Process research on the preparation of 1-(3-trimethylsilylphenyl)-2,2,2- trifluoroethanone by a friedel-crafts acylation reaction

Article abstract of DOI:10.1021/op700199p

Zifrosilone (1-(3-trimethylsilylphenyl)-2,2,2-trifluoroethanone) (3) is a Cholinesterase inhibitor that has been studied for the treatment of Alzheimer's disease. Process research has been carried out on a route to convert phenyltrimethylsilane to 3 by Friedel-Crafts acylation using trifluoroacetic anhydride. Kinetics and products analyses suggest that the optimal conditions for this reaction are noncatalytic amounts of aluminum chloride, dichloromethane solvent and as low a temperature as can be practically used in a scaled-up process. Significant separation challenges to isolate 3 from the isomer byproduct 1-(4-trimethykilylphenyl)-2,2,2-trifluoroethanone) (4) remain. These challenges were investigated using vapor-liquid equilibrium studies.

Full text of DOI:10.1021/op700199p

Organic Process Research & Development 2008, 12, 23–29
Full Papers
Process Research on the Preparation of
1-(3-Trimethylsilylphenyl)-2,2,2-trifluoroethanone by a Friedel–Crafts acylation
Reaction
Richard A. Wolf*
Dowpharma, The Dow Chemical Company, 1710 Building, Midland, Michigan 48674, U.S.A.
Abstract:
should be noted.11 An earlier version of the present research
was reported in a U.S. Patent application.12
Zifrosilone (1-(3-trimethylsilylphenyl)-2,2,2-trifluoroethanone) (3)
is a cholinesterase inhibitor that has been studied for the treatment
of Alzheimer’s disease. Process research has been carried out on
a route to convert phenyltrimethylsilane to 3 by Friedel–Crafts
acylation using trifluoroacetic anhydride. Kinetics and products
analyses suggest that the optimal conditions for this reaction are
noncatalytic amounts of aluminum chloride, dichloromethane
solvent and as low a temperature as can be practically used in a
scaled-up process. Significant separation challenges to isolate 3
from the isomer byproduct 1-(4-trimethylsilylphenyl)-2,2,2-tri-
fluoroethanone) (4) remain. These challenges were investigated
using vapor–liquid equilibrium studies.
The aluminum chloride catalyzed electrophilic aromatic
substitution (trifluoroacylation-deprotonation) of arenes by
trifluoroacetic anhydride (TFAA) has recently been cited in the
literature.13–16 However, the reaction of such trimethylsilyl arenes
as phenyltrimethylsilane (1) and 1,3-bis(trimethylsilyl)benzene17–22
(2) with TFAA is not so clear-cut (Scheme 2). The trimethylsilyl
group has been known to be displaced from aromatic rings by
acyllium ions. Thus, 2 has been reacted with acetyl chloride
and aluminum chloride in carbon disulfide solvent to form
3-trimethylsilyl acetophenone in a yield of 44%.23 Another paper
cites this reaction (with acetic anhydride) as occurring with
”excellent yields”.24
Ipso substitution of 2 has been reported with dichlorometh-
ylmethyl ether and aluminum chloride (in dichloromethane
solvent) to form, after workup with water, trimethylsilyl
benzaldehydes (Scheme 1).25 The regiochemistry for the reac-
tion, however, was only 89% meta product, with 9.5% of
p-trimethylsilylbenzaldehyde also formed. Compound 2 has
been reacted with (hydroxy(tosyloxy)iodo)benzene in acetoni-
trile solvent to effect an ipso electrophilic aromatic substitution
1. Introduction
Zifrosilone (1-(3-trimethylsilylphenyl)-2,2,2-trifluoroetha-
none) (3) is a cholinesterase inhibitor that has been studied for
the treatment of Alzheimer’s disease.1–8 The preferred synthetic
method has been a bis-metalation process, starting with 1,3-
dibromobenzene and sequentially attaching the trimethylsilyl
and trifluoroacetyl groups, to form 3.9,10 A warning regarding
the potential explosiveness of 3-bromophenyllithium, a potential
byproduct in this and some other potential routes to Zifrosilone,
(11) Leleu, M. J. Cah. Notes Doc. 1977, 88, 367.
(12) Wolf, R. A. PCT Int. Appl. WO 95-US4837, April 18, 1995.
(13) Liu, K.; Xu, L.; Berger, J. P.; MacNaul, K. L.; Zhou, G.; Doebber,
T. W.; Forrest, M. J.; Moller, D. E.; Jones, A. B J. Med. Chem. 2005,
48, 2262–2265.
* To whom correspondence should be addressed. Telephone: 859 261-3743.
E-mail: wolfrap@hotmail.com.
(1) Lane, R. M.; Polymeropoulos, M. H. PCT Int. Appl. WO 2003-EP8719
20, August 6, 2003.
(2) Coe, J. W.; Sands, S. B.; Harrigan, E. P.; O’Neill, B. T.; Watsky,
E. J., U.S. Pat. Appl. 2001-760966, January 16, 2001.
(3) Tracey, K. J.; Pavlov, V. A. PCT Int. Appl. WO 2005-US3029, January
27, 2005.
(14) Mahindroo, N.; Huang, C.-F.; Peng, Y.-H.; Wang, C.-C.; Liao, C.-
C.; Lien, T.-W.; Chittimall, S. K.; Huang, W.-J.; Chai, C.-H.; Prakash,
E.; Chen, C.-P. J. Med. Chem. 2005, 48, 8194–8208.
(15) Adams, A. D.; Hu, Z.; Langen, D. von; Dadiz, A.; Elbrecht, A.;
MacNaul, K. L.; Berger, J. P.; Zhou, G.; Doebber, T. W.; Muerer, R.;
Forrest, M. J. Bioorg. Med. Chem. Lett. 2003, 13, 3185–3190.
(16) Corrie, J. E. T.; Munasinghe, V. R. N.; Rettig, W. J. Heterocycl. Chem.
2000, 37, 1447–1456.
(4) Dow, J.; Dulery, B. D.; Hornsperger, J.-M.; Di Francesco, G. F.;
Keshary, P.; Haegele, K. D. Arzneim.-Forsch. 1995, 45, 1245–1252.
(5) Cutler, N. R.; Seifert, R. D.; Schleman, M. M.; Sramek, J. J.; Szylleyko,
O. J.; Howard, D. R.; Barchowsky, A.; Wardle, T. S.; Brass, E. P.
Clin. Pharmacol. Ther. 1995, 58, 54–61.
(17) Freeburger, M. E.; Spialter, L. J. Am. Chem. Soc. 1971, 93, 1894.
(18) Clark, H. A.; Gordon, A. F.; Young, C. W.; Hunter, M. J. J. Am.
Chem. Soc. 1951, 73, 3798.
(19) Eaborn, C.; Jackson, P. M. J. Chem. Soc. B 1969, 21.
(20) Bourgeois, P.; Calas, R. J. Organomet. Chem 1975, 84, 165.
(21) Schlosser, M.; Choi, J. H.; Takagishi, S. Tetrahedron 1990, 46, 5633.
(22) Mochida, K.; Hiraga, Y; Yakeuchi, H.; Ogawa, H. Organometallics
1987, 6, 2293.
(6) Zhu, X.-D.; Giacobini, E.; Hornsperger, J.-M. Eur. J. Pharmacol. 1995,
276, 93–99.
(7) Hornsperger, J.-M.; Collard, J.-N.; Schirlin, D.; Dow, J.; Heydt, J.-
G.; Dulery, B. Alzheimer Dis. 1994, 13, 1–136.
(8) Hornsperger, J.-M.; Collard, J.-N.; Heydt, J.-G.; Giacobini, E.; Funes,
S.; Dow, J.; Schirlin, D. Biochem. Soc. Trans. 1994, 22, 758–763.
(9) Gandolfi, R.; Klipa, D. K.; Fetner, N. J. Eur. Pat. Appl EP 95-401478,
June 21, 1995.
(23) Yamakawa, T.; Kagechika, H.; Kawachi, E.; Hashimoto, Y.; Shudo,
K. J. Med. Chem. 1990, 33, 1430.
(24) Effenberger, F.; Spiegler, W. Angew. Chem., Int. Ed. Engl. 1981, 20,
265.
(10) Schirlin, D.; Collard, J.-N.; Hornsperger, J.-M. Eur. Pat. Appl. EP
89-401775, June 22, 1989.
(25) Calas, R.; Gerval, J. C. R. Acad. Sci., Paris 1987,t. 305, Series II,
1423.
10.1021/op700199p CCC: $40.75
Published on Web 01/03/2008
2008 American Chemical Society
Vol. 12, No. 1, 2008 / Organic Process Research & Development
•
23

Scheme 1. ipso substitutions of compound 2 with
dichloromethylmethyl ether25 and with
Scheme 3. Reaction of trimethylsilylbenzene with
trifluoroacetic anhydridea
hydroxy(tosyloxy)iodobenzene (Koser’s reagent)26
a Compound 5 could be formed directly from 1 by ipso substitution or via
benzene intermediate by protonation-desilation of 1, followed by trifluoroacyl-
ation–deprotonation. Compounds 3 and 4 can be formed from 1 by direct
trifluoroacylation-deprotonation.
Scheme 2. Potential pathways from phenyltrimetnylsilane
or 1,3-bis(trimethylsilyl)benzene to
1-(3-trialkylsilylphenyl)-2,2,2-trifluoroethanone
Scheme 4. Reaction of 1,3-bis(trimethylsilyl)benzene with
trifluoroacetic anhydride
(Scheme 1).26 A 24% isolated yield of (3-trimethylsilyl)phenyl,
phenyl iodonium tosylate (the ipso substitution product) was
obtained. Electrophilic ipso substitutions of aryltrimethylsilanes
have been reviewed.27 Before the present study was initiated,
it was unclear whether the target product 3 could be made by
the Friedel–Crafts acylation of 1 by trifluoroacetic anhydride
in an economically competitive process.
para substitution close to zero.28 This would explain the nearly
2 to 1 isomer ratio of meta to para substitution (twice as much
3 formed as 4). Because of the large size of the trimethylsilyl
group, one would not have expected any substitution ortho to
the TMS group.
2. Results and Discussion
At the beginning of this study, the expected starting material
to prepare 3 was to be either phenyltrimethylsilane (1) or 1,3-
bis(trimethylsily1)benzene (2) (see Scheme 2). The amounts
of products and byproducts observed under the most favorable
reaction conditions for the conversion of 1 to 3 are given in
Scheme 3. The amounts of products and byproducts observed
under the most favorable reaction conditions for the conversion
of 2 to 3 are given in Scheme 4. Throughout this publication
the yields cited are solution yields by gas chromatography
analyses.
Conversion of 1 to 3 via Friedel–Crafts acylation. The
results for 14 experimental trials for the reaction of 1 with
TFAA, using aluminum chloride as Friedel–Crafts catalyst, are
given in Table 1. The trimethylsilyl group is known to act like
a proton as far as directing electrophilic aromatic substitution,
with Hammett σ values for the TMS group for both meta and
Comparison of Methylene Chloride and Cyclohexane
Solvents. A comparison of conversion of phenyltrimethylsilane
to products and byproducts of this study is made in Table 1 for
the solvents dichloromethane and cyclohexane. The reactions
to form 3 and 4 in cyclohexane went through an induction
period, which could be a real practical concern upon scale up.
This induction period was not observed in dichloromethane
solvent. The reaction in cyclohexane required a 15–20° higher
temperature, as compared to the reaction in dichloromethane.
The conversion of 1 to 3 was somewhat higher in dichlo-
romethane, and the reaction seemed cleaner, as compared to
the reaction in cyclohexane. However, relatively less 4 byprod-
uct was made in cyclohexane solvent. In cyclohexane solvent
the product molar ratios of 3 to 4 were 2.4:1 (see entries 4, 6,
(26) Koser, G. F.; Wettach, R. H.; Smith, C. S. J. Org. Chem. 1980, 45,
(28) Gordon, A. J.; Ford, R. J. The Chemist’s Companion: A Handbook of
Practical Data, Techniques and References; John Wiley: New York,
1972, pp 147–153.
1543.
(27) Snieckus, V. Chem. ReV. 1990, 90, 879.
24
•
Vol. 12, No. 1, 2008 / Organic Process Research & Development

Table 1. Experimental trials for the reaction of phenyltrimethylsilane with trifluoroacetic anhydride (reactions carried to
completion)a
mol/mol
AlCl3/TFAA
mol fraction of starting 1b
d
entry
solventc
T, °C
TFAA/1
PhH
1
3
4
5
ArCF3
1
CH2Cl2
CH2Cl2
CH2Cl2
C6H12
CH2Cl2
C6H12
C6H12
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
C7H14
CH2Cl2
CH2Cl2
TFAA
-10
-11
-12
-10
-10
8
1.10
1.10
0.33
0.40
0.50
0.50
0.50
0.50
1.00
0.50
0.50
0.50
1.50
1.00
5.32
1.00
2.73
0.97
2.00
2.00
2.00
2.00
2.00
2.00
2.00
2.00
1.94
1.33
1.00f
0.38
0.39
0.39
0.16
0.27
0.43
0.35
0.23
0.41
0.42
0.33
0.29
0.60
0.43
0.45
0.12
0.16
n.d.
0.20
0.21
0.09
0.18
0.24
0.09
0.22
0.22
0.23
0.18
0.16
0.07
0.22
0.13
0.03
0.10
0.10
0.04
0.08
0.12
0.04
0.09
0.11
0.11
0.09
0.08
0.03
0.11
0.07
0.01
0.15
0.30
0.03
0.11
0.14
0.05
0.19
0.15
0.24
0.04
0.05
0.05
0.24
0.32
0.36
0.45
0.61
0.16
0.37
0.50
0.18
0.50
0.48
0.58
0.31
0.29
0.15
0.57
0.52
0.40
2
3
0.68
0.36
0.06
0.47
0.27
0.11
0.00
0.37
0.43
0.24
0.01
0.00
0.47
4
5
6
7
2
8
-18
-18
-34
-34
2
9
10e
11e
12
13
14
15
-18
22
23
a Analyses of products solutions were done on a Hewlett-Packard Series II 5890 gas chromatograph, using a silica column. b With dichloromethane solvent, up to 4%
benzene converted was to diphenylmethane, and less than 1% of benzene was converted to (m-trifluoroacetyl)phenyltrimethylsiloxyldimethylsilane and
3-benzyl-trifluoroacetophenone. With cyclohexane solvent, up to 4% of benzene was converted to dimethyldiphenylsilane, up to 4% of benzene was converted to
phenylpentamethyldisiloxane, and less than 1% of benzene was converted to diphenyltetramethyldisiloxane. The numbers in this table have been normalized not to include
these silicon-derived related products. c Dichloromethane (CH2Cl2), cyclohexane (C6H12) or methylcyclohexane (C7H14). d Total of trifluoroacetylated aromatics. e Entries 10
and 11 were experiments run under the same conditions (repeat experiments). These reactions were not run to completion. f Also contained FeCl3 at initial molar ratio 1.00
versus TFAA.
and 7, Table 1). The reaction in cyclohexane seemed to have a
very narrow temperature range within which to operate the
reaction. Below 0° the reaction was too slow, and above 10°
the reaction gave lower yields of 3.
Reaction Condition Variables for Trifluoroacylation
Route to Form Compound 3. Experiments are summarized
in Table 1. Mol fractions of products and byproducts from late
aliquots are given in the right six columns of Table 1. The
following conclusions may be suggested:
(1) Increasing the initial molar ratio of AlCl3 to TFAA from
1.0 to 2.7 did not affect significantly the conversion of 1 to 3
or 4 (compare entries 1 and 2). The main effect of this increase
was to convert more of 1 to benzene and to 5. When a molar
excess of aluminum chloride to TFAA was used, the yield of
3 and 4 was nearly independent of the initial molar ratio of
TFAA to 1 for dichloromethane solvent (compare entries 8, 9,
and 13).
(2) If less than 50 mol % of TFAA (versus 1) was used,
less 3 and 4 were formed (compare entries 3, 4, and 5, Table
1). Increasing the initial molar ratio of TFAA to 1 above 0.5
did not increase the conversion of 1 to 3.
(3) A series of experiments using methylene chloride were
done at three temperatures, -10 (entry 5), -18 (entry 8), and
-34 °C (entries 10 and 11). On the basis of conversion of 1,
these results suggest that lower temperatures were advantageous
for the reactions to form 3 and 4 in dichloromethane solvent.
(4) The use of methylcyclohexane as solvent seemed to
accelerate the protiodesilation side reaction, as compared to
using cyclohexane solvent (compare entries 6 and 12). The
reaction in methylcyclohexane was very complicated, resulting
in many new unassigned peaks detected by gas chromatography.
(5) An attempt to use TFAA as the solvent (run 40) resulted
in a low yield of 3 (compare entries 14 and 15).
higher yields of 3. In all of these tests, the reaction conditions
were 20 °C, with initial 1:1 molar ratio of TFAA:1 and with
dichloromethane or cyclohexane as solvents. The catalyst trials
fell into two groups. Ferric chloride, boron trifluoride diethyl
etherate, stannic chloride and titanium tetrachloride gave
essentially no reaction (almost quantitative reactant 1 recovered).
Ferric bromide, polyphosphoric acid, boron trichloride and
trifluoromethane sulfonic acid gave almost total conversion of
1 to benzene. None of these agents catalyzed the formation of
either 3 or 4 in amounts greater than 1%.
Conversion of 2 to 3 via Protiodesilation/Friedel–Crafts
acylation and/or Trifluoroacyldesilation (ipso) Substitution.
By analogy to the reported ipso acyldesilation by acetyl
chloride23 and acetic anhydride,24 the reaction of trifluoroacetic
anhydride with 2, in the presence of aluminum chloride, was
expected to give 3. When this reaction was tried, the major
product was the trisubstituted 3,5-bis(trimethylsilyl)trifluoro-
acetophenone (7, see Scheme 4). This product would result from
a normal Friedel–Crafts reaction. Related products 1 and
benzene could be explained by protiodesilation reactions on 2.
These related products could in turn be reacted in the normal
Friedel–Crafts mechanism to form 5 (from benzene) and 3/4
(both from 1). Some additional 3 (but not 4) might have been
formed by protiodesilation of 7. The results of the aluminum
chloride catalyzed reactions of trifluoroacetic anhydride and
reactant 1 in dichloromethane (see Table 1) support the
combination of protiodesilation and normal Friedel–Crafts
trifluoroacyldeprotonation as major reaction pathways from
reactant 2.
Another, less likely, possibility is that competing ispo
trifluoroacyldesilation reactions occurred to form 5 (directly
from 1) and 3 (directly from 2). If the ipso reaction were the
only mechanism, one would have expected a final molar ratio
of 3 to 4 of 20:1.29 If the combined protiodesilation/trifluoroacyl-
Screening of Potential Catalysts for the Conversion of
Phenyltrimethylsilane to Compound 3. Several other Lewis
acid type catalysts were tried in order to determine if they gave
(29) Deans, F. B.; Eaborn, C. J. Chem. Soc. 1957, 498.
Vol. 12, No. 1, 2008 / Organic Process Research & Development
•
25

Table 2. Rate constants determined using SimuSolv software
The best fits for the experimental data versus calculated rate
constants were made for the model above, in which the unit
processes to convert 1 to 3, 4 and benzene are all first-order
overall (first order in 1). The kinetic model predicts that the
unit process to convert benzene to 5 is second order overall
(first order in benzene and first order in trifluoroacetic anhy-
dride). It is important to emphasize that just because the best
curve fitting was obtained using these assumptions, this does
not prove the unit processes or mechanisms for the formation
of 3, 4 and 5. Therefore, one should not speculate, in the absence
of more experimental results, about the possible differences in
mechanisms for the trifluoroacylation of 1 versus the trifluoroac-
ylation of benzene, based solely on these curve fitting estima-
tions. As discussed earlier, one cannot rule out the simultaneous
competition of an ipso substitution pathway from 1 to make 5.
Given these caveats, one can make some speculations
regarding the possible ”optimal” conditions to run a potential
process to convert 1 to 3, by using TFAA/A1C13 in dichlo-
romethane solvent. The calculations summarized in Table 3
suggest that lower temperatures for the reaction would enhance
the formation of 3 and 4 slightly, as compared to the protiode-
silation reaction to form benzene. However, the trifluoroacyl-
ation of benzene would be more significantly retarded at lower
temperatures, since this unit process has the largest Arrhenius
activation energy, by 20 J/mol. The larger preexponential factor
for the protiodesilation reaction of 1 to form benzene, as
compared to the trifluoroacyldeprotonation reactions to form 3
and 4, suggests less steric crowding at the highest energy
transition state for the protiodesilation process.
The activation parameters calculated and shown in Table 2
were used to calculate rate constants for the four unit processes
at -47 °C, which are given in column seven of Table 2. Using
four sets of rate constants at four temperatures, one can estimate
final yields of 3 and the three related products. The results of
these calculations at -34 and -47 °C are given in Table 3.
These calculations suggest that the optimal temperatures to
operate the proposed process to make 3 might be as low as
-47 °C. The yield of 3 from 1 would maximize at around 32%.
The formation of 4 and benzene would still be significant side
reactions. However, at -47 °C, the trifluoroacylation of benzene
would be predicted to occur so slowly that related product 5
would be formed at less than 1%. This depressed side reaction
would free up more trifluoroacetic anhydride to be used in the
desired reaction to form 3. However, a serious downside to
running the reaction at -47 °C would be the predicted long
time for the reaction to go to completion (around 150 h).
Other Related Products from the Friedel–Crafts Tri-
fluoroacylation of 1. Attempts were made to identify any
related products formed in the proposed trifluoroacylation
process in dichloromethane or cyclohexane solvents. The
identification of these other related products were done by gas
chromatography/mass spectrometry (Table 4). The account-
abilities of starting 1 for the reactions at larger scales for the
two solvent systems were excellent. This suggests that most
all of the related products in the two experiments have been
identified.
fitting of kinetic data of entries 8 and 9 of Table 1
at
at
kxy calcd
at -47 °C
-18 °C -34 °C
Ea
rate constant (× 105) (× 105) units (J/mol) ln A (× 105)
k13 (1 f 3)
k14 (1 f 4)
4.75 0.999 s-1
2.23 0.501 s-1
46.7 12.0 0.262
44.8 10.3 0.137
52.2 15.3 0.401
k1B (1 f PhH) 10.42 1.83 s-1
kB5 (PhH f 5) 18.83 1.67 M-1 s-1 72.6 23.2 0.021
deprotonation were the only mechanism, one would have
expected a final molar ratio of 3 to 4 of 2:1, unless some 3
forms from 7. Since the experimental molar ratio of 3 to 4 from
2 was 4.08, one could estimate that 79% of the trimethylsilyl-
trifluoroacetophenone products come from the protiodesilation/
trifluoroacyldeprotonation reaction pair, and 21% of these
products come from the direct ipso reaction on 2. An estimation
of the competition between the protiodesilation and trifluoroa-
cyldesilation (ipso) reactions might be made by comparing the
amounts of protiodesilation products (benzene and phenyltri-
methylsilane plus 79% of trifluoroacetophenone and 79% of
the trimethylsilyltrifluoroacetophenone products) to the possible
ipso products (21% of 5 and 21% of the trimethylsilyltrifluo-
roacetophenone products). This estimation predicts a 12%
contribution of ipso reaction, in the competitive reaction
pathways from reactant 2.
Kinetics Experiments in Dichloromethane Solvent. The
highest solution yields of the desired product 3, based on
reactant 1, were around 23–24%, in dichloromethane solvent
(Table 1). These yields are slightly less than but still comparable
to the yields of 3 from 1,3-dibromobenzene via the bis-
metalation routes.9,10 Kinetic analyses were done on the tri-
fluoroacylation of 1, using SimuSolv software, for two of the
higher yield reactions for 3, entry 8 (run at -18 °C) and entry
10 (done at -34 °C; see Table 1). The following reactions were
analyzed using the numerical integration methods of SimuSolv
software. This software optimizes the fit of calculated rate
constants to the experimental amounts of reactant and products
by the method of maximum likelihood function.30 The conven-
tion of the rate constants is to signify the reactants and products.
Thus, k13 refers to the rate constant for the reaction of 1 going
to product 3. The following kinetic expressions were proposed
in order to determine the value of the rate constants for
experiments 8 and 10:
-d(1)/dt ) k13(1) + k14(1) + k1B(1)
where B refers to the byproduct benzene
d(3)/dt ) k13(1)
d(4)/dt ) k14(1)
d(B)/dt ) k1B(1) - kB5(B)(TFAA)
d(5)/dt ) kB5(B)(TFAA)
In the kinetic expressions above, k13, k14 and k1B are first-
order rate constants. Rate constant kB5 is second-order. The
results obtained from the SimuSolv software-fitting of the kinetic
data of runs 8 and 10 are given in Table 2. The plots of
experimental data versus fitted curves for runs 8 and 10 are
given in Figure 1.
(30) Reilly, P. M.; Blau, G. E. Can. J. Chem. Eng. 1974, 52, 289.
26
•
Vol. 12, No. 1, 2008 / Organic Process Research & Development

Figure 1. SimuSolv software curve fittings of experimental kinetics data from experiments 8 and 10. Run 8, dichloromethane at
-18 °C; run 10, dichloromethane at -34 °C.
Table 3. Experimental and calculated (using rate constants
of Table 2) amounts of product and byproducts (mol %
from 1) for TFAA acylation of phenyltrimethylsilane at
several temperatures
Table 4. Gas chromatographic and mass spectroscopic
identification of major products and byproducts from
reaction of TFAA and phenyltrimethylsilane in
dichloromethane and cyclohexane solvents
MS ion peaks (g/mol)
at -10 °C at -18 °C at -34 °C at -47 °C
GC retention
time (min) parent largest largest largest
second third
reaction hours
6.0
7.0
30
150
calcd
32.2
16.9
1.3
49.1
0.4
compound
expt or calcd
expt
23.8
11.6
6.3
expt
22.2
10.7
10.5
40.8
15.4
calcd
27.9
14.0
2.6
3
5
3
4
3.5a
174
246
246
320
168
264
246
224
246
212
286
105
231
231
255
168
195
231
209
231
197
271
77
232
232
305
167
264
232
210
232
212
193
51
91
4
6.8a
1
7.3a
134
151
165
165
91
PhH
5
43.7
14.1
47.2
8.4
b
C13H19O2F3Si2
9.6a
diphenylmethane
11.1a
15.2a
23.6d
23.8d
24.8d
36.2d
39.6d
3-Bn-TFAc
3
e
C11H20OSi2
97
Analyses of Distillation Fractions from the Larger-Scale
Preparations of 2. If the trifluoroacylation route from 1 to 3
were done in a commercial process, a crucial part of the process
would be the successful isolation of the final product by some
separation process, such as distillation or preparative chroma-
tography. Such a separation would be dependent on the relative
volatility differences between 3, 4 and the other related products
produced in the reaction. Compounds 3 and 4 were removed
easily from the higher boiling related products produced in the
proposed trifluoroacylation process by flash distillation. These
higher boilers would include diphenyldimethylsilane and diphe-
nyltetramethyldisiloxide (from the reaction in cyclohexane) and
diphenylmethane and two other related products (from the
reaction in dichloromethane). A low-boiling related product
(phenylpentamethyldisiloxane) would also be separated readily
from 3 and 4.
4
134
198
28
DMDPSf
C16H22OSi2
g
a Dichloromethane solvent. GC Conditions: start at 90 °C (0 time hold), ramp
°C/min to 180 °C, then 10 min hold. b (m-Trifluoroacetyl)phenyl-
8
trimethylsiloxyldimethylsilane. c 3-Benzyl-trifluoroacetophenone. d Cyclohexane
solvent. GC Conditions: start at 70 °C (15 min hold), ramp 5 °C/min to 240 °C,
then
0
time hold. e Phenylpentamethyldisoxlane. f Dimethyldiphenylsilane.
g Diphenyltetramethyldisiloxane.
of 5 is about 10 times higher than that of 3, and 5 was separated
easily as a forecut in the distillation of 3. Product 3 is 30-60%
more volatile than its isomer 4. Early fractions (3–5) have
greater than a 2:1 ratio of 3 versus 4 (Table 5). Assuming a
relative volatility of 3 to 4 of 1.60, it is estimated that a
separation of the two isomers, to obtain 99.9% pure 3, would
require a 36-stage distillation column.31,32
The results of a flash vacuum distillation of the product
isolated from experiment 5 (largest scale trial) are given in Table
5. The relative volatilities of components of various distillation
cuts from two experiments are given in Table 6. The volatility
(31) The author thanks K. A. Cobb, L. S. Green, R. Srivastava and P. Au-
Yeung of the Department of Analytical Sciences of The Dow Chemical
Company for the Vapor-Liquid Equilibrium measurements of the
compounds of this study.
Vol. 12, No. 1, 2008 / Organic Process Research & Development
•
27

Table 5. Flash distillation fractions of products from
experiment 5 (by GC analyses)
4. Experimental Section
Gas chromatograms were obtained with a Hewlett-Packard
Series II 5890 gas chromatograph. Gas chromatograms/mass
spectra were done with a Hewlett-Packard Series II 5890 gas
chromatograph, connected to a Hewlett-Packard 5971 Series
mass selective detector, with Series 59822E ionization gauge
controller. Nuclear magnetic resonance spectra (H1 and C13)
were taken on a Bruker AC 300 spectrophotometer. NMR
spectra of some of the compounds of this study are given in
Supporting Information.
molar fractionsa
(8 torr)
fraction weight (g) temp (°C)
5
3
4
DPMb
1
4.61
6.10
7.12
6.04
8.67
6.46
6.86
3.71
0.89
7.07
22–47
47–74
74–85
85
85–87
87
87–89
89–104
104–117
0.979 0.015 0.005 0.001
0.911 0.066 0.022 0.002
0.086 0.681 0.217 0.016
0.006 0.722 0.252 0.020
0.004 0.687 0.283 0.026
0.001 0.636 0.330 0.034
2
3
4
5
6
7
n.d.
n.d.
n.d.
0.538 0.398 0.049
0.310 0.488 0.159
0.071 0.326 0.519
8
Largest Scale Friedel–Crafts Trifluoroacetylation of
Phenyltrimethylsilane (1); Experiment 5, Table 1. A reaction
flask containing aluminum chloride (66.67 g, 500 mmol) and
185 mL of chloromethane was cooled to -10 °C, and
trifluoroacetic anhydride (52.50 g, 250 mmol) was added to
the contents, in 5 min. The temperature of the reaction slurry
increased to +5 °C. After the slurry had cooled again to -10
°C, neat phenyltrimethylsilane (75.00 g, 500 mmol) was added,
over a 10 min period. The temperature of the resulting slurry
rose to -3 °C and then fell back to -10 °C after 40–45 min.
This suggests the presence of an exothermic reaction.33
The reaction slurry was stirred at -10 °C for 370 min, and
nine aliquots were taken during this time to be analyzed by
GC. At the end of the reaction, the brown reaction solution
was decanted away from remaining solids into a suitable flask,
and the reaction flask was rinsed with 100 mL of fresh
dichloromethane. The organic material was poured into a stirred
and chilled mixture of 400 g of ice/water, in such a way as to
allow the temperature of the resulting mixture to reach only 13
°C. A dichloromethane rinse (75 mL) of the flask originally
containing the organic material was also added to the cold
organic/aqueous slurry. The organic phase (cloudy and yellow)
was separated from the acidic aqueous phase. The organic phase
was stirred three times with 450 mL of cold fresh water. The
pH readings of the aqueous phases after the washes were
successively 1.5, 2.8 and 3.8. The resulting dichloromethane
layer was dried with anhydrous magnesium sulfate. The dried
dichloromethane layer was isolated by filtration, and the solvent
was removed by evaporation under vacuum. Volatiles were
stripped at a final vacuum of 20 torr at 13 °C. A dark orange-
brown oil remained (59.3 g). The organic material after
evaporation was distilled under vacuum, with most of the
distillation at 8 torr. Nine distillation fractions were collected
and analyzed by GC (Table 5). The combined weights of the
fractions and distillation pot residue were 57.5 g (97.0% material
balance).
9
pot
undistilled n.d.
n.d.
0.003 0.669
a Molar fractions of starting compound 1. b Diphenylmethane.
Table 6. Relative volatilities of product 3 versus related
products from Friedel–Crafts trifluoroacylation of 1,
measured from distillation fractions from reactions 4 and 5
(expressed as volatility of product 3 over volatility of related
product)
relative volatilities at 130 °C (3/X)
3/4 (mol/mol)
b
fraction in fraction
4
5
DPMa C11H20OSi2 DMDPSc
2
3.0d
3.1d
2.9d
2.4d
1.9d
1.4d
0.6d
1.66 0.05 n.d.
1.60 0.09 10.9
1.59 0.10 10.1
1.58 0.10 10.8
1.61 0.12 11.2
1.60 n.d. 12.4
1.57 n.d. 12.5
1.60 0.09 11.3
1.57 0.09 17.9
1.29 n.d. n.d.
1.24 n.d. n.d.
1.26
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
3
4
5
6
7
8
ave
avf
1
3.2g
2.6g
0.77
0.58
0.68
0.58
12.8
29.2
21.0
2
avh
avi
1.20
a Diphenylmethane. b Phenylpentamethyldisoxlane. c Dimethyldiphenylsilane.
d From reaction in dichloromethane solvent. e Average from seven fractions
measured at 130 °C. f Average from seven fractions measured at 80 °C. g From
reaction in cyclohexane solvent. h Average from two fractions measured at 130 °C.
i Average from two fractions measured at 80 °C.
3. Conclusions
The Friedel–Crafts trifluoroacylation of phenyltrimethylsilane
is a potential route to make Zifrosilone. The advantages to this
process versus the existing bis-metalation process from 1,3-
dibromobenzene are that the Friedel–Crafts process would be
only one step from readily available phenyltrimethylsilane, and
that the reactants, intermediates and products are relatively less
hazardous. The overall yields of 3 from the two processes are
comparable. The disadvantage of the proposed Friedel–Crafts
process is the relatively large amount of 4, the isomer to 3,
which would necessitate a challenging purification of 3 by
distillation or chromatography. Vapor–liquid equilibria and
kinetics studies of the proposed Friedel–Crafts process enabled
an analysis of the challenges raised in the present study. Because
Zifrosilone was never manufactured at Dowpharma, the final
choice between the existing two-step process and the one-step
process of the present study was never made.
NMR Spectra of Reactants, Related Product and Crude
Product. Reactant 1: 1H NMR (300 MHz, CDCl3) δ 7.52–7.49
(m, 2H), 7.32–7.30 (m, 3H), 0.26 (s, 9H); 13C NMR (75 MHz,
CDCl3) δ 140.6, 133.5, 129.0, 127.0, -0.70. Related product
5: 1H NMR (300 MHz, CDCl3) δ 8.05 (s, 2H), 7.69–7.51 (m,
3H); 13C NMR (75 MHz, CDCl3) δ 135.6, 130.2, 129.2. Crude
product mixture: 1H NMR (300 MHz, CDCl3) δ 8.15–8.05 (m,
6.5H), 7.50–7.03 (m, 22H), 4.11–4.00 (m, 8.4H), 0.20/0.16/
(33) A group additivity thermal calculation, using CHETAH software
created by The Dow Chemical Company, estimated that the conversion
of 1 to 3 released 46 J/mol of energy. The author thanks J. Downey,
Department of Analytical Sciences, The Dow Chemical Company for
this calculation.
(32) P. Au-Yeung, The Dow Chemical Company, personal communication,
based on VLE results in Table 6.
28
•
Vol. 12, No. 1, 2008 / Organic Process Research & Development

0.09 (3xs, total 9H); 13C NMR (75 MHz, CDCl3) δ 150.0, 139.5,
130.6, 130.2, 129.8, 129.6, 129.4, 129.25, 129.19, 128.8, 128.6,
128.3, 127.6, 126.9, 116.7, 115.0, 42.1, 1.7, 1.3, 0.1.
Relative Volatilities of Reaction Products. Vapor–liquid
equilibria were measured for the various products and related
products for these two reactions, using distillation fractions from
experiments 4 and 5.31 These results are summarized in Table
6.
Small-Scale Trials of Friedel–Crafts Trifluoroacetylation
of Phenyltrimethylsilane (1). Many variations on the experi-
mental conditions to prepare 3 via a Friedel–Crafts process using
aluminum chloride catalyst were tried. These reactions were
done at the 10–70 mmol (for reactant 1) scale. Reaction
variables such as solvent, concentration, temperature, molar
ratios of reactants and time of reaction were studied. The results
of these studies are given in Table 1.
mmol) in 10 mL of dichloromethane was added to the reaction
flask, and the resulting slurry was cooled again to -10 °C. A
solution of 2 (1.16 g, 4.8 mmol) in 10 mL of dichloromethane
was added in 1 min to the stirred and chilled reaction slurry.
The reaction was run at -12 °C for 500 min, with a total of
16 aliquots removed to be analyzed by gas chromatography.
The composition in molar fractions of the last aliquot (at 500
min reaction time) was the following: Benzene (0.199), 1
(0.082), 2 (0.008), 3 (0.211), 4 (0.052), 5 (0.094), 6 (0.004)
and 7 (0.350).
Supporting Information Available
Additional chromatographic and NMR information. This
material is available free of charge via the Internet at
http://pubs.acs.org.
Attempted ipso Substitution of 1,3-Bis(trimethylsilyl)ben-
zene To Prepare 3. A flask containing aluminum chloride (2.00
g, 15 mmol) and 30 mL of dichloromethane was cooled below
-10 °C. A solution of trifluoroacetic anhydride (1.10 g, 5.5
Received for review September 3, 2007.
OP700199P
Vol. 12, No. 1, 2008 / Organic Process Research & Development
•
29