Some reaction safety aspects of ruthenium-catalyzed allylic oxidations of Δ-5-steroids in the pilot plant

Article abstract of DOI:10.1021/op9700587

The ruthenium-catalyzed allylic oxidation of Δ-5-steroids with tert-butyl hydroperoxide to the corresponding unsaturated ketones was studied. During the scale-up to the pilot plant level, serious safety problems were detected, and the identified runaway potential of this reaction was studied. By modifying the original oxidation procedure, it was possible to establish this reaction safely in the pilot plant at 17-kg scale. Furthermore, the procedure was applied successfully to several other steroids on the laboratory scale.

Full text of DOI:10.1021/op9700587

Organic Process Research & Development 1998, 2, 100−104
Some Reaction Safety Aspects of Ruthenium-Catalyzed Allylic Oxidations of
∆-5-Steroids in the Pilot Plant
Michael Harre, Reinhard Haufe, Klaus Nickisch, Peter Weinig, and Hilmar Weinmann*
Chemical DeVelopment, Schering AG, D-13342 Berlin, Germany
William A. Kinney and Xuehai Zhang
Magainin Pharmaceuticals Inc., 5110 Campus DriVe, Plymouth Meeting, PennsylVania 19462
Scheme 1
Abstract:
The ruthenium-catalyzed allylic oxidation of ∆-5-steroids with
tert-butyl hydroperoxide to the corresponding unsaturated
ketones was studied. During the scale-up to the pilot plant level,
serious safety problems were detected, and the identified
runaway potential of this reaction was studied. By modifying
the original oxidation procedure, it was possible to establish
this reaction safely in the pilot plant at 17-kg scale. Further-
more, the procedure was applied successfully to several other
steroids on the laboratory scale.
Introduction
Squalamine (1) is a novel polyaminosteroidal sulfate
which was isolated recently from tissues of the dogfish shark,
Squalus acanthias.1 The interesting biological activity of
squalamine against Gram-negative and Gram-positive bac-
teria and, especially, its potent tumor-inhibiting properties
triggered some synthetic efforts in this class of compounds.2
During studies concerning the scale-up of the intermediate
4 (Scheme 1) for the preparation of squalamine, we faced
the problem of an allylic C-H oxidation at C-7 of a ∆-5-
steroid to the corresponding unsaturated 7-ketone. This
transformation is an old problem in steroidal chemistry3 for
which there were no really satisfying solutions until recently.
Several methods for such an allylic oxidation are described
in the literature. The classic and probably best-known
variation of this theme is the chromium(VI) oxide-mediated
oxidation, which can also be accomplished with the more
modern reagents pyridinium chlorochromate (PCC) and
pyridinium dichromate (PDC).4 Although the yields are
moderate to good, this method suffers from the high excesses
of the chromium reagents, which sometimes make the
isolation of the products difficult and environmental problems
arise. Catalytic variations of the chromium oxidation with
chromium hexacarbonyl and tert-butyl hydroperoxide in
acetonitrile or benzene5 seemed also not so attractive to us
from an environmental and process point of view. The more
environmental friendly oxidation by molecular oxygen and
N-hydroxyphthalimide as the catalyst6 seems to be restricted,
more or less, to the laboratory scale until now. Besides the
technical and safety problems connected with a reaction
under an oxygen atmosphere in the pilot plant, we found
the scale-up difficult. In our case, experiments in the range
of 500 g and above did not proceed satisfactorily.
* To whom correspondence should be addressed. Telefax: (international) +49-
30 4681 8036. E-mail: Hilmar.Weinmann@SCHERING.DE.
(1) Moore, K. S.; Wehrli, S.; Roder, H.; Rogers, M.; Forrest, J. N.; McCrimmon,
D. D.; Zasloff, M. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1354.
(2) (a) Moriarty, R. M.; Tuladhar, S. M.; Guo, L.; Wehrli, S. Tetrahedron Lett.
1994, 35, 8103. (b) Moriarty, R. M.; Enache, L. A.; Kinney, W. A.; Allen,
C. S.; Canary, J. W.; Tuladhar, S. M.; Guo, L. Tetrahedron Lett. 1995, 36,
5139. (c) Pechulis, A. D.; Bellevue, F. H.; Cioffi, C. L.; Trapp, S. G.; Fojtik,
J. P.; McKitty, A. A.; Kinney, W. A.; Frye, L. L. J. Org. Chem. 1995, 60,
5121. (d) Rao, M. N.; McGuigan, M. A.; Zhang, X.; Shaked, Z.; Kinney,
W. A.; Bulliard, M.; Laboue, B.; Lee, N. E. J. Org. Chem. 1997, 62, 4541.
(3) Dauben, G. W.; Lorber, M.; Fullerton, D. S. J. Org. Chem. 1969, 34, 3587.
(4) (a) Parish, E. J.; Chitrakon, S.; Todd, K. L. Org. Prep. Proced. Int. 1985,
17, 192. (b) Parrish, E. J.; Wei, T. Synth. Commun. 1987, 17, 1227.
(5) (a) Kumar, V.; Amann, A.; Ourisson, G.; Luu, B. Synth. Commun. 1987,
17, 1279. (b) Pearson, A. J.; Chen, Y.-S.; Han, G. R.; Hsu, S.-Y.; Ray, T.
J. Chem. Soc., Perkin Trans. 1 1985, 267.
(6) Ishii, Y.; Nakayama, K.; Takeno, M.; Sakaguchi, S.; Iwahama, T.;
Nishiyama, Y. J. Org. Chem. 1995, 60, 3934.
100
•
Vol. 2, No. 2, 1998 / Organic Process Research & Development
S1083-6160(97)00058-3 CCC: $15.00 © 1998 American Chemical Society and Royal Society of Chemistry
Published on Web 02/24/1998

For these reasons, we found the recently published
protocol by Miller et al. for a ruthenium-catalyzed allylic
oxidation with tert-butyl hydroperoxide the most attractive
alternative.7
Results and Discussion
The ∆-5-ketal 3, which was the starting material for our
oxidation studies, can be easily prepared from the com-
mercially available 20-hydroxy-4-pregnen-3-one (2) in a few
steps. With this material in our hands, we first checked the
original oxidation method described by Miller et al. and were
glad to find the reaction working very smoothly in the 5-g
scale. With cyclohexane as solvent, using 10 molar equiv
of tert-butyl hydroperoxide and 0.62 mol % RuCl3‚H2O at
20 °C, 3 was transferred to the unsaturated ketone 4 and
isolated in 85% yield as a white solid. This result is
especially remarkable since the structure of the ∆-5-steroid
3 contains a fragile ketal and tert-butyldimethylsilyl ether
moiety. We found no significant difference in yield and
reaction time when we changed from tert-butyl hydroper-
oxide in isooctane to tert-butyl hydroperoxide in water
(70%). Cyclohexane and MTBE were the solvents of choice.
Acetone, ethyl acetate, dichloromethane, and toluene gave
lower yields or no reaction at all. Because of electrostatic
problems connected with reactions in pure cyclohexane, we
decided to use the aqueous tert-butyl hydroperoxide solution
for the scale-up experiments and the reactions in the pilot
plant. For the same reason, we used MTBE in the first
experiments on the laboratory scale. On a 50-g scale, we
were able to run the reaction successfully in MTBE, although
the yield was slightly decreased to 76% of 4.
During our initial studies, the addition time of the peroxide
to the reaction mixture was found not to be a critical
parameter and had been, more or less, neglected. Although
it is common knowledge that peroxides are catalytically
destroyed by heavy metals, we had not detected any
exothermic reaction in the laboratory scale. The routine
safety assessment of our laboratory procedure showed a
safety hazard risk because of accumulation of the tert-butyl
hydroperoxide, which gives rise to a dangerous runaway
potential. A calorimetric study of the originally desired
process on an 80-g scale, using MTBE as solvent at 20 °C,
was carried out to determine the overall heat of reaction
(∆Hr ), the rate of heat production, and the thermal
accumulation potential in the reactor due to unreacted
material, using isothermal reaction calorimetry (Mettler’s RC-
1). The heat of reaction measured is strongly exothermic,
∆Hr ) -582 kJ/kgreaction mass or -2257 kJ/mol of 3 due to
the high excess of peroxide which is decomposed in the
course of the reaction by the catalyst. The exothermic
reaction was finished after ca. 6 h. The reaction energy
implies a direct measure of the potential for a severe
runaway, i.e., the destructive potential.8 Usually, an indicator
for this is the adiabatic temperature increase, ∆Tad. From
Figure 1. RC1: synthesis of 4, TBHP dosage 1 h (mdos )
balance, therm. conv. ) thermal conversion).
Figure 2. Phe-Tec II: runaway scenario, original laboratory
procedure.
the measured heat of reaction, an adiabatic temperature
increase of ∆Tad ) 263 K was estimated, indicating a high
potential for a severe runaway. A maximum thermal
accumulation of up to 84% of the total heat output was
determined at the end of the peroxide addition, indicating a
heat output not proportional to the addition rate of the
peroxide (Figure 1). To improve the evaluated results of
the runaway scenario, a batch addition of peroxide was
considered as a worst-case scenario. Thus, a further study
using adiabatic reaction calorimetry (HEL’s Phi-Tec II) on
an 8-g scale was conducted (40 mL of MTBE, 0.025 g of
RuCl3 in 4.8 mL of H2O, addition of 22.5 mL of tert-butyl
hydroperoxide solution). The peroxide was added in one
portion, and the reaction under adiabatic conditions was
monitored. Within less than 10 min, the runaway up into a
heat explosion occurred, with a maximum rate of temperature
rise, dT/dt, up to 11 000 °C/min and a maximum rate of
pressure rise, dP/dt, up to 5000 bar/min (Figure 2). Hereby,
a maximum pressure of p ) 55 bar was measured, and ∆Tad
) 236 K was determined (Figure 3). In a blank experiment,
the stability of tert-butyl hydroperoxide in the reaction
mixture without the steroidal alkene 3 was investigated using
adiabatic reaction calorimetry, following the same protocol
as described above (40 mL of MTBE, 0.025 g of RuCl3 in
4.8 mL of H2O, addition of 22.5 mL of tert-butyl hydrop-
eroxide solution). In this case, a runaway occurred also
within less than 10 min, with a maximum rate of temperature
rise, dT/dt, up to 1500 °C/min and a maximum rate of
pressure rise, dP/dt, up to 1200 bar/min. A maximum
pressure of p ) 45 bar and ∆Tad ) 220 K were determined.
After these studies revealed major safety concerns for running
(7) (a) Miller, R. A.; Li, W.; Humphrey, G. R. Tetrahedron Lett. 1996, 37,
3429. (b) Miller, R. A.; Thompson, A. S.; Bakshi, R. K.; Corley, E. G. Int.
Pat. Appl. PCT/US95/06004, WO 95/32215, 1995.
(8) Steinbach, J.; Ebertz, A.; Stoessel, F. In Ullmann’s Encyclopedia of Industrial
Chemistry; VCH: Weinheim, 1995; Vol. B8, p 343.
Vol. 2, No. 2, 1998 / Organic Process Research & Development
•
101

After this catalytic oxidation method had been adjusted
to the safety issues in the manner described above, we
transferred the procedure to the pilot plant. Two batches
with 17 and 15.72 kg of the ketal 3 were run in a 250-L
enamelled reaction vessel. To ensure a fast and safe reaction,
the reaction mixture was stirred vigorously while nitrogen
was bubbled through the solution. Seven molar equivalents
of the tert-butyl hydroperoxide were added during 7 h, and
the mixture was stirred for another 10 h. After this time,
no peroxides could be detected any longer. After the usual
aqueous workup, the cyclohexane phase was concentrated
and cooled to 5 °C to induce crystalization. Concentration
of the mother liquor and purification by chromatography
yielded a second crop. The total yield of the two batches
was 17.46 kg (52%) of 4 as a white solid. Although the
yield was decreased compared to the good results in the
laboratory scale (85%), it was shown that it is possible to
perform this oxidation successfully in a safe way in the
multikilogram scale. The moderate yield is probably con-
nected with the inefficient mixing of the heterogeneous
reaction mixture in the pilot plant reaction vessel compared
to that in the laboratory.
Figure 3. Phi-Tec II: runaway scenario, original laboratory
procedure.
the pilot plant batches with the procedure we followed
successfully at the laboratory scale, a new experimental
program was started in process research to find a method to
control the evolving reaction heat.
As we had found a slight increase of impurities with
MTBE as solvent when we oxidized larger amounts of the
steroid, we switched back to cyclohexane. In our first
attempts, we tried to avoid the accumulation of peroxide by
performing the oxidation at higher temperatures (60 or 80
°C) and adding the tert-butyl hydroperoxide (10 molar equiv)
over longer intervals (2-6 h). Although these methods
worked and gave an improvement concerning the accumula-
tion, all of these reaction mixtures contained more impurities
and unreacted starting material than the original room-
temperature variation. For these reasons, we returned to the
original protocol and tried to minimize the large excess of
aqueous tert-butyl hydroperoxide (10 molar equiv). With
only 2.5 molar equiv of the peroxide and an addition time
of 6 h, the conversion was incomplete and 50% of the starting
material was left unreacted after a 24-h reaction time. Under
the same conditions using 5 molar equiv, 20% of the starting
material was left. With longer reaction times, the number
and amount of impurities increased. The optimum conditions
which we found finally were to add 7 molar equiv of the
aqueous tert-butyl hydroperoxide during 7 h at 20 °C under
vigorous stirring to the ∆-5-steroid 3 in cyclohexane and to
quench the reaction after an additional hour at this temper-
ature. Lower amounts of peroxide gave only incomplete
conversion, whereas higher amounts are connected with a
severe safety hazard risk because of potential accumulation.
This new protocol led to a maximum thermal accumulation
of approximately 20% of the tert-butyl hydroperoxide. In a
control experiment (8 g of steroidal alkene 3, 40 mL of
cyclohexane, 0.025 g of RuCl3 in 4.8 mL of water, 4.5 mL
()2 molar equiv of tert-butyl hydroperoxide) using adiabatic
reaction calorimetry, the exotherm reaction occurred within
20 min, reaching a maximum rate of temperature rise, dT/
dt, of 65 °C/min and a maximum rate of pressure rise, dP/
dt, of 70 bar/min. A maximum pressure of p ) 1 bar and
∆Tad ) 67 K were determined, indicating only a moderate
runaway potential, which is in the range of common organic
reactions. During the course of this run, no further exo-
thermic reaction was detected (even after heating to 90 °C
in the adiabatic reaction calorimeter).
As described above, we were interested not only in an
oxidation procedure for the preparation of the intermediate
4 but also in a general application of this allylic oxidation
for the preparation of ∆-5,7-ketosteroids. For this reason,
we applied the developed oxidation procedure to several
other steroids and received moderate to good isolated yields
of the reaction products. These results are summarized in
Table 1.
Conclusions
The ruthenium-catalyzed allylic oxidation method7 re-
cently described by Miller et al. was successfully adjusted
to our transformation problem, considering the described
safety study data. The oxidation of 3 to 4 was accomplished
in a safe way in the multikilogramm scale in the pilot plant.
Due to scale-up problems, the yield was reduced compared
to that at the laboratory scale. Although we got a smaller
amount of the product than we had expected, we had
sufficient material for the performance of the next steps. The
generality of the adjusted oxidation method was shown with
several other steroidal compounds. In our opinion, this
transformation is a typical example of the general difficulties
one encounters on the safe scaling-up of catalytic heteroge-
neous reactions with reagents of high energy content. A
combined effort of chemical and technical solutions is needed
to overcome such problems.
Experimental Section
Commercial reagent grade solvents were used as obtained.
Melting points were measured with a Bu¨chi 510 melting
point apparatus and are uncorrected. IR spectra were
measured on a Bruker FT-IFS 25 spectrometer. 1H NMR
spectra were recorded on a Bruker AC 300 (300 MHz), and
δ values are given in ppm relative to tetramethylsilane as
internal standard. Mass spectra were determined with a
Fisons Instruments VG 70-70 E spectrometer at 70-eV
102
•
Vol. 2, No. 2, 1998 / Organic Process Research & Development

solid; mp 129 °C; [R]25D -43.0° (c ) 1, CHCl3). IR (KBr):
2950, 1660, 1100 cm-1. 1H NMR (CDCl3): δ 0.0 (s, 6 H),
0.68 (s, 3 H), 0.86 (s, 9 H), 0.98 (d, 3 H, J ) 7 Hz), 1.18 (s,
3 H), 1.0-2.5 (m, 18 H), 2.65 (dd, 1 H, J ) 2, 15 Hz),
3.16-3.28 (m, 1 H), 3.5-3.6 (m, 1 H), 3.8-4.0 (m, 4 H),
5.64 (d, 1 H, J ) 2 Hz). MS (m/z) (relative intensity): 504
(47, M+ + 1), 503 (100, M+), 99 (7), 52 (13). Anal. Calcd
for C30H50O4Si: C, 71.66; H, 10.02; Si, 5.59. Found: C,
71.24; H, 9.91; Si, 5.25.
Table 1. Ruthenium-catalyzed allylic oxidations of
∆-5-steroids
∆-5,7-Ketosteroid 6. Yield: 0.433 g (42%) of a white
solid; mp 182 °C; [R]25D -106.0° (c ) 1, CHCl3). IR (KBr):
2950, 1730, 1670, 1250 cm-1. 1H NMR (CDCl3): δ 0.7 (s,
3 H), 0.75-0.9 (m, 9 H), 1.02 (d, 3 H, J ) 7 Hz), 1.2 (s, 3
H), 1.05-2.6 (m, 23 H), 2.07 (s, 3 H), 4.65-4.8 (m, 1 H),
5.03 (dd, 1 H, J ) 10, 15 Hz), 5.18 (dd, 1 H, J ) 10, 15
Hz), 5.7 (d, 1 H, J ) 1 Hz). MS (m/z) (relative intensity):
469 (32, M+), 410 (39), 409 (100). Anal. Calcd for
C31H48O3: C, 79.44; H, 10.32. Found: C, 79.42; H, 10.02.
∆-5,7-Ketosteroid 8. Yield: 0.730 g (71%) of a colorless
oil. IR: 2950, 1730, 1710, 1670, 1280, 1240 cm-1. 1H NMR
(CDCl3): δ 0.82 (s, 3 H), 1.24 (s, 3 H), 1.4-2.0 (m, 12 H),
2.07 (s, 3 H), 2.25-2.75 (m, 6 H), 4.65-4.8 (m, 1 H), 5.09
(d, 1 H, J ) 5 Hz), 5.78 (d, 1 H, J ) 2 Hz), 7.4-7.49 (m,
2 H), 7.5-7.6 (m, 1 H), 8.03 (d, 1 H, J ) 7 Hz). MS (m/z)
(relative intensity): 468 (46, M+ + NH3), 451 (19, M+),
392 (41), 391 (100). Anal. Calcd for C28H34O5: C, 74.64;
H, 7.61. Found: C, 74.29; H, 7.61.
∆-5,7-Ketosteroid 10. Yield: 0.716 g (69%) of a white
solid; mp 184 °C; [R]25D -86.6° (c ) 1, CHCl3). IR (KBr):
2950, 1740, 1670, 1240 cm-1. 1H NMR: δ 0.9 (s, 3 H),
1.25 (s, 3 H), 1.3-2.0 (m, 10 H), 2.07 (s, 3 H), 2.1-2.2 (m,
1 H), 2.35-2.68 (m, 5 H), 2.75-2.9 (m, 1 H), 4.55-4.7
(m, 1 H), 5.78 (d, 1 H, J ) 2 Hz). MS (m/z) (relative
intensity): 362 (12, M+ + NH3), 345 (10, M+), 286 (35),
285 (100). Anal. Calcd for C21H28O4: C, 73.23; H, 8.19.
Found: C, 72.96; H, 8.18.
∆-5,7-Ketosteroid 12. Yield: 0.528 g (51%) of a white
solid; mp 147 °C; [R]25D -72.9° (c ) 1, CHCl3). IR (KBr):
2950, 1730, 1705, 1670, 1240 cm-1. 1H NMR: δ 0.68 (s,
3 H), 1.24 (s, 3 H), 1.28-2.0 (m, 12 H), 2.07 (s, 3 H), 2.14
(s, 3 H), 2.2-2.3 (m, 1 H), 2.4-2.6 (m, 5 H), 4.65-4.8 (m,
1 H), 5.74 (d, 1 H, J ) 2 Hz). MS (m/z) (relative
intensity): 373 (36, M+), 314 (37), 313 (100). Anal. Calcd
for C23H32O4: C, 74.16; H, 8.66. Found: C, 73.88; H, 8.46.
Procedure for the Preparation of ∆-5,7-Ketosteroid 4
in the Pilot Plant. In a 250-L enamelled reaction vessel,
to 17 kg (34.77 mol) of ∆-5-steroid 3 was added 49 g (0.22
mol, 0.62 mol %) of ruthenium trichloride monohydrate
dissolved in 5 L of deionized water. After addition of 85 L
of cyclohexane, the reaction mixture was stirred vigorously,
and nitrogen was bubbled through the solution. Next, 33.3
L of a 70% aqueous tert-butyl hydroperoxide solution (243.0
mol, 7 equiv) was added slowly over 7 h, during which time
the internal temperature was controlled between 20 and 25
°C. The reaction mixture was stirred for another 10 h. At
that time, the peroxide test was negative. Next, 30 L
deionized water was added, and the mixture was stirred for
ionizing voltage, using NH3 in case of chemical ionization
(CI). Microanalytical data were provided by Schering
analytical department. TLC analyses were performed on
Merck F254 silica gel plates. The peroxide test was performed
with the Merckoquant peroxide test kit.
General Procedure for the Ruthenium-Catalyzed Al-
lylic Oxidation of ∆-5-Steroids. ∆-5,7-Ketosteroid 4. To
5 g (10.2 mmol) of ∆-5-steroid 3 (M ) 488.8 g/mol) and
14.3 mg (0.063 mmol, 0.62 mol %) of ruthenium trichloride
monohydrate (M ) 225.4 g/mol) was added under nitrogen
25 mL of cyclohexane. To the vigorously stirred suspension
was added at room temperature over 7 h 9.8 mL of an
aqueous 70% tert-butyl hydroperoxide solution (71.6 mmol,
7 molar equiv). After an additional hour of stirring, the
peroxide test was negative, and the reaction mixture was
quenched with water (30 mL). Extraction of the aqueous
phase with cyclohexane (40 mL) and concentration of the
combined organic phases yielded a yellow oil, which was
further purified by silica gel chromatography with ethyl
acetate/hexane. Yield: 4.36 g (8.67 mmol, 85%) of a white
Vol. 2, No. 2, 1998 / Organic Process Research & Development
•
103

30 min. After separation of the phases, the organic phase
was filtered over a pad of 25 kg of silica gel, and the filter
was washed with 100 L of cyclohexane. The solution was
concentrated to 90 L, cooled to 5 °C, and stirred for 1 h.
The product crystals were filtered off, and the mother liquor
was concentrated to 35 L. Purification by silica gel
chromatography (450 kg of silica gel) with 95:5 hexane/
ethyl acetate yielded a second crop. The combined yields
of this batch and a second batch with 15.72 kg of starting
material gave 17.46 kg (34.72 mol, 52%) of 4 as a white
solid, which was fully identical with the material prepared
in the laboratory.
Acknowledgment
Excellent technical assistance by Mr. Klaus Koenig is
gratefully acknowledged. We thank the Schering Analytical
and Spectroscopic Departments for measuring the spectro-
scopic data and for HPLC support.
Received for review October 10, 1997.
OP9700587
104
•
Vol. 2, No. 2, 1998 / Organic Process Research & Development