close

Вход

Забыли?

вход по аккаунту

?

Organocobalt Compounds in the Synthesis of PyridinesЦAn Example of Structure-Effectivity Relationships in Homogeneous Catalsis.

код для вставкиСкачать
Organocobalt Compounds in the Synthesis of PyridinesAn Example of Structure-Effectivity Relationships in
Homogeneous Catalysis* *
By Helmut Bonnemann"
Dedicated to Professor Giinther Wilke on the occasion of his 60th birthday
The cocyclization of alkynes with cyano compounds using organocobalt catalysts of the
type [YCoL] has evolved into a versatile and technically useful method for synthesizing pyridine and its derivatives. An important advance came with the realization that the organo
group Y remains attached to the cobalt throughout the catalytic cycle. This opened u p the
possibility of optimizing the catalyst by varying the controlling ligand Y.
q
1. Introduction
Pyridine and its derivatives are technically important
fine-chemicals. Their isolation from coal tar is decreasing,
whereas their manufacture by synthetic methods has increased rapidly since 1950: in 1958, ca. 3700 t of pyridine
and its derivatives were produced in the USA; by 1978, the
yearly production had reached 14000 t, and, in 1980, it was
already in excess of 16000 t. In the coming years the annual growth is expected to be 4%.''I Aside from the parent
compound most importance attaches to 2-methyl-, 2-ethyl-,
2-vinyl- and 2-amino-pyridine, all of which can be prepared selectively and in one step from acetylene and the
appropriate cyano compound using soluble organocobalt
catalysts [Eq. (l)].
111
%
-
R
QC0Q
b
3
2
4
The Japanese authors later turned to cobaltocene 5,[5.61
since the phosphane complex 1 proved to be unsuitable
for practical purposes. Hardt (Lonza AG) developed various procedures for preparing pyridine derivatives with 5
as the catalyst.[71An elegant model reaction by Yamazaki
and Wakatsuki[']showed that cobaltocene can be regarded
as the precursor for the actual catalyst complex 6 : prior to
the catalysis, 5 is converted by alkyne into an $-diene complex [Eq. (2)].
[COl
+
coQ
QR
H<=C-H
2
c o
It = 1 1 . CH3, HC=CH2, NH2
[ C o ] = soluble cobalt catalyst
H
As long ago as 1876, Sir William Ramsay[21obtained
small amounts of pyridine on passing acetylene and hydrogen cyanide through a red-hot tube. In 1973, Yamazaki
and Wak~tsuki['~
first reported the homogeneous catalytic
cycloaddition of alkynes and nitriles using the phosphanestabilized cobalt(iI1) complex 1.
Vollhardt et a1.[8.'1incorporated a p d i y n e s and nitriles
in cobalt-mediated [2 2 2]-cycloadditions [Eq. ( 3 ) ] .
+ +
R
It
FCH
I
+
(CH2)n
R
H
6
5
\
C=CH
n = :j,4,5;
C
[Co(CO),(nS€5H5)I
7
(3)
Ill
N
R = alkyl, a r y l , e t c .
8
Plt3 R
At the same time we[41observed the catalyzed cocyclization [Eq. (I)] o n cobalt catalysts prepared in situ, as well as
on easily accessible organocobalt-diolefin complexes of
the type 2, 3, and 4 .
[*] Prof. Dr. H. Bonnemann
Max-Planck-lnstitut fur Kohlenforschung
Postfach 01 1325, D-4330 Mulheim a. d. Ruhr (FRG)
[**I
Based on the work of Helmut Bonnemann, Werner Brijoux, Ruiner Brinkmann. Willi Meurers. Michael Radermacher, and Srephan Wendel. The
author thanks his co-workers for their important contributions to this
aork.
248
0 VCH Verlaysgesellschaft mbH, D-6940 Wernheim. 1985
Using this variant of the cobalt-catalyzed cycloaddition,
Schleich et aI.["l opened u p a new route to pyridoxine (Vitamin BJ as hydrochloride 11 (Scheme 1).
In our experience, substituents in the alkyne and in the
cyano component can be widely varied, so that we have
been able to develop the basic reaction [Eq. (l)] into a general synthetic method for preparing mono- to penta-substituted derivatives (for preparative results see [ ' I , I l l ) .
We concentrated on the development of highly reactive
organocobalt(i) complexes of the type [YCoL], in particular for the homogeneous catalytic synthesis of 2-alkylpyr-
0570-0833/85/0404-0248 $ 02.50/0
Angew. Chem. I n / . Ed. Enyl. 24 (1985) 248-262
2.1. Y = q5-Cyclopentadienyl Derivative
10
9
In contrast to ferrocene and similar compounds,[l5I in
the case of (q5-cyclopentadieny1)Co1half-sandwhich complexes only a few examples of ring substitution reactions
are known. Thus, the acetyl- and benzoyl-derivatives 12['']
are obtained on Friedel-Crafts acylation of the complex 7
[Eq. (411.
11
Scheme I . Synthesis of substituted pyridinea. X = O : R ' = K'=(CH,).Si;
R'=C,H,OCO, R2=(CH,)$3: ratio 9 : 10= 16:l. Pyridoxine hydrochloride
11 is obtainable by secondary reactions.
idines and 2-vinylpyridine according to equation (1). Since
the actual catalytic cycle occurs at a core-complex of the
type [YCo], it is possible to change the properties of the catalyst by varying the organyl ligand Y.[13.141
As shown in
Scheme 2, a large number of catalyst complexes having a
wide variety of Y and neutral ligands have been synthesized. The catalytic properties of these complexes were
tested under standardized conditions. At first, the role of
the neutral ligand was thoroughly investigated; in the subsequent catalyst-screening, the catalytic turnover number
(TON) of the "controlling ligand" Y was determined.
R
= CH3, C5H5
For a general entry to (R-C,H,)Co complexes, it is first
necessary to prepare appropriately substituted cyclopentadiene derivatives 13 and then in a second reaction step to
complex these and the neutral ligand to cobalt [Eq. (Sa),
(5~1.
(5a)
13
A classic method for the complexation step (5b) involves
the reaction of C O ~ ( C O 14
) ~ with monomeric cyclopentadiene derivatives, which leads to the corresponding dicarbony1 complex 15 in 70-90?41 yield"'. I x l [Eq. (611.
I
C o 2 ( C O ) o + 2 R'
&-
R2
2
.I&
14
C O ( C O ) 2 + 4 C'O
(6)
15
COD = I , 5-cyclooctadiene
K'
= ii; 11' = H , CH3, (CH3)3S1, CEH5
It' = R 2 = (CH3),Si
Scheme 2. Synthesis of substituted pyridines with complexes of type [YCoL].
Y IS the controlling ligand, L is the neutral ligand.
The final step involves optimization of the reaction conditions for the synthesis of 2-alkyl- and 2-vinylpyridine.
2. Synthesis of Complexes of the Type [YCoLl
In order to obtain as broad an insight as possible into
the control function of the group Y attached to cobalt, we
prepared cobalt complexes having mono- and multiplysubstituted 7'-cyclopentadienyl, q5-indenyl groups as well
as boron-containing cyclopentadienyl and q-cyclooctenyl
groups. Ethylene, various diolefins, aromatic hydrocarbons, carbon monoxide, and phosphorus-containing compounds served as the neutral ligands.
Anqew. Chem. I n [ . Ed. Engl 24 (1985) 248-262
The carbonyl groups can be cleaved either thermally or
photochemically at room temperature. In the presence of
1,5-cyclooctadiene (COD), diolefin complexes 16 (type 4 )
are obtained in high yields [Eq. (7))
The two-step carbonyl method [Eq. (6), (7), and (S)]
proved to be a very versatile catalyst synthesis on a laboratory scale.["] The complex 17 can be prepared from either
1,2,3,4,S-pentamethyl~yclopentadiene~~*~
'"I
or
methyl
1,2,3,4,5-pentamethylcyclopentadienylketone'"' according
to equation (6). The subsequent transformation of 17 into
the diolefin complex 18 is achieved in high yield by irradiation in the presence of COD[*'] [Eq. (S)].
In order to stabilize the thermally unstable cyclopentadiene derivatives 13 as salts and then to react these to give
the corresponding cobalt complexes under mild conditions, we adopted the approach developed b y Rausch et
249
Me
Me
1
-H
Me
13%
H Me
Me
The carbonyl complexes 15 were converted into the corresponding COD-complexes as shown in equation (7). The
[YCoL] complexes which were prepared using the Rausch
method are listed in Table 1.
We obtained the binuclear cyclooctadiene(fulva1enediy1)cobalt complex 20 in high yield and avoiding carbonyl intermediates by protolytic substitution of 2 with dihydrofulvalene [Eq. (lo)].
Me
17
- CH3CO
60%
Me
hv
17 + C O D
78%-
[ ( M e s C 5 ) C o ( c o d ) l 18
2
[( c o d ) C o(C5H4-C5H,)C o(cod)1
20
al.[23’ 14 reacts with an equimolar amount of iodine at
25°C in tetrahydrofuran (THF) to give the intermediate
19, which is allowed to react further with cyclopentadienide salts to yield derivatives of type 15 [Eq. (9)].
20
C O ~ ( C O +) ~I2
THF, 25‘C
------+
19
+
+
2 “ICo(CO),”
(8
- 2n)CO
19
14
M@C5H,RQ - - + R a C o ( C 0 ) 2
+ (n- 2)C0 +
MI
(9)
15
M = Li, Na, K , T 1
R = C1, B r , I , ( C G H ~ ) ~ C
Table 1. Synthesis of [(q’-RC5H,)Co] complexes [see Eq. (9) and (7)].
[XCo(PR3)31 + L i Y
R
CI
Br
I
HCO
CHXO
CH,CHOH
H,C=C(CH,)
C2HsC0
CH,OCO
HOOC
ClCO
CbHSNHCO
(HSC5)FeC5H4-C0
PI
(CoH,hC
ICI
(CaHs)2P [dl
(C,H,)zP(S)
54
57
12
36
45
86
26
21
42
13
70
62
5
16
24
42
4
10
1231
(231
“231
1231
123, 251
”231
[23, 271
1241
118, 231
1231
“231
[231
1231
119, 23, 261
1191
1191
1191
I191
90
52 67
86
80
53 64
87
92
91
95
79
[a1
90
54
20 87
79
62
52
[el
1191
[19]
1191
(241
[14, 241
[24]
[24]
[24]
1241
[24]
[24l
[24]
(241
1191
1191
[I91
[19]
[a] Decomposition on irradiation. [b] Fulvalene derivative, cf. 20. [c]
(CaH5)&H instead of RCSH4. [d] “Monocarbonyl dimer” bridged via P,
having the formula [(PhZPC5H4)Co(CO)I2.[el Not investigated.
250
Using an analogous method to that developed by Rausch
via “ICo(CO),” 19, we devised a new synthetic route to
[YCo(cod)] complexes. This approach is based on the reaction of halotris(triorganophosphane)cobalt(i) complexes
with cyclopentadienide salts and subsequent displacement
of both triorganylphosphane ligands with
[Eq.
(1 111.
+
[ Y C O ( P R , ) ~+
] LiX
+
PR,
L_1
R = OC3H7, C6H5; X = C1, B r
In our search for a one-step synthesis of diolefin-complexes we developed the approach shown in Scheme 3.
In this direct reduction salts of di- and trivalent cobalt
are allowed to react with various reducing agents in the
presence of the substituted cyclopentadiene RC5H5 and
the neutral ligand L to give half-sandwich complexes.
Thus, Co(acac), reacts in the presence of cyclopentadiene
and COD with AI(C,H,), to give 4 in 77% yield.[281Elementary magnesium is better suited as the reducing agent
in a technical direct synthesis of the catalyst complex according to Scheme 3. The magnesium is both economical
Angew. Chem. I n t . Ed. Enql. 24 11985) 248-262
COX,
+
Cobalt
salt
[CO]
mRed
Reducing
agent
+
RCsHs
[Co]
+
+L
+ Red,X,
The resulting “naked cobalt” is stabilized by complexation with olefins. The direct reduction of cobalt salts with
anthracene-activated magnesium in T H F in the presence
of cyclopentadiene or its derivatives and diolefin neutral
ligands, in particular COD, represents the currently most
economical route to [(qs-RC,H,)Co(q4-diolefin)] complexes (Table 2).
Two routes to give [(q5-C5H5)CoL]systems by varying
the neutral ligand L involve .transformations of cobaltocene 5 . The displacement of one cyclopentadienyl moiety
from 5 with potassium in the presence of ethylene, as described by Jonas et al., smoothly yields the bis(ethy1ene)complex 21, which reacts with 2-butyne to give the
[Eq. (12)].
20-electron system 22[321
“Naked”
cobalt
* [(T~~-RC~H~)COL]
Catalyst complex
L = diolefin, e.g. COD, norbornadiene, RCSHS,cyclohexadiene
X = CI, Br, I, OCZH5,acac
n, m = 2 , 3
Red= AI(C,H,),, AI(C,H,),, Na, K, Mg, Ca
Scheme 3. Synthesis of catalyst complexes with “naked” cobalt.
and easy to handle. The effectiveness of the reductant is
however dependant on its active surface area: if magnesium is employed in Scheme 3 without any special precautions, then the catalyst complex is obtained in only ca. 10%
yield due to only partial reduction of the cobalt salt. A
substantial improvement was achieved when it was found
that the addition of 3-6 mol-Yo of anthracene to the magnesium powder acts as a phase transfer catalyst to dissolve
the magnesium. This “organic soluble” magnesium (Mg*)
is quasi-atomic and is therefore able to totally reduce transition metal
(Scheme 4).
f
c OD
1 Mp-powder<
1LI e
22
Me
4
0.1 mm
1
Scheme 4. Preparation of anthracene-activated magnesium (Mg*).
Reaction of 5 with CH-acidic organic reagents in the
presence of oxygen according to the method developed by
Kojimu and Hugiharu et al.[331leads to [(q5-CsH5)Co]-complexes which contain substituted q4-cyclopentadiene
groups as stabilizing neutral ligands [Eq. (13)].
complexes using anTable 2. Direct synthesis of [(11’-RCSH,)Co(r14-diolefin)]
thracene-activated magnesium (in analogy to Scheme 4). Numerical values:
mmol.
R
H
Diolefin
Cyclopentadiene 50
H
1,3-Cyclohexadiene
H
cod 4
CH,
5-Methyl1,3-cyclopentadiene [a]
CH,
cod 51
rBu
cod
C 6 H 5 cod
Me,Si cod
Me,Ge cod
Co(acac), Mg Anthra- RCsHs Diole- Yield Ref.
cene
fin
1%1
100
300 6.2
800
-
63
100
300 6.2
100
200
53
I00
Ill
400
250
100
300 6.2
300 6.2
19
62
100
50
100
100
5
300
I50
300
300
15
110
57
90
250
125
250
250
11
23 + 2 R H
-a
2
23
Co O
R
+
(13)
Hz02
24
6.2
3.1
6.2
6.2
0.3
-
R H = H s C ~ - C = C H , HsC-CO-CH,,
( 5 6 7 0 , 7070, 8 7 % , 5170, 97%)
HsC-CN, H ~ C - C H Z - C N , HCCl,
71
42
[a] Starting from 1 methyl- 1,3-cyclopentadiene.
The reaction proceeds via the peroxy bridged complex
23. In the presence of CH-acidic compounds, the peroxy
bridge is cleaved with liberation of H z 0 2 and formation of
the q4-R-C5H4 complex 24. With the intention of convert-
Angew. Chem. lnt. Ed. Engl. 24 (198s) 248-242
25 1
110
5.5
1
70
30
ing cobaltocene 5 into active catalyst complexes of type 24
using simple reagents, we extended the experiments of the
Japanese a ~ t h o r s ~ ~to' ' reactions of 5 with cyclopentadiene, methylcyclopentadiene, indene, and methanol in the
presence of oxygen [Eq. (14)]. In this way the transformation of an $-cyclopentadienyl group into an q4-cyclopentadiene system having exo-substituents was achieved in all
cases (yields 30-60%)."9,341
Table 3. Synthesis of [(qi-R',RZ,R'C~Hq)Co(~~-diolelin))
complexes using
anthracene-activated magnesium (in analogy to Scheme 4). C,H, = indenyl.
The complexes with I-manosubstituted indenyl moiety were prepared from
I-substituted indene. Exception: R' = CnHSfrom 3-phenylindene. Numerical
values: mmol.
K'
k3
R'
~
The same result was obtained by reaction of 5 with H 2 0
in the presence of CH-acidic corn pound^."^^ The intermediate cobaltocenium hydroxide 25 is as basic as the alkalimetal hydroxides[351and abstracts the proton from the
acidic reagent RH to give water. The anion R" undergoes
nucleophilic addition to the cobaltocenium ion, and the
formation of 24 is accompanied by a change in color from
yellow to red.
R'
R3
H
Diole- Co(acac),
fin
Norbornadiene
H
H
H
cod 26
H
cod
CH~ n
CH, cod
CH,
H
cod
C2H5 H
H
cod
IBU
H
H
Me,Si H
H
cod
cod
C,H, n
H
NMel H
H
cod
H I00
100
100
40
100
74
100
50
50
Mg An- In- Di- Yield Ref.
thra- dene ole- [%I
cene
fin
300 6.2
250
280 31
1371
6.2
6.2
2.8
6.2
4.6
6.2
3.1
3.1
250
250
253
100
248
188
250
125
125
1371
1371
(241
[24]
1241
1371
1241
[24]
300
300
120
300
328
300
I80
150
107
40
1I I
81
115
61
55
85
48
20
85
I1
51
26
1
with halotris(phosphite)cobalt(i) complexes and subsequent exchange of the phosphite with COD[241must be resorted to [Eq. (lS)].
R'
R1
2.2. Y = Annelated q5-Cyclopentadienyl Derivative
The q5-indenyl complex 26 (Fig. 1) was first prepared
(in SO% yield) by Samson by protolytic substitution of 2
with indene (cf. [361). The reduction of Co(acac), with anthracene-activated magnesium in the presence of indene
and C O D in an analogous manner to Scheme 4 affords 26
in good yield.'37,381
.
I
-\
fiY
The $-fluorenyl complex 29 is obtained in almost
quantitative yieid on reaction of the bis(phosphite) comp l e ~ [28
~ ~with
' excess COD at 100°C [Eq. (16)). The bonding in 29 has been confirmed by X-ray structure analys ~ s [ (Fig.
~ " ~ 1).
\
28
ca
2.3. Y = Boron-Containing Ligand
1
26
29
Fig. I . Structure of ~4-cyclooctadirnr(1)-indcn) I k u h , d ~20 j {bl dnd q4-cyclooctadiene(q'-fluoreny1)cobalt 29 1401 in the crystal.
Analogues of 26 having substituted indenyl groups are
readily prepared from indene derivatives, diolefins and cobalt salts by direct reduction with anthracene-activated
magnesium (cf. Table 3).
Limitations to the direct synthesis are encounted when
certain groups on the indene (e.g. amino- and halo-substituents) react with the activated magnesium. In these cases
the indirect approach involving reaction of indenide salts
252
q6-Borininato ligands were first used as Cp-analogous
6n-electron ligands for cobalt by Herberich et al. in 1972'4'1.
Using his($-borininato)cobalt as starting material, the diamagnetic COD-complexes 30 and 31 were prepared according to equation (17a).
In an alternative approach we prepared 30 in ca. 40%
yield from I-phenyl- 1,4-dihydroboraben~ene,~~*~
C O D and
Co(acac)3 by direct reduction with soluble Mg*"" [Eq.
(17b)i.
Siebevt et al.1431have described the preparation of numerous transition metal complexes having q5-dihydro-1,3diborolyl ligands. The preparation of the complex 33 was
Angew. Chem. Int. Ed. Engl. 24 (1955) 245-262
Co
verted into the diolefin complex 39 [Eq. (21b)], which,
after chromatography, separated as dark red crystals. 39
was completely characterized by its N M R and mass spectra.
+Li-
C o ( a c a c ) 3 + R-B
3
+ Mg + COD
42%
’
THF, 65 OC
(17b)
R - B o Co
39
R = OC3H7
achieved by Siebert and Bochmann by reaction of the dihydrodibor~le[~
32~ Iwith Co,(CO), 14.1451Better yields were
obtained by protolytic substitution of 34 with 32‘461[Eq.
( 1811.
We obtained the crystalline, red, carbonyl-free complex
35 by reduction of CoCI, with NaBH(C,H& in the presence of the ligand 32 and a n t h r a ~ e n e l [Eq.
~ ~ ’ (19)].
The structure of 35 was deduced from the I3C-NMR
data. An analogous reaction using naphthalene as the stabilizing ligand gave a mixture of 36 and 37 in the ratio
20:80[471[Eq. (ZO)]. 36 and 37 are deep-red crystalline
compounds; their structures could be unequivocally determined by I3C-NMR spectroscopy. If the indene system is
Et
COClz
+ 32 +
8[
NaBHEt3
Besides the CpCo-analogous complexes discussed
above, one can also use q3-allyl-Co1complexes in the pyridine synthesis [Eq. (l)]. In this case, the catalysis cycle occurs at an q3-allylcobalt core-complex with 12 outer elect r o n ~ [ ’ ~ (Fig.
- ’ ~ ] 2).
33
32
34
2.4. Y = q3-Cyclooctenyl
co
E
’
t3:
8
8
(20)
THF, 20°C
E:
37
extended by introduction of a boranediyl group, then formally “boranaphthalenes” result.14s1In collaboration with
Paetzold and Finke[491we were able to bind such a system
to cobalt by the routes shown in equation (21a) and
(21b).
Reaction according to equation (21a) furnishes the phosphite-stabilized complex 38. This intermediate is conA n y e w . Chrm. In!. Ed. Engl. 24 (1985) 248-262
Fig. 2. ~’-Allylcabaltcare-complex.
COT
We were able to prepare the long known””] catalyst complex 3 in 70% yield from Co(acac), and COD by direct reduction with A1(CzH5)3.”41A convenient synthesis of 3
based on cobaltocene has recently been developed by
Jonas et al.’”] When 3 is allowed to react with hydrogen in
the presence of arenes, the allylic C8-ring is eliminated as
cyclooctane and one obtains a new class of crystalline
[Co(q‘-arene)(q 1,q2-cyclooctenyl)]complexes such as 40 44 [Eq. (22)][”].
v
40
41
42
43
44
3
40 -44
Arene
Yield [Yo]
Benzene
Mesitylene
Hexamethylbenzene
Aniline
Anisole
62
41
12
76
69
Using 13C-NMR spectroscopy it can be shown that the
eight-membered ring in the new IS-electron complexes
40-44 is not bonded to cobalt as an allyl-system, but as
a n q‘,q*-system.
3. The Role of the Neutral Ligand L
As shown in Scheme 2, the initial step in the cobalt-catalyzed pyridine synthesis is the displacement of the neutral
ligand L from [YCoLI-complexes by alkyne/nitrile. From
c o m p l e x - ~ h e m i c a l and
’ ~ ~ ~kine ti^^^^.^^] findings it is evident
253
I
that the coordination sites of the core-complexes [YCo] are
occupied by two alkyne molecules to give 45.
45
As can be deduced from equation (23), L merely functions as a “place-reserver” which stabilizes the catalytic
central unit [YCo]as isolable complexes. The liberation of
the propagating species 45 depends on the complexing
ability of the neutral ligand. The nature of L is therefore of
prime importance for the “starting behavior” of the various [YCoLI-complexes in the pyridine synthesis [Eq.
(2411.
0
co
T=1250C
COD
1
+2HC-CH,
1
0c c ]
Fig. 3. Continuous flow apparatus lor the optimization of homogeneous
catalytic processes. A: catalyst solution, B: starting compounds, C : thermostated reactor, D: trap, E: gas-chromatograph, F: data evaluation.
Thereafter, the temperature-dependence of the propyneconversion was monitored and the temperatures at which
each [YCoLI-system continually gave 65% propyne conversion in the test reaction [Eq. (25)] were determined. Figure
4 shows five such conversion/temperature curves.
(24)
4
c
-
.-0
L?
30-
$
20-
al
c
a
/
% 10Ethylene is already displaced from the cobalt of 21 at
room temperature in the presence of the substrates,[32d1
whereas cyclobutadiene and its derivatives cannot be displaced from the central metal, either thermally or photochemically.[91 The chelating COD-ligand is a compromise
between the weakly stabilizing neutral ligand and the
“blocking” systems, since, as evidenced by GC, it can be
displaced in a temperature range of practical importance
(1 10- 130°C). The major influence of the various ligands
on the initial step of the catalysis [Eq. (23)], and, therefore,
on the “starting behavior” of the various [YCoLI-complexes in the catalyzed pyridine synthesis, was investigated
using a test reaction [Eq. (25)]. In this study certain con[YCOL]
EtCEN
+
2 MeC=CH
Me
Me
Me
46
47
48
49
centration ratios were adopted and a continuous flow apparatus (Fig. 3)[’4,54.551
was used”’] (catalyst: 4.3 mmol/L;
propyne: 262 g/L=6.6 mol/L; propiononitrile: 213 g/
L=3.9 mol/L; toluene: 345 g/L).
The temperature of the reactor was slowly increased
from 25 to a maximum of 180°C and the propyne conversions on the various catalysts were determined on-line. The
temperature range in which the first propyne-conversion
could be measured indicated the “start” of the catalysis.
254
?
a
-
0 -,
30
50
70
90
iio
130
150
-
170
T [“C]
Fig. 4. Propyne conversion and reaction temperature for various [TI’C,H5)CoL]-complexes as catalysts.
As can be seen in Figure 4, the “start behavior” of the
individual (C5Hs)Co compounds is clearly dependent on
the complexing ability of the neutral ligands: the ethylene
complex 21 already liberates the propagating (C5H5)Cospecies at room temperature (curve a) : the hexamethylbenzene complex 22 causes the first propyne-conversion at
50-60°C (curve c); the q4-cyclopentadiene- and CODcomplexes 50 and 4, respectively, first cause propyne-conversion at 75--85°C and 120--125”C, respectively (curves
b and d). The conversion/temperature curves a-d of the
(C5H5)Co-catalysts 21, 50, 22, and 4 have the same characteristics: the influence of the neutral ligand on the liberation of the “active species” 45 [Eq. (23)] is noticeable in
the temperature range in which 0-25% propyne-conversion takes place. Even the bis(ethy1ene)-complex 21 only
achieves 25% propyne-conversion at ca. 120°C. Above
120°C the conversion/temperature curves a-d become
more similar. In this temperature range the neutral ligands
have been almost completely displaced, so that the propyne-conversion, which is now independent of the neutral
ligand, depends on the catalytic activity of the (C5H5)Counit. At 150”C, the catalyst complexes 21, 50, 22, and 4
continually convert 65% of the propyne pumped in. With
the dicarbonyl complex 7 , the catalytic propyne converAngew. Chem. Int. Ed. Engl. 24 (198Sj 248-262
sion only starts above 130°C, and even at 150°C only 5%
conversion occurs. It can be seen from the conversion/
temperature curve e that thermal cleavage of the carbonyl
ligands is incomplete even at 150°C. If the carbonyl
groups are displaced p h ~ t o l y t i c a l l y [ from
~ ~ ’ the (C5H5)Counit as shown in equation (26), then the catalytically active
species can be generated even below 0°C. The temperature/conversion behavior of 7 then closely corresponds to
curve a.
R’
I
If low reaction temperatures have to be used in the pyridine-synthesis because of thermally labile starting compounds, then the bis(ethy1ene)-complex 21 o r the photolytic elimination of the carbonyl groups in the dicarbonyl
complex 7 should be chosen. The reaction rates, however,
are slow at room temperature, so that long reaction times
or high catalyst concentrations are required. Using the test
reaction (25) the individual [YCoLI-complexes can be
characterized by the following parameters :
1) Starting temperature
(temperature-interval in which the first propyne-conversion is detected)
2) Reaction temperature
(required for 65% propyne conversion)
3) Regioselectivity
(molar ratio 46 to 47 or 48 to 49)
[(C5H5)Co(diolefin)]-complexes convert 65% of the propyne in the continuous flow system under comparable
conditions. After complete displacement of the neutral ligands, all the (C5H5)Cosystems have the same catalytic activity. Changes in the Y-ligands d o however have a major
influence on the “starting temperature” and the reaction
temperature required for 65% conversion. This is evident
on comparing the C O D complexes 4, 51, and 26.
The regioselectivity, which, within experimental error, is
the same for all the (C5H5)Co-catalysts, changes immediately when the Y-ligand is varied.
I n Figure 4 it can be seen that replacement of olefinic ligands L by stronger complexing carbonyl groups hinders
the liberation of the propagating species 45. The blocking
of the (C,H,)Co-moiety by strongly coordinating ligands
even at 150°C is reflected by the propyne conversions
listed in Table 5.
Table 5 . Starting temperature and propyne-conversion at 150 “C using
[(CIH5)CoL]-systems.
Starting
temperature [“C]
Catalyst
Propyne conversion [%I
65
5
16
20
A change of the Y-ligand at the cobalt while keeping the
same neutral ligand has a major effect on the “starting behavior” and the activity in the pyridine synthesis (cf. Fig.
5).
Table 4 summarizes the results obtained for some
[YCoL]-comple~es.~’~]
Table 4. Data for [YCoL] complexes in the test reaction (25). Apparatus, see
Figure 3; for conditions see text.
Catalyst
Y
L
2 CHZ=CH2 21
Me& 22
5-Methyl- [57]
I .3-cyclopentadiene
2-Pyron [58]
C5H, 50
3-(2,4-Cyclopentadieny1)indene
5-Methoxy- 1,3cyclopentadiene
cod 4
cod 51
cod 26
Starting
temperature [ T I
<30
55-65
70-80
T[”C]
[a]
148
150
148
46
48
47
49
1.60
1.72
1.66
2.59
2.58
2.60
2.65
2.70
2.63
65-75
75-85
90-95
149
147
1.66
1.70
1.62
115-120
149
1.70
2.64
120- I25
120-12s
95-100
147
I65
127
1.71
2.02
1.48
2.65
2.06
2.64
148
[a] Reaction temperature for 65% propyne conversion.
Depending on the stabilizing effect of the ligand L in the
[(C,H,)Co(diolefin)] series the “starting temperature” varies between 30 and 125°C. In the q4-cyclopentadienyl ligands, substituents with + I effect promote the displacement of ligand, whereas substituents with - I effect hinder
the displacement, so that the “starting temperature” is increased. Inspection of Table 4 also shows that all
Angew. Chem. Int. Ed. Engl. 24 (1985) 248-262
0 40
L.....
100
110
120
130
140
150
-
160
T [“C]
Fig. 5 . Propyne conversion and reaction temperature for various [YCo(cod)]complexes as catalysts.
Replacement of an q5-cyclopentadienyl by an q5-indenyl
group as Y-ligand at the cobalt lowers the “starting temperature” by 25°C. This indicates that Y has an influence
on the dissociation of L according to equation (23). A comparison of the conversion/temperature curves in Figure 5
shows that, after dissociation of the neutral ligand from
[YCo(cod)], separate curves a-c result at higher temperatures. Whereas a uniformly effective YCo-system is generated after dissociation of the neutral ligands L from the
[(C,H,)CoL]-complexes (as seen from the trend toward
255
congruence of the curves at 150°C and 65% conversion in
Fig. 4), the results given in Figure 5 for various [YCo(cod)]complexes show that the Y-ligand has a major influence
on the "starting behavior" and on the activity of the catalyst. As can be seen in Table 4, the selectivity of the cobalt
catalyst is also effected by Y. Comparison of the results
obtained with methylcyclopentadiene as the neutral ligand
L and as ligand Y clearly illustrates the roles played by
these types of ligands. The result obtained with [(11'C5H5)Co(q4-MeC,H,)] is no different from that obtained
with other [(qr-C,H,)CoL]-systems (Table 4) as far as the
reaction temperature for 65% propyne conversion and the
product distribution is concerned. If, however, the methyl
group is introduced into the cyclopentadienyl part of the
catalyst, which remains attached to the metal as the Y-ligand in the whole catalysis cycle, and therefore effects its
reactivity, then the substituent effect is reflected in the activity and selectivity of the catalyst.
The role of the neutral ligand in [YCoLI-systems consists
therefore in stabilizing the propagating central unit [YCo]
by complexation and in effecting the "starting behavior"
by liberating the active species at various temperatures. If
the species 45 has been formed [Eq. (23)], the liberated
neutral ligand has no further influence on the activity and
selectivity in the catalytic cycle. In order to develop a catalyst which is reactive and selective one must therefore vary
the ligand Y.
4. The Controlling Function of the Ligand Y
4.1. Catalyst-Screening Using Test Reaction (25)
In order to compare various [YCoI-systems, we investigated a series of [YCo(cod)]-complexes in the test reaction
(25) under identical conditions in the continuous flow apparatus (Fig. 3).1241Differences in the product distributions
are due to the influence of Y, as all experiments were carried out under stationary conditions and hence kinetic effects or influence of the neutral ligand can be ruled out.
The reaction temperature T required for 65Yo propyne conversion in the continuous system is a measure of the activity of the [YCoI-moiety. The lower the required temperature, the more reactive is the particular [YCoI-system. The
molar ratio of pyridine- to benzene-derivatives gives the
chemoselectivity and the ratio of the pyridine isomers 46
to 47 characterizes the regioselectivity.""'41 Table 6 summarizes the results obtained.
Inspection of Table 6 reveals that the three above-mentioned parameters are strongly controlled by the ligand Y.
Table 6. Reaction temperature T (for 65% propyne conversion), chemo- and regioselectivity. and "CO-NMR shift o f [YCo(cod)] catalysts in dependence of Y . For
conditions see Eq. (25); A T a n d S,,,, see text. ~ ( " C O )of K3Co(CN),=0.
Y
T
["CI
AT
I"C1
Chernoselectivity
46 + 47
__
48
Y= r$CyclopentadienyI
+ 49
46
-
Regioselectivitv
46 [%,I
47
S('YC0)
6...,
- 1413
-237
["'"I
47
derivatives
(CH,)Ks 18
C,H5-NH-CO-C,H,
HOOC-CIH,
Bicyclo[3.3.0]octadienyl 54
CH,-C,H,
CH3-CHOH-CsHU
CI-C5H4
CsH5
tBu-C5H4
Me3Si-C5H,
H>C=C(CH,)-CIHn
C,H,-C,H,
l,2-(Me,Si)Z-CsH3
CH3CO-CsH4 53
Ph,-CsH
C2HSCO-CSH4
CH30CO-CsH4
OHC-C5Ha
ChHsCO-C5Ha
220
> 200
185
I80
162
182
170
147
152
I44
185
140
138
123
124
I24
125
129
1I 9
+ 73
+ 38
+ 33
+ 15
135
+ 23
> +53
0
+ 5
- 3
38
- 7
- 9
- 24
- 23
- 23
- 22
- 18
- 28
+
1.9
3.4
2.0
1.9
6.2
2.4
6.9
2.5
2.4
1.8
2.4
3.51
2. I 9
1.98
2.50
2.02
1.73
1.79
1.71
1.77
1.67
1.74
1.73
2.25
I .46
2.70
1.42
1.57
1.43
1.22
77.8
68.7
66.4
71.4
66.9
63.4
64.2
63. I
63.9
62.5
63.5
63.4
69.2
59.3
73.0
58.7
61.1
58.8
55.0
22.2
31.3
33.6
28.6
33.1
36.6
35.8
36.9
36. I
37.5
36.5
36.6
30.8
40.7
27.0
41.3
38.9
41.2
45.0
0.9
2.3
2.0
1.9
2.3
1.7
2.5
2.2
3.65
2.56
1.94
2.08
I .48
3.42
1.30
2.3 1
78.5
71.9
66.0
67.5
59.7
77.4
56.5
69.8
21.5
28.1
34.0
32.5
40.3
22.6
43.5
30.2
I .7
I .6
1.5
1.9
1.6
I .4
1.8
2.3
[bl
[bl
- 1261
- 1227
- 1212
- 1199
- I176
- II66
~
I I49
- 1127
- 1088
[bl
- 1055
M
- 1051
- 1047
- 1033
- 1001
-
85
51
- 36
- 23
0
10
27
49
88
+
+
+
+
+ I21
+ 125
+ 129
+ 143
+ 175
Y= qs-/ndenyl derivatives
2-(CH3)*N-C9H6
I,~-(CH~)GHS
I-CH,-C9H,
I-CZH5-C9Hh
C9H7 26
l-tBu-CpHo
I-ChHS-C9Hh
I-Me3Si-C9Hh
175
165
I46
144
126
125
123
110
+ 49
+ 39
+ 20
+I8
0
- I
- 3
- 16
W
- 1018
- 931
- 930
- 847
~
826
Ibl
- 688
- 171
- 84
- 83
+
0
21
+ 159
Y=q'-Fluorenyl
Ci,H, 29
170 [a]
0.9
2.46
71.1
28.9
- 384
Y = 7f-Borininato derivative?
l-C2H5-C7H5B 31
l-C6H5-C5H5B 30
I24
I20
2.6
2.5
3.28
2.52
76.6
71.6
23.4
28.4
- 1018
0.5
2.32
69.9
30.1
- 270
~
975
Y = q-'-CyclooctenJ'l
[a] 15% propyne conversion. [b] No '"Co-NMR spectrum obtained.
356
Anyen'. Chem. Inr. Ed Eiigl. 24 I l Y X Y ' 4 h
'I,'
The influence of Y on the charge distribution at the cobalt
is eaaily determined from the 5 ’ C ~ - N M R chemical
shifts‘’’ 5 y i (Fig. 6). If one relates the 59Co-NMR-shiftsin
R
selectivity so that mainly benzene derivatives are
formed.“31With q‘-borininato groups as the controlling ligand Y attached to cobalt highly reactive catalysts with excellent chemo- and regioselectivity are obtained. A comparative study of [YCo(cod)] systems in a continuous flow
system therefore allows a differentiation of molecular catalyst properties.
4.2. Correlation of the Control with 59Co-NMR Data”4.241
59Co-NMR
big. 6 . Determination of ligand effect by ”Co-NMR spectroscopy
the series of q’-cyclopentadienyl and q5-indenyl complexes with the values of the unsubstituted parent compounds (6,,,(59C0)= 0), then a correlation between the relative chemical shifts and the electronic character of the
ring substituent is found: alkyl groups which act as electron donors cause a larger deshielding of the metal center,
whereas phenyl or acetyl groups decrease the deshielding.
The influence of the substituents is clearly additive, so that
6,,,(“Co) for Y = M e s C s is five times higher than that
found for Y=MeC5H4. In the case of Y=cyclopentadienyl or indenyl one can refer to them as “catalyst families”
in which the electronic character of the parent systems can
be widely varied by introducing further ring substituents
(fine-control). This fine-control of the catalyst properties
by additional substituents is mainly due to electronic effects. Thus, for example, in the q’-C5Hs series an additional acetyl group on the ring induces an increase in the
catalyst activity in comparison to a 1-hydroxyethyl group,
which has nearly the same steric requirements. In the (q5cyclopentadieny1)Co-series, the pentamethyl system shows
the lowest activity in the test reaction (25), whereas the
highest activity at 65% propyne conversion is found for the
benzoyl-substituted system (factor 1000 times more reactive than the pentamethyl system). Mesomeric substituent
effects influence the activity more than inductive effects,
so that replacement of a methyl- by a chloro-substituent
only has minor influence on the activity. The regioselectivity is in general inversely proportional to the catalyst activity. Exceptions are found when Y = 1,2-(Me3Si),C5H3and
Ph4CsH. In these cases, both high regioselectivity and activity are found. This is probably due to the accumulation
of sterically demanding substituents. Changing from (q5cyc1opentadienyl)Co to (q5-indenyl)Co catalysts lowers the
reaction temperature required for the same propyne conversion by 20°C. This indicates that the latter complexes
are four times more reactive. Additive substituent effects in
the ‘“CO-NMR spectra are also found in these “catalyst
families”, as can be seen by comparing the 6,,,-values for
the mono- and dimethyl substituted derivatives. The finecontrol of the activity and selectivity by additional indenyl
ring substituents is as found in the q5-cyclopentadienyl series. The results obtained with Y=fluorenyl and q3-cyclooctenyl in the test reaction (25) listed in Table 6 cannot
be directly compared with the other results as only 15%
propyne conversion was achieved and an increase in reaction temperature resulted in catalyst decomposition. The
substitution of an q5-cyclopentadienyl by an q’-cyclooctenyl group attached to cobalt causes a change in the chemoA n q e n . C h e m lnr. Ed. Engl. 24 (19851 248-262
If one compares the change in the catalyst properties
caused by ring substituents in the cyclopentadienyl- and
indenylcobalt series with the relative chemical shifts
6re1(59C~)
given in Table 6, then parallel trends become
evident, especially between the activity and the signal position in the ’9Co-NMR spectrum. Furthermore, the increase
in activity found on going from Y = cyclopentadienyl to
Y = indenyl is reflected in a downfield shift. Y =q‘-borininato lies between the two “families” as far as activity and
”CO-NMR shift are concerned. A relationship between the
regioselectivity and the relative chemical shift 6,el(’9C0) is
also found: the more negative the 6,,l(s’Co) value the
greater the control of ligand Y in favor of the symmetrically substituted pyridine derivatives 46. If one plots the
arc,(‘9C0) values against the reaction temperature T which
is needed for 65% propyne conversion in the continuous
flow system for the [(C,H,)Co(cod)] family, then an almost
linear correlation is found for the 15 members of this “catalyst family” (Fig. 7).
-220 -180 -1LO -100 -60
-20
%,I
20
60
100
IhO
180 220
A
Fig. 7. Catalytic activity o f [ Y C o(cod))~complexca( Y =cyciopentadienyl derivative) as function of the ”Co-NMR shift (reaction temperatures T i n the
test reaction (25) under standard conditions; 0-0see Table 6).
Substituted [YCo(cod)]-systems (Y = cyclopentadienyl
resonances
~
which are shifted
derivative) which have 5 9 c
to higher field (6,,,(’9C0) is negative) are less active than
[(C,H,)Co(cod)], whereas derivatives having downfield
shifts (S,,,(59C0) is positive) show enhanced activity. The
257
linear relationship between the reaction temperature T and
6,,1(59C~)
in this catalyst family can also be expressed by
the regressional equation (27).
T ["C]= 131.6 - 0.17 .6,,1(59C~)
significance level 95%, Fisher test:
correlation coefficient r =0.98
T [ T I = 160.6-0.25-6,,,(19C0)
significance level 95%, Fisher test:
correlation coefficient r = 0.9 1
[YCo(cod)]-complexes (Y = cyclopentadienyl derivative)
with large positive 6 r e , ( 5 9 C ~values
)
are therefore particularly active. A similar quasi-linear correlation between regioselectivity and 59Co-NMR shift is found in this series.
The pentamethyl derivative preferentially forms the symmetrically substituted isomer 46, whereas the benzoyl-substituted catalyst produces most of the asymmetric isomer
47. Both catalysts are found at the extremities of the
6,,1(59C0) shift scale. The linear relationship found between regioselectivity and 59Co-NMR shift in this series
can also be described by regressional equations (28a) and
(28b).
46
-
20
60
100 160 180
@-a,
47 [%]=35.07+0.05~6,,,(59C0)
significance level 95%, Fisher test:
correlation coefficient r=0.95
This correlation is shown graphically in Figure 8.
/
**
t
u
t
1
20,
-220 -180 -140 -100 -60 -20
6rei
20
60
An increasing shielding of the cobalt-cores by alkyl substituents is found in the $-cyclopentadienyl or $-indeny1
systems. Increased electron density at the cobalt core
therefore causes a reduction in the catalytic activity. In
contrast, electron-withdrawing substituents lower the electron density at the cobalt core, resulting in deshielding and
higher catalytic activity.
The empirically found correlations [Eq. (27)-(29)] have
hitherto proven valid for the cyclopentadienylcobalt- and
indenylcobalt-core complexes. We consider the 59C0NMR, and transition metal-NMR spectroscopy in general
to be a very promising method for efficient catalyst screening because of the wide chemical shift scale and the sensitivity towards substituent effects.
4.3. Relationship between Crystal Structure and the
Catalytic Activity
LO
100
140
180 220
Fig. 8. Regioselectivity of [YCo(cod)j-complexes ( Y = cyclopentadienyl derivative) a s a function of the "Co-NMR shift (relative yield of 46 and 47 in
the test reaction ( 2 5 ) under standard conditions; 0-0,
see Table 6).
A direct relationship between catalytic properties and
s9Co-NMR shifts also appears to exist in other [YCo(cod)]
catalyst families. In the indenylcobalt series we have 59C0N M R data for the parent compound and five derivatives
thereof. As shown in Figure 9, a linear relationship between catalyst activity and 6,,1(59C0) is also found in this
case [cf. regression equation (29)].
258
6re1
Fig. 9. Catalytic activity of [YCo(cod)J-complexes(Y = indenyl derivative) as
a function of the "Co-NMR shift (reaction temperatures T i n the test reacsee Table 6).
tion ( 2 5 ) under standard conditions;
[Yo]= 64.93 - 0.05.6,,~('9C~)
*
-180 -140 -100 -60 -20
For purposes of comparison, several X-ray structure determinations were carried out in the q5-cyclopentadienyl
and $-indeny1 seriesL401.
Some results are shown in Figure
10.
The X-ray structure data
that the average
C-C-distance in the $-cyclopentadienyl rings and the average C=C-distance in the complexed C O D remain constant. The distance between the Co atom and the center of
the cyclopentadienyl ring or between the Co atom and the
center of the C=C bond in C O D d o not vary to any relevant extent. The compounds investigated can, however, be
classified into two groups according to the position of the
substituents on the q5-system relative to the arrangement
of the C = C bond of the cyclooctadiene (Table 7). In the
case of structure type I the substituent on the cyclopentadienyl ring crosses a double bond of the complexed COD,
i.e. relative to the cobalt atom it is therefore transoid to the
second double bond of the complexed COD. In the strucAngew. Chem. Int. Ed. Engl. 24 (198s) 248-262
,qF+
5. Application : Maximization of the Catalytic
Turnover Number TON
A
Y
I‘
w
5.1. 2-Ethylpyridine
,
co
‘CO
With the intention of achieving maximal catalytic turnover numbers (TON,,) for the industrially interesting synthesis of 2-ethylpyridine [Eq. (30)] we systematically investigated ca. 60 [YCoLI-complexes in a batch reactor.”’]
w
53 Ila
52 I
54
Ilb
Fig. 10. Structure types of [YCo(cod)] complexes, Y =cyclopentadienyl derivative (see Table 7).
ture type IIa the bond of the substituent on the cyclopentadienyl ring is approximately parallel to both double bonds
of the complexed COD. Structure type IIb also has a parallel orientation, but contains sterically very demanding
substituents. If one correlates the structures with the 59C0NMR findings, then it becomes evident that substituents
Table 7. Structure types of [YCo(cod)] complexes (cf. Fig. 10).
EP
The catalytic screening in the continuous flow system
(see Table 6) enables a characterization of the catalyst
properties. For preparative purposes, the alkyne concentration and temperature also have to be optimized for the
various catalysts. Since two acetylene molecules are coordinated to the cobalt in the rate determining ~ t e p , [ ’ ~the
-’~~
catalytic conversion has to be performed with a high stationary alkyne concentration.
The concentration of acetylene in organic solvents increases overproportionately with increasing
This is particularly the case when nitriles are used as solvents between 5 and 25 bar (Fig. 11).
14
Type IIa
O.7
Type Ilb
0.7CI-CsH4
Br-CSH4 52
(CH3)&-C5H,
CH3CO-C5H. 53
C9H726
(C~HS)~C-CSH~
Bicyclo[3.3.0]octadienyl 54
[a] [YCo(cod)] = [(cod)CoCsH,-CsH,Co(cod)].
0.6-
-t 0.50)
.c
L
.-
z
with strong donor ability (6,,,(59C0) is negative) belong to
structure type I. Complexes of this type have a lower activity in the catalysis than the unsubstituted ligand Y. Cyclopentadienyl or indenyl derivatives which have a positive
6re,(59C~)
value all belong to the structure group IIa. Compounds in this group have electron-withdrawing substituents and are more active than the parent compounds in
the catalysis. Sterically demanding donor substituents
force a parallel orientation of substituent and COD double
bond (structure type IIb). Apart from these exceptions
(type IIb), obviously induced by steric effects, correlations
are found between structure type I or IIa, sign of the relative 59Co-NMRshifts, and catalytic activity. The linear relationship between 59Co-NMRshift and catalytic activity
of the cyclopentadienyl- and indenylcobalt catalysts indicates that a relationship between electron distribution at
the cobalt and the activity of various complexes in the catalysis is to be expected. In order to shed light on this, high
resolution X-ray structure analyses are currently being carried out involving determination of deformation density by
the X-X- and X-N-methods[601 on the complexes 4, 53,
and 54.
Angew. Chern. Int. Ed. Engl. 24 11985) 248-262
~
p 0.4t
a ,
z1
.
c
2 0.3a
0
)
-
0.2-
0.1 -
1
5
-
10
15
prbarl
20
25
Fig. 11. Solubility of acetylene in nitriles and related compounds at 25°C as a
function of pressure. a: acetonitrile: b: isobutyronitrile: c: propiononitrile:
d : acrylonitrile; e: butyl cyanide: f methyl thiocyanate; g: benzonitrile; h:
decyl cyanide: i: undecyl cyanide; j: benzyl cyanide.
The series of optimization experiments was carried out
in a batch reactor‘621without inert gas using acetylene pressures of up to 60 bar. The safety precautions mentioned in
“Technical Rules Acetylene”L631
were adhered to. The maximal TON values for the 2-ethylpyridine synthesis [Eq.
259
(30)] for some of the [YCoI-catalysts are given in Table 8.
The values vary between 50 and 7000 product forming
steps per cobalt, i.e. by a factor of 140. As expected from
the catalyst screening involving variation of Y (Table 6),
maximal TON values are obtained with such [YCoI-systems having electron-withdrawing groups on the ligand Y .
Conversely, alkyl substituents lower the TON values drastically. In a reaction time of 2 h nitrile conversions between 2 and 42% are achieved. Catalysts which achieve
propyne conversions in excess of 20% can be regarded as
Table 8. Synthesis of 2-ethylpyridine (EP) on [YCo(cod)] and related complexes according to reaction (30) Best values in bold print, poorest values in
italics.
Catalyst
0.14 mmol [a]
T
Acety["C] [c]
lene
[gl [bl
(Ph,C,H)Co(cod)
(CH,OCO-C,H&
Co(cod)
(CH,CO-C,H,)Co(cod) 53
(CeHsCO-C,HI)Co(cod)
(CI-C5H4)Co(cod)
(C,H,NHCOC5H,)Co(cod)
(I-Me3Si-C9H,)Co(cod)
(C,H,)Co(C,H,) 50
(MelSi-C5H4)Co (cod)
(Br-GHJCo (cod) 52
(C5H,)Co(cod) 4
(C,
H ,)Co(cod) 26
[ 1 ,2-(Me,Si),C,H3]Co(cod)
(1 -CH,-C,H,)Co(cod)
(CHKHOHCsH4)Co(cod)
(rBu-C5H,)Co(cod)
(CIH,)CO(C,H,) 5
(I-C5H,)Co(cod)
(CH - C5H4)Co(cod) 51
(Me,Ge-C,H,)Co(cod)
(C~H~)CO(CO
7 )~
(I-C~HI-CSHSB)Co(cod) 30
(Bicyclo[3.3.O]octadienyl)Co(cod) 54
(Mesitylene)Co(CyH,,)
(1 -Ethoxy-3,4benzoborininato)Co(cod) 39
( Me2N -C,H,)Co(cod)
(MeEt,-Dihydrodiborolyl).
Co(tetrahydroanthracene) 35
(I-CI-C,H,)Co(cod)
(Ph2P-C.5H4)Co(cod)
(MeEt,-Dihydrodiboro1yl)CO(CO)? 33
(CsH5)Co(PPhs)2
(Fluorenyl).
Co(cod) 29
(CUH,,)Co(cod)
(Me5C,)Co(cod) 18
62
60
Nitrile EP
EP
TONp,
Yield Benzene [d]
conversion
110-165 25.7
100-160 41.6
91.8
94.5
7.9
6.2
6946
6060
75
90-160
36.5
93.4
4.9
5261
93
95-185 40.0
97.7
4.6
4612
62
65
115-160 30.5
100-180 30.5
91.9
88.7
6.4
4.4
4597
4140
86
70-142 27.8
91.8
3.3
3735
75
42
72
50
48
48
60-165
130-201
110-140
130-180
100-120
120-165
24.5
23.0
22.9
22.5
23.5
17.4
91.3
92.8
91.1
80.0
84.3
93.5
3.3
7.1
5. I
5.4
6.5
10.7
3532
3499
293 I
2926
2761
261 1
53
130-180 20.3
85.6
3.3
2323
52
115-200 23.2
89.8
6.7
2214
50
45
52
43
150-180
110-185
105-145
140-176
19.7
16.3
14.6
18.6
83.1
73.4
80.4
64.9
4.5
7.1
6.2
6.0
2065
2024
I765
1732
44
135-181 13.8
83.6
4.3
1653
41
52
125-196 13.3
to 122 8.8
75.1
91.9
5.3
5.7
1 646
1604
26
150-202 13.6
67.8
8.1
1527
54
105-180 12.7
82.2
3.5
1262
44
135-180
9.0
77.4
2.6
1122
54
59
115-161
to 140
9.7
6.4
75.5
58.9
2.2
1.8
872
668
41
39
26
110-168
130-194
to 180
5.6
6.0
7.4
60.2
69.8
66.3
4.1
2.4
10.2
505
493
460
35
52
180
152
6.5
3.6
62.0
72.4
2.3
to
1.4
417
344
56
29
to
to
130
3.4
185
1.9
to
26.7
15.4
-
3.1
2. I
5.2. 2-Vinylpyridine
The 2-vinylpyridine synthesis [Eq. (31)] must be carried
out below 130- 140"C, since acrylonitrile and 2-vinylpyridine undergo thermal
When using
[YCoI-systems which only catalyze this reaction above
130"C, the reaction must be carried out in dilute benzene
or toluene solutions so that the TON values d o not exceed
ca. 500.
I I3
50
-
[a] I n 150 mL propiononitrile. [b] Acetylene reservoir at 20 "C in 150 mL propiononitrile. [c] Reaction temperature range. [d] Based on 2 h reaction time.
260
preparatively useful. The 2-ethylpyridine yield (based on
nitrile conversion) varies between 15 and maximally 98%.
Best chemoselectivity is found for the 1,2-bis(trimethylsi1yl)cyclopentadienylcobalt catalyst, which gives ca. I I
moles of 2-ethylpyridine derivatives per mole of benzene.
In the series Y=Me1E-CSH4 ( E = C , Si, Ge) the TONmaximum is found when E = Si. In the series Y = halocyclopentadienyl the best TON was found for the chloro substituent. The q5-chlorocyclopentadienyl ligand proved to
be noticeably more effective in the batch reactor than
would be expected from its position in the "activity scale"
(Table 6). This result again demonstrates that an optimization of the reaction parameters is vital prior to practical
use of the catalysts.
Comparison of the TON values of the [YCoL] complexes having various neutral ligands, verifies that strongly
complexing ligands such as triphenylphosphane or C O
hinder the liberation of the propagating species. The maximal TON values for these complexes, which are stabilized
by more easily displaced ligands, are all of the same order
of magnitude.
The maximum amount of acetylene which can be first
introduced at room temperature, so that the maximum
safety limit of 60 bar is not exceeded at the reaction temperature, is of importance for the preparative success of
the cobalt-catalyzed 2-alkylpyridine synthesis [Eq. (30)].
Highly reactive catalysts quickly use u p the acetylene reservoir, so that the pressure stabilizes after reaching a maximum between 40 and 60 bar and rapidly sinks at the end of
the reaction. When using [YCoI-systems of low reactivity
only small amounts of acetylene in the reservoir are used,
so as not to exceed the safety limit of 60 bar. The temperature control is also of importance in the 2-alkylpyridine
synthesis. A temperature control is used to slowly increase
the temperature until the catalysis sets in and the strongly
exothermic reaction is then moderated by water cooling.
The thermal stabilities of the individual [YCo]-systems
vary widely; the trimethylsilylcyclopentadienylcobaltsystem remains equally active at 200°C, whereas the indenylcobalt system already decomposes at 140°C under the catalysis conditions. In order to make full preparative use of
the activity, especially for the highly reactive [YCo]-systems, the amount of acetylene and the temperature control
have to be carefully optimized. Minor deviations from the
optimal combination of the parameters can lead to TON
value fluctuations of 1000 for one and the same catalyst.
VP
Angew. Chem. hi. E d . Engl. 24 (IY85) 248-262
Only very active catalysts can be used for achieving the
conversion (3 1) in pure acrylonitrile. We investigated all
the catalysts that were active below 125°C according to the
"activity scale" (Table 6) in a batch reactor for achieving
reaction (31). A solution of the catalyst in pure acrylonitrile was saturated with acetylene at ca. 20 bar and then
heated to 130°C. The TON values found in this way after
two hours' reaction time are summarized in Table 9.
Table 9. Synthesis of 2-vinylpyridine (VP) o n [YCo(cod)] and related complexes according to Eq. (31) i n the temperature range u p to 130°C.
Catalyst
0.14 mmol [a]
VP
Acety- NitrileVP
TONp,
lene
conversion Yield Benzene [cl
[gl [bl IW
[%I
30 58
57
(Ph,C, H)Co(cod)
58
(CH30CO-CsH4)Co(cod)
64
(C,H,CO-C,H,)Co(cod)
(CH,CO-C,H,)Co(cod) 53 58
67
(C,H,CO-C5H,)Co(cod)
75
(I-Me,Si-CpH,)Co(cod)
63
(C,H,)Co(cod) 26
68
(MeEt,-Dihydrodiborol y l ) c o ( c o ) l 33
46
(I-Ethoxj 3,4-benzoborimndto)Co(cod) 39
12.9
11.9
7.1
8.9
5.4
61.6
73.7
69.2
80.4
81.1
78.9
68.2
38.6
53.4
6.6
8.5
5.4
6.1
5.8
6.0
5.1
6.0
4.3
2164
1625
1513
1421
1345
1286
648
568
346
3.0
31.8
7.4
265
14.2
15.5
14.9
( l-C,Hs-C,H5B)Co(cod)
15.1
[a] In 150 mL acrylonitrile. [b] Acetylene reservoir at 20 "C in 150 m L acrylonitrile. [c] Based on 2 h reaction time.
The best results to date were obtained with the q6-phenylborininato complex 30. The outstanding position of
this catalyst is apparently due to the fact that the catalytic
vinylation reactions (32) and (33) are largely suppressed by
it. All other catalyst systems named in Table 9 cause the
reactions of acrylonitrile and 2-vinylpyridine to give appreciable amounts of 55 and 56 or 57. These activated
olefins can compete with acetylene for cobalt coordination
sites and they therefore act as catalyst poisons.
H2C=CH C N + 2 HC-CH
--f
(32)
V
C
55
+ HC-CH
N
+
-L
W
C
N
56
(33)
51
Since indenylcobalt-core complexes are generally more
reactive than the cyclopentadienylcobalt systems in the
pyridine synthesis, it is to be expected that the introduction of boron into the indenyl system might have a similar
positive effect on the catalytic activity in the 2-vinylpyridine synthesis, as was previously found on going from the
cyclopentadienyl- to the q5-borininato ligand."31 The use
of the complex 39 with Y = 3,4-benzoborininat0~~~~,
however, only gave relatively low TON values in the 2-vinylpyridine synthesis [Eq. (31)]. Similar results were also found
with the 1,3-dihydroborolyl complex 33. It is possible that
the introduction of the boron is compensated by an activity-reducing secondary effect of the alkyl substituents or
ethoxy group. This question will first be answered o n
Angew Cheni Int Ed Engl 24 11985) 248-262
carrying out the vinylpyridine synthesis [Eq. (3 I)] using
boron-containing control ligands Y having activating substituents.
The author and his co-workers wish to thank the director
of the Max-Planck-Institut fur Kohlenforschung, Professor
Dr. G . Wilke, for his continued support of this research. We
are grateful to the analytical departments of the institute f o r
their expert help: Dr. R . Benn, Dr. R . Mynott (NMR-spectroscopy); Dr. D . Henneberg (mass-spectroscopy); Pro$ Dr.
C . Kriiger, Dip1.-Chem. K . Angermund (X-ray structure
analyses); Priv.-Doz. Dr. G . Schomburg (chromatography)
and Dr. K . Seevogel (IR-spectroscopy).
Received: November 30, 1984 [A 527 IE]
German version: Angew. Chem. Y7(1985) 264
[ I ] A. Budzinski, Chem. Ind. (Diisseldof) 33 (1981) 529.
[2] a) W. Ramsay, Philos. Mag. [ 5 ] 2 (1876) 269; 4 (1877) 24; b) Ber. Dtsch.
Chem. Ges. 18 (1885) 431; c) R. Meyer. A. Tanzen, h i d . 46 (1913)
3186.
[3] H. Yamazaki, Y. Wakatsuki, Tetrahedron Lelr. 1973. 3383.
[4] a) H. Bonnemann, R. Brinkmann, H. Schenkluhn, S-ynthesis 1974. 575;
b) H. Bonnemann, H. Schenkluhn, DBP 2416295 (April 4, 1974). USPat. 4006 149.
[5] Y. Wakatsuki, H. Yamazaki, Synthesis 1976, 26.
[6] H . Yamazaki, Y. Wakatsuki, Jap. 0s 7725780 (1975): Chem. Ahstr. 87
(1977) 68 168.
(71 P. Hardt, a ) DOS 2615309 (April 8, 1976). Swiss Pdt.-Appl. 12 139-75
(September 18, 1975); b) DOS 2742541 (April 20, 1978). Swiss Pat:
Appl. 13079-76 (October 15, 1976): c) US-Pat. 4196387 (April I. 1980).
Lonza AG; d) DOS 2742542 (February 15, 1979). Swiss Pat.-Appl. 947177 (August 2, 1977).
[S] a) A. Naiman, K. P. C. Vollhardt, Angew. Chem. 89 (1977) 758: Angew.
Chem. Ini. Ed. Engl. 16 (1977) 708: b) D. J. Brien, A. Naiman, K. P. C.
Vollhardt, J . Chem. Soc. Chem. Commun. 1982, 133.
[9] K. P. C. Vollhardt, Angew. Chem. 96 (1984) 525; Angew. Chem. Int. Ed.
Engl. 23 (1984) 539.
[lo] R. E. Geiger, M. Lalonde, H. Stoller, K. Schleich, Helu. Chim. Acta 67
(1984) 1274.
[ I l l H. Bonnemann, Angew. Chem. 90 (1978) 517: Angew. Chem. I n t . Ed.
Engl. 17 (1978) 505.
[I21 H. Bonnemann. W. Brijoux in R. Ugo: Aspects of Humogeneous Catalysis. Vol. 5. D. Reidel, Dordrecht 1984, p. 75-196.
[I31 H. Bonnemann, W. Brijoux, R. Brinkmann, W. Meurers, Helr:. Chrm.
Acta 67 (1984) 1616.
[I41 H. Bonnemann, W. Brijoux, R. Brinkmann, W. Meurers, R. Mynott, W.
von Philipsborn, T. Egolf, J . Organornet. Chem. 272 (1984) 231.
(151 Gmelins Handhuch der anorganischen Chemie, Vol. 14. Part A , Ferrocene
1-7. Springer, Berlin 1974-1980.
116) a) J. Kozikowski, US-Pat. 2916503 (December 8, 1959): Chem. Absrr. 54
(1960) 5693; b) W. P. Hart in [23d], p. 79-83.
[I71 M. D. Rausch, R. A. Genetti, J . Org. Chem. 35 (1970) 3888.
[IS] N. E . Schore, J . Organomet. Chem. 173 (1979) 301.
1191 M. Radermacher, Dissertation, Technische Hochschule Aachen 1985.
[20] R. B. King, M. B. Bisnette, J . Organornet. Chem. 29 (1973) 227.
[21] R. B. King, A. Efraty, J. Am. Chem. SOC.93 (1971) 4950.
122) H. Bonnemann, W. Brijoux in 1121, p. 152f.
a) M. D. Rausch, W. P. Hart, D. W . Macomber, J . Am. Chem. Soc. I02
(1980) 1196; b) J . Mucromol. Sci. Chem. 16 (1981) 243: c) D. W. Macomber, W. P. Hart, M. D. Rausch, Adu. Organomet. Chem. 21 (1982) I ; d)
W. P. Hart, Ph.D. Dissertation, University of Massachusetts, Amherst
1981; e) D. W. Macomber, Ph.D. Dissertation, University of Massachusetts, Amherst 1982; fi 8 . G. Conway, MS Thesis, University of Massachusetts, Amherst 1981: g) M. D. Rausch, private communication.
W. Meurers, Dissertation, Technische Hochschule Aachen, in preparation.
W. Meurers, Diplomarbeit, Technische Hochschule Aachen 1982.
K. P. C. Vollhardt, T. W. Weidman, Organometallics 3 (1984) 82.
J . Altman, G. Wilkinson, J . Chem. Soc. 1964. 5654.
H. Bonnemann, W. Brijoux in [12], p. 149f.
a) H. Bonnemann, B. BogdanoviC, DOS 3205501 (February 17, 1982);
Eur. Pat.-Appl. 83 101 246.3 (February 10, 1983); US-Pat. 4469638 (September 4, 1984), Studiengesellschaft Kohle; b) H. Bonnemann, B. Bogdanovib, R. Brinkmann, D. W. He, B. Spliethoff, Angew. Chem. 95
(1983) 749: Angew. Chem. lnt. Ed. Engl. 22 (1983) 728.
H. Bonnemann, W. Brijoux in [12], p. 150-154.
H. Bonnemann, R. Brinkmann, unpublished.
a) D. Habermann, Dissertation, Universitat Bochum 1980, p. 58-70; b)
K. Jonas, C . Kriiger, Angew. Chem. 92 (1980) 513: Angew. Chem. Int. Ed.
26 I
Engl. 19 (1980) 520; c ) K. Jonas, Adu. Organomel. Chem. 19 (1981) 97;
d ) K. Jonas, E. Deffense, D. Habermann, Angew. Chem. 95 (1983) 729;
Angen,. Chem. I n t . Ed. Engl. 22 (1983) 716; Angew. Chem. Suppl. 1983.
1005.
[33] a) H. Kojima, S. Takahashi, H. Yamazaki, N. Hagihara, Bull. Chem. Soc.
Jpn. 43 (1970) 2272; b) H. Kojima, S. Takahashi, N. Hagihara, J. Chem.
Sot. Chem. Commun. 1973, 230.
(341 H. Bonnemann, M. Radermacher, C. Kriiger, H. J. Kraus, Helu. Chim.
Acla 66 (1983) 185.
(351 a) G. Wilkinson, J. Am. Chem. Suc. 74 (1952) 6148; b) E. 0. Fischer, R.
Jira, Z . Nuturforsch. 8 8 (1953) 1.
[36] H. Bonnemann, M. Samson, DBP 2840460 (September 16, 1978); USPat. 4266061 (May 5 , 198 I), Studiengesellschaft Kohle.
[37] H. Bonnemann, W. Brijoux in [12], p. 154-158.
[38] H. Bonnemann, M. Samson, C. Kriiger, L. K. Liu, unpublished.
[39] P. Diversi, A. Giusti, G. Ingrosso, A. Lucherini, J. Orgunornet. Chem.
205 (1981) 239.
[40] a) K. Angermund, Dissertation, Universitat Wuppertal, in preparation;
b) C. Kriiger, K. Angermund, unpublished results.
(41) G. E. Herberich, G. Greiss, Chem. Ber. I05 (1972) 3413.
(421 a) A. J. Ashe, P. Shu, J . A m . Chem. Soc. 93 (1971) 1804; b) E. Raabe,
Dissertation, Technische Hochschule Aachen 1984.
[431 a) J. Edwin, M. C. Bohm, N. Chester, D. M. Hoffman, R. Hoffmann. H.
Pritzkow, W. Siebert, K. Stumpf, H. Wadepohl, Organometallics 2 (1983)
1666; b) W. Siebert, Adu. Organomel. Chem. 18 (1980) 301.
(441 P. Binger, Angew. Chem. 80 (1968) 288; Angew. Chem. I n t . Ed. Engl. 7
(1968) 286.
(451 M. Bochmann, W. Siebert, Angew. Chem. 89 (1977) 483; Angew. Chem.
/nt. Ed. Engl. 16 (1977) 468.
1461 M. Bochmann, K. Geilich, W. Siebert, Chem. Ber. 118 (1985) 401.
(471 H. Bonnemann, R. Brinkmann, unpublished results (1984).
[48] R. Boese, N. Finke, J. Henkelmann, G. Maier, P. Paetzold. H. P. Reisenauer, G. Schmid, Chern. Ber. 118 (1985), in press.
(491 H. Bonnemann, N. Finke, P. Paetzold, M. Radermacher, unpublished.
[SO] a) C. Grard, Dissertation, Universitat Eochum 1967: b) G. Wilke, Kagaku Kugyo 20 (1967) 1308, 1310: c) S . Otsuka, M. Rossi, J. Chem. Soc.
A 1968, 2630; d) S. Koda, A. Tanaka, T. Watanabe, J. Chem. Soc. Chem.
Commun 1969, 1293; e) H. Lehmkuhl, W. Leuchte, E. Jansaen, J . Organomet. Chem. 30 (1971) 407.
[51] K. Jonas, Angew. Chem. 97 (1985) 292; Angew. Chem. I n t . Ed. Engl. 24
(1985) 295.
[52] S. Wendel, Dissertation, Technische Hochschule Aachen, planned for
1986.
(531 H. Yamazaki, Y . Wakatsuki, J . Organornet. Chem. 139 (1977) 157.
1541 W. Brijoux, Dissertation, Universitit Dortmund 1979.
(551 a ) H. Bonnemann, W. Brijoux, K. H. Simmrock. ErdGI Kohle 33 (1980)
476-479; b) H. Bonnemann, W. Brijoux in [12], p. 133-140.
(561 H. Bonnemann, W. Brijoux, W. Meurers, unpublished.
1571 H. Lehmkuhl, H. Nehl, Chem. Ber. 117 (1984) 3443.
(581 M. Rosenblum, B. North, D. Wells, W. P. Giering, J . Am. Chem. Snc. 94
(1972) 1239.
[59] T. Egolf, Dissertation, Universitat Zurich 1983.
[60] R. Goddard, C. Kriiger in P. Coppens, M. B. Hall: Electron Distribution
and the Chemical Bond, Plenum Press, New York 1982.
[61] H. Bonnemann, W. Brijoux in [12], p. 125f.
(62) H. Bonnemann, W. Brijoux in 1121, p. 123f.
1631 a) TRAC (Technical Rules Acetjlenei. TRAC 203: Compressors; TRAC
204: Capillaries; TRAC 206: Acetylene bottles; TRAC 207: safety devices and safety valves; Carl Heymanns Verlag, Koln 1980; b) H. B. Sargent, Chem. Eng. (New York) 64 (1957) No. 2, p. 250; c) B. A. Ivanov, S.
M. Kogarko, Dokl. Akad. Nairk SSSR 142 (1962) 631.
[64] R. Brinkmann: Beirrage zur Entwicklung der Cobalf-katalysierten r?,ridinSynrhese: MPI fur Kohlenforschung, Mulheim a. d. Ruhr 1982.
Catalytic Synthesis of Organolithium and Organomagnesium
Compounds and of Lithium and Magnesium HydridesApplications in Organic Synthesis and Hydrogen Storage**
By Borislav Bogdanovic*
Dedicated to Professor Giinther Wilke on the occasion of his 60th birthday
A recent development in homogeneous catalysis is the discovery of catalysts that are active
for the lithiation of I-alkenes to alkenyllithium compounds and lithium hydride as well as
for the hydrogenation of lithium and magnesium under mild conditions. The catalytically
prepared magnesium hydride is highly reactive and adds to 1-alkenes to give diorganomagnesium compounds and can also be used in the preparation of, for example, silane and “active” magnesium. The use of metal hydrides in hydrogen storage is discussed: hydrogenatioddehydrogenation experiments show that the catalytically prepared magnesium hydride
(which can be doped with a second metal) can be used as a high-temperature hydrogen storage material.
l. Introduction
For decades organolithium and organomagnesium compounds (mainly in the form of Grignard reagents) have
[*I Prof. Dr. B. Bogdanovic
~~
[**I
Max-Planck-Institut fur Kohlenforschung
Kaiser-Wilhelm-Platz I , D-4330 Miilheim a. d. Ruhr I (FRG)
Based on work carried out by Ekkehard Bartmann. Borislav Bogdanovid.
Alexis Cord;. Gudrun Koaoetsch. Oleu Kuzmin. Shih-rsien Lioo. Puolo
Locutelk Meenakshr Maruthamuthv, Klaus Schlichte, Munfred Sch wickardi. Peter Sikorsky. Bernd Spliethoff, Halszka Stepowska. Bernd Wermeckes, Uwe Westeppe, and Ursula Wilczok. The author thanks these coworkers for their dedicated and enthusiastic assistance.
1 1
262
0 VCH Verlagsges~llschaffmbH, 0-6940 Weinheim, 1985
been among the most frequently used organometallic reagents for organic synthesis.‘l1 They are usually prepared by
with organic
reaction of metallic ]ithiurn or magnesium
ha[ides,f’-3~
In the following article it is shown that dialkylmagnesium compounds can be prepared catalytically in a manner
analogous to the synthesis of trialkylaluminum compounds.’41 An important role in the catalytic activation of
lithium is probably played by ‘‘poly-lithium complexes”
which are formed from lithium and 1.6.6ah4-trithia~enta1(Section
the
hydrogenation Of magnesium involves the activation of the metal through the reI
,
2)1
0570-0833/85/0404-0262 $ 02.50/0
Angew. Chem. I n t . Ed. Engl. 24 11985j 262-273
Документ
Категория
Без категории
Просмотров
1
Размер файла
1 280 Кб
Теги
structure, synthesis, example, compounds, homogeneous, effectivity, organocobalt, pyridinesцan, relationships, catalsis
1/--страниц
Пожаловаться на содержимое документа