Ind. Eng. Chem. Res. 1989,28, 1125-1130 Shing, K. S.; Chung, S. T. Computer Simulation Methods for the Calculation of Solubility in Supercritical Extraction Systems. J. Phys. Chem. 1987,91, 1674. Smith, R. D.; Tingey, J. M.; Blitz, J. P.; Fulton, J. L. Pressure Dependent Properties of Reverse Micelles and Microemulsions in Supercritical Fluids. Am. Chem. SOC.,Symp. Ser. 1989, No. 406. Squires, T. G., Paulaitis, M. E., E&. Supercritical Fluids: Chemical and Engineering Principles and Applications; ACS Symposium Series 329; American Chemical Society: Washington, DC, 1987. Stahl, E.; Quirin, K. W.; Gerard, D. Dense Gases for Extraction and Refining; Springer-Verlag, Berlin, 1988. Stanley, H. E. Introduction to Phase Transitions and Critical Phenomena; Oxford New York, 1971, p 47. Townsend, S. H.; Abraham, M. A.; Huppert, G. L.; Klein, M. T.; Paspek, S. C. Solvent Effects during Reactions in Supercritical Water. Znd. Eng. Chem. Res. 1988,27, 143. Tsonopoulus, C.; Heidman, J. L. High-pressure Vapor-Liquid Equilibria with Cubic Equations of State. Fluid Phase Equilib. 1986, 29, 391. Twu, C. H.; Gubbins, K. E. Thermodynamics of Polyatomic Fluid Mixtures 11. Polar, Quadrupolar and Octopolar Molecules. Chem. Eng. Sci. 1978, 33, 879. Van der Haegen, R.; Koningsveld, R.; Kleintjens, L. A. Solubility of Solids in Supercritical Solvents IV. Mean-Field Lattice Gas De-
1125
scription for the P-T-X Space Diagram of the System Ethylene-Naphthalene. Fluid Phase Equilib. 1988,43, 1. Walsh, J. M.; Ikonomou, G. D.; Donohue, M. D. Supercritical Phase Behavior: The Entrainer Effect. Fluid Phase Equilib. 1987,33, 295. Weeks, J. D.; Chandler, D.; Anderson, H. C. Role of Repulsive Forces in Determining the Equilibrium Structure of Simple Liquids. J . Chem. Phys. 1971,54(12), 5237-5247. Whiting, W. B.; Prausnitz, J. M. Equations of State for Strongly Nonideal Fluid Mixtures: Applications of Local Compositions Toward Density-Dependent Mixing Rules. Fluid Phase Equilib. 1982, 9, 119. Wong, J. M.; Johnston, K. P. Solubilization of Biomolecules in Carbon Dioxide Based Supercritical Fluids. Biotech. Prog. 1986, 2, 29. Wong, J. M.; Pearlman, R. S.; Johnston, K. P. Supercritical Fluid Mixtures: Prediction of the Phase Behavior. J. Phys. Chem. 1985,89, 2671-2675. Wu, A. H.; Stammer, A,; Prausnitz, J. M. Extraction of Fatty-Acid Methyl Esters with Supercritical Carbon Dioxide. Proc Znt. Symp. Supercritical Fluids 1988, 107. Received for review April 4, 1989 Accepted May 15, 1989
KINETICS AND CATALYSIS Synthesis of [2.2lParacyclophane via Hofmann Elimination: A Kinetic Study of the Homogeneous Reaction Paolo Beltrame,*-tPier Luigi Beltrame,? Paolo Carniti,t Pietro Delogu,f Carlo Pina,* and Giovanni Zurettit Dipartimento di Chimica fisica ed Elettrochimica, Uniuersitci di Milano, Via C. Golgi 19, I-20133 Milano, Italy, a n d Istituto G. Donegani, Via Caduti del Lavoro, I-28100 Novara, Italy
Paracyclophane (PCP) and byproducts were obtained from (p-methylbenzy1)trimethylammonium hydroxide (QOH) by reaction with NaOH in concentrated water solutions. T h e system presents a miscibility gap: selectivities and kinetics were separately studied in a NaOH-rich (”Na”) region, a QOH-rich (“Q“) region, and a “dilute” region, a t temperatures in the range 70-110 “ C . The reaction rate is much higher in the Na region, particularly a t high levels of NaOH concentration, but selectivity to PCP is better in the other regions. The dependence of the reaction rate on the alkali concentration is discussed, relating it t o the poor solvation of OH- ions. Thin polymeric coatings, characterized by excellent properties, are obtained from the cyclic p-xylylene dimer (or [2.2]paracyclophane) by pyrolysis, vapor-phase polymerization, and vacuum deposition on solid surfaces (KirkOthmer, 1980). Such coatings are used, for instance, on electronic devices in heavy-duty applications. The dimer was previously produced by p-xylene pyrolysis in the vapor phase a t high temperatures and low pressures (Pollard, 1964); now a condensed-phase reaction is preferred, using the Hofmann elimination of a (p-methylbenzy1)trimethylammonium derivative, such as the chloride or the hydroxide, with aqueous alkali. In the common version of this process, the reaction is carried out in the presence of an organic solvent, immiscible with water. The process is thus typically heterogeneous. Byproducts, in particular, large quantities of a linear polymer, are obtained besides paracyclophane. The resulting low selectivity in the useful product is responsible for its high cost and, as a conse-
quence, for the quite limited number of economic applications. In order to contribute to a better understanding of this process, we studied the ternary system “(p-methylbenzy1)trimethylammonium hydroxide (QOH) + sodium hydroxide + water” a t 25 OC; even without a specific organic solvent, a phase splitting was recognized, giving rise to two phases, both aqueous, one richer in QOH (“Q phase”), the other in NaOH (“Na phase”). Within the regions of existence of the homogeneous phases, the Hofmann elimination reaction was studied under kinetic conditions, determining the product selectivities and measuring the reaction rate, in the temperature range from 70 to 110 “C.
Experimental Section Materials. An industrial aqueous solution of (pmethylbenzy1)trimethylammonium chloride (QC1, ca. 2.6
0888-5885/89/2628-1125$01.50/00 1989 American Chemical Society
1126 Ind. Eng. Chem. Res., Vol. 28, No. 8, 1989
mol/L) was washed in sequence with p-xylene, dichloromethane, and diethyl ether and then concentrated at reduced pressure; finally QCl was crystallized from methanol: mp 225 "C, as in the literature (Winberg, 1960). A solution of (pmethylbenzy1)trimethylammonium hydroxide (QOH) was obtained from aqueous QCl by adding excess Ag,O, filtering, and washing. The solution was kept in the dark. When required, concentrated QOH solutions, up to 50 wt % , were obtained by water evaporation under reduced pressure. [2.2]Paracyclophane (PCP), p-methylbenzyl alcohol (MBAL), p-methylbenzaldehyde (MBAD), 1-methylnaphthalene (MN), and 2,6-di-tert-butyl-4-methylphenol (DTBMP) were commercial products with declared purities of 97%, 98%,99%, 98%, and 99%, respectively. Bis(p-methylbenzyl) ether (MBE) was prepared by heating MBAL in dimethyl sulfoxide: mp 63 "C, as reported (Emert et al., 1977); correct elemental analysis; lH NMR (200 MHz, CCl& 6 2.33 (CH,), 4.41 (CHZO), 7.10 (Ar); hydrogen ratios 6:4:8, as expected for MBE. Analytical Methods. Total alkalinity was determined by titration with aqueous HCl. Two methods were employed for analysis of QOH: (i) UV absorption at 264 nm (molar extinction e = 239 L mo1-lcm-l) for solutions containing no other organic compound, that is for room-temperature experiments; (ii) a colorimetric method, based on the formation of a complex between QOH and bromophenol blue (Longman, 19671, for measurement of fractional conversion during product determination and kinetic runs. The last method has been found to be reliable for a quantitative determination based on the Lambert-Beer equation, provided the following procedure is carefully followed. The initial solution contained concentrations of (7-9) X lo-, mol/L QOH and ca. 0.1 mol/L NaOH. Bromophenol blue (25 mg) and NaCl (600 mg) are dissolved in 2 mL of the solution. By extraction five times with CHCl, (20 mL each time), most of the colored complex is taken out of the water phase, in a reproducible way. Absorption of the organic phase, diluted as required, is measured a t 604 nm. The procedure was calibrated by appropriate blanks. In a few cases, when QOH concentration was below the stated limit, the method was still applied, but the dependence of the extinction coefficient on the concentration had to be considered for quantitative analyses. Most reaction products were determined by GC of the xylene solutions on a SE-30 column (2 m long), with a TC detector, H2 (30 mL/min) as carrier, isothermally a t 150 "C. Products were partly identified by GC-MS; in any case, peaks were assigned by comparison with pure Samples. The latter were also used for calibration of the quantitative analysis, using MN as the internal standard. were found to be a function of the Calibration factors concentration of the correspondingproduct: this was taken into account by expressing fi as a linear function of the ratio area of peak ilarea of peak MN. Approximate retention times (minutes) were as follows: p-xylene, 0.52; MBAD, 0.87; MBAL, 1.05; MN, 1.85;DTBMP (when used as inhibitor), 3.7; PCP, 10.9; MBE, 17.5. In a few cases, the GC analysis was repeated at 200 "C; in these conditions, MBE had a retention time of 2.9 min, and a new peak, with ca. 30-min retention time, was observed. The product was recognized as the cyclic trimer of p-xylylene (TI, by isolating a sample from the reaction mixture, as follows. The solvent was completely evaporated from the p-xylene solution of the soluble product mixture. The residue was partially dissolved by quick treatment with a small amount of hexane (Errede and
vi)
//
'
\c2,5,22
/
L .
NaOH
Figure 1. System QOH-NaOH-H20: schematic representation of the upper part of the phase diagram. The approximate positions of the mixtures used in the product determinations and kinetic runs are numbered as in Tables I11 and IV (runs at 90 "C, without inhibitors).
Cassidy, 1960). After evaporation of the solvent and crystallization from methanol, column chromatography on silica gel, with petroleum ether (40-60 "C):diethyl ether (1OO:l) as eluent, brought a pure crystalline product: mp 164 "C (from acetone); 'H NMR (200 MHz, CDC13) 6 2.95 (CHJ, 6.72 (Ar), hydrogen ratio 1:l. These properties are close to those described for tetracyc10[14.2.2.2~,'.2~~J~]tetracosa-4,6,10,12,16,18,19,21,23-nonaene (Errede and Cassidy, 1960). Further elution with diethyl ether and solvent removal gave a product with the retention time of MBE. For the determination of product T by GC, a constant calibration factor (measured in the appropriate concentration range) was employed. Phase Diagram. A separating funnel, jacketed for thermostating purposes and equipped with a thermowell for temperature control, was used to determine the tie lines. To a given amount of aqueous QOH of known concentration, chilled to 16-17 "C, a weighed amount of solid NaOH was gradually added, keeping the temperature under 25 "C. The biphasic system was left a t 25 "C for 30 min and then the phases were separated and analyzed for total alkali and for QOH (by UV absorption). Sodium hydroxide was evaluated by difference. In order to outline the boundaries of the miscibility curve in its upper part, a few additional points were determined by "titrating" with water a biphasic system of known overall composition, up to the appearance of complete miscibility. In order to have a tool able to interpolate between experimental data, liquid-liquid data reduction was performed according to the NRTL model (Renon and Prausnitz, 1968). Several objective functions (OF) have been proposed for the parameter estimation (Sarensen et al., 1979). It has been shown that all functions give the same results when they are weighed according to the maximum likelihood principle (Carvoli et al., 1982). In this work, we used the "concentration" OF (eq I), which is a "maximum likelihood" OF when all the experimental variances are equal (x = molar fraction). OF =
no. tie lines 2 mx&%CH,l - xf$8H,l)2 +
c
k=l
1=1
(x&b,c
- x3&,O2I (1)
Computations were performed by the computer program obtained from Sarensen et al. (1979). Product Determination and Kinetic Runs. Reaction mixtures were obtained by mixing the components in such amounts as to fall in the monophasic regions of the phase diagram (Q, Na, or dilute region, as shown in Figure 1). When required, a polymerization inhibitor (I) was added as a solid.
Ind. Eng. Chem. Res., Vol. 28, No. 8, 1989 1127 Table 1. Ternary System QOH-NaOH-H,O at 25 "C. Tie Lines and Additional Points at Phase Boundaries (x = Molar Fraction) Na phase Q phase
G C F A
B E A3 A1 A2
XQOH,~
XN~OH,I
xH*O,l
xQOH,2
xNaOH,2
xHa0.2
0.13063 0.09691 0.08394 0.07043 0.06535 0.04915 0.0435 0.0307
0.05787 0.04796 0.04245 0.03780 0.03684 0.04389 0.0396 0.0471
0.81150 0.85513 0.87361 0.89177 0.89781 0.90696 0.9169 0.9222
0.00159 0.00176 0.00205 0.00295 0.00323 0.00838
0.26119 0.20321 0.16088 0.12802 0.10533 0.07807
0.73722 0.79503 0.83707 0.86903 0.89144 0.91355
0.0136
0.0639
0.9225
In detail, to obtain mixtures in the Na region, aqueous solutions with known concentrations of NaOH and QOH were mixed, while mixtures in the Q region were prepared by using solid NaOH. It has been noticed that adding NaOH as an aqueous solution (instead of a solid) to QOH aqueous solutions may give clear solutions even at overall compositions for which phase separation is expected from the phase diagram. Thus, metastable behavior is involved. A few runs were carried out with these metastable monophasic systems. For product determinations, most of the solution was poured into Pyrex tubes (ca. 50 mL) with screw caps and PTFE gaskets, while a minor amount was poured (in 1mL portions) into small flasks (ca. 2 mL) adapted with similar caps and gaskets. For kinetic runs, only the small flasks were needed. Reactions were carried out a t constant temperature to hO.1 "C. When the fractional conversion of QOH had to be measured, at a given time a flask was taken out of the thermostat, the reaction quenched by quick cooling in iced water, the content diluted with water, and (for the Na region only) ca. 90% of the alkali neutralized with HC1. Determination of QOH was effected according to the described colorimetric method, after dilution and alkalinity correction, if required. In the case of kinetic runs, the QOH concentration as a function of time was plotted as -(ln C OH) versus t , obtaining linear plots from which pseulo-first-order coefficients (k) were evaluated. When k values at two different temperatures were available, the apparent activation energy was calculated. When products were looked for, a t an appropriate time, while a flask was withdrawn for measuring the QOH fractional conversion, the tubes were withdrawn for analysis. The reaction in the tubes was quenched by cooling; the contents were neutralized by HC1 to pH 5 and then refluxed with p-xylene under N2 during 24 h, as required for the complete extraction of reaction products, apart from an insoluble polymer. The hot system was filtered, the white solid washed with water, p-xylene, and acetone, dried, and weighed. After calcination for 2 h a t 600 "C, the solid was weighed again, the difference being taken as p-xylylene polymer, while the residue was silica. The xylene phase was separated, concentrated by evaporation, and brought to a known volume; 50 p L of a solution of MN in p-xylene, containing 1.78 mg of MN, was added to 1 mL of the product solution; GC analysis followed. The mass of each product, as given by GC, was expressed by its content of p-xylylene and referred to the total amount of QOH converted in the product tubes. The polymer mass from gravimetric analysis was then added, and the selectivities were calculated. Results and Discussion By studying the system QOH + NaOH + HzO, data were obtained about the tie lines and about additional
f"ZO
V
0.3
0.7
/'
/
/ QOH
0.1
/
1
0.2
0.3
---L
s
d NaOH
NaOH
Figure 2. System QOH-NaOH-H20: upper part of the phase diagram a t 25 "C. Figures are molar fractions. Circles: experimental or without ( 0 )tie-line determination]. Tie lines: points [with (0) full, experimental; broken, calculated with the NRTL model. Table 11. Values of NRTL Parameters Obtained for the Liquid-Liquid System QOH-NaOH-H20 at 25 "C
'(Y
i
i
$OH QOH H2O
HzO NaOH NaOH
Tij,
K
-1557 -350 508
K
(Yo
2922 2369 -503
0.2 0.2 0.3
Tji,
values were assigned; 7 values were optimized.
points defining the phase boundaries. The results are presented in Table I and in Figure 2. Since the experimental points refer in any case to x N ~ O H< 0.3 and xQOH < 0.15, only the upper part of the ternary plot was actually determined. In order to correlate the results, at least to an empirical level (i.e., neglecting the electrolyte nature of components QOH and NaOH), a NRTL correlation program was employed. The calculated tie lines in Figure 2 correspond to the parameter values collected in Table 11. Of course, this correlation holds only for compositions with large water content, avoiding the region where solid phases are involved. The determination of the reaction products was carried out by a run a t 70 "C and several runs a t 90 "C. The results are given in Table 111. A secondary reaction was always present, giving rise to p-methylbenzyl alcohol (MBAL) and the corresponding ether (MBE). The alcohol was often accompanied by p-methylbenzaldehyde (MBAD), usually in small amounts. The formation of MBAL and MBE is justified by nucleophilic attack at the benzylic carbon of the Q' ion, with substitution of trimethylamine, as in Scheme I. When R = H, this reaction gives MBAL; when R = CH3C6H4CH2(that is, RO-is the anion of MBAL), the reaction gives MBE. Alcohols have been known for a long time as secondary products of the Hofmann "exhaustive methylation" (Hanhart and Ingold, 1927).
1128 Ind. Eng. Chem. Res., Vol. 28, No. 8, 1989 Table 111. Product Determinations at 90.0 OC (Unless Otherwise Specified) conversion initial concn, mol/L of QOH run QOH NaOH I t, h YQOH MBAD MBAL Na Region 40 144 0.0063 12.0 0.614 6.8 5 48 0.816 0.3 1.9 0.0900 12.0 6 48 0.0079 12.0 0.860 5.5 3.2 0.05 x 10-ab 48 I 0.795 2.2 0.0811 12.0 8 48 0.780 0.3 3.0 0.49 X 12.0 0.0786 2.06 X 48 0.707 0.4 2.9 9 0.0798 12.0 48 0.701 0.4 2.9 4.14 X 12.0 10 0.0789 14 48 3.87 X 0.795 0.4 2.4 0.0794 12.0 16 48 0.719 0.3 2.5 15.5 X 0.0798 12.0 0.811 2.4 22 48 0.0789 12.0 2 0.809 2.4 24 0.0827 16.5
selectivities, % MBE PCP T
polymer
(E)
4.1 4.8 1.2 4.9 6.1 7.3 8.0 4.7 6.0 6.6 5.7
12.1 11.8 6.3 13.4 17.7 20.3 22.9 16.0 18.0 10.9 11.4
nd nd nd nd nd nd nd nd nd 2.1 2.2
70.5 76.1 78.6 73.6 65.8 62.0 58.9 69.7 65.9 77.1 77.4
(93.5) (95.0) (94.8) (94.2) (93.0) (92.9) (93.1) (93.2) (92.6) (99.1) (99.1)
0.2
11.7
22.3
4.1
60.8
(99.1)
Q Region 23
2.95
1.18
168
0.422
13 15 17 18
0.64 1.82 0.65 0.64
2.30 1.27 2.27 2.32
264 360 360 360
Dilute 0.264 0.354 0.302 0.333
4.8 4.5 5.7 5.5
10.8 10.5 13.4 13.0
28.3 26.5 34.8 29.9
nd nd nd nd
47.5 49.1 35.9 41.7
(92.0) (91.1) (90.5) (90.7)
12 21
1.99 2.01
2.30 2.29
168 216
Metastable Systems 0.421 0.4 3.2 0.559 5.4
8.7 12.5
24.9 24.2
nd 4.3
54.6 52.9
(91.8) (99.2)
45.6 4.56
X X
Region 0.5 0.5 0.7 0.6
OTemperature, 70.0 "C. bInhibitor = DTBMP. 'Inhibitor = AgN03.
Scheme TI -NICH3$
OH-+ H
,
C
G
CH,NICH1l,
+
- H,O
The aldehyde MBAD is considered to be an oxidation product from MBAL. The procedure of product separation, involving a prolonged p-xylene extraction, could be the origin of a t least part of MBAD and MBAL by oxidation of p-xylene, if carried out in air, as experimentally verified. However, the extraction was performed under nitrogen; in these conditions, it has been checked that MBAD and MBAL are not produced from p-xylene. The main products can reasonably be attributed to alkaline attack at the p-methyl hydrogen, elimination of trimethylamine with formation of p-xylylene, and further reaction of the latter, according to Scheme 11. The products that have been identified are [2.2]paracyclophane (PCP), the cyclic p-xylylene trimer (T),and an insoluble polymer. In most cases, T was not determined; when a modified analytical procedure was applied, in order to include T among the measured products, it was found that a few percent of the overall product was represented by this trimer. The mass balance of the runs where T had been neglected was somewhat deficient (91-95% recovery); when T was determined, a more satisfactory mass balance was found (99% recovery). It can be noticed (Table 111) that the main derivative of p-xylylene was always the (presumably 'linear) polymer. In order to improve the yield of PCP, by hsing polymerization inhibitors, addition of 2,6-di-tert-butyl-4-methylphenol (DTBMP) (Ungarelli et al., 1987) or silver nitrate was tried. Although they also slightly lower the overall fractional conversion of QOH, these inhibitors improve the PCP selectivity. Taking run 5 as the reference (no in-
Table IV. Kinetic Runs for Overall Conversion of QOH initial concn, mol/L run temp, OC QOH NaOH 103k, h-' Na Region 4.18 i 0.06 0.074 10.0 c1 90 0.076 12.1 c2 90 32.6 f 0.6 c3 90 0.081 14.3 158.2 f 3.4 842 i 14 c4 90 0.083 16.7 c5 110 0.074 10.0 14.6 f 0.1 277 i 3 c7 70 0.083 16.7 3.24 f 0.02 20.8 i 0.3 3.28 i 0.03
Dilute Region 0.64 2.30 1.82 1.27 1.95 1.15
1.19 i 0.13 1.19 f 0.03 1.18 f 0.01
Metastable Region 1.99 2.30 1.98 2.30 AE., kJ/mol C4 and C7 57.5 C1 and C5 72.4 C9 and C10 107.4 12 and C6 107.9
3.55 f 0.61 23.0 i 0.3 COON-.mol/L 16.7 10.0 4.33 4.49
90 110 90
13 15 C8
90 90 90
12 C6 from from from from
Q Region 2.98 2.98 3.37
1.36 1.34 0.98
c9 c10 c11
90 110
hibitor; selectivity N 12%), it can be seen that increasing amounts of DTBMP gradually enhanced the selectivity 23%, while values up to 18% were reached by up to adding AgN03, for reactions in the Na region. As for reactions in the dilute region, run 13 can be taken as the reference (selectivity 28%) and addition of DTBMP is seen to increase the PCP selectivity up to 35%, the larger addition giving the larger effect (Table 111). Run 17 is noticeable, since its PCP selectivity almost reached the polymer selectivity. Rate measurements were carried out for the overall reaction of QOH. First-order rate coefficients, tZ (Table IV), measured at different levels of alkali concentration, proved
Ind. Eng. Chem. Res., Vol. 28, No. 8, 1989 1129
~~
10
15
20
in
c:~.
25
30
Figure 3. Pseudo-first-order coefficient 12 versus cooH- (full circles, N a region; empty circles, other regions).
to depend very strongly on such concentrations. Equation 2 was applied to the results. In it, CooH-represents in
k = k,(CooH-)"
log k = log k, + n log
COOH-
(2)
general the overall initial concentration of alkali (=C'N,OH C'QOH), while it is CooH- N CoNaOH for runs in the Na is negligible with respect to C'N,OH. region, where COQ~H Evidence that, in other cases, the overall alkali concentration has to be considered is given by k values in Table I V runs 13,15, and C8, having different values of CoNaOH and C o Q o H but close values of C'OH- (~2.9-3.1mol/L), h-'1. An showed the same k values t(1.18-1.19) x analogous comparison is possible of run C9 with C11. For runs at 90 "C in the Na region, eq 2 gives a value of n = 10.4 for the apparent order of the reaction with respect to alkali, For reactions at 90 "C in other regions, considered all together, such an apparent order was found to be ca. 3. The logarithmic plots are presented in Figure 3. It should be appreciated that the metastable system of run 12 was close to the Q region. A few kinetic runs were carried out a t temperatures other than 90 "C. This allowed the evaluation of the apparent activation energy for four couples of runs, characterized by different levels of alkali concentration (Table IV). From runs with C0oH- = 16.7 and 10.0 mol/L, values of AE, of 57.5 and 72.4 kJ/mol, respectively, were obtained; from runs with CooH- N 4.4 mol/L, a AE, of 107-108 kJ/mol was found. So the apparent activation energy proves to decrease when the alkali concentration is increased. A reasonable interpretation of the experimental results is based on the consideration that OH- ions are poorly or very poorly solvated in the solutions considered here. For the runs in the Q and the dilute regions, the initial composition of the reaction mixtures assures roughly from 3 to 9 water molecules for each ion (Na+ or Q' or OH-). This gives rise to a situation of poor solvation so that an increase in the ions concentration involves a marked decrease in the solvation of OH- and therefore a considerable increase in its reactivity (apparent reaction order 3). The situation in the Na region is even worse. In run C1, the composition approximately corresponds to 19.8 Na+ ions and 20.1 other ions (mainly OH-) every 100 H 2 0 molecules; taking into account that the most probable value of the primary solvation number of Na+ is 4 (Bockris and Reddy, 1970), it appears that there is insufficient
+
solvation. In run C4, there are 38.0 Na+ ions and 38.3 other ions every 100 H 2 0 molecules, so there is even less water than required for the primary solvation of Na+: the case is one of very poor solvation. The observed apparent order of ca. 10 with respect to OH- is born out of these conditions. The poorly solvated OH- ion meets, in these cases, lower energy barriers in its reaction with Q+, as confirmed by the apparent activation energy values. For the sake of comparison, one may consider the Hofmann elimination reaction of tetrahexylammonium hydroxide studied by Landini and Maia (1984) under phase-transfer catalysis conditions (chlorobenzene-aqueous NaOH). In that case, the hydration state of quaternary ammonium hydroxide in the chlorobenzene phase was measured at 25 "C and found to vary from n = 11.0 to n = 3.5 at increasing NaOH concentration in the water phase (from 4.8 to 20.0 mol/L): a strong increase in reactivity of OH- was observed on reducing its hydration sphere (Landini and Maia, 19841, corresponding to an apparent reaction order of ca. 6 with respect to NaOH. Conclusions The reaction was found possible both in the water-poor Na region and in the Q or dilute regions. In the former case, the reaction rate is larger, and it strongly increases with NaOH concentration; however, selectivity to PCP is only around 11% (23% with a polymerization inhibitor). Better selectivities were obtained in the Q or dilute regions, up to 28% without inhibitors and up to 35% in the presence of an inhibitor. In the case of a heterogeneous process, when both a Na phase and a Q phase are present, the largest contribution is given by the reaction in the former. Since a better selectivity is primarily looked for, one should avoid the heterogeneous process and should operate in the Q or dilute regions. Unfortunately, in these regions, the specific rate of production of PCP can be lower by 2 orders of magnitude and is unacceptable from a practical point of view. So, either because of slow kinetics or because of low selectivity, none of the conditions of the present work are suitable for commercial application. However, a better knowledge of the behavior of the different phases has been reached, and it should help in finding reaction conditions of practical interest. It can be noticed that the Q region, where selectivity is much better, is characterized by a medium that is (by weight) roughly half aqueous and half organic, considering the organic nature of QOH, while the solvent in the Na region is almost only water. Therefore, a possible way out of the difficulties encountered is the search for a mixed aqueous-organic solvent, able to improve the mutual solubilities, in order to have acceptable levels of both selectivity and reaction rate. On this line, some suggestions have recently appeared in the patent literature (Bornengo et al., 1988). Acknowledgment
Dr. F. Guerrieri is thanked for his suggestions as to the colorimetric analysis of QOH. Dr. A. Rampoldi is thanked for the separation of product T by column chromatography. Literature Cited Bockris, J. OM.; Reddy, A. K. N. Modern Electr0chemistry;'Plenum Press: New York, 1970; Vol. I, p 131. Bornengo, G.; Maiacrida, A.; Campolmi, S.; Beretta, M. A. Eur. Pat. Appl. EP 253 191, 1988. Carvoli, G.; Cori, L.; Delogu, P. h o c . 3rd Austrian-ltalian-Yugoslau Chem. Eng. Conf. 1982, 1, 188.
Ind. Eng. Chem. Res. 1989, 28, 1130-1140
1130
Emert, J.; Goldenberg, M.; Chiu, G. L.; Valeri, A. J. Org. Chem. 1977, 42, 2012. Errede, L. A.; Cassidy, J. P. J. Am. Chem. SOC.1960, 82, 3653. Hanhart, W.; Ingold, C. K. J . Chem. SOC.1927, 997. Kirk-Othmer Encyclopedia of Chemical Technology, 3rd ed.; Wiley: New York, 1980; Vol. 10, p 253. Landini, D.; Maia, A. J. Chem. SOC.,Chem. Commun. 1984, 1041. Longman, G. F. The Analysis of Detergents and Detergent Products; Wiley: London, 1967; p 262. Pollard, D. F. US Pat. Appl. 3 149 175, 1964.
Renon, H.; Prausnitz, J. M. AIChE J . 1968, 14, 135. S0rensen, J. M.; Magnussen, T.; Rasmussen, P.; Fredenslund, A. A. Fluid Phase Equilib. 1979, 2, 297; 1979, 3, 47; 1980, 4, 151. Ungarelli, R.; Beretta, M. A.; Sogli, L. Ital. Pat. Appl. 20 059A/87, 1987. Winberg, E. J . Am. Chem. SOC.1960,82, 1428. Received for review January 3, 1989 Accepted May 15, 1989
Activation Studies with a Promoted Precipitated Iron Fischer-Tropsch Catalyst Dragomir B. Bukur,*it Xiaosu Lang,?Joseph A. Rossin,+William H. Zimmerman,? Michael P. Rosynek,t Eshan B. Yeh,* and Chiuping Lil Kinetics, Catalysis and Reaction Engineering Laboratory, Department of Chemical Engineering and Department of Chemistry, Texas A&M University, College Station, Texas 77843
T h e effect of reductant type (CO, Hz, or Hz/CO = 0.68), activation temperature (250, 280, and 310 "C), duration (8 and 24 h), and pressure (0.1 and 1.48 MPa) on the activity and selectivity of a promoted precipitated iron catalyst (100Fe/3Cu/0.2K) was studied in a fixed bed reactor. I t was found that activation parameters have a strong effect on the catalyst behavior during Fischer-Tropsch (FT) synthesis a t 250 "C, 1.48 MPa, 2 nL/g of catalyst/h, and H 2 / C 0 = 1. Activations in CO led t o catalysts that had higher initial activity and better selectivity (less methane and other light hydrocarbons) than H, activated catalysts. However, the activity of CO-reduced catalysts declined with time on stream, whereas the activity of Hz-reduced catalysts remained constant or increased during 120 h of testing. Activation in CO a t 280 "C and 0.1 M P a for 24 h was the most desirable on the basis of overall catalyst activity, selectivity, and stability. The most common catalysts for CO hydrogenation (Fischer-Tropsch (FT)synthesis) are group VI11 elements: cobalt, nickel, ruthenium, and iron. Before synthesis, a catalyst precursor is subjected to a pretreatment, the purpose of which is to bring the catalyst into an active form for synthesis. Cobalt, nickel, and ruthenium are almost always reduced in flowing H2 a t 200-450 "C to the zerovalent metallic state. During the synthesis, under a variety of process conditions, these catalysts remain in the zerovalent state (Anderson, 1956). However, the purpose of pretreatment for iron catalysts is not so clear. Reduction in H2 may lead to the zero-valent state, but during the synthesis, the metallic iron is rapidly converted to a carbide phase or phases (e.g., Amelse et al. (1978), Raupp and Delgass (1979), and Dry (1981)). At high H2 + CO conversions, the reaction mixture becomes more oxidizing (relatively high HzO/H, and CO2/CO ratios) and magnetite (Fe304)is also formed (e.g., Anderson (19561, Dry (1981), and Satterfield et al. (1986a,b)). Other pretreatments have also been employed, such as CO reduction (activation), synthesis gas treatment (induction), and/or H2 reduction followed by CO or vice versa. These activations often yield a better catalyst than that obtained by Hz reduction, but the catalyst composition still changes during synthesis. Numerous studies have been published concerning correlations between phases present in the iron catalyst and its reaction behavior, as summarized, for example, by Dweyer and Hardenbergh (1984) and Satterfield et al. (1986a). However, there is no clear consensus as to which of the phases is responsible for catalyst activity. The two extreme views are that the active phase(s) is(are) either (a) iron carbides (e.g., Amelse et al. (1978), Raupp and
* Author t o whom correspondence should be addressed. Department of Chemical Engineering. *Department of Chemistry.
Delgass (1979), and Niemantsverdriet et al. (1980)) or (b) iron oxides (Reymond et al., 1982; Blanchard et al., 1982). Based on the extensive studies at the US Bureau of Mines and the previous German work, the following iron phases were found to be catalytically active: metallic a-Fe, various carbides, nitrides, and carbonitrides, and Fe304 (Hofer, 1956, p 434). During synthesis, industrial catalysts consist of mixtures of carbides and magnetite, and in general, there is no clear correlation between the catalyst bulk composition and its activity and/or selectivity (Dry, 1981, p 197; Anderson, 1984, p 56). Thus, the objectives of the catalyst pretreatment in the case of iron are not clear. The general goals of a successful pretreatment are to obtain high activity, the desired selectivity, and long life (high stability). The effective activation procedures for iron catalysts have been developed empirically. There have been only a few studies on the effect of activation parameters on subsequent catalyst activity, selectivity, and stability during the synthesis. In early work with alkalized precipitated catalysts at the Kaiser Wilhelm Institute in Germany, it was found that reduction with H, at 360 "C was not effective, and further studies focused on activations with carbon monoxide and CO-rich synthesis gas (Anderson, 1956, pp 176-180). They investigated the effects of activation temperature (255-450 "C) and pressure (0.01-1.5 MPa) on catalyst activity and stability. The optimal activation parameters were CO a t 0.01 MPa (subatmospheric pressure) and 325-345 "C. These activation conditions gave high H2 + CO conversions and constant catalyst activity for 80-120 days in fixed bed reactor tests. In studies with a precipitated iron catalyst a t the US Bureau of Mines, several activation procedures were emCO, and H2/C0 ployed with different reducing agents Hz, = 1 or 2 (Anderson, 1956, pp 183-184; Anderson, 1984, pp 56-58). It was found that catalysts activated with syngas
0888-5885/89/2628-1130$01.50/0 0 1989 American Chemical Society