3848
Ind. Eng. Chem. Res. 1996, 35, 3848-3853
Kinetics of Ethoxylation and Propoxylation of 1- and 2-Octanol Catalyzed by KOH M. Di Serio,† G. Vairo,† P. Iengo,† F. Felippone,‡ and E. Santacesaria*,† Dipartimento di Chimica, dell’Universita` Federico II di Napoli, Via Mezzocannone 4, 80134 Napoli, Italy, and Pressindustria SpA, Via Porta D’Arnolfo 35, 20046 Biassono Milano, Italy
Ethoxylation and propoxylation reactions are often performed together or in alternation to obtain surfactants with particular properties or random and block copolymers. Both reactions are normally performed in the same reactor, in the presence of an alkaline catalyst, at relatively low temperature, 120-130 °C, to avoid the intervention of the side reactions that are typical of the propoxylation. The propoxylation of a primary fatty alcohol is slower than the corresponding ethoxylation and gives place to a secondary hydroxyl terminal group that is still less reactive. On the contrary, ethoxylation restores more reactive primary hydroxyl terminal groups. Therefore, it is important for optimizing the described industrial operations to know the reactivity of ethylene and propylene oxide with, respectively, primary and secondary hydroxyls, in the presence of the most used KOH catalyst. In this paper, the kinetics of both the ethoxylation and propoxylation of 1- and 2-octanol catalyzed by KOH have been studied for this purpose. We will show, first of all, that propylene oxide ring opening occurs selectively giving only secondary hydroxyls as terminal groups. The ratio of addition rate of ethylene oxide to primary and secondary alcohol with respect to that of propylene oxide is always greater than 1. Kinetic data collected have been interpreted by using a kinetic model able to simulate during the time the consumption of both the octanol and the alkoxide and the evolution of the oligomer distributions. The kinetic model and related parameters can be easily extrapolated to different industrial situations in which ethoxylation and propoxylation occur together or in alternation. Introduction Ethoxylation and propoxylation of fatty alcohols are reactions of relevant importance in the surfactants industry. As a matter of fact, nonionic surfactants are usually produced by reacting ethylene oxide with hydrophobic substrates containing a mobile hydrogen, such as fatty alcohols, alkylphenols, or fatty acids. Propylene oxide can frequently be added to ethoxylated molecules to obtain less foaming products. On the contrary, the addition of propylene oxide directly to the fatty alcohols before ethoxylation can be made to increase the hydrophobic character of the surfactant (Kosswing, 1994) allowing the use of fatty alcohols of lower molecular weight. At last, as ethylene-propylene oxide copolymers are normally produced on an industrial level, it is of great interest for all the mentioned cases to know the difference of reactivity of these epoxides with, respectively, primary and secondary alcohols, remembering that propoxylation originates mainly secondary alcohols as chain terminals that are less reactive, while ethoxylation restores primary alcohols. Both ethoxylation and propoxylation are usually performed at 2-5 atm, in the presence of an alkaline catalyst, i.e. KOH or NaOH, but at temperatures that are usually lower for the propoxylation (120-130 °C) in respect to the ethoxylation (150-180 °C), to avoid the intervention of side reactions such as hydroxyl dehydration. When the two reactions are performed together or in alternation the temperature is normally kept at the lower level (120-130 °C). In previous papers, Santacesaria et al. (1990, 1992a, 1992b) and Di Serio et al. (1994) studied the ethoxyla* To whom correspondence should be addressed. Tel.: +39 +81 5476544. Fax +39 +81 5527771. E-mail: Santacesaria@ chemna.dichi.unina.it. † dell’Universita ` Federico II di Napoli. ‡ Pressindustria SpA.
S0888-5885(96)00200-X CCC: $12.00
tion kinetics of nonylphenol, 1-dodecanol, and dodecanoic acid, respectively. In said papers, the authors showed that the ethoxylation kinetics of the different substrates can be described with a unique kinetic model and that the different behaviors observed can be explained as a consequence of the different acidity of the used substrates and of the nucleophilicity of their conjugated bases. Moreover, the authors stressed the role of the correct description of the epoxides solubility in the kinetic approach (Di Serio et al., 1995) as well as the importance of this aspect for the safety of the industrial reactor management (Santacesaria et al., 1995). Very few papers have been published, on the contrary, about the comparison of the reactivities of ethylene and propylene oxides with primary and secondary alcohols, respectively. In the paper by Gladkovskii and Ryzhenkova (1971), for example, the reactivities of ethylene and propylene oxide with CH3OCH2CH2OH and CH3OCH2CH(CH3)OH were compared by measuring the rates of pressure decrease. The kinetic data collected in this way can be interpreted with difficulty because propagation rates are in some cases different from initiation ones, changing, for example, the type of chain terminal group. The rate ratio of the addition of ethylene oxide to a primary alcohol with respect to propylene oxide (r1) and the rate ratio of addition of propylene oxide to a secondary alcohol with respect to ethylene oxide (r2) are reported in some papers devoted to the study of the EO/ PO copolymer synthesis. These rates, often determined by 1H NMR or 13C NMR analysis [see, for example, M. Adal et al. (1994); Heatley et al. (1991)] show a broad range of scattering (r1 ) 1.34-4.8, r2 ) 0.14-1.49) probably because the reactions have been performed by using different solvents and/or temperature conditions. Moreover, kinetic data obtained by indirect methods such as those mentioned are often affected by large © 1996 American Chemical Society
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 3849 Table 1. List of the Kinetic Runs Performed run
alcohol
epoxide
KOH (mol %)
T (°C)
P (atm)
1 2 3 4 5 6 7 8 9 10 11
1-octanol 2-octanol 2-octanol 2-octanol 2-octanol 1-octanol 1-octanol 1-octanol 2-octanol 2-octanol 2-octanol
EO EO EO EO EO PO PO PO PO PO PO
2 2 2 2 2 2 2 2 2 2 2
120 100 120 130 160 100 120 130 100 120 130
2 2 2 2 2 2 2 2 2 2 2
experimental errors. At last, these kinetic data have often been collected in a temperature range being far from the industrially interesting one. To overcome the lack of ethoxylation kinetic data about the reaction with a secondary alcohol, and that of propoxylation, in this paper we have studied and compared the kinetics of ethylene and propylene oxide addition to 1-octanol and 2-octanol, respectively. Kinetic data have been obtained both by measuring epoxide consumption and by analyzing with a gas chromatograph some samples taken at different reaction times. The choice of these low molecular weight alcohols has been made for their commercial availability and for the possibility to analyze easily the alkoxide derivatives by gas chromatography without derivatization. Kinetic data, that is, the evolution with time of the epoxide consumption, of the unreacted alcohol and of the oligomers distribution have been interpreted through a kinetic model also considering the influence of the epoxides solubility. The kinetic parameters of the four systems considered have been determined by mathematical regression analysis of the experimental data. As these parameters are scarcely affected by both the chain length of the starting fatty alcohol and the length of the growing chain, the results can be extrapolated to the systems of industrial interest provided that the related alkoxide solubility data are available. Experimental Section Kinetic Runs and GC Analysis. Kinetic runs were performed in a reactor that is well described elsewhere (Santacesaria et al., 1992a). The control of pressure, temperature, and epoxide feeding the reactor was fully automated by a computer. Epoxide consumption was directly measured and recorded at each instant with a weight balance measuring the weight loss of the epoxide bottle. The epoxide in the bottle was pressurized with nitrogen at about 10 atm and automatically fed to the reactor to keep the preset pressure value (2 atm) constant. KOH was used as catalyst. Samples of the reaction mixtures were taken at different times and analyzed by the gas chromatographic technique. A 0.2 µL portion of solution (0.03 g of the reaction sample dissolved in 4 cm3 of CHCl3) was injected on a 25 m × 0.32 mm i.d. HP1 (100% dimethylpolysiloxane gum) column and analyzed by taking the temperature at 80 °C for 1.5 min and then heating at a rate of 10 °C/min until 320 °C. A FID detector kept at 350°C was used. The oligomer response factors have been calculated as suggested by Milwidsky and Gabriel (1982) and have been verified on the basis of the mass balance of the epoxide consumption. The list of the kinetic runs performed and of the operative conditions applied is reported in Table 1. 13C NMR Analysis. The asymmetric characteristic of the propylene oxide molecule could generate primary
Table 2. Relationship Giving Density as a Function of the Mean Number of Epoxide Adducts per Mole of Starting Alcohol and Related Parameters starter 1-octanol 1-octanol 2-octanol 2-octanol ad
epoxide
A
EO PO EO PO
0.860 0.860 0.826 0.826
B
C
∆n
∆T (°C)
0.07 0.05 0.14 0.06
9.0 × 9.0 × 10-4 7.0 × 10-4 7.0 × 10-4
0-1 0-1 0-1 0-1
100-130 100-130 100-160 100-130
10-4
) A + Bn - CT (g/cm3) (T, °C).
or secondary alcohol when ring opening occurs as consequence of nucleophile attacks to the methylene (1) or methyne group (2), respectively: CH3 O RO–M+
+ CH2 1
1
CH 2
CH3
ROCH2
CHO–M+
(1)
CH2O–M+
(2)
CH3
2
ROCH
In order to verify the contribution of these two reactions, samples of propoxylated 1- and 2-octanol were analyzed by 13C NMR. The areas of the signals concerning the primary alcohol (δ ) 62.4 ppm) and secondary alcohol (δ ) 66.0 ppm) for the samples dissolved in CDCl3 were determined using a Bruker 270 MHz apparatus operating at 67 MHz. Acquisition times of 6 s without delay between the pulses were applied. The pulse angle was 90 °C, and the spectral width was 1360 Hz. Data were processed with an exponential weighting function (weight ) 0.3 Hz) before transformation in order to decrease the noise. Chemical shifts were referred to tetramethylsilane as external reference. In the analysis of propoxylated 2-octanol, no signals related to primary alcohols were observed, while in the analysis of propoxylated 1-octanol the fraction of primary alcohol measured by the 13C NMR technique corresponds to the unreacted alcohol determined gas chromatographically. Therefore, it can be concluded that the reaction at the methyne group is negligible in comparison with that occurring at the methylene group, in agreement with the observation by Shilling and Tonelly (1986) for propylene oxide polymerization and by Gee et al. (1959, 1961) for methanol and isopropanol propoxylation. Density Measurement. In order to correctly evaluate the reaction volume, it is necessary to know the density of the reaction mixture at any reaction time. For this purpose, we measured with a pycnometer the density of the starter as well as of the corresponding ethoxylated and propoxylated products, with an average number of epoxide groups ranging between 0 and 1, at different temperatures. The relation to determine the densities for different EO and PO amounts and temperatures containing three parameters was derived by processing the experimental data. This relation is reported in Table 2 together with the numerical values of the parameters. The epoxide densities in the liquid phase were evaluated through the method suggested by Yeen and Woods (1966); the calculated values at different temperatures are reported in Table 3. Determination of Epoxide Solubilities. The molar fractions of ethylene oxide and propylene oxide in the two different alcohols were evaluated by calculating the activity coefficients in the liquid phase through the UNIFAC method (Fredeslund et al., 1977) and consider-
3850 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996
Figure 1. Equilibrium pressure of epoxide as a function of the liquid molar fraction in dodecanol at 100 °C (dots are experimental; lines are calculated). Table 3. Density of Epoxides Calculated by the Yeen and Woods Relation (1966) T (°C)
dEO (g/cm3)
dPO (g/cm3)
100 120 130 160
0.74 0.70 0.68 0.60
0.78 0.74 0.72 0.65
Table 4. Epoxide Concentrations Calculated by UNIFAC Using Data of Tables 2 and 3 at 2 atm concn (mol/cm3 × 103) substrate
epoxide
100 °C
120 °C
130 °C
1-octanol 1-octanol + 1EO 1-octanol 1-octanol + 1PO 2-octanol 2-octanol + 1EO 2-octanol 2-octanol + 1PO
EO EO PO PO EO EO PO PO
1.4 1.4 2.7 2.5 1.4 1.6 2.6 2.0
0.91 0.86 1.8 1.2 0.89 1.0 1.6 1.2
0.77 0.73 1.4 1.0 0.79 0.87 1.3 0.99
Figure 2. (a) Ethylene oxide consumption related to the initial moles of alcohol (EO/S) in the ethoxylation of 1-octanol (O run 1 of Table 1). Dots are experimental; line is calculated. (b) Experimental and calculated oligomer distributions of run 1 at different reaction times.
160 °C
0.42 0.49
ing the vapor phase as ideal:
xEO ) P/(γEOP°EO)
(3)
xPO ) P/(γPOP°PO)
(4)
The vapor pressures of the two epoxides were calculated using Antoine’s equations (Prausnitz and Anderson, 1980):
P°EO ) exp[16.74 - 2568/(T - 29.01)]/760
(5)
P°PO ) exp[15.32 - 2108/(T - 64.87)]/760
(6)
Di Serio et al. (1995) have shown that the UNIFAC method is reliable in describing the ethylene oxide solubility in the ethoxylated alcohols with a low number of adducts. In order to verify the reliability of this conclusion for propylene oxide too, we measured the propylene oxide solubility at 100 °C, in dodecanol, as described in a previous paper (Santacesaria et al., 1992a). The experimental values obtained are compared with the calculated ones in Figure 1. The same figure also shows the agreement obtained for ethylene oxide solubility still calculated by the UNIFAC method. Applying the UNIFAC method and the density values of Tables 2 and 3, it is thus possible to evaluate the epoxide concentration in the reaction mixture. Table 4 reports an example calculated concentration of both epoxides in 1- and 2-octanol, at different temperatures.
Figure 3. (a) Propylene oxide consumption related to the initial moles of alcohol (PO/S) in the propoxylation of 2-octanol (+ run 9, 9 run 10, O run 11 of Table 1). Dots are experimental; lines are calculated. (b) Experimental and calculated oligomer distributions of run 11 at different reaction times.
Reagents. All the reagents were supplied by FLUKA Co. at the maximum purity commercially available; ethylene oxide was supplied by SIAD Co. Results and Discussion Figures 2a-5a report the epoxide consumptions as function of time for all the runs reported in Table 1, while Figures 2b-5b reported examples of the evolution
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 3851
ethoxylated chain, while oligomer distribution mainly depends on the proton transfer equilibria (Santacesaria et al., 1992a, 1992b, 1992c). The same behavior can be observed for secondary alcohol propoxylation, as it can be seen in Figure 3a, where propylene consumption is reported as a function of time for the propoxylation of 2-octanol, at different temperatures. 1-Octanol ethoxylation and 2-octanol propoxylation have a similar kinetic behavior because in both cases the structure of the reaction product is the same as that of the reacting alcohol. As a matter of fact, a primary alcohol is obtained by primary alcohol ethoxylation, while the secondary alcohol is obtained by secondary alcohol propoxylation (as it was shown in the paragraph about 13C NMR): O CH2O–M+ + CH2
R
CH3 R
Figure 4. (a) Ethylene oxide consumption related to the initial moles of alcohol (EO/S) in the ethoxylation of 2-octanol (+ run 2, 9 run 3, O run 4, × run 5 of Table 1). Dots are experimental; lines are calculated. (b) Experimental and calculated oligomer distributions of run 4 at different reaction times.
CH2
k11
CH2OCH2
R
O
CHO–M+ + CH2
CH
CH3
k22
R
CH2O–M+ (7)
CH3
CH3
CHOCH2
CHO–M+ (8)
On the contrary, reaction rates increase with time in the reaction of ethylene oxide with a secondary alcohol, as it can be seen in Figure 4a, where the consumption of ethylene oxide in the ethoxylation of 2-octanol, at different temperatures, is reported as a function of time. Ethylene oxide addition produces a primary alcohol in this case, too: CH3 R
CH3
O
CHO–M+ + CH2
CH2
k21
R
CHOCH2
CH2O–M+ (9)
but as the rate of addition of ethylene oxide to a primary alcohol is greater than the rate of addition to a secondary one (Gladkowskii and Ryzhenkova, 1971), the ethoxylation rate increases after the monoadduct formation. The strong difference in the ethoxylation rates for the initiation and propagation step, respectively, produces a broad oligomer distribution, as it can be seen in the example of Figure 4b related to run 4 of Table 1. In this case, oligomer distributions are affected by both the proton transfer equilibria and the relative rates in the ethylene oxide addition to primary alcohol with respect to secondary one (k11/k21 > 1). In the case of primary alcohol propoxylation the behavior is opposite, as it can be seen in Figure 5a, which reports the consumption of propylene oxide for the propoxylation of 1-octanol at different temperatures; as it can be observed, the rate of reaction decreases with the time. This behavior is related to the formation of a secondary alcohol as a consequence of both the initiation and propagation step of the reaction (as it has been shown in the paragraph about 13C NMR): O
Figure 5. (a) Propylene oxide consumption related to the initial moles of alcohol (PO/S) in the propoxylation of 1-octanol (+ run 6, 9 run 7, O run 8 of Table 1). Dots are experimental; lines are calculated. (b) Experimental and calculated oligomer distributions of run 6 at different reaction times.
of oligomer distribution during the time. In Figure 2a, which is related to the ethoxylation reaction of 1-octanol at 120 °C, it can be noted that the ethylene oxide consumption rate is roughly constant during time. This behavior is quite similar to that observed in dodecanol ethoxylation and, as in that case, the kinetic constant would be considered independent of the length of the
R
CH2O–M+ + CH2
CH
CH3
k12
CH3 R
CH2OCH2
CHO–M+
(10)
Secondary alcohols are less reactive than primary ones, i.e. k22/k12 < 1 (Gladkovskii and Ryzhenkova, 1971), and the reaction rate decreases, consequently. As the reaction with primary alcohol occurs faster than the successive reactions with propoxylated oligomers, distributions as the one reported in Figure 5b are obtained that are narrower than the ones related to the propoxy-
3852 Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996
lation of a secondary alcohol as the example reported in Figure 3b. Since the mechanism of the reaction can reasonably be assumed to be the same for both ethoxylation and propoxylation, a general kinetic scheme similar to that previously reported by Santacesaria et al. (1992a, 1992b) has been adopted, that is: in situ catalyst formation Ri
Ri
CHOH + MOH
R
CHO–M+ + H2O
R
(11)
initiation reaction Ri
O
CHO–M+ + CH2
R
CH
Rj
kij
R
Ri
Rj
CHOCH2
CHO–M+ (12)
propagation reactions
R
Ri
Rj
CHOCH2
CHO–M+ + CH2
O
R
R
CH
kjj
Rj
Figure 6. Kinetic constants arranged in an Arrhenius type plot.
Ri
Rj
Rj
CHOCH2
CHOCH2
CHO–M+
Ri
Rj
Rj
CHO(CH2
CHO)k–2CH2
CHO–M+ + CH2
R
O CH
(13)
The concentrations of the different ionic species can be determined by the following equilibrium equations
kjj
Rj
Keij =
Ri
Rj
Rj
CHO(CH2
CHO)k–1CH2
CHO–M+
Then, proton transfer equilibria must be considered. When using alkaline catalysts, equilibria would be considered independent of the alkoxylated chain length; therefore, we can write (Santacesaria et al., 1992a)
[R
[R
Ri
Ri
Rj
Rj
CHOH] [R
CHO(CH2
CHO)k–1CH2
CHO–M+]
Ri
Ri
Rj
Rj
CHO–M+] [R
CHO(CH2
CHO)k–1CH2
CHOH]
Ri [R
Ri
Rj
Rj
CHO–M+ + R
CHO(CH2
CHO)k–1CH2
CHOH
R
R
[R
Ri
Rj
Rj
CHOH + R
CHO(CH2
CHO)k–1CH2
CHO–M+
(14)
with i ) 1 or 2, j ) 1 or 2, R1 ) H, and R2 ) CH3. On the basis of this kinetic scheme, the following kinetic model can be written which is able to simulate all the kinetic runs of Table 1 Ri dm0 dt
= –kij [R
Ri dmk dt
[R
CHO(CH2
O
CHO–M+] [CH2 CH
Rj]
Rj
Rj
CHO)k–2CH2
CHO–M+] –
Rj
Rj
O
CHO)k–1CH2
CHO–M+])
= kjj ([R CHO(CH2
Ri
[CH2 CH
(15)
Rj]
k = 1, n (16)
where
mk = [R
CHO–M+] =
Ri
Rj
Rj
CHO(CH2
CHO)k–1CH2
CHOH] +
Ri
Rj
Rj
CHO(CH2
CHO)k–1CH2
CHO–M+
[R
B °m0
(18)
C + m0
Ri
Rj
Rj
CHO(CH2
CHO)k–1CH2
CHO–M+] =
Keij
Ri
(17)
together with the mass and charge balance equations (Santacesaria et al., 1992a)
proton transfer Ri
k = 1, n
mk B °m0 Keij m0 C + m0
k = 1, n
(19)
where B° is the catalyst concentration and C ) ∑k)1,nmkKeij. This model has been used to simulate all the kinetic runs reported in Table 1, applying the density values of Tables 2 and 3 and the epoxides concentration values of Tables 4. The kinetic and equilibrium constants have been determined by mathematical regression analysis on all the experimental data available. The values of the kinetic constants evaluated from the runs of Table 1 have been arranged in an Arrhenius type plot (see Figure 6) in which it clearly appears that k11 . k12 > k21 . k22. Equilibrium constants are roughly constant with temperature in the range considered. In Figures 2-5 the fitting obtained in the simulation of the kinetic runs for both epoxide consumption and oligomer distributions can be appreciated. In Table 5 preexponential factors and activation energies determined by regression are reported together with the constants of proton transfer equilibria. The activation energy calculated for k11 agrees with the value previously obtained by Di Serio et al. (1995) for dodecanol ethoxylation, 13.2 ( 0.3 kcal/mol, and with that reported for 1-butanol, 14.4 ( 0.6 kcal/mol, by Gee et al. (1959).
Ind. Eng. Chem. Res., Vol. 35, No. 11, 1996 3853 Table 5. Kinetic Constants for the Ethoxylation and Propoxylation of 1-Octanol and 2-Octanol, Respectively, and Proton Transfer Equilibria Constants for Four Cases Considered kinetic constants
ln A (cm3 mol-1 s-1)
E (kcal/mol)
k11 k12 k21 k22
20.52 ( 0.25 22.69 ( 0.03 23.83 ( 0.14 25.10 ( 0.19
13.0 ( 1.8 15.6 ( 0.5 16.8 ( 1.0 19.0 ( 1.9
Proton Transfer Equilibrium Constants Ke11
Ke12
Ke21
Ke22
2.0 ( 0.3
3.5 ( 0.4
2.2 ( 0.2
2.5 ( 0.3
Conclusions In this paper, the kinetics of ethoxylation and propoxylation of, respectively, 1- and 2-octanol, catalyzed by KOH, have been studied. The reaction mechanism seems to be the same for all the considered alkoxylations. Therefore, kinetic data have been interpreted by applying a previously developed kinetic model able to describe, for all the runs performed, the evolution with time of both epoxide consumption and oligomer distribution along the time. Observing the kinetic parameters obtained, it can be pointed out that the ratio of the addition rate of ethylene oxide to a primary alcohol with respect to propylene oxide r1 ) k11/k12 is always >1; on the contrary, the ratio of the addition rate of propylene oxide to a secondary alcohol with respect to ethylene oxide r2 ) k22/k21 is always
