Dedicated to 60th anniversary of Dr. Yurii Pustovit
Published as part of the ISySyCat2019 Special Issue
There are many efficient reagents for organic synthesis known from the classical textbooks,
but by no means are all of them popular among chemists for real application in laboratory
practice. Gaseous or volatile compounds which possess extremely high toxicity, like
CH2N2, HCN, COCl2, and MeNCO, are among the most characteristic examples. The Bhopal disaster, where
approximately 200,000 people were exposed to MeNCO and around 20,000 died as a result,
has clearly demonstrated such reagents as actually dangerous.[1] Lately some of the above-mentioned reagents are experiencing a renaissance due to
achievements in flow technology. For example, during the last 10 years the safe flow
method using CH2N2
[2] and HCN[3] has been developed. Another, more common way of obtaining the same results, as in
the case of using the dangerous reagents, is development of their less toxic, more
convenient, and safe synthetic equivalents. Thus Me3SiCHN2,[4] Me3SiCN,[5] triphosgene,[6] and MeNHCO2CH2CF3
[7] were successfully introduced into organic synthesis. But despite the great achievements
in modern reagent and technique developments, some synthetic transformations, which
require extremely toxic and hazardous gaseous reagents are still remaining. SF4 is not so common reagent in comparison with the discussed above, but it is a key
compound in organofluorine chemistry.[8] The compound is a colorless, highly reactive, and corrosive gas (bp –38 °C), possessing
extreme toxicity (LD50 = 19 ppm (86 mg/m3, 4 h, rats[9])). Also, SF4 causes burns on unprotected skin due to formation of HF and SOF2 as a result of hydrolysis. Of course, such properties of SF4 significantly limited its application in synthesis, especially in regular laboratories.
Nevertheless, unique properties of SF4 in substitution of carbonyl oxygen with two fluorine atoms are very attractive. This
is making development of more safe and convenient SF4-based analogues like DAST (Et2NSF3) and XtalFluor-E ([Et2N+=SF2]BF4
–) or other similar reactants like fluoro-amine reagents (FAR) very important.[10] Such replacement of reagents is successful, but it does not always happen. Fluorination
of carboxylic acids to CF3 derivatives is one of the most known examples to the contrary. This process proceeds
smoothly under SF4 treatment, but in the case of DAST or XtalFluor-E the reaction stops at the fluoroanhydride
formation step. There is only one successful example described – Fluolead (4-tert-Butyl-2,6-dimethylphenylsulfur trifluoride), which is used for fluorination of carboxylic
acids to CF3 derivatives instead of SF4.[11] However, Fluolead is a rather expensive reagent, therefore this approach does not
find further application. In this work we describe another example of utilizing SF4 as unique deoxofluorinative reagent, like cited above. As a part of our ongoing efforts
on design and synthesis of advanced reagents for medicinal chemistry[12] and especially functionalized gem-difluoro derivatives[13], we chose β-keto esters, the precursors for β,β-difluorocarboxylic acids, promising
building blocks for medicinal chemistry, as substrates for the fluorination.
The products of deoxofluorination of β-keto esters are corresponding β,β-difluorocarboxylic
acids, building blocks of high value to medicinal chemistry. Some recent representative
examples A–F of such building blocks from medicinal chemistry programs related to different therapeutics
areas are shown in Figure [1].[14] In spite of wide use of β,β-difluorocarboxylic acids as building blocks by big pharma
and biotech companies, direct and efficient approaches to their synthesis are still
unknown. The analysis of compounds presented in the literature reveals, that many
of them are known, but available only from commercial sources without any information
about synthetic routes and procedures.
Figure 1 Example of bioactive compounds based on β,β-difluorocarboxylic acids
The first attempts of deoxofluorination of β-keto esters were made in the early 1980s
by L. M. Yagupol’skii and co-workers.[15] As were shown in these seminal researches, the reaction was accompanied by side
dehydrofluorination processes, the impact of which could be decreased by reducing
the temperature (Scheme [1]). Therefore, the reaction in HF media at room temperature could be considered as
preparative.[16] Nevertheless, all attempts to replace SF4 by DAST failed. Unexpectedly, in this reaction DAST, introducing an additional fluorine
into the molecule, formally oxidizes the substrate.[17] In our previous investigations we also tried to optimize the reaction and replaced
SF4 with DAST-type reagents, but all our attempts failed as well. In consequence an alternative
synthetic route was proposed.[18] The strategy was based on three-step transformation of the ester function into a
nonacceptor CH2OAc group, which allowed DAST-based deoxofluorination. The further deacylation/oxidation
led to desired β,β-difluorocarboxylic acids. In spite of successful realization of
the strategy additional six-step sequence was needed, so the total yields were in
14–16% range. Such avoiding of SF4 is justified for the small-scale synthesis but inefficient for the further scale-up.
Therefore, we decided to test diverse deoxofluorination reactions of β-keto esters
with hazardous SF4 in autoclave conditions and scale them up to hundred grams.
Firstly, we tested the reaction of deoxofluorination by SF4 with and without addition of HF at different temperatures and different ethylacetoacetate/SF4 ratios using the simplest ethylacetoacetate (1a) as a model compound. It was found that in the absence of HF, the reaction proceeded
nonselectively with predominant dehydrofluorination to the product 3 at 100 °C, as well as at 25 °C. The fraction of dehydrofluorination was dramatically
decreased by addition of HF to the system, and at 25 °C a significant selectivity
of formation of β,β-difluorocarboxylic ester 2a was achieved (Scheme [1]). Further optimization showed that the most favorable was the amount of HF of 0.8
mL per 1 g of ethylacetoacetate, the ratio of SF4/keto ester = 1.7:1, and the reaction time of 10 h. Using these conditions, we performed
the reaction on 100 g scale of ethylacetoacetate in 1.2 L Hastelloy autoclave. The
level of dehydrofluorination was less than 5% and as a result the desired β,β-difluorocarboxylic
ester 2a was isolated in preparative 70% yield.
Scheme 1 The synthesis of β,β-difluorocarboxylic acid
For the investigation of scope and limitation of the developed protocol, a diverse
set of substrates were chosen. Acetoacetic ester derivatives 1a–g, their mono- and dialkyl-substituted analogues 1h–k and 1l–o, respectively, functionalized acetoacetic ester derivatives 1p–s and cyclic β-keto esters 1t–y were presented among them. The nonenolizable dialkylated derivatives 1h–k were added to the set for checking the influence of possible enol formation as the
reaction occurs. Also, the set of functionalized acetoacetic esters 1p–s were tested for the group-tolerance determination, and derivatives 1t–y to examine the impact of conformational restriction (Figure [2]).
Figure 2 The set of substrates for SF4 based deoxofluorination
These substrates were tested in deoxofluorination reaction with SF4/HF system according to the aforementioned optimized protocol for ethylacetoacetate
(1a).[19] The procedure appeared to be suitable for most β-keto esters except for the substrates
highlighted in boxes in Figure [2]. Treatment of compounds 1d, 1r, and 1s with SF4/HF led to complex undefined mixture of products, the desired dehydrofluorinated compounds
were not observed. The cyclopropane derivative 1d probably decomposed via cyclopropylmethyl/cyclobutyl cation rearrangement, which we had observed during fluorinations
earlier.[20] Decomposition of 1r in the reaction conditions was unexpected due to our previous successful experience
with DAST fluorination of TFA-protected amino ketones,[18] while decomposition of the substrate 1s was anticipated. According to our previous expertise fluorination of compounds containing
PhCH2O fragment by SF4 led to debenzylation with subsequent unselective decomposition. It should be noted
that compound 1w, bearing an ether fragment, also did not give the desired product in the SF4/HF system. In this case the compound having m/z [M+] = 286 in GC–MS and m/z [M + 1] = 287 in positive mode in APCI HPLC MS was observed as a major product, but
we were not able to determine its structure based on these results as well as on NMR
data. Nevertheless, the product 1w was successfully deoxofluorinated by SF4 in the absence of HF at 60 °C in 61% preparative yield,[21] which was a rare exception to our procedure. The rate of dehydrofluorination in
this case was less than 10%. Substrates 1e, 1j, and 1v also reacted unselectively under the optimized conditions, but the corresponding
deoxofluorinated products were registered at about 10%, making the procedure nonpreparative.
The results of deoxofluorination of β-keto esters 1 are summarized in Table [1]. The preparative yields obtained using the above-mentioned substrate set are high
(from 55–90%) and comparable for both enolizable and nonenolizable keto esters. Considerable
rate of dehydrofluorination was observed in the case of using substrates 1u (up to 20%), 1c, 1h (up to 10%), and 1b, 1i (up to 5%).
Table 1 Yields of Deoxofluorination of β-Keto Esters with Subsequent Hydrolysis to the Corresponding
Carboxylic Acids
|
|
Fluorination
|
Hydrolysis
|
Entry
|
Substrate
|
Product
|
Scale (mol)a
|
Yield
(%)
|
Protocol
|
Bp
(°C/mmHg)
|
Product
|
Yield
(%)
|
Protocol
|
Bp (°C/mmHg)
|
1
|
|
|
0.6
|
70
|
A1
|
126–127/760
|
|
78
|
A2
|
70–72/10
|
2
|
|
|
0.3
|
82
|
A1
|
44–46/20
|
|
83
|
A2
|
77–78/10
|
3
|
|
|
0.3
|
78
|
A1
|
67–69 / 20
|
|
84
|
A2
|
87–89/10
|
4
|
|
|
0.9
|
75
|
A1
|
57–59 / 20
|
|
69
|
A2
|
85–88/10
|
5
|
|
|
0.6
|
70
|
A1
|
35–37/20
|
|
74
|
A2
|
61–62/10
|
6
|
|
|
0.3
|
81
|
A1
|
61–62/20
|
|
80
|
A2
|
77–80/20
|
7
|
|
|
0.6
|
77
|
A1
|
77–72/20
|
|
80
|
A2
|
86–88/10
|
8
|
|
|
0.6
|
73
|
A1
|
50–52/20
|
|
70
|
A2
|
71–72/10
|
9
|
|
|
0.9
|
88
|
A1
|
65–66/20
|
|
82
|
B2
|
88–90/10
|
10
|
|
|
0.9
|
70
|
A1
|
55–56/20
|
|
69
|
B2
|
70–72/20
|
11
|
|
|
0.6
|
85
|
A1
|
75–76/20
|
|
88
|
B2
|
105–107/10
|
12
|
|
|
0.6
|
90
|
A1
|
58–59/0.3
|
|
88
|
B2
|
95–97/0.3b
|
13
|
|
|
0.9
|
86
|
A1
|
46–47/0.3
|
|
67
|
A2
|
–c
|
14
|
|
|
0.6
|
48
|
A1d
|
95/10
|
|
73
|
A2d
|
–c
|
15
|
|
|
0.6
|
85
|
A1
|
52–53/20
|
|
87
|
A2
|
99-101/10b
|
16
|
|
|
0.3
|
55
|
A1
|
56–57/20
|
|
84
|
A2
|
107–110/10b
|
17
|
|
|
0.6
|
61
|
B1
|
64–67/10
|
|
79
|
A2
|
91–92/0.3b
|
18
|
|
|
0.3
|
86
|
A1
|
67–68/20
|
|
90
|
B2
|
44–45/0.3b
|
19
|
|
|
0.3
|
91
|
A1
|
88–82/20
|
|
85
|
B2
|
62–63/0.3b
|
a Amount of starting β-keto ester.
b Crystallized after cooling, mp (10o) 90–91 °C; mp (10t) 68–69 °C; mp (10u) 74–75 °C; mp (10w) 78–79°C; mp (10x) 72–73°C; mp (10y) 77–78°C.
c Solid compounds, crystallized from hexane, mp (10p) 162–163°C; mp (10q) 58–59°C.
d 3 equivalents of SF4 were used in step A1; 6 equivalents of HCl without formic acid were added in step
A2.
The reaction was scaled up to 50–150 g (0.3–0.9 mol) of starting material from one
synthetic run without changing the protocol. Such amounts required operating with
significant quantity, up to 175 g, of SF4 for one run. These operations were performed in the special well-ventilated laboratory
with strictly limited staff access due to safety reasons. The staff always wear the
personal-protection equipment including single-filter, full-face masks during operation
in accordance with international safety regulations.[22] The damper technique applied for loading of SF4 into the autoclave was shown in Figure [3]. The standardized damper chambers were used, which contained 25±1 g of SF4 at atmospheric pressure. The excess of SF4, along with the other gaseous byproducts, is vented from the autoclave through a
KOH solution after the reaction is complete.
Figure 3 Equipment for SF4-based dioxofluorination. (a) Opened Hastelloy autoclave 1200 mL; (b) loading of SF4 to vacuum autoclave from the balloon through damper chamber; (c) releasing of the
excess of SF4 and gaseous byproducts into KOH solution. 1 – vacuumed autoclave loaded with substrate
and anhydrous HF; 2 – tank with liquid nitrogen; 3 – damper chamber filled with SF4; 4 – balloon with SF4; 5 – canister with 15% aqueous solution of KOH.
All obtained β,β-difluoroesters 2 were subjected to subsequent hydrolysis into the corresponding acids. Taking into
account unsustainability of enolizable β,β-difluoroesters to fluorine anion elimination,[23] the acidic conditions[24] were chosen for hydrolysis. In the case of nonenolizable β,β-difluoroesters 2e,f and 2l–o more convenient alkali hydrolysis was applied.[25] The corresponding acids were obtained in good preparative yields under both conditions
(Table [1]).
The next milestone of the investigation was the elaboration of an efficient method
for the synthesis of Medicinal chemistry relevant difluorinated cyclic amino acid
derivatives type 11 starting from readily available compounds type 12 (Scheme [2]). Earlier, this methodology was applied only for the 3,3-difluoroproline derivative
11c. Recently, the preparative deoxofluorination of the corresponding precursor, where
PG = Cbz and R = t-Bu, was described using DAST as a reagent.[26] The approaches to derivatives of amino acids 11b and 11c were also described based on another methodology. The 3,3-difluoroisonipecotic acid
derivatives were obtained via multistep synthesis starting from ethyl bromodifluoroacetate
as CF2 moiety source.[27] In the case of 4,4-difluoro-β-proline the core was assembled by [3+2] cycloaddition
of azomethine ylide with benzyl 3,3-difluoroacrylate.[28]
Scheme 2 The synthesis of gem-difluorinated cyclic amino acid derivatives
At first, we chose the NBn-protected compounds 12a–d as potential substrates for deoxofluorination. These compounds were examined in a
standard protocol with SF4 in HF. Among them substrates 12a–c gave the corresponding difluoro derivatives 11a–c in good preparative yields (from 68–83% on 0.6 mol scale of starting materials).
But compound 12d unexpectedly gave dehydrofluorinated compound 13d as the major product under the reaction conditions according to 19F NMR and 1H NMR analysis of the reaction mixture and the crude product (Scheme [2]). Unfortunately, all attempts to isolate the reactive compound 13d in a pure state failed. The deprotected fluorinated amino acids 14 could be quantitatively hydrolyzed in acidic conditions[29] to the corresponding acids 16 as hydrochloric salts. It was illustrated by the synthesis of Bn-protected amino
acids 16a,b. Amino acids 11a,b were formed by catalytic hydrogenation of Bn-protected derivatives 16a,b at room temperature and 1 atm hydrogen pressure over Pd on carbon in MeOH–H2O media[30] as hydrochloric salts. These compounds were easily transformed into Boc-protected
derivatives 18a,b
[31] that are more convenient for utilizing as building blocks in parallel synthesis
in comparison with Bn-protected derivatives. The orthogonal benzyl deprotection from
the compound type 14 could be also accomplished by catalytic hydrogenation.[32] It was demonstrated by synthesis of the amino ester 15c. In the case of 4,4-difluoro-β-proline derivatives replacement of Bn protection group
with TFA in substrate 19d changed the behavior of deoxofluorination. The standard SF4/HF protocol gave the desired difluoro derivative 20d in 71% preparative yield on 80 g scale of the used starting material. TFA deprotection
by classical nucleophilic methodologies with such reagents as NH2NH2, NH2OH, or NaOMe is incompatible with enolizable β,β-difluoroesters. Therefore, an alternative
mild acidic deprotection method was applied. It was found that ethanolic solution
of anhydrous HCl, generated by addition of acetyl chloride into EtOH,[33] selectively cleaved TFA amide, leaving the ester function intact.[34] In the case of compound 20d the method led to amino ester 15d in 90% yield as hydrochloride (Scheme [2]).
A preparative solution for substrate 1j and its analogues was found. The ester group in these compounds could be exchanged
to a nitrile. Deoxofluorination of keto nitriles 21a, 21j, and 21t proceeded smoothly in SF4/HF system according to protocol A1 at room temperature (Scheme [3]). The desired β,β-difluoronitriles 22a, 22j, and 22t were formed in moderate to good yields (46–78%) on 0.15–0.3 mol scale. The possibility
of hydrolysis of such nitriles was demonstrated on the intermediate 22j. The acid 10j was obtained through acidic hydrolysis of nitrile 22j by H2SO4 at 90 °C[35] in 74% yield.
Scheme 3 Deoxofluorination of β-keto nitriles by SF4 in HF
Finally, the reaction of deoxofluorination of β-keto esters and β-keto nitriles by
SF4 in anhydrous HF media was investigated. The scope and limitation of the reaction
were determined. Substrates having steric hindrance at the keto group, bearing ArCH2O–, –NHTFA, and fragments capable of cationic-like rearrangements, are out of the
scope of the procedure. The reaction was scaled up to 0.9 mol of starting material
using 1200 mL Hastelloy autoclave. Work with such quantity of SF4 required a special technique and equipment, which was also demonstrated. The promising
building blocks for medicinal chemistry, β-difluorinated acids, were produced by hydrolysis
of the appropriate esters on 100 g scale. Despite the serious difficulties of using
toxic, hazardous SF4 towards special lab space, equipment, personal protection, and staff skills, the
elaborated methods are substantially more efficient in comparison with multistep sequences
based on less hazardous fluorine sources. Moreover, the proposed protocol can be easily
introduced into the production cycle at industrial facilities that use SF4.