0 Equipment, reagents and operating condltrons

The dehydrochlormatron of propylene chlorohydrms in
aqueous basic solution 1s a fast reaction occurrmg with
The basic medmm necessary for the second reaction can
the ehmmatlon of hydrogen chloride As previously
be obtamed by using slaked hme or a caustic soda
mentioned, the basic envuonment can be obtamed by
solution
usmg slaked lame or caustic soda In the present paper
In spite of the fact that this process 1s quite old, very
the kmetlcs of propylene oxide formation, both in the
httle mformatlon IS avadable on the kmetlcs of the
presence of slaked lune and of caustic soda, have been
dehydrochlonnatlon reactlon[3,4] Together with the
studied In the first case, the reaction rate has been
reactlon of propylene oxide formatlon, hydrolysis always
determined by contmuous potentlometr?c evaluation of
occurs, according to the scheme
the amounts of chlonne released dunng the reactlon
accordmg to the method described m[S, 71 A concentra-
tOH-J
CH2-CH-CH, + HzO- CHz-CH-CHs tion electrochermcal cell, with two Ag/Ag Cl electro-
\/ des has been used for the determmation One electrode
0 d I-I OH was unmersed m the reaction solution and the other m
a K2S04 saturated solution, separated from the reaction
The kmetlcs of this reactlon IS also poorly described m solutlon by a porous disk of fntted Pyrex glass, which
the literature [4,5] Smce the hydrolysis reaction lowers allows electrlcal continuity The electrodes were pre-
the yields, propylene oxide must be removed as quickly pared by anodlc attack of a salver rod m a 0 1 N HCl
as possible from the reaction envuonment solution at very low current mtenslties. as suggested by
For these reasons, the mdustrlal productIon of propy- Ives et al [S] The AgCl coatmg was renewed before each
lene oxide IS performed m a dlstdlatlon column, and run m order to prevent the effects of electrode agmg[8]

1123
1124 S CARR.4et al

Before each experiment, the response of the electrodes RESULTS

to the chloride ion concentration was calibrated by ad- When the propylene oxide formation occurs m the
ding measured quantltles of HCI to an aqueous solution presence of slaked hme, Ca CIZ IS formed and the solu-
contammg an excess of Ca (OH)* at the reactlon tem- tion 1s buffered In fact, If we consider the relations
perature
In the presence of caustic soda, the reactlon rate has ICa*][OH-]* = k, solublhty product (1)
been determined by usmg a glass electrode to measure
the decrease of the hydroxyl Ion concentration as a [Cal= i [OH-] + k [Cl-] electroneutrahty (2)
consequence of the reaction In this case, a saturated
calomel electrode was used as reference electrode
It IS possible to write also
In both cases, the reaction took place m a ther-
mostated stirred batch reactor Fifty cm of basic solu-
[Ca] = l/2 {[OH-] + XC} (3)
tion contammg about 4g of Ca(OH)z or 10e3 -
10-l moles/l of NaOH was put mto the reactor When the
where X IS the degree of advancement of the reactlon,
reaction temperature was reached, 1 cm3 of the organic
and C the uutlal concentration of the organic reagent It
reagent was InJected into the soiutlon with a glass synnge,
follows that
and the potentlometrlc response was recorded
This response was quick m both cases and the dead
[OH-l3 + XC0[OH-]2 - 2K, = 0 (4)
time IS negligible with respect to the time of the reaction
Runs made with the two pure chlorohydnns isomers
The concerttratton of hydroxyl Ions decreases very
have shown IdentlcaI reactivity The kmetlc data collected
slowly after the uutlal stage of the reaction Therefore, m
m this papel refer to an mdustnal mixture of the two
this case, we can consider the hydroxyl concentratttlon
Lsomers mentloned These mixtures contam secondary
during the reaction as constant The Ag/AgCl electrodes,
and primary alcohol m a ratlo of about 9 1 [l]
described m the previous sectlon, give potentlometrlc
The hydrolysis of propylene oxide was studied wrth a
curves which can be used for measuring the increase of
different procedure, because there 1s no ehmmatlon of
the chloride ton concentration as a result of the reactlon
chloride ions or change of pH, and the reaction IS rather
These curves can be easily rearranged m order to obtam
slow
the conversion degree as a function of time In Fig 1 the
The reaction rate, m this case, has been determmed by
experlmentaI values of conversion versus time are
using an NMR spectrometer Jeol JNM-C-4OHL to
reported for several temperatures The experImental data
measure the decrease of propylene oxide concentration
reported In Fig 1 are well fitted, as shown m the Fig,
Runs were made at several temperatures and pH
assummg pseudo first order krnetlc reactton model of the
NaOH 1 N was added to 50 cm3 of aqueous solutlon
type
contammg 6% m weight of NaCl m order to adJust the
PH (5)
Then, 20mg of CHKOONa were added as Internal
reference to 2 cm3 of this solution and, after freezing at [PC] bemg the concentration of propylene chlorohydrm
about 0C 70~1 of propylene oxide was introduced The use of this slmphfied model 1s Justified because the
0 5 cm of this solution were then transferred to a test solution, as already mentioned, IS buffered and the
tube and submitted to NMR analysis, at a given tem- hydroxyl concentration can be considered nearly con-
perature stant
The dead time for reaching this temperature IS negh- The kmetlc parameters obtained from the experlmental
grble with respect to the reactlon time runs are reported tn Table 1 In Fig 2, an Arrhenms type
For long runs, the NMR test tube was kept soaked m a plot, IS reported for this reaction
thermostated vessel and analyzed from time to time at If the dehydrochlormatlon IS made m the presence of
regular Intervals The spectral peaks used for analytical caustic soda, the hydroxyl ion concentration decreases
purposes were those related to the methyl protons of rapidly during the reaction The change of hydroxyl
propylene oxide (doublet at 6 = 1 33) and the smglet concentrat!on dunng this tame has been determined by a
methyl of CHsCOONa at 8 = 1 97 Also the peaks of the glass electrode Smce the reaction rates slow down with
propylene glycol at 6 = 1 15 have been determmed the decrease of pH, they depend on the hydroxyl con-
The measurements were performed m the 3 Hz for cm centration
expanded range The experlmental results, m fact, can be well
The hydrolysis reactlon rates have been determmed by represented by assuming a second order kmetm model,
observing the decrease m the mtenslty of the regstered of the type
peaks of the propylene oxide
Runs were made in the pH range 6-13 and m the r<d,, = kz[OH-][PC] (61
temperature range 25~90C The experrments at low pH
values (610) were made m the absence of CH,COONa, The experimental data obtained for runs made m
cahbratmg with accuracy the uutlal concentration of different operating condmons are reported m Table 2
propylene oxide The kmetrc constants have been calculated on the basis
 Synthesis of propylene oxide from propyIene chlorohydrms--I 1125

Table 1 Kmetwz parameters for the dehydrochlormatlon reactton made m the presence of slaked hme and causttc
soda

Adopted
Basw mednun A EI (kcal/mole)
klnetlc model 1

= kl [PC]
rlItTI~ -1
Slaked lune 3 02x 109s 15 860
-E,/RT
k, = A1 e

r soda = k2 [PCllOH-1 I
12
Caustic soda 503x10 -- 16 422
-E2/RT
moks
= x2 e
k2

37 C
0

0 1 2 3
tlmc ( m~nules)

Rg I Trend of conversIon vs time for the dehydrochlonnatlon reaction of propylene chlorohydrms made m slaked
lime suspension

Fig 2 The Arrhemus plot for the kmetlc constants of the dehydrochlonnatlon reaction in the presence of slaked lime
and cawhc soda soluhon

of the second order kmetlc model The kmetlc and k, IS

parameters are summarrzed m Table 1 In Fig 2, an
Arrhenms plot for this reaction IS also reported From
Ftg 2 one can compare the kmetlc constants obtamed for
the reaction made m the presence of slaked hme and
caustic soda The relation between the two constants ks
ka = k, [OH-] (7)
Therefore kl/kl = [OH-] IS the hydroxyl oncentratlon m
the Ca(OH)z saturated solution, at a given temperature
In Table 3 the ratios k2/kl and the related [OH-] concen-
1126 S Cm et al

Table 2 Kmetlc data obtamed for the dehydrochlormatron reachon made m caustic soda solutions The kmehc
constants have heen calculated assummg a second order kmetx model The mrtud concentration of chlorohydrm IS
0 23 moles/l
I
iOH-10 Time X
RUTI

I T (%
moles/l 1 (sex)
0
Conversions

0
%
k2
l/mole set

I 5
10
1
1
087
956
0
0
251
262
0 01
20 3 034 0 265
30 3 626 0 266
40 3 913 0 257

=L
0 0
10 13 043 0 167
20 22 174 0 178
2 30 01
30 28 261 0 184
40 33 043 0 197
50 36 086 0 203
0 0
10 1 608 0 203
20 2 695 0 214
3 27 0 01
30 3 261 0 205
I 40 3 608 0 197
3 869 0 197
0
14 739 0 130
4 25 -q--g- 23 608 0 131
29 043 0 130
33 043 0 131
0
0 089 0 190
5 28 0 109 0 180
0 123 0 175

---I-?
0 132 0 168
0
1 856 ! 223

I
3 2 461 I 226
4 45 0 01
55 3 269 1 237
10 4 002 1 130

---l-e
4 339 I 390
0
1 787 0 232

I
20 2 a04 0 229
7 28 0 01
30 3 374 0 221
40 3 730 0 217

---i-T-
3 952 0 214
0
2 265 3 244
2 3 374 3 223
8 I 59 o 01
33 3 826 3 650
4 4 030 2 930
4 134 2 700

Table 3 A comparison between the ratios &Jkl and the related [OH-] concentration values, at several temperatures
I I I I
I T (Cl I K,'K2

50 3 595 x 10 -2 3 277 x 10 -2

90 2 5631x 10 -2 (y) 2 110x 10 -2

x Extrapolated data

trataon values are compared at several temperatures The the two basic media must be attibuted to the Ca(OH)z
values of [OH-] concentrations m saturated Ca(OI-02 solubdlty, whrch decreases when the temperature IS
solutions have been calculated by usmg the data reported Increased 191
by Stephen[7] As, can be seen from the data reported m We can then conclude that the propylene oxide reac-
Table 3, the ratios k$k, and the correspondmg hydroxyl tlon rate m the presence of slaked hme can be expressed
concentrations are not much different Thus confirms the as
kmetlc models adopted
The dfierences m the actlvatlon energies observed m rlime= 3 02 X lo e-5B60RTCpCJ (8
 Synthesis of propylene oxide from propylene chlorohydnns--I 1127

The same reaction, m caustic soda solution, follows the the fact that at low hydroxyl concentratrons @H = 6 -
kmetlc model 10) the kmetlc constant becomes independent of this
concentration, allow us to write
rda = 5 03 x 10 e- 422/RT [PC][OH-] (9)
kn = kk + kc,, [OH-] (10)
Propylene oxide hydrolysis lowers the yields of the
process It 1s therefore unportant to know the kmetlc Hence the hydrolysis of propylene oxide obeys the
parameters of thts reaction Runs made with an NMR general acid-base catalyzed kmetlcs
spectrometer at several pH and temperatures show that The relation of the reaction rate becomes, m this case
hydrolysis IS a first order reaction An effective kmetlc
constant can be evaluated by the determmatlon of the rhyd. = Ikh + kch [OH-W01 (11)
half life time In Tables 4-7 the percentages of
hydrolyzed propylene oxide for several operatmg con- where [PO] = propylene oxide concentration
ditions are reported as a function of tune The effective The kmetlc parameters have been determined by a
kinetic constants k, are reported for each run in the same no&near regression calculation and are k,, = 1, 48 X
tables lo exp (- 16,388)IRT) and kk =268X Wexp(-
Figures 3 and 4 show the values of -In (I -x) vs tune l6,253/RT)
for runs made at dtierent temperatures The linearity of If the pH IShigher than 12, k,, becomes newble and the
the curves justdies the assumption of the iirst order for kinetic expression can be srmpldied as follows
the orgamc reagent In Fig 5, the values of the effective
kinetic constant, obtamed for the runs made at the same Trh ydr = 2 68 X 108 e-6Z3RT[PO][OH-] (12)
temperature #WC), but at different pH values, are
reported as a function of hydroxyl concentration As can Tlus slmphfied kmetlc model IS important for mdustnal
be seen there IS a linear trend between ks and [OH-] purposes, because the pH of the solutlon m the reactor
Moreover, the presence of an intercept m Fig 5 and always IS kept higher than 12

Table 4 Expertmental data obtamed for the propylene oxtde hydrolysis performed at a hydroxyl concentrahon of
0 I moles/l, changmg the temperature

I-to3 T1XTI.Z [PO1

Q- 1)
(mmutes) COlWer6lOZl (IWC

25 8 2
100 27 6
145 34 9 -5
28 5 067 x IO
205 45 4
308 63 0
407 69 9

10 6 b
35 16 6
-5
36 65 25 8 7 49 x 10
120 42 9
150 50 3

lb 11 1
30 18 5
60 36 0 -4
45 1 208 x 10
90 46 6
150 bb 0
190 74 7

10 30 1
22 55 3
72 0
62 33 6 103 x lO-4
42 79 5
52 85 0
62 89 3

10 2 31 0
43 0 -4
65 13 0 7 602 x 10
lb 5 50 2
18 3 52

5 41 3
9 63 0
-3
76 13 78 5 2 061 x 10
19 89 4
25 94 4
 S Cma.4 et al

Table 5 Expertmental data obtamed for the propylene oxide hydrolysis performed at hydroxyl concentration of
0 01 moles/l, changmg the temperatures

T tC) TIIYW IPOI

(mmutes) Conversion
%

90
1050
25 2500
5443
13395

28
105
-5
60 197 46
28 35 I 5 456 x 10
244
382

20
45
-5
67 75 36
27 5I I 9 387 x 10
82 40 0
115 49 6 I
26 30 4 I
37 40 5
-4
83 50 63 9 2 95Y x 10
75 70 4
96 79 5 I
13 28 0
23 45 9
-4
86 34 58 5 -4483 x 10
44 67 0
48 73 0
I

15

10
x

5
7

05

0
a 100 200 300 LOO
t Ime ( mmutes )
Fig 3 A plot of -In (l-x) vs time for the propylene oxide hydrolysis reactlon performed at pH = 12
 Synthesis of propylene oxide from propylene chlorohydrms-I 1129

Table 6 E!xpc wmental data obtamed for the propylene oxide hydrolysis. at low hydroxyl concentration (pH = 6-10)

T ?C) l-4
Conversion
%

9 1
18 8
-5
56 26 5 I 427 x 10
31 1
66 7

90 11 5
185 29 8
-6 -5
56 10 270 35 2 2 126 x 10
38 7

56
---l-fL
IO
-5
177
255
83

27
33
9
4

3
7
4 2 000 x 10
-5

320

I 1355
39
81
1
8

40 8 3
166 24 7

I
-4 -5
56 10 250 33 2 1 913 x 10
340 39 8
I335 79 7

14 5 a
49 17 4
-6 6 439 x 10 -5
75 10 120 37 2
175 49 8
245 61 0

11 12 7
25 27 1
-6 -4
90 10 48 43 5 1 961 x 10
66 54 3
87 64 8

Table 7 Expertmental data obtamed for the propylene oxide hydrolysis performed at several hydroxyl concen-
tratlons

T(OC) [OH-I Tame [PO1

k3- 1)
moles/l (mmutes) COI-lVf2rSlOn (6.X

25 I 5 2
63 15 6
-4 -5
60 3 16x 10 175 26 4 24 x 10
370 47 0
1357 86 5

22 J 34 I

60 10
-3
74
161 I 13
26
5
5 I 3 03 x 10 -5

378
249 I 49
36 71 I
12 17 3
25 32 4
-2 42 3 2 646 x 10 -4
60 3 98 10 36
53 56 8
60 63 0

8 21 1
15 30 7
-2 -4
60 708 10 23 48 4 4 558 x 10
31 60 3
39 6~ 5
1130 s CARRAetal

z
z
5

0
0 ml 200
ltme (mm&r)
Fig 4 A plot of -In (I - X) vs tune for the propylene oxide hydrolysis reactlon performed at pH = 13

Fu 5 The trend of the effective lunehc constant k~with the hydroxyl concentration for runs performed at 60C

In Fig 6, an Arrhenrus type plot for the effective The rates of reactions 1 and 2 are identical, as has been
constant of this reaction IS drawn seen experimentally
For the epoxlde formation, the rmg closure reaction,
DlSCUS!%ON occurs, as described by Patit and Weissberger [31,
The experiments described m this paper, allowed the according to an mtemal nucleophdlc substitution (Z&2)
determination of the kinetic parameters of the reactions mechanism preceded by a base catalyzed drssoclatlon
involved m the propylene ox&de production process The equlhbrmm
scheme of the reactions studied IS as follows The reactzon mechanism can be represented as fol-

CH~-CHXHI

&l AH \ +I-ZtCB
-HCl ~-------,CH2-_CHXH,

yH,TH-CH3
/
CH\YH-CH
0
3
AHAH
 Synthesis of propylene oludc from propylene chiorohydrms-I 1131

lows

-,C-C<+Cl-+HzO

The reaction rate, m this case, IS and the activation energy depends also on the enthalpy
change of calcium hydroxide dissolution
r = kK[R][OH-] (13) That IS
E ,-E+AHR+1/3AH. (15)
where [R] IS the reagent concentration, k IS the kinetic
where E IS the true activation energy, AHn IS the enthalpy
constant and K IS the equdtbrmm constant for the for-
change m intermediate ion formation. AH, IS the enthalpy
mation of the intermediate Ion
change of calcmm hydroxide dlssolutlon The last term of
this expression does not appear m the actlvatlon energy for
Ill k, the reaction In caustic soda solution
The rmg opening or hydrolysis reaction IS a base
catalyzed reaction occurring accordmg to the scheme
12 - suggested by Patai [4]

I I I
CHOH CHOH
10 r 0 + OH-m +H*CJ
HO- ___, ci HOH +OH-
I

8- For this mechamsm, a kinetic law of the followmg type IS

Justified
rhydr = kLXJR][OH-1 (16)
6-
where kih IS the kinetic constant, K, IS the equdlbrmm
I I 1 I
constant, and R IS the organic reagent concentration
20 25 30 35
lO/l This kinetic equation IS true for pH values higher than
Wg 6 An Arrhemus type plot for the effective kmettc constant 12 When the pH IS lower than 10, the reaction rates
k3 hardly change with the hydroxyl concentration A kmetrc
model of the followmg type can then be suggested
The reaction rate strongly depends on this equlhbrmm
constant and therefore depends also on the enthalpy rhydr = {kh + kchW-1HRl (17)
change for the mtermedlate ion formation
The two chlorohydrm isomers which can give propy- which shows the mterventlon of a parallel uncatalyzed
lene oxide have, as already mentioned, the same reac- first order reaction This reactlon IS also present m the
tivity This IS due to the mductive effect of the methyl acid catalyzed hydrolysis, as reported by Patal[4], and
group, which favours the reaction m either case, mcreas- we can conclude that the reaction obeys the general
mg the chlorine mobMy or the negative charge on the acid-base catalyzed kinetics Finally, if we consider that
oxygen at IWC, which IS the reactton temperature of the m-
dustraal process, the ratio between the rmg closure and
the rmg opening reaction rates IS 830 we can conclude
that, for low contact time, yields of propylene oxide are
hardly affected by the hydrolysis

The probabdlty of mtermedlate Ion formation 1s NOTATION

evidently the same
Ai preexponentlal factors
The dlfferenees m actlvatlon energy values for the C uutlal concentration of chlorohydrms m water
reaction m the presence of slaked lime and of caustic apparent actlvatlon energy for the synthesis
El
soda 1s due to the solublllty of Ca(OH)*, which performed m slaked hme solution
decreases when the temperature IS increased The
EZ apparent actlvatlon energy for the synthesis
effective constant m the presence of slaked lime could be
performed m caustic soda solution
expressed as E true activation energy for the propylene oxide
kr-kK+Ks (14) synthesis
1132 S CARRA et al

k, effective kinetic constant for the hydrolysis Greek symbols

reaction AHn enthalpy change m intermediate Ion formation
kh effective kinetic constant for the non catalyzed durmg the synthesis reaction
hydrolysis AHs enthalpy change for the calcium hydroxide dls-
k ch effective kinetic constant for the acid-base solution
catalyzed hydrolys:s reaction
K, Ke equlhbrmm constants for the intermediate ions REFERENCES

in the synthesis and in the hydrolysis of pro- 111 Horsley H L , Encycfopedra of Chemical Technology (Edited
pylene oxide respectively by Othmer K ). p 595 Wiley, New York (1965)
k, k:,, kinetic constants for the synthesis and hydroly- El Lowenhelm F A, Moran M K , Industnal Chemrcals, p 692
Wllev-Interscience. New York (1975)
SISof propylene oxide respectively [31 Wersiberger A , The Chemistry .of Heterocyclrc Compounds
KS solublhty product of slaked hme Interscience, New York 1964
kl kinetic constant for the propylene oxide syn- [41 Pam S , The Chemistry of Functronol Groups InterscIence,
thesis m slaked lime solution New York 1967
PI Smrth L , Zertschr f Physlk lo18 93 59
k2 kmetlc constant for the propylene oxide syn-
[61 Carra S Santacesarra E , Morbldelh M Schwarz P and DIVO
thesis rn caustic soda solutlon C , SubmItted to Ind Engng Chem Proc Des Dev
rw propylene chlorohydrms concentration r71 Carra S , Santacesarla E , Morbldelh M , Schwarz P and
[PO1 propylene oxide concentration DIVO C , Submitted to Ind Engng Chem Proc Des Dee
r reaction rates PI Ives D J G and Janz G J , Reference Electrodes Academic
Press, New York (l%l)
R general constant of gases r91 Stephen H , Sohbhtres of Znorgonrc and Organrc Com-
T absolute temperature pounds Pergamon Press, Oxford 1964
X degree of advancement of the reaction