THE INVESTIGATION OF FACTORS
INFLUENCING THE STEREOCHEMISTRY
OF THE WITTIG REACTION
JEROME THOMAS KRESSE
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The author wishes to express his appreciation to his research director,
Dr. George B. Butler, for his guidance and encouragement during the execution
of this work.
The author also expresses his gratitude to his fellow graduate students and
associates for their helpful suggestions and criticisms.
Particular thanks are due Mrs. Frances Kost and Mrs. Thyra Johnston
for their conscientious typing of this dissertation.
The author also thanks his wife for her patience, encouragement and under-
standing. Without her cooperation this work would not have been possible.
The financial support of the Petroleum Research Fund is also gratefully
TABLE OF CONTENTS
ACKNOWLEDGMENTS .......................... ii
LIST OF TABLES ............................. v
LIST OF FIGURES ................... ........ vi
I INTRODUCTION ....................... 1
Historical Background .................. 1
Stereochemistry ...................... 9
Statement of the Problem ................ 19
Method of Attack ..................... 20
I DISCUSSION AND RESULTS . . . . . . .... 22
Solvents . . . . . . . . ... ..... .22
Temperature Effects ................... 27
Reaction Times . . . . . . . .... ... .. 33
Concentration Effects . . . . . . . .... .34
Substituent Effects ......................... 37
Anion Effects ....................... 49
1, 4-Addition and Isomerization . . . . ... .51
TABLE OF CONTENTS
III EXPERIMENTAL ...................... 54
Equipment and Data ...... ........... 54
Source and Purification of Materials . . . . . 54
Solvents . . . . . . . . . . . . . 55
Aldehydes . . . .. .. . . . . . . .. . 57
Miscellaneous Chemicals . . . . . . .... .57
Preparation and Purification of Phosphonium Salts . 59
Apparatus . .. . . . . . . . .. . . 161
Reactants . . . . . . . . . . . . 61
The Reaction .................... ... 65
Analysis . . . . . . . . . . . . 67
Calibration of the Internal Standard . . . .... .67
Qualitative Analysis ............. ... 69
Precision of Chromatographic Analysis . . . ... 72
IV SUMMARY ........ .. ............ . 75
LIST OF REFERENCES .......... .......... .... 76
BICGRAPHICAL SKETCH ................. ** ....... 79
LIST OF TABLES
1 Solvent and Halide Ion Effects 11
2 Combination Effects on Stereochemistry 13
3 Solvent Effects 23
4 Temperature Effects 28
5 Reaction Times 33
6 Concentration Effects 35
7 Substituent and Anion Effects 38
8 Ring Opening Reactions of 4-Octene Oxide and Stilbene Oxide 47
9 Anion Effects 50
10 Reactions of Methyltriphenylphosphoranes with Crotonaldehyde 52
11 Physical Constants of Phosphonium Salts 62
12 Infrared Absorption Phosphonium Salts 73
LIST OF FIGURES
1 Energy Profile for the Reaction of a Stable Ylide 8
2 Postulated Reaction Paths of the Wittig Reaction 17
3 Possible Betaine Configurations 41
4 Reaction Manifold 63
5 Calibration Curve for trans-1, 3-pentadiene 70
6 Chromatogram of Reaction Products 71
Since Wittig and Geisslerl first reported the olcfin synthesis which has be-
come known as the Wittig synthesis literally hundreds of papers have been published
concerning. the synthetic and mechanistic aspects of the reaction.
This synthesis in its general form involves the reaction of a phosphorane
derived from a phosphonium salt with an aldehyde or ketone. The phosphonium
salts are usually prepared from triphenylphosphine and an alkyl or arylalkyl halide.
The phosphoralce is produced from the phosphonium salt by removal of an a-hydro-
gen by a suit-ble base. Addition of a carbonyl containing compound to a solution of
the phosphorane forms an intermediate betaine which then collapses to produce the
oiefin and uriphenylhphshine oxide.
0 P: + R R2CHX->[ P-CHR1R2]+ X-
I + B: -> BH + 3P-C-R1R
II + R3R4CO-> R R C-CR R
4 1 1P+ 3 4
I;->RtR,C= C3R + 03 P->O
Although the reaction appears straightforward synthetically, mechanistically
the reaction is exceedingly complex. The structure of the phosphorane is apparent-
ly a resonance hybrid of canonical forms IV and V. Form IV is also referred to as
an ylene and form V as an ylide. Form IV requires the assumption of d w7 bonding,
03P=CRIRa2 P CR1R2
between carbon and phosphorus. That such-bonding exists for phosphorus and.
o0therelements of-the third period-has been demonstrated by base catalyzed
a deuterium exchange studies. Form V is that of a carbanion, the extremely strong
conjugate base of the phosphonium salt.
If we assume that structures IV and V approximate the limiting forms of a
phosphorane we can see that the degree to which this structure resembles either IV
or V will be dependent on the nature of the substitutents on phosphorus and carbon.
Jaffe3 has calculated that one of the criteria for d v bonding is the existence of a
positive charge on phosphorus in its singly bonded structure. It follows that groups
on phosphorus which would be electron donating, e.g. alkyl groups would lessen the
importance of the contribution of form IV to the hybrid. Likewise the incorporation
of electron withdrawing groups on carbon, e.g. carbethoxy would tend to make form
IV more important. Regarding the reactivity of the phosphorane we find that groups
increasing the contribution of form IV to the hybrid produce a less reactive species.
This can be seen in comparing the properties of triphenylfluorenylidenephosphorane4
and tributylethylidenephosphorane. The former contains groups capable of delocal-
izing the charge on the carbon and intensifying the positive charge on the phosphorus.
It is extremely unreactive. It can only be hydrolyzed by refluxing with strong
base. The latter compound contains groups which tend to diminish the positive
charge on phosphorus and increase the carbanion character of the structure. It is
hydrolyzed rapidly on exposure to the atmosphere. Both compounds will react with
carbonyl compounds but the conditions for carrying out the reaction are much more
vigorous for the fluorenylidenephosphorane.
Between these two extremes lie almost a continuum of compounds with vary-
ing reactivities. It is this range of reactivity which has made the study of the
Wittig reaction both interesting and challenging. Work done on one system cannot
be directly correlated with a more or less reactive system. This is especially
true of the stereochemistry of the reaction.
The mechanism of the reaction as postulated by Wittig involves an attack by
form V of the phosphorane on the electrophilic carbon of the carbonyl to give a
P+ O P+ O-
I II I I
C- + C C C
R1 R2 3 4 R1 R2R3 4
The betaine then may collapse via a four membered cyclic transition state to
olefin and triphenylphosphine oxide. In attempting to elucidate the mechanism
Wittig6 studied the reaction between methylenetriphenylphosphorane and benzalde-
'P O- O3P- >
I I +
C C --> \ C''
1 2 R R R1 3 R4
hyde. During the reaction Wittig was able to trap the betaine by the addition of
hydrogen iodide giving 2-phenyl-2-hydroxyethyltriphenylphosphonium iodide which
aP= CH2 + QCHO -> 9CH- CH2
on heating gave styrene and 93P->O. He also refluxed benzophenone with the
betaine hydroiodide, but was unable to obtain 1, 1-diphenylethylene, the product one
might expect if betaine formation were reversible. Wittig therefore concluded that
betaine formation is irreversible.
However, Filszar, Hudson and Salvadori found that the lithium bromide--
ether complex of the betaine isolated by Wittig yielded no olefin products on heating
but rather benzaldehyde. These workers concluded that betaine formation is rever-
sible and that decomposition of betaine to olefin and phosphine oxide is rate deter-
(CH=CH + 0P->O
7- B: 2 3
I+ C O+ BH I
LO P~C 9C=CH2
Speziale and Bissing have demonstrated the reversibility of betaine forma-
tion by reacting triphenyl and tributylphosphine with cis and trans-ethyl phenyl-
glycidate in the presence of meta-chlorobenzaldehyde.
R3P + 1CH-CHCO C H5---> R P+-CHCO2C2H
225 3 22 5
X -> R3 P->0 + qCCH = CHCOCH5
X -> R3P = CHCO C H5 + 1CHO
XII + m-C1C H CHO-> R3P->O + m-CIC H4CH=CHCO C 2H
They found using a 1:1:1 molar ratio of triphcnylphosphine: trans-ethyl
phenylglycidate: meta-chlorobenzaldehyde that they obtained 48.7 per cent cis
ethyl cinnamate; 17.4 per cent trans ethyl cinnamate; 3.0 per cent cis meta-chloro-
cinnamate and 30. 9 per cent trans meta-chlorocinnamate.
These workers further showed through a kinetic study of the reaction between
a series of substituted benzaldehydes with carbomethoxymethylenetriphenylphos-
phorane (a stable ylido) that the rate of reaction is second order, first order in
ylide and first order in aldehyde. They found that the reaction rate is increased
by increases in solvent polarity, substitution of butyl for phenyl on the phosphorus
of the ylide and by electron withdrawing substitutents on the aldehyde.
The mechanism they postulated based on kinetics and the epoxide ring open-
ing reactions mentioned previously is slow, reversible betaine formation followed
by rapid decomposition of the betaine to olefin and phosphine oxide.
House and Rasmusson had previously suggested a mechanism for the reaction
of acetaldehyde with carbomethoxymethylenetriphenylphosphorane to give methyl
tiglate and methyl angelate in which there is a rapid reversible formation of betaines
with slow preferential decompositions of the stereoisomeric betaines to products.
Concerning reactive ylides, Wittig, Weizmann and Schlosser6 had observed
that though betaine formation with reactive ylides is rapid, taking place within a
few minutes, the decomposition of betaine requires standing at room temperature or
heating for several hours. They concluded that betaine decomposition is the slow
step in this type of reaction.
Bergelson and Shemyakin0 interpreted the stereochemistry observed in the
reaction of stable ylides (giving mostly trans olefin) as opposed to unstable or
reactive ylides (giving mostly cis olefin) by using two different reaction paths.
For reactions involving stable ylides they believed that because of the decreased
electron density about the carbon attached to the phosphorus of the ylide, attack
would take place by the carbonyl oxygen on the phosphorus rather than by the car-
banion of the ylide on the carbonyl carbon.
R' H R' H
P P -0 C -- trans olefin
I I C H R'
R HR H
This would be followed by rotation either about the phosphorus-oxygen or carbon-
oxygen bonds to give the more stable (trans) betaine which would then decompose to
olefin and phosphine-oxide. This mechanism involves irreversible betaine formation
in direct conflict with the results of Speziale and Bissing.
Reactions of reactive ylides would take place by the accepted path involving
attack of the carbanion on the carbonyl carbon.
Trippett11 recently discussed the work of Speziale and Bissing and his own
results in contrasting the kinetics and mechanism involving stable and reactive phos-
phoranes. Since stable phosphoranes produce intermediate betaines which cannot
be isolated Trippett proposes a potential energy diagram in which the "valley"
representing the betaine is quite shallow. He believes that the transition states T1
leading to the betaines from reactants and T2 leading from the betaines to products
resemble the structures of the betaines more than they resemble either the pro-
ducts or reactants.
Using these assumptions and a relative rate equation based on them he con-
cludes that the most important factor in determining the ratio of isomers in the
reaction of stable ylides is the relative rate of betaine formation with smaller
Figure 1. Energy Profile for the Reaction of a Stable Ylide
contributions from the relative rate of betaine dissociation and decomposition all
acting in the same direction.
For the case of very reactive phosphoranes, e.g., CH2=P-<3 he maintained
that betaine formation is still reversible but now the difference lies in the isomeric
betaines. The one giving cis olefin dissociates faster than it decomposes to
products and thie one giving trans olefin decomposes to products faster than it
Although wve have already touched on the steroechemistry of the Wittig re-
action in, our discussion of the mechanism considerably more work has been done
o: this phac.e of the synthesis. If one starts with an aldehyde or an unsymmetrical
ketone and an unsymmetrical phosphorane the olefin resulting can exist in either of
two geometrical forms. The olefins receiving most attention
+ RI /R3
q3P-CR R2 + R3R4C=O0--P3P--O + C = C
have been those in which either R1 or R2 is H and R3 or R is H. Wittig and
Schollkoop in some early work observed that the reaction between benzaldehyde
and benzylidenetriphenylphosphorane give cis and trans stilbene in a 30:70 ratio
respectively, whereas the reaction of benzaldehyde with allylidenetriphenylphos-
phorane give cis and trans 1-phenylbutadiene in a 45:55 ratio respectively.
Ketcham and coworkers conducted a reaction between p-nitrophenylmethyl-
enetriphenylphosphorane and anisaldehyde and p-methoxyphenylmethylenetriphenyl-
phosphorane and p-nitrobenzaldehyde. The product mixture in the former reaction
consisteL of .ii trans olefin, while in the latter reaction the product contained 48
per cent cis and 52 per cent trans olefin. They reasoned that the less reactive
p-nitro ylide reacts reversible with the unreactive anisaldehyde while the reactive
p-metLhoxy yiide reacts irreversibly with the very reactive p-nitrobenzaldehyde
giving an almost stJtistically controlled product mixture.
Speziale and Rattsl3 also found that they obtained all trans olefin from p-nitro-
benzaldehyde and the stable ylide carbethoxychloromethylenetriphenylphosphorane.
Wailes14 however reported that the reaction of dodecylidenetriphenylphos-
phorane (a reactive ylide) and propynal give after treatment with ethylmagnesium
bromide and carbon dioxide the enynoic acid containing 80 per cent cis isomer.
Truscheit5 and coworkers using butylidenephosphorane and 12-acetoxy-
dodec-2-enal obtained 70 per cent of the cis diolefin. Using ethylidenephosphorane
they obtained 67 per cent cis diolefin.
Kucherov et al. using the stable carbethoxymethylenetriphenylphosphorane
and 2,4, 6-octatriene-1, 8-diol obtained a 57 per cent yield of all trans olefin. They
also obtained all trans olefin using the same aldehyde and 3-carbethoxyallylidene-
There are in the literature many more references concerning the stereo-
chemistry of the Wittig reaction but like most of those above they have not been
either intensively or extensively studied. However, there appeared in the last three
years several stereochemical studies of the reactions which deserve attention.
10, 17, 18, 19
Shemyakin and Bergelson10,1 in a series of papers have published an
abundance of useful, though sometimes controversial information concerning the
chemistry of moderately unstable ylides. The system used in their early experi-
ments was the reaction of propionaldehyde with benzylidenetriphenylphosphorane to
give phosphine oxide and P-ethylstyrene.
Using a series of solvents they obtained the results shown in Table 1. Their
explanation of the isomer ratios involves a solvent-ylide complex which may be con-
sidered either a coordination compound or a solvate. Whatever its exact nature they
say that this complex produces a "mutual inaccessibility of the phosphorus and oxygen
SOLVENT AND HALIDE
* Base = CH CH20
% cis % trans
in the prereaction complex" and that "under such conditions the betaine" (leading to
the cis olefin) "forms more readily than its diastereoisomer. "10 The order of the
effects noted in the solvent they relate in the case of the oxygen containing com-
pounds to relative nucleophilicity while in the amine series they feel steric factors
are of particular importance. Dimethylformamide, they believe, though weakly
basic, possibly interacts with the phosphorus of the ylide through its strongly
polarized oxygen rather than its nitrogen.
These workers also found that the addition of lithium halides to the reaction
in benzene and DMF gives increased yields of the cis isomer as shown in Table 2.
The rationalization of these effects again involves a halide complex with the
positively charged phosphorus of the ylide. This diminishes its electrophilicity
thus favoring the formation of the betaine leading to the cis isomer "due to electro-
static repulsion between the halide and oxygen electronic shells. The selective
formation of cis olefins shows that this effect is considerable. "10
In their later papers Shemyakin and Bergelson have clarified their reasoning
on the interactions leading to the stereochemistry of the Wittig reaction in the
presence of Lewis bases. They propose coordination of the Lewis base with the
phosphorus which is facilitated by a transition of the phosphorus from a tetrahedral
to a trigonal bipyramidal configuration in which the three phenyl substituents be-
come coplanar. They predict that as a result of repulsion between the electronic
clouds of the phosphorus and oxygen the betaine leading to the trans olefin is "de-
stabilized by steric repulsion of the skewed R and R' substituents" thus leading to
predominant cis olefin. They point out, however, that though this might be the case
with some systems it does not prevail in all. Considering the case in which betaines
COMBINATION EFFECTS ON STEREOCHEMISTRY
Ylide:RCHO % Yield Cis:Trans
are formed faster than they decompose, they reason that now the steric course of
the reaction "depends on the relative energies of the stereoisomeric betaines not
only in the most stable conformation, but also in the eclipsed reacting conforma-
tion closely allied to the four-membered transition state. If the reacting betaine
conformations are sufficiently well differentiated energetically, the over-all
equilibrium will be shifted in the direction of the betaine that most readily decom-
poses into olefin and phosphinoxide. This can lead to stereoselectivity, even when
the diastereoisomeric betaines in the most stable conformation differ little in energy."
In the case of the reaction of benzylidenetriphenylphosphorane with propionaldehyde
in benzene which gives 80 per cent trans-P-ethylstyrene they slate that the selec-
tivity illustrated here is hard to explain on the basis of the small differences in
non-bonded interactions observed in the two betaines. They believe that a more
important factor in this case is the stabilization of the incipient double bond by the
phenyl group which would only be possible in the betaine leading to the trans olefin.
They9 also consider the possibility of steric control by changing the concen-
tration of reactants, thereby reducing the reversible dissociation of the betaines.
While equimolar amounts of ethylidenetriphenylphosphorane and benzaldehyde in
benzene in the presence of lithium iodide give 34 per cent cis-P-ethylstyrene the
doubling of either the aldehyde or ylide concentration practically doubles the amount
of cis isomer found.
In another group of experiments these investigators demonstrated that sub-
stantial amounts of trans olefin could be obtained from moderately unstable ylides
contrary to the usual experience. For example, using propionaldehyde and 3-ethyl-
allylidenetriphenylphosphorane they obtained the results shown in Table 2.
House and Rasmusson investigated the reaction between acetaldehyde and 1-
carbomethoxyethylidenetriphenylphophohrane and between ethylidenetriphenylphos-
phorane and ethyl pyruvate to give mixtures of methyl angelate and methyl tiglate.
They found that the reaction between the stable phosphorane gave 96. 5 per cent
trans ester whereas the unstable phosphorane gave 68 per cent trans ester. They
rationalized these results by postulating an equilibrium in the formation of betaines
and more rapid decomposition of the betaine leading to the trans ester because of
increased stabilization in the transition state. This increased stabilization could
arise because only in the trans betaines would the carbomethoxy group be able to
become planar with the incipient double bond. In the cis betaine coplanarity of the
carbomethoxy group would be prevented by interference with the adjacent methyl
group. The increased amount of trans isomer obtained with the ethylidenephos-
phorane they said was caused by the increased reactivity of this species which
opposed the formation of the equilibrium between betaine and reactants allowing
the stereochemical outcome to be determined by the relative ease of betaine forma-
House, Jones and Frank20 recently reported the results of a series of re-
actions involving both stable and unstable ylides in different solvents in the pre-
sence of added inorganic salts and with two aldehydes of differing reactivity. They
found that stereochemically the reaction of carbomethoxymethylenetriphenylphos-
phorane and acetaldehyde is practically unaffected by changes in the polarity of the
solvent in going from methylene chloride to 1, 2-dimethoxyethane to chloroform to
dimethylformamide. They did find that solutions of lithium salts regardless of the
anion give increased yields of the cis isomer. However, they also found that a
protonic solvent such as ethanol is even more effective than added salts in in-
creasing the proportion of cis olefin. They furthermore found that chloroacetal-
dehyde gives increased amounts of the cis isomer compared to acetaldehyde.
To explain these results they propose coordination of the carbonyl oxygen by
a Lewis acki (either RGH or Li -) which could then effect the stereochemical out-
come in the following way (Figure 2).
The interconversion of the intermediate solvated betaines by either a rever-
sal of the formation reaction or through some intermediate ylide resulting from a
loss of a proton fron- either C or D may be slower than the interconvcrsion of A
and B. If the rates of decomposition of the betaine remain unaltered then the
reaction would be less stereoselective. House pointed out that even if the rate of
interconversion of the betaines is not retarded by solvation the concentration of the
solvated betaines should be different than the concentration of unsolvated betaines
because the stabilities of the solvated betaincs would be more nearly equal than the
unsoivated betaine in which the trans would be more stable. He based these con-
clusions on a consideration of the interactions of the non-bonded groups in the pre-
House also reported in this paper a repetition of work done by Shemyakin and
Bergelson in which he finds that the latter's results for the reaction of benzylidene-
triphenylphosphorane and propionaldehyde in the presence of added LiBr and LiI are
much too high. On repeating these experiments Shemyakin and Bergelson found
their results to be closer to those of House but that significant differences, partic-
ularly in the case of the dimcthylformamide solvent system still exist.
HC C C
CH3CHO + C = CKCO CH
CR3 2 / 3
O-- - H
2O-----O H Pea
RO -I O H CO CH
11 \/ 2 3
H3C H P03
RO-H---O 0 /
H3 C\ /c 2c3
SC ---- C
Fig-ure 2. Postulated Reaction Paths of the Wittig Reaction
In 1964 Drefahl, Lorenz and Schnitt21 conducted a study of the effects of
various solvents, bases, reaction temperatures, anions, reaction times and re-
actant concentrations on the stereochemistry of the Wittig reaction. These effects
were studied on one or all of the following reactions:
SCH P q + 0CHO cis + trans stilbene
XVI + cis + trans-l-phenyl-2-(a-naphthyl)-
CH P -0
0 + XVII -- cis + trans-1, 2-bis-(a-naphthyl)ethylene
These workers found that changes in the reaction temperature, anion, reaction
time and reactant concentrations have no effect on the stereochemistry of the re-
sulting olefin. However, they observed that the bases sodium carbide and sodium
amide in benzene and tetra hydrofuran respectively produced a marked decrease in
the amount of cis isomer whereas butyl lithium in benzene produced a slight increase
in the cis isomer, all compared to sodium ethoxide in the respective solvents. Their
study of solvents showed that ethanol, methanol and aniline give approximately the
same cis:trans ratio (58:42), compared to these solvents chloroform gives an in-
creased amount of cis isomer and ethyl ether, tetra hydrofuran, dioxane, benzene,
pyridine, methylene chloride, carbon tctrachloride, acctonitrile, nitrobenzene and
dimethylformamide give a decreased amount of cis isomer. They offer no expla-
nation of these results.
In summary it appears that few conclusions or generalizations can be drawn
concerning either the mechanism or the stereochemistry of the Wittig reaction at
the present time. The studies thus far conducted have produced controversy rather
than clarity--testimony to the complexity of the reaction. Furthermore, large
gaps still exist in our knowledge of certain aspects of the synthesis. This is
particularly true of the reactions of unstable or reactive ylides which is the subject
of the research to be described.
Statement of the Problem
As a result of research2' in Dr. G. B. Butler's group on the synthetic
aspects of the Wittig synthesis a number of interesting observations were made
which had previously received little or no attention in the literature. In surveying
the literature it became apparent that the stereochemistry of the reaction had also
been largely untouched. It seemed appropriate then to study these several aspects
of the Wittig reaction at the same time since they are logically related.
The objective of this research was to study the effect of certain factors on
the stereochemistry of the Wittig reaction. The system chosen was one leading to
the isomeric 1,3-pentadienes. The factors to be studied were:
3) reaction time
4) relative concentration of reactants
5) nature of the anion
6) nature of substituents.
In addition, the possibility of a 1,4-addition of ylide to a conjugated
carbonyl system was to be studied.
Method of Attack
The choice of the 1, 3-pentadiene system was based on the following considera-
tions. First, the olefin is well characterized. Second, it is highly volatile which
was an essential property in our experimental procedure. Third, since it is the
first member of the homologous 1, 3-dienes exhibiting geometrical isomerism it
should be of general interest. Fourth, no work had previously been reported on
As it turned out the choice was fortunate since the reactivity of the allylidene-
triphenylphosphorane lies between the more widely studied stable ylides and the
unstable or reactive ylides. This should permit the ready correlation of existing
data across the spectrum of ylide reactivities encountered in the Wittig reaction.
The approach was to synthesize 1, 3-pentadiene by the Wittig reaction in three
different ways as shown. Each of the three ylides shown could be prepared from
1) CH2=CH-CH-P-( + CH3CHO CH2=CH-CH=CHCH3
2) CH3CH-P03 + CH2=CH-CHO CH2=CH-CH=CHCH3
3) CH2-P-% + CH3CH=CHCHO CH2=CH-CH=CHCH3
three different phosphonium salts. These nine salts were obtained or prepared
03-P-CH2CH=CH2 I + BuLi BuH + Lil + 0 -P-CH-CH=CH3
0P-CII2CH=CHII Br + BuLi--BuH + LiBr + -P-CH-CH=CH3
03P-CH2CH=CH2 C1 + BuLi -- BuH + LiC1 + 0 -P-CH-CH=CH3
and used in investigating the nature of the anion effect.
The reaction starting with allyltriphenylphosphonium bromide, acetaldehyde
and butyl lithium in equimolar amounts was chosen as a standard and used to in-
vestigate the other factors enumerated. The standard solvent was tetrahydrofuran,
the standard reaction temperature was 00 C. and the standard reaction time was
120 seconds. These conditions were chosen for convenience and through experience
gained in preliminary work with the system.
Because of the number of reactions to be run a semi-micro technique was
employed using 1.0 millimole quantities of reactants and analyzing the products
by previously standardized vapor phase chromatography. The reactions were con-
ducted in an all glass manifold under nitrogen and no mechanical transference of
the products was required until the time of analysis.
A few auxiliary experiments were carried out so that appropriate comparisons
could be made with a somewhat different system.
DISCUSSION AYD RESULTS
Reactions in which the effect of the solvent was determined were carried out
in the standard procedure except a test solvent rather than THF was added to the
The results of these experiments are listed in Table 3. It is obvious that
the effect of solvent on the stereochemistry with but two exceptions is negligible.
On the other hand, the effect of solvent on the yield is considerable. In going from
decalin to N, N-diethylaniline one observes a 400 per cent increase in the yield.
One notes that though there is a crude correlation existing between yield and the
dielectric constant of the solvent several exceptions occur.
Both n-butyl and t-butyl alcohols give only traces of 1, 3-pentadiene under
these reaction conditions. This is not altogether unexpected since the allylidenetri-
phenylphosphorane is a very strong base, much stronger than the conjugate bases
of these two alcohols. It appears that the addition of these alcohols initiates an
acid-base reaction in which most of the ylide is converted to phosphonium salt.
CH CHRCHRCHlOH +03 -P-CH-CH=CH ---3-P-CH CH=CH + CH3 CH CHCH 0
3 2 22 3 2 3 2 2 2 3 22 2
Cis: Trans Dielectric Constant
*20 Min. reaction time
Three observations were made concerning the reaction in the butyl alcohols.
First, the fact that some 1,3-pentadiene is obtained at all seems to indicate that
an equilibrium does exist but that it is displaced far to the right. Second, further
evidence for the equilibrium is obtained in the reaction of t-butyl alcohol in which
the reaction time is lengthened from two minutes to twenty minutes. Though this
is a small change it is in the expected direction. Third, the fact that the yield of
product is greater in t-butyl alcohol than in n-butyl alcohol is in keeping with the
proposed equilibrium since the t-butoxide ion is a stronger base than the n-butoxide
ion and should cause the equilibrium to be shifted to the left.
Concerning the stereochemistry of the above two reactions it would seem
unwise to place much significance on the cis-trans ratios observed since the yields
are so low and the relative error so high.
The case of N,N-dimethylformamide (DMF) is peculiar in two respects. It
produces a 25 per cent increase in the amount of cis 1,3-pentadiene in the product
mixture whereas the yield is reduced to practically one-third of that observed in
tetrahydrofuran (THF) the next most polar solvent. This anomalous behavior is
rather difficult to explain. Shemyakin and Bergelsonl0 felt that the enhancement
of the yield of the cis isomer which they observed for reactions carried out in DMF
vs THF is caused by an interaction between the phosphorus of the ylide and the
strongly polarized oxygen of the DMF. This proposal seems unlikely for two reasons.
First, DMF, because of resonance stabilization represented by forms XX and XXII
exists as a planar molecule.24 As a planar molecule with high 7relectron density
(forms XX and XXI) on the oxygen its mode of solvation of an electrophilic center
would be through a line perpendicular to the plane of the molecule. The solvation of
HI3C 0 HI3C 0 II3C 0
S\ \+ / 3\ +/
N-C N=C N-C
/ / /
H3C HC H3C
XIX XX 7 XXI
HC /0 0/
the ylide phosphorus would appear difficult from steric considerations if the DMF
molecule assumes the same configuration as that described by Bergelson and
Shemyakin for other Lewis bases.
Second, it has been shown that DMF is a particularly effective solvent25 for
cations such as alkali metal cations. It therefore seems likely that the species
solvated would be the lithium cation rather than the phosphorus of the ylide.
If this is the case and if the ylide exists as an ion-pair as Schlosser26 has
suggested, the amount of "free" ylide would increase in DMF in contrast to less
specific cation solvents. This would result in a more rapid but less selective
attack of ylide on aldehyde. This seems to be a more plausible explanation than
that of the Russian workers.
The low yields with DMF have been observed previously by House20 and
Shemyakin. Both of these workers found yields in DMF to be lower than those in
benzene. However, neither of them has offered an explanation of this behavior.
The possibility of side reactions seems improbable since we observed no peaks in
our chromatograms different from those seen using THF. Also other workers
have obtained under somewhat different conditions and using different starting
materials, yields in excess of 90 per cent. The only speculation that can be
offered at present is that the DMF may form a relatively stable complex with the
intermediate betaines thereby decreasing the rate of betaine decomposition to
The case of the amines is also difficult to explain. There appears to be a
steric effect operating in the solvation process which favors the somewhat less
hindered DMA over DEA but no other conclusions can be drawn at this time. One
added observation was made during the course of the solvent study. When the ylide
is first generated in decalin it is an orange-red suspension. The addition of
toluene and DEA has a negligible effect on deepening the color or increasing the
solubility of the ylide. However all of the other solvents except the alcohols
caused increased or complete solution of the ylide and an intensification of the
color to a very deep red. The alcohols each reacted a bit differently. The
n-butanol formed a yellow-orange solution that contained a small amount of white
solid while the t-butyl alcohol produced a brownish-red solution containing a larger
amount of solid. The increased solubility in the more polar solvents is to be
expected but more study is necessary before conclusions can be drawn regarding
the bathochromic shifts observed.
These reactions were carried out in the standard procedure except that the
ice bath surrounding the reaction tube was replaced by the following baths. For
the room temperature bath water which had come to equilibrium at room tempera-
ture was used. For the bath at -200C a mixture of magnesium nitrate hexahydrate and
crushed ice was used. For the -780C bath Dry Ice and acetone was used.
From the data in Table 4 one may see that over the temperature range
covered no significant change in the cis-trans ratio takes place. However, the yield
is increased by 30 per cent as one proceeds from room temperature to -780C. This
is probably the result of diminishing the rate of side reactions relative to the rate
of olefin formation. The nature of the side reaction or reactions was not extensively
explored in the present work but Butler and Hauser22 have postulated a competing
aldol condensation as being responsible for the low yields generally obtained with
A side reaction that was noted in these experiments was the production of
Temperature % Yield Cis: Trans
27 C. 41 42:58
0 C. 45 42:58
-200C. 49 43:57
-780C. 53 43:57
cis and trans 2-hexene. As one goes from higher to lower temperatures one notes
that the relative amount of the 2-hexenes produced decreases from 3. 6 to 0. The
number 3. 6 represents approximately 2 mole per cent of isomeric 2-hexenes.
Therefore this is not the only side reaction taking place.
The production of the 2-hexene may be rationalized in at least two ways.
Seyferth27 and co-workers have found evidence for pentacovalent phosphorus
intermediates in the formation of ylides and proposed the mechanism shown below.
The fact that both deuterobenzene and the hydrogen substituted benzene are produced
P-CH3 Br- 4 D< Li > Li Br + D
XXV CH =3 + D 0+ H
is indicative of either a structure such as XXV or an exchange mechanism. Seyferth
favors the formation of a pentacovalent intermediate. Whether an analogous
mechanism is responsible for the 2-hexenes is open to question since Seyferth noted
that the reaction of methyl lithium with a benzyltriphenylphosphonium salt gave no
benzene. He concluded that in certain cases only hydrogen abstraction took place.
The only supporting evidence we have for such a reaction sequence besides the
product is the detection of small amounts of propene in our product mixture.
However this propene could arise from the reaction of unreacted ylide with the
water quench. That such reactions take place has been demonstrated by Fenton and
P '-CH CH=CH2 Br + Bu Li >- + Li Br
SCH 2CH 2CH3
S CH2CHLCH CH -- > (PP-CHCH CH CH + CH CH=CH
c3P-CHCHCH2CH + CH3CHO 93P-->O + 2-Hexene
3 2 2 3 2 3
03P-CHCH=CH2 + H20 --- 3P(OH)CH2CH=CH2
q3P(OH)CH2CH=CH2 3P--O> + CH3CH CH
and Ingold. 28 These workers in examining a series of phosphonium hydroxides found the
order in which groups are displaced is (CH2 >p>alkyls.. Since we also found benzene
in our product mixtures it is difficult to decide for or against this path.
The other route to the 2-hexenes involves the addition of butyl lithium to
acetaldehyde and the subsequent elimination reaction as outlined.
CH CCH CHCH Li + CH3CHO--- CH3CH2CH2CH2CH(CH3)0 Li
3 2 2 2 3 3 2 2 2H O
Q3P-CHCH=CH2 + CHH3CH2CH2C2CH(CHI) 0 Li --> 3 P-CHCH=CH
2H 1o H 2
SP HCHCI-2C CH C--> P->O + CH CH=CH + 2-Hexene
3 '-- CH 2 2 3 '3 2 2
Speziale and Ratts have postulated a similar path for the reaction of dichloro-
methylenetriphenylphosphorane in t-butyl alcohol which yielded considerable
amounts of isobutylene. An objection to this path would be that no 1-hexene was
observed along with 2-hexene. However considering the small amount of the more
stable 2-hexene formed this would be understandable.
One might conclude that if the second reaction path is in effect, the produc-
tion of the 2-hexenes would always be accompanied by the formation of propene
and vice versa. If this is true we can rule out this path since several cases were
noted where propene was liberated but no 2-hexene could be detected. Returning
to the first reaction path one might reason similarly except that here there is an
optional reaction path. Instead of the allyl group abstracting the proton from the
butyl group, a phenyl could be used in which case benzene would be a by-product
rather than propene. Benzene is ob served to be a product but the nature of the
ylide is difficult to rationalize unless we take certain geometrical characteristics
of the species into consideration. The allyltriphenylphosphonium cation probably
has tetrahedral syrametry. If we proceed through a pentacovalent structure this
species must arise from the attack of the butyl lithium on one of the sides labeled
A, B, C or D.
cD CH2CH=C H=CH2
C CH3CH2CH2CH2 CII2CH=CH2
XXV:II XXIX XXX
If the attack takes place on sides A, B or D we obtain the trigonal bipyramid XXIX.
If the attack occurs on side C we obtain structure XXX. If we take into account the
relative basicities of the three species we find that they decrease in this order
n-butyl > 0 > allyl. The acidities would conceivably be in the reverse order how-
ever the phenyl carbon attached to phosphorus has no hydrogen to donate leaving
just allyl and butyl groups to compete in this decomposition to ylide. Considering
structure XXIX it seems probable that the more acidic allyl group would donate its
proton to the butyl group giving butane. In structure =XX the strongly basic butyl
group is no longer in position to accept a proton from the allyl group and so their
would be a competition between the two acidic groups allyl and butyl to donate a
proton to a phenyl. Since the acidity of the allyl group is greater than that of the
butyl more allylidenebutyldiphenylphosphorane would be formed than butylidene-
allyldiphenylphosphorane. This is also true of structure XXIX but the probability
would undoubtedly be reduced because of the favorable positions of the acidic allyl
and basic butyl groups. Also one notes that structure XXIX is statistically favored
by 3:1 over XXX. Whether all the ylide is produced through intermediates XXIX
and XXX is an open question as is the existence of such intermediates.
Perhaps the best evidence for this reaction path is based on the stereochem-
istry of the 2-hexenes formed. In fourteen experiments under varying conditions
ten gave an excess of the cis-2-hexene, two gave an excess of trans-2-hexene and
two were about 1:1 ratios. The predominance of cis isomer seems to be diagnostic
of a reactive ylide which the butylidenephosphorane is. If the other mechanism
were correct one might expect either a statistical distribution or a preponderance
of the more stable trans isomer.
In Table 5 are the results for reactions carried out under standard conditions
but with variable reaction times. One observes that no change in the stereochem-
istry of the products takes place over the range of times covered. The yield shows
about a 10 per cent increase in going from the 30 second to the two minute reaction
time but apparently levels out beyond this.
Of greatest significance in this series of reactions is the extremely fast rate
at which the reaction proceeds. It was hoped that rough kinetics could be obtained
for the reaction during this series but this proved to be impossible under the
experimental conditions used. In the absence of kinetic data a great deal can not
be said concerning the mechanism or the transition states of this reaction except by
A kinetic study is of utmost importance in obtaining an understanding of the
mechanisms and stereochemistry of the reaction. It would appear to be the next
logical step in this research area.
REACTION TIME % YIELD CIS:TRANS
30 sec. 40 42:58
120 sec. 45 42:58
1200 sec. 46 42:58
During this series of experiments all the standard conditions were in effect
except that varying amounts of phosphonium salt, acetaldehyde and butyl lithium
were used. The results of these experiments is given in Table 6.
From the Table it is evident that varying the reactant concentrations produces
no significant change in the stereochemistry of the resulting olefin. The very slight
increase in the cis olefin produced by the doubling of both aldehyde and phosphonium
salt concentrations may or may not be significant since it lies just beyond the range
of the average deviation.
Of more interest is the fluctuation in yield that one observes. Doubling the
concentration of butyl lithium produces a sharp drop in olefin yield for three possible
reasons. First, the excess butyl lithium may add across the carbon-oxygen double
bond. Second, being a strong base it may initiate aldol type condensation reactions.
Third, it may polymerize the 1,3-pentadiene produced. These reactions represent
well known modes of reaction of organolithium compounds, and could occur singly
or together. During the runs in which the butyl lithium concentration was doubled
large quantities of viscous polymeric materials were observed to form in the
reaction tube which would be indicative of the last two reactions. However, no
attempt was made to isolate or identify such by-products.
Reactions in which the acetaldehyde concentration was doubled or the butyl
lithium and phosphonium salt concentrations were doubled though they appear com-
parable proved to be otherwise from a yield standpoint. Whereas the excess
aldehyde resulted in an increased yield of olefin compared to the reaction using
equimolar amounts of reactant, the excess amount of ylide caused a slight decrease
Concentration (mmoles/6 ml. solvent)
03-P+-CH CH=CH Br n-Bu Li CH3CHO % Yield Cis:Trans
1 1 1 45 42:58
2 1 1 50 43:57
1 2 1 16 43:57
1 1 2 59 43:57
2 2 1 41 43:57
2 1 2 75 45:55
in the yield of olefin. The latter result may be rationalized by noting the strongly
basic properties of the ylide and its ability to induce base catalyzed condensations
of the aldehyde. The increase in yield produced by doubling the aldehyde concentra-
tion was expected. If the rate expression for this class of ylides is similar to the
expression obtained by Speziale and Bissing for stable ylides then the rate would
be proportional to the ylide concentration multiplied by the aldehyde concentration.
We might expect twice the amount of product for this reaction provided there were
no side reactions and that the reaction was still in its early stages where this kinetic
expression would be linear. However, as was observed in our experiments on the
reaction time, this reaction is very fast. It begins to level off in approximately 30
seconds. Under these circumstances we would not expect to observe a doubling of
yield with a doubling of the aldehyde concentration but an increase of variable magnitude.
Doubling the concentration of phosphonium salt would appear to have little
effect on the yield since it should not alter the concentration of either reactant.
However, the data shows that this conclusion is not completely correct.
An examination of the chromatograms from the reaction of equimolar amounts
of ylide and aldehyde under varying conditions invariably reveals the presence of
small amounts of n-butane. The exact amount of n-butane cannot be ascertained
because of the overlap of the n-butane peak with the propene peak which is almost
always present and the lack of calibration for n-butane with the internal standard.
However, to make a very rough estimate the n-butane would probably be around 5
mole per cent plus or minus 3 per cent. This represents the n-butyl lithium left
after the reaction has been completed. The n-butane is formed by the hydrolysis
of this unreacted n-butyl lithium with the aqueous quench. Although the chromat-
ogram from the reaction using double salt concentration does not reveal a decrease
in the amount of n-butane it shows an increase in the amount of propene produced.
This is comparable to an increase in the amount of ylide which would account for
the increased yield. The point of this rationale is that the reaction between the
phosphonium salt and n-butyl lithium does not go to completion under the reaction
conditions used in this work. This leaves a fair amount of n-butyl lithium (perhaps
more than 10 mole per cent) free in solution to react with the aldehyde thus
decreasing the yield. The additional phosphonium salt converts part of this
unreacted n-butly lithium to ylide in which state it can then react with aldehyde to
give an increased amount of product. The observation that no 2-hexene is formed
in this particular reaction is probably caused by increasing the odds for normal
hydrogen abstraction from the allyl group rather than the butyl group by increasing
the number of phosphonium ions present.
The highest yield was observed to occur when both the phosphonium salt and
acetaldehyde concentrations alone led to increased yields of 1,3-pentadiene. The
magnitude of the increase suggests a synergistic effect which may be explained by
reasoning similar to that used for the increase accompanying the doubling of the
salt concentration. Since the increased amount of salt would form more ylide and
we now have a large excess of aldehyde with which it may react, one would expect
a substantial increase in yield.
In this series of reactions standard conditions were used except that the nature
SUBSTITUENT AND ANION EFFECTS
No. R-P-33 X R'-CHO Yield Cis: Trans
R X R'
1 CII 2CHCH2- I CH3- 42 48:52
2 CH 2CHCI2- Br CH 45 42:58
3 CH =CHCH2- Cl CI 3- 21 42:58
4 CH3CH2- I CH2=CH- 17 61:39
5 CH3CH2- Br CH2-CH- 40 59:41
6 CH3CH2- Cl CH2=CH- 47 74:26
7 CH2C-I2CH2- Br CH3- -- 71:29*
8 CH3CH2- Br CH3CH2- -- 72:28*
* Product is 2-pentene
of the phosphonium salts and aldehydes was varied. From Table 7. one observes a
very striking change in the stereochemical outcome of the reaction as one changes
either the aldehyde or the salt. When one exchanges the "R" group attached to the
aldehyde and ylide in reaction 1 through 6 the cis to trans isomer ratio is reversed.
In reactions 7 and 8 we can see the effect of replacing a vinyl group in the
allyl ylide or acrolein with an ethyl group. The result is a 75 per cent increase
in the cis isomer for allyl ylide and a 20 per cent increase for acrolein. To be
sure these systems are not strictly comparable since the product of allyl ylide
and acetaldehyde and ethyl ylide and acrolein give 1,3-pentadiene while ethyl ylide
and propionaldehyde and propyl ylide and acetaldehyde give 2-pentene as the product.
However the two systems are similar enough so that the concusions regarding the
stereochemistry involved should be fairly valid.
Since all of the ylides used in this work are "reactive" ylides we may assume
that the initial step in the reaction, betaine formation is very fast. It follows then
that this process will be unselective giving equal amounts of "cis" and "trans"
betaines. Since the ratio of products in no case shows this exact 1:1 cis to trans
ratio it would seem probable that the equilibrium postulated for unreactive and
less reactive ylides must also be operating here.
Assuming the attainment of an equilibrium between ylide and aldehyde and the
isomeric betaines one must conclude that the factor determining the stereochemistry
is the relative magnitude of the rate constants for dissociation and decomposition
of betaine as expressed by Speziale and others. The key word here is magnitude
since similar arguments have been made in rationalizing products containing greater
than 95 per cent trans isomer down to 50:50 ratios. If one makes several assump-
tions one can arrive at a fairly satisfactory rationalization of the entire range of
First it is assumed that in the initial attack of ylide upon aldehyde, unreactive
ylides are more selective giving a predominance of "trans" betaine compared to
reactive ylides. In comparing conformation of the ensuing betaines (Figure 3)
one observes that the largest effect on the orientation of ylide and aldehyde is
played by the electrostatic attraction between the developing opposite charges.
For unreactive ylides, conformations leading to XXXLII and XXXVII would be the
least likely of those shown. Of the remaining possibilities conformation leading
to "trans" betaines would be favored over those leading to "cis" betaines. The
case for reactive ylides would be quite different since betaine formation is very
rapid and hence less selective perhaps to the point of being statistical.
Second, in the absence of any stabilization, other than weakly inductive effects,
the decomposition of "cis" betaine will be favored over the "trans." Further, the
placement of the stabilizing group on the carbon attached to the phosphorus will
have a greater effect than if attached to the carbonyl carbon. These stabilizing
groups described will either exhibit a (+R) and a (-I) effect e.g. Cl or -0 or a
(-R) and (-I) effect e.g. CEN or COOR. However, if one is dealing with an
unreactive ylide the addition of an electron withdrawing group to the aldehyde will
generally increase the amount of "cis" betaine since it will increase the reaction
rate and lower the selectivity of betaine formation.
The postulate that a "cis" betaine will decompose to olefin (cis) faster than
the "trans" betaine when not stabilized is not without precedent. There are cases
of steric acceleration in the literature although none are strictly analogous to the
+ + p+ 0-^
H R 0 H R
H R' H R H R H
R H O
XXXI XXXII XXXIII XXX1V
Sp Z P 043 P3
R H 0 R H
II R' H R H R IL
H R O
XXXV XXXVI XXXVII XXXVIII
Fig. 3 Possible Bctainc Configurations
present system. Brown and Bcrncis30 showed that the solvolysis of halide XXXIX
is twenty-one times as fast as t-butyl chloride XLI and that XL is five hundred
times as fast. Since one could not possibly ascribe such differences to inductive
or hyperconjugative effects Brown reasoned that the phenomenon could best be
C-C-C-c1 c-C-c c-C-ci
C C C-C-C1 C
explained by the release of steric strain. The chlorides on ionization give carbonium
ions in which the dihedral angle between the alkyl substituents is increased from
1090 to 1200
If we consider the two possible betaine configurations involved in the forma-
tion of 2-pentene we can see that as the methyl group eclipses the ethyl group in
CH3 CH3CH2 CH3 H
IH H IH H I CHCH3
C C C C
O" PO O P^
0- 3 0- 13
cis XLII trans XLIII
the "cis" configuration the repulsive interaction produced will increase the bond
angles a and P. At the same time angles y and 6 will decrease slightly and the
distance between negatively charged oxygen and positively charged phosphorus
will decrease. In the case of the "trans" betaine this extra push toward phosphorus-
oxygen bond formation is not structurally possible. Since there exists an
equilibrium between the two betaines via reactants, the observed isomer ratio can
thus be rationalized.
Turning to the reaction of allyl ylide and acetaldehyde in which the products
consisted of a 42:58 ratio of cis to trans 1, 3-pentadiene we would predict that
the allyl ylide though very reactive is less reactive than an alkyl ylide. Therefore,
the initial attack on aldehyde would be somewhat slower and a bit more selective,
giving more "trans" than "cis" betaine. In this case since we have a stabilizing
group present, located on the carbon attached to phosphorus, it should have a
substantial effect in altering the stereochemical outcome of the reaction. XLIV
represents a possible transition state leading from betaine to olefin. This structure
shows the cyclized betaine in the process of breaking a phosphorus-carbon bond.
H3CC C CH=CH2
That this bond should break rather than the carbon-oxygen would be predicted on
the difference in bond energies which is about 7 kcal/mole, the release of
steric strain and the probability that this is an axial bond making it somewhat
weaker than a normal tri- or tetra-covalent carbon-phosphorus bond. As charge
is developed on carbon (2) the ability of the vinyl group to delocalize it will be
dependent on its ability to lie in a plane passing through carbons (1) and (2) and
perpendicular to the departing phosphorus. XLV and XLVI show that stabiliza-
tion of the developing negative charge and the double bond can be best accomplished
for steric reasons when the vinyl and methyl groups are trans to one another.
H3C I Hi H3CC H
H2C =HC H H CH=CH2
cis XLV trans XLVI
The 50 per cent increase in cis isomer obtained on exchange of the vinyl
and methyl groups between aldehyde and ylide as shown in reactions 4, 5 and 6
can be explained in a similar manner. We now have a more reactive ylide than
the allyl ylide which should give us less selectivity in betaine formation with an
increased amount of "cis" betaine. A transition state similar to that pictured for
the allyl ylide-acetaldehyde case would show that since the vinyl is not attached
to a center of primary charge-formation the stabilization of the intermediate
through a coplanar vinyl group is of less importance than the steric acceleration
promoting decomposition of the "cis" betaine.
Using the foregoing assumptions one can justify other examples of predom-
inant cis olefin as reported in the literature in a similar manner.
There is oie point which we have neglected in the foregoing argument--the
dissociation of botaine. XLVII represents a possible transition state for betaine
dissociation. We assume the conformation shown since this structure should lead
state fo betaine dt ti p
S H=-CH CII=CH
i e C +C C
SO H CH=CH,
"*aus" betain T. S. XLVII reactants
to the least electrostatic attraction between the oxygen and phosphorus and thus
be the least likely state for betaine decomposition to olefin. In this proposed
transition state we have the partial breaking of the C1 C2 bond with a shifting
oi the boidi4-. pair of electrons to C the ylide carbon. This situation necessari-
ly means that C1 ill be somewhat electron-deficient and C2 somewhat electron-
rich. Stabili-ation of such a transition state and an accompanying increased rate
of dissociation would result if electron withdrawing groups were attached to C2
and electron-donating groups were attached to C .
Using this reasoning we would conclude that in the series studied in this
work, the rate of dissociation would decrease in the following order:
allyl ylide + acetaldehyde > ethyl ylide + acrolein > ethyl ylide + propionaldehyde =
propyl ylid +- acetaldehyde. The relationship of these rates to the stercochem-
ical outcome of the reaction arises from the ensuing equilibrium between reactants
and betaine. However conclusions cannot be drawn without taking into account
the rate of betaine decomposition. Furthermore, the relative rates of dissoci-
ation of the isomeric bctains must be considered. Bissing and Spczialc31 have
recently published work in which they have determined the relative rates of
dissociation vs decomposition for three systems. Starting with isomeric 4-octene
oxides they treat these with triphenylphosphine generating the betaine in situ. If
an excess of the more reactive m-chlorobenzadehyde is present any butylidene-
triphenylphosphorane formed through equilibration would react with it in preference
to butyraldehyde giving 1-(m-chlorophenyl)-l-pentene. The ratio of l-(m-chloro-
phenyl)-l-pentene to 4-octene would then be an indication of the ratio of betaine
dissociation to betaine decomposition. They find for the 4-octene oxides this
3 P + C3H7CH-CHC3H C3HCH-- CHC3H
XLVIII P 0P-O + C H CH=CH C3H
XLVIII C3H7CH=P 3 + C3H7 CHO
XLIX + Cl CH=Cl
SCHO CH=CH H H
ratio to be quite small indicating that the rate of betaine dissociation for unstabi-
lized betaines is low. They also found that this ratio increases for stilbene oxide
and ethyl phenylglycidates which we would also expect. Their work on the 4-octene
oxides and stilbene oxides is particularly interesting since it also reveals infor-
nation concerning the relative rate of dissociation of "cis" and "trans" betaines.
Table 8 shows the results of their work with the pure isomeric epoxides. They
assumed that a cis epoxide would lead to trans betaine and trans epoxide would
lead to cis betaine.
RING OPENING REACTIONS OF 4-OCTENE OXIDE AND STILBENE OXIDE
Epoxide % 4-Octene % l-(m-chlorophenyl) -1- pentene
cis trans cis trans
cis 4-octene oxide 31 55 6 9
trans 4-octene oxide 73 12 7 9
cis stilbene oxide 21 23 12 44
trans stilbene oxide 22 22 16 39
If all of their assumptions are correct only three products should result from
each isomeric epoxide. In the case of cis-4-octene oxide one should obtain trans-
4-octene and cis and t-rans l-(_-chlorophenyl)-l-pentene. However we see that
a fairly large amount of cis-4-octene is also produced and the second isomer
arises in the other two reactions as well. The authors state that the "forbidden"
isomer may arise from either direct attack of the triphenylphosphine on oxygen of
the epoxide or from a proton exchange.
H C3 3 H H C3H7
3- 3 7
HC 3 7 H C3 7
The point to be observed here is that the ratio of betaine dissociation to
betaine decomposition is the same for "cis" and "trans" betaine derived from
4-octene-oxide and from stilbene oxide. Though this conclusion may be correct
for the stilbenes it is not correct for the octenes. The fact that the "cis"-4-octene
betaine gives 73 per cent cis 4-octene while the "trans"-4-octene betaine gives
only 53 per cent trans -4-octane is indicative that the rates of betaine decompo-
sition are quite different. The authors state that the reliability of the dissocia-
tion to decomposition ratio is probably not too high so it seems probable that the
rate of betaine decomposition is higher for the "cis" than the "trans" betaine
whereas for dissociation the reverse is true.
We can rationalize the latter phenomenon by considering models of a
possible transition state leading to dissociation for the isomeric betaines. Since
no stabilizing groups are attached to the carbon atoms making up the bond about
to be broken the difference in the energy of the transition state should be a
reflection of steric considerations alone. In the "trans" structure there will
result a skew interaction between the adjacent n-propyl groups which is absent
in the "cis" structure. If we invoke steric acceleration we would predict the
"trans" isomer to dissociate to reactants faster than the "cis." However, when
one or both of the n-propyl groups are replaced by stabilizing groups such as
vinyl, carbomethoxy, etc., the reverse would be true since the ability of that
group to stabilize the charged center would be dependent on its ability to become
co-planar. It could accomplish this much more easily in the "cis" structure than
in the "trans".
In looking at the data in Table 7 we note two reactions which gave higher
proportions of the cis-1,3-pentadiene than the others. The reaction using
allyltriphenylphosphonium iodide gives a 48:58 ratio. The reaction of ethyl-
triphenylphosphonium chloride gives a 74.26 cis :trans ratio while the other ethyl-
phosphonium salts give a 60 :40 ratio. Whether we are observing a halide effect
or an experimental error cannot be said with certainty. The latter is a possibility
only because the explanation of the former postulation is so difficult.
Shemyakin and Bergelsonl9 reported a result similar to ours for reactions
4 and 6. They reacted propylidenetriphenylphosphorane with benzaldehyde in
ethyl ether to give P-ethyl styrene in the presence of added lithium halides.
Their results are shown in Table 9. Their results exhibit the same trends
in yield and stereochemistry as ours.
CH3CH2CH=P(C 3 + qCHO CH3CIICH=CH0 + 03P O
Solvent Salt Ylide:RCHO Yield Cis:Trans
EtOEt LiC1 1:1 70% 80:20
EtOEt LiI 1:1 46% 22:78
lH LiC1 1:1 62% 79:21
IH Lil 1:1 33% 35:65
In another series of experiments32 these investigators reacted the ylide
derived from benzyltriphenylphosphonium chloride and iodide with propionaldehyde
in benzene. With the chloride they obtained a 20 : 80 cis to trans ratio while with
the iodide they obtained a 41: 59 ratio. They gave no yield data. This reflects
a similar increase in the amount of cis isomer we experienced in comparing
reactions (3) and (1). The benzyl and allyl ylide are roughly comparable and we
have demonstrated the absence of a solvent effect for this system so that the
analogy is not without some merit.
Even if we can justify our results by comparing it with the work of others
there still remains the need of a better explanation or a satisfactory mechanism.
Regarding the question of a "halide effect" we have two opposite schools of thought.
Shemyakin and Bergelson feel the effect is caused by the halide ion acting on the
phosphorus of the ylide while House believes that it is caused by a lithium cation
acting on the oxygen of the carbonyl function.
Whatever the mechanism of this effect it seems clear from our results
and those of the Russians that it is of a synergistic nature. The presence of a
specific halide (cation) only has an effect on the stereochemistry when a specific
ylide is used. When used with a different ylide no effect on the stereochemistry
There appears to be no explanation for these results at the present time.
Too few data are available on the solubilities and other factors which certainly
have a bearing on this phenomenon. Until we have such data it seems best to
avoid unsubstantiated postulates.
1,4-Addition and Isomerization
The reactions in Table 10 were carried out under standard conditions
except the specified mcthylphosphonium salts and crotonaldehyde were used. The
commercially available crotonaldehyde is apparently almost all trans isomer.
V.p.c. analysis showed but one peak on two different columns but there is still the
possibility that neither column could resolve the mixture of isomers.
From the Table 10 one observes a constant cis-trans isomer ratio of 5 : 95.
This indicates that if the starting aldehyde was a 5 : 95 mixture of cis and trans
isomers no isomerization took place under the reaction conditions used, that is,
REACTIONS OF METHYLTRIPHENYLPHOSPHORANES
Sal: % Yield cis-trans
Me P I 19 5:95
MeP Br 17 5:95
Me P Cl 25 5:95
The increased yield with the chloride may be indicative of a "halide effect"
as discussed in the previous section.
The other purpose for running this series of experiments was to see if a
possible i,4-addi-ion of the ylide to aldehyde would take place. There are several
references 3334,3 in the literature to such reactions but none have been reported
for s.uh a simple system as crotonaldehyde. If 1,4-addition occurred it would
do so by the path shown.
CH =C -CH=CH2
No evidence for such a reaction occurring was concluded by the observation
that no isoprene is formed.
To further show that no isomerization took place under our reaction con-
ditions a small amount of pure trans 1, 3 -pentadiene was added to the reaction
between methylenetriphenylphosphorane and acetaldchyde. No cis 1,3-pentadiene
was observed among the products.
Equipment and Data
All temperatures are reported in degrees centigrade and are uncorrected.
Pressures are reported in millimeters of mercury as determined by either a
Zimmerli or McLeod gauge.
Infrared spectra were either obtained with a Perkin-Elmer Infracord or a
Beckman IR 10 under the conditions noted.
Refraction indexes were obtained using a Bausch and Lomb Abbe 34 Refrac-
tometer equipped with an achromatic compensating prism.
Vapor phase chromatography (v.p. c.) was carried out on a F and M Model
700 chromatograph using a hydrogen flame-ionization detector and the columns
and conditions stipulated.
Elemental analyses were performed by Galbraith Laboratories, Knoxville,
Source and Purification of Materials
The absence of water and purity of the solvents and aldehydes were checked
by infrared and v.p.c. analysis on two columns having the following specifications:
8 ft. 10% Carbowax 20M on 60-SO mesh Diatoport S
8 ft. 10% Silicone Fluid (Nitrile) on 60-80 mesh Diatoport S
The temperatures and flow rates for the v.p. c. analysis were selected to
give optimum resolution.
Distillations at atmospheric pressure were carried out under nitrogen.
Tetrahydrofuran was J. T. Baker reagent grade material. It was pre-dried
over "Dri-Na" for several days, then refluxed with and distilled from lithium
aluminum hydride. B.p. 66.3-66.4C., n20 1.4078; lit. (36)b.p. 65.4C., n2
1.407. The infrared spectrum showed a trace of water which may have been intro-
duced into the exceedingly hygroscopic solvent during sampling. V.p.c. on both
columns showed a single peak.
Decalin was Fisher Laboratory Grade--a mixture of cis and trans isomers.
It was washed successively with three portions of concentrated sulfuric acid, three
portions of 5 per cent aqueous sodium hydroxide and four portions of distilled water.
It was dried overnight with anhydrous calcium chloride or "Dri-Na" and distilled
under vacuum, a 10 per cent forerun being discarded. Infrared examination showed
water to be absent and v.p. c. showed just two peaks.
Triethylamine was "white label" grade obtained from Distillation Products
Industires, Division of Eastman Kodak Company. It was refluxed with and distilled
from anhydrous barium oxide. B.p. 88.7-88. 80 C., nD 1.4005; lit. (37) b.p.
89.30 C., nD 1.4003. Infrared showed a trace of water and v.p. c. showed just
Toluene was J. T. Baker reagent grade material. It was distilled, discarding
the first one-third of the distillate. B.p. 110. 00 C., nD 1.4968; lit. (38) b. p.
110.60 C., n 1.4969. Infrared showed no water to be present and v.p. c.
showed but one peak.
N, N-diethylaniline was "white label" grade "mono-free" material obtained
from Distillation Products Industries Division of Eastman Kodak Company. It was
refluxed with and distilled from anhydrous barium oxide under reduced pressure.
o 20.8 22.3
B.p. 52.6-52.7 C./0.8 mm n 1.5417; lit. (39) b.p. 217. 5 C./760 mm n22
1. 5411. Infrared showed no water or primary and secondary amine to be present.
V.p.c. showed a single peak.
N, N-dimethylaniline was "white label" grade "mono-free" material obtained
from Distillation Products Industries of Eastman Kodak Company. It was refluxed
with and distilled from anhydrous barium oxide under reduced pressure. B.p.
63. 50C./2. mm n1 1.5581; lit. (40) b. p. 194. 20 C./760 mm, nD 1.5587.
Infrared showed the absence of water or primary and secondary amines. V.p. c.
showed a single peak on the Carbowax column and a small shoulder of less than 1
per cent total area on the Silicon column.
N, N-dimethylformamide was reagent grade material obtained from Fisher
Scientific Company. It was refluxed with and distilled from anhydrous barium
oxide, b. p. 149. 5 C., n0 1. 4301; lit. (41) b.p. 1530C., n 41.4294. Infra-
red showed a trace of water present. V.p. c. showed a single peak.
n-Butyl alcohol was reagent grade material obtained from J. T. Baker Chemi-
cal Company. It was refluxed with and distilled from sodium and di-n-butyl
phthalate. B.p. 118.5-118.7C., n204 1.3995; lit. (42) b.p. 118 C., n2 1.3991.
V.p. c. showed a single peak.
tert-Butyl alcohol was J. T. Baker reagent grade material. It was refluxed
with and distilled from sodium metal. B.p. 83.7-83.80 C., nD 1.3867; lit. (42)
b.p. 82.60C., n0 1.3878. V.p.c. showed a single peak.
Acetaldehyde was "white label" grade material obtained from Distillation
Products Industries Division of Eastman Kodak Company. It was distilled through
a 12-inch vacuum-jacketed Vigreux column. The condenser temperature was kept
at approximately 100 C. by external cooling. B.p. 22-230 C. ; lit. (43) b.p. 20. 20 C.
V.p. c. showed a single peak.
Acrolein was "white label" grade material obtained from Distillation Pro-
ducts Industries Division of Eastman Kodak Company. It was dried with anhydrous
magnesium and sodium sulfates then distilled. B.p. 52. 5-52.70 C. ; lit. (44) b.p.
52. 50 C. Infrared showed a trace of water. V.p. c. on the Carbowax column showed
two small extra peaks of total area approximately 1 per cent. Only one small peak
besides that of acrolein was noted on the silicone column.
Crotonaldehyde was "white label" grade material obtained from Distillation
Products Industries, Division of Eastman Kodak Company. It was dried with
anhydrous magnesium sulfate and distilled. B.p. 102.9-103.20 C., nD 1.4370;
lit. (45) b.p. 102.20C., n205 1.4362. Infrared showed a trace of water to be
present. V.p.c. showed a single peak.
Samples of cis and trans-1, 3-pentadiene, cis and trans-2-hexene and cis
and trans-2-pentene were obtained from Chemical Samples Company. V.p. c.
analysis of these compounds was carried out on a 25 ft. e, e-oxydipropionitrile
Cis-1,3-pentadiene shows characteristic infrared absorption54 at 771 cm-1
V.p.c. gives a single peak. n 1.4367; lit. (52) nD 1.4363.
Trans-1, 3-pentadiene exhibits characteristic infrared absorption54 at 814
cm V.p.c. shows a small peak in addition to the main peak. This small peak
precedes the larger and constitutes approximately 5-10 per cent of the larger peak.
nD 1.4298; lit. (52) nD 1.4301.
Cis-2-hexene exhibits characteristic infrared absorption3 at 691 cm
V.p.c. shows a very small peak preceding the main peak. nD 1.3974; lit. (52)
nD -1. 3977.
Trans-2-hexene exhibits characteristic infrared absorption5 at 964 cm-
V.p. c. shows a small peak preceding the main peak. The area of the small peak
is approximately 3-5 per cent of the main peak. n2 1. 3941; lit. (52) nD 1.3935.
Cis-2-pentene exhibits characteristic infrared absorption at 696 cm
V.p.c. shows a very small peak preceding the main peak. n 7 1.3832; lit. (52)
Trans-2-pentene exhibits characteristic infrared absorption5 at 964 cm-1
V.p.c. shows a single peak. n17 1.3791; lit. (52) nD 1.3793.
Ethyl acetate, 2-propanol, benzene and carbon tetrachloride which were used
as solvents in the recrystallization of the phosphonium salts were Fisher reagent
The methyl isobutyl ketone and chloroform used in recrystallizations were
reagent grade chemicals obtained from J. T. Baker Chemical Company. The 95 per
cent ethanol used in recrystallizations was obtained from Union Carbide Chemical
All of the above were used without further purification.
Ethyl iodide was obtained from Columbia Organic Chemical Company and
was used without further purification.
Triphenylphosphine and methyl iodide were obtained from Peninsular
ChemResearch, Incorporated and was used without further purification.
Allyl chloride, ethyl bromide and n-propyl bromide were obtained from
Distillation Products Industries Division of Eastman Kodak Company. The allyl
chloride and ethyl bromide were redistilled before use; the n-propyl bromide was
used as received.
The Nalcite SAR 20-50 mesh anion (chloride) exchange resin was obtained
from National Aluminate Corporation.
The butyl lithium in hexane was supplied by Foote Mineral Company. The
butyl lithium was assayed using the procedure of Kamienski and Esmay.
Preparation and Purification of Phosphonium Salts
Methyltriphenylphosphonium bromide, allyltriphenylphosphonium chloride
and allyltriphenylphosphonium bromide were on hand having been used in previous
research by Dr. Butler's group.
Methyltriphenylphosphonium iodide was prepared from methyl iodide and tri-
phenylphosphine in benzene according to the procedure of Wittig and Schollkopf5
in 54 per cent yield.
Ethyltriphenylphosphonium bromide was prepared from ethyl bromide and
triphenylphosphine in benzene by the procedure of Wittig et al. in .0 per cent yield.
Ethyltriphenylphosphonium iodide was prepared from ethyl iodide and tri-
phenylphosphine in benzene by the procedure of Wittig et al. in 75 per cent yield.
Methyltriphenylphosphonium chloride was prepared in two ways. First,
18.7 g. (0.1 mole) of silver nitrate were dissolved in 50 ml. of distilled water
followed by treatment with an excess of 6 M HC1. The precipitated silver chloride
was filtered and washed with distilled water, then added to a suspension of 40.4 g.
(0.1 mole) of methyltriphenylphosphonium iodide in 200 ml. of water. The suspen-
sion was refluxed with stirring for four hours. The suspension was allowed to cool,
then filtered with suction. The aqueous filtrate containing the methyltriphenyl-
phosphonium chloride was evaporated to dryness on a rotary evaporator. The
solid remaining was dissolved in a small amount of chloroform, transferred to an
evaporating dish and the solvent evaporated. The amorphous white salt was pro-
duced in 101 per cent yield (if anhydrous).
Second, to a 4 ft. by 1 in. O. D. Pyrex column charged with 0. 5 lbs. of
Nalcite SAR anion exchange resin in the chloride form was added 5.14 g. of methyl-
triphenylphosphonium bromide in one litre of distilled water. The effluent was
evaporated to dryness on a rotary evaporator giving a small amount of white,
amorphous methyltriphenylphosphonium chloride.
Ethyltriphenylphosphonium chloride was prepared using the first procedure
described above for methyltriphenylphosphonium chloride.
Allyltriphenylphosphonium iodide was prepared by the procedure of Wittig
et al. using triphenylphosphine and allyl iodide in benzene. The yield was 96 per
cent. The allyl iodide was prepared by the procedure of Letsinger and Traynham46
from allyl chloride and sodium iodide in acetone.
n-Propyltriphenylphosphonium bromide was prepared by the procedure of
Wittig et al. using triphenylphosphine and n-propyl bromide in benzene.
All data concerning the physical constants, analyses, crystallization sol-
vents etc., for the above salts are summarized in Table 11.
The reactions were run in the all glass apparatus shown in Figure 4. The
stopcock (A) connects the manifold to a tank of Linde "high purity" nitrogen. The
nitrogen is passed through a drying tower containing indicating Drierite. Stop-
cock (B) connects the manifold to the reaction tube (G). Stopcock (C) connects the
manifold to the receiver (H). Stopcock (D) connects the manifold to the vacuum
pump through either a Dry Ice and acetone or liquid nitrogen cooling trap. Out-
let (E) leads to a Zimmerli gauge and stopcock (F) connects the system to the
atmosphere through a silicone oil bubbler and mercury trap in series.
The reaction tubes, receivers, syringes, etc., were all dried for several
hours at 1300 C. before use.
The phosphonium salts which had already been dried once were weighed to
the nearest milligram into a tared weighing bottle on an analytical balance. The
salts were then dried in a vacuum oven overnight at 1000 C., removed to a desic-
cator and stored there until used. The weight loss incurred on this second drying
was usually less than 0. 5 milligrams but in the case of the chlorides almost a
stochiometric amount of water was removed from the monohydrate. Appropriate
corrections were made in the procedure for these salts.
PHYSICAL CONSTANTS OF PHOSPHONIUM SALTS
R-P+'-03 X- Recrystallization *m.p. 0OC. Lit. m.p. (Ref.) Analyses (found) Analyses (calcd.)
R X- Solvent(s) C H P C H P
1 st. 2 nd.
CH =CH-CH2 I F G 227-230 -- 58.36 4.58 7.27 58.63 4.69 7.20
CH2=CH-CH2 Br- C A 219-220 209-214 (47) 65.79 5.34 8.07 65.80 5.26 8.08
CH2=CH-CH2 Cl B D 227-228 225-227 (21) 72.50 6.62 4.03 74.46 5.95 9.15
CH CH2 I- A A 168-169 164-165 (48) 55.88 4.62 7.14 57.43 4.82 7.41
57.50 4.93 7.40
CH3CH2 Br C A 207-208 203-205 (49) 62.65 5.44 8.40 64.70 5.43 8.34
CH CH2 Cl E D 240-241 234-236 (50) 73.17 6.17 9.46 73.50 6.17 9.48
CH3 A A 186-187 188-189 (1) 56.62 4.25 7.84 56.46 4.49 7.67
CH3 Br A A 231-232 229-232 (1) 63.95 5.17 8.75 63.89 5.08 8.67
CH3 Cl E D 221-222 212-213 (48) 72.59 5.98 9.98 72.69 5.80 9.91
CH CH CH Br A 236-237 229-230 (51) 65.65 5.60 7.99 65.12 6.24 8.00
* All salts were dried at 1000 C. for from 12-24 hours in a vacuum oven. A = isopropyl alcohol; B = tert. butyl alcohol;
C = 1:1 isopropyl alcohol:methyl isobutyl ketone; D = 1:3 isopropyl alcohol:ethyl acetate; E = 3:3:1:1 ethyl acetate:methyl
isobutyl ketone:chloroform:n-butanol; F = 2:3 isopropyl alcohol:abs. ethyl alcohol; G = 95% ethyl alcohol.
Figure 4. Reaction nifold
Figure 4. Reaction Manifold
J = "Drierite" tower
K = Silicone oil bubbler
I, = Mercury trap
M = Zimmerli gauge
N = Magnetic stirrer
The aldehyde solutions in decalin were made up by weight on an analytical
balance. For acetaldehyde because of its volatility, five milliliter samples were
made up enough for two runs. For acrolein and crotonaldehyde both five and
ten milliliter samples were used. Acetaldehyde solutions were used the same
day and kept in the refrigerator between runs as were the other aldehydes. In
the case of acrolein, its solutions were protected from light until the sample was
withdrawn from introduction into the reaction tube. Since the solutions of aldehydes
were made up in volumetric flasks this work was carried out in the dark room at
approximately 200 C. The aldehyde solutions were withdrawn from the volumetric
flasks using a five milliliter syringe which had previously been flushed with nitro-
gen several times. The syringe was overfilled, inverted and the plunger brought
to the 2.0 ml. mark. Since the same syringe was always used along with the
foregoing procedure only small errors in the amount of aldehyde introduced should
The butyl lithium solution was kept in a 100 cc. amber bottle under nitrogen
sealed with a syringe cap. The bottle was kept inclined with the cap up so that
salts resulting from decomposition or hydrolysis of the reagent would collect at
the bottom of the bottle. Care was taken to obtain clear solutions of the reagent in
the one milliliter syringe perviously flushed with nitrogen. Again the same syringe
was always used for the butyl lithium solutions.
The tetrahydrofuran was kept in a 50 milliliter 2-necked flask under nitrogen.
When a sample was to be withdrawn a stream of nitrogen was first led into the flask
at a moderate rate before the second stopper was removed and the THF withdrawn
with a syringe.
Other solvents were used immediately after distillation and such rigid pre-
cautions were not found necessary.
The following description is what may be called a standard procedure because
it was this procedure or a variation of it that was used throughout this work. This
standard procedure involves the reaction between allylidenetriphenylphosphorane
(derived from allyltriphenylphosphonium bromide) with acetaldehyde in equimolar
amounts using tetrahydrofuran as the solvent. The reaction time is 120 seconds,
the reaction temperature is 00 C. The other reactions in this work are but varia-
tions of one of these parameters--a different temperature, etc. However, in any
experiment only one parameter is varied. The remainder are kept constant.
One millimole of phosphonium salt is transferred by a funnel from the weigh-
ing bottle to the reaction tube containing a small stirring bar. The side arm is
closed with a serum cap and 2. 0 ml. of decalin is introduced by pipet. The
reaction tube is then attached to the manifold and with stopcocks (A) and (F)
closed and (B), (C) and (D) open the system is evacuated. During this evacuation
the stirrer is activated to aid in degassing the suspension. After approximately
five minutes stopcock (D) is closed and stopcock (A) opened carefully to readmit
nitrogen to the system. This process is repeated two more times. After the
nitrogen has been admitted the third time stopcock (F) is opened and the flow rate
of nitrogen decreased.
0. 595 ml. (1 millimole) of butyl lithium is now added to the reaction tube with
a syringe through the serum cap. The suspension rapidly changes color from
yellow to orange to brick red. After two minutes stopcocks (A) and (F) are closed
and the system is evacuated. During this 30-minute evacuation period n-butane
and the hexane that was used as a solvent for the butyl lithium is removed. The
boiling point of decalin is sufficiently high so that little of it is removed in this
evacuation process. The allylidenetriphenylphosphorane and unreacted butyl
lithium are non-volatile. At the end of this evacuation, stopcock (A) is opened
and nitrogen readmitted to the system. Stopcock (F) is then opened to restore the
system to atmospheric pressure.
2. 0 ml. of tetrahydrofuran are then introduced into the reaction tube by means
of a syringe and the blood red solution is allowed to stir for ten minutes. At the
end of the ten minute period a slush ice bath is placed around the reaction tube and
the reaction mixture is allowed to equilibrate for two minutes. Stopcock (B) is then
closed and 2. 0 ml. of acetaldehyde ( 1 millimole) in decalin is introduced by syringe
into the reaction tube over a one minute period. The addition of the aldehyde solu-
tion generally results in the formation of a white solid and a lightening of the red
color. After two minutes 1. 0 ml. of distilled water is added by syringe. This
produces complete decolorization almost immediately with the formation of a two
phase liquid-liquid system. After two minutes the reaction tube is frozen out with
liquid nitrogen for three minutes. The system is evacuated after opening stopcock
(B) and closing stopcocks (A) and (F). The evacuation to less than 1 mm takes
approximately four minutes after which stopcock (D) is closed and the liquid nitro-
gen bath is transferred from the reaction tube (G) to the receiver (H).
As the reaction tube warms the volatile materials distill or sublime into the
receiver. The distillation is continued for 40 minutes. About half-way through the
distillation period the stirrer is activated to promote the transfer of the volatile
materials. At the completion of the distillation nitrogen is readmitted through
stopcock (A), stopcock (F) is opened and the liquid nitrogen bath removed from the
receiver. When the contents of the receiver have melted the receiver is discon-
nected from the manifold, sealed with a serum cap and removed to the dark room
(200 C.) where 30 p1l. of cyclohexene the internal standard is added. The mixture
is then either analyzed immediately by v.p. c. or the receiver is kept in a Dry
Ice and acetone bath until the analysis is performed.
The analyses were all run on the F and M Model 700 using a 25 ft. x 1/4 in.
O. D. column containing 10 per cent by weight of e, e-oxydipropionitrile on 60-80
mesh Chromosorb P. The column temperature was 250 + 10 C. and the flow rate
approximately 75 ml. per minute. The size of the sample injected was 0.3 pIl.
Calibration of the Internal Standard
A series of synthetic mixtures containing known amounts of trans-1,3-
pentadiene and cyclohexene, the internal standard, were prepared in tetrahydro-
furan in concentrations such that the sensitivity setting of the chromatograph would
be the same or very close to those used during the actual runs. The samples were
made up by volume at 200 C. For example, the "known" containing 50 mole per
cent 1, 3-pentadiene was prepared in the following way. Into a small vial flushed
with N2 and sealed with a serum cap was placed 0.7 ml. of tetrahydrofuran.
Next 10.04 j1. of cyclohexene was added followed by 5. 0 4-l. of trans-1, 3-penta-
diene. The mixture was shaken well and then at least two 0. 3 p1. portions were
chromatographed under the same conditions as those used for the actual runs.
In between runs the vials were kept in a Dry Ice-acetone bath. The calculations
used to arrive at the volume of internal standard and pentadiene are as follows:
Gram Molecular weight cyclohexene = 82. 1 g.
Density at 200 C. cyclohexene = 0. 811 g. /ml.
Gram Molecular weight 1, 3-pentadiene = 68.1 g.
Density at 200 C. of 1, 3-pentadiene = 0. 676 g. /ml.
1. 000 l. trans-1, 3-pentadiene = ? gram-moles
= 1 p1. x 10-3 ml. l. x 0.676 g.
ml. x 1 mole 68.1 g.
1. 000 l. = 9.93 x 10-6 moles trans-1, 3-pentadiene
9.93 x 10-6 moles cyclohexene = 7 1.
Pl. = 9.93 x 10-6 moles x 82.1g. mole- x 1 ml. (0.811 g.)- x
1 l. 10-3 ml.-1
9.93 x 10-6 moles = 1.004 P1. cyclohexene
Therefore if we want a 50 mole per cent solution of 1, 3-pentadiene we would
use 0. 500 1'l. 1, 3-pentadiene for each 1. 004 p1. cyclohexene. Since we used 5.00
vl. of 1, 3-pentadiene we would use 10.04 4l. of cyclohexene.
Using the calculations above a curve was plotted of the concentrations of
trans-1, 3-pentadiene vs. the ratio of the area under the pentadiene peak over the
area under the cyclohexene peak or
%area under pentadiene peak
% 1, 3-pentadiene area under cyclohexene peak
This was done for five concentrations of 1, 3-pentadiene, the resulting curve is
shown in Figure 5. From this curve and the areas of the 1, 3-pentadiene and
cyclohexene peaks it is possible to determine the concentration or yield of 1, 3-
pentadiene in the volatile mixture and the cis to trans isomer ratio.
To make sure that the distillation of the 1, 3-pentadiene was quantitative the
chromatogram of the residue was taken. In all cases the remaining 1, 3-pentadiene
constituted less than 1 mole per cent of the theoretical yield.
Since trans-1, 3-pentadiene was used for the calibration a possible error
might arise if the detector response differed for the cis isomer. Therefore a 20
mole per cent solution of the cis-1, 3-pentadiene was made up and analyzed using
the calibration curve based on trans-1, 3-pentadiene. The calculated amount of
cis isomer was 19. 8 per cent which is an error of 1 per cent. Therefore it was
assumed that the detector response was the same for both isomers and the one
curve was used to calculate the yields of both cis and trans-1, 3-pentadiene.
For the quantitative analysis of the 2-pentene system the isomer ratios were
calculated both by area and peak height. The two methods agreed within 2 per cent.
The reproduction of the chromatogram of the product mixture resulting from
the reaction of allylidenetriphenylphosphorane and acetaldehyde under standard
conditions is shown in Figure 6. The peaks are numbered from (1) through (11).
Peaks (8) (9) (10) and (11) were obtained at approximately 450 C. while peaks (1)
through (7) were obtained at 250 C.
0 0.20 0.40 0.60 0.80 1.00
trans-1, 3-pentadiene peak area
cyclohexene peak area
Figure 5. Calibration Curve for trans-1, 3-pentadiene
Temperature ( C. )
10 15 20 25 60
Figure 6. Chromatogram of Reaction Products
Peaks (10) and (11) were cis and trans decalin. They showed increased peak
height on the addition of a few drops of pure decalin (mixed isomers) to the product
mixture. Peaks (8) and (9) are tetrahydrofuran and benzene respectively which
were identified by the above procedure. Peak (7) is cyclohexene the internal
standard. Peak (6) is cis-1, 3-pentadiene and peak (5) is trans-1, 3-pentadiene
identified by the addition of authentic samples of the pure isomers to the product
mixture as above. Peaks (3) and (4) are trans and cis-2-hexene respectively also
identified by the addition of authentic samples of the pure isomers to the mixture.
Peak (8) is n-butane identified by the enhanced peak height observed on addition of
a THF solution of n-butane prepared by the hydrolysis of n-butyl magnesium bro-
mide. Peak (1) is propene identified by the addition of bromine to the product mix-
ture and then noting the propylene bromide peak enhancement following the addition
of known propylene bromide.
Precision of Chromatographic Analysis
A measure of the precision of the determination of the cis:trans isomer ratio
may be obtained by taking the average of the average deviations for the duplicate
runs. This is t 0.7 with the largest average deviation being 1 1.7.
The precision of the yield determination as given by taking the average of
the average deviations for the duplicate runs is t 2. 0 with the largest average
deviation being t 5. 0.
A summary of the infrared absorption bands for the phosphonium salts used
in this research is shown in Table 12. As a means of identifying individual phos-
phonium salts, infrared is impracticable. However the alkyl group of the lower
INFRARED ABSORPTIONS OF PHOSPHONIUM SALTS
1979 w, 1903 w, 1820 w, 1780 w, 1671 w, 1620 w, 1595 m-s, 1490 m-s, 1443 s,
1420 m-s, 1400 m, 1339 m, 1318 m, 1306 sh, 1118 s, 998 m-s, 900 s cm-1
1978 w, 1902 w, 1821 w, 1779 w, 1671 w, 1620 w, 1594 m-s, 1489 m-s, 1442 s,
1419 m-s, 1394 m, 1338 m, 1316 m, 1304 w-m, 1116 s, 998 m-s, 899 s cm-1
1976 w, 1902 w, 1821 w, 1779 w, 1671 w, 1620 w, 1593 m-s, 1488 m-s, 1441 s,
1418 m-s, 1394 m, 1337 m, 1318 m, 1302 w-m, 1116 s, 998 m-s, 896 s cm-1
Ethyltriphenylphosphonium 1974 w, 1901 w, 1820 w, 1778 w, 1590 m, 1480 m-s, 1437 s, 1420 m-s, 1388 m,
chloride 1332 m, 1312 m, 1108 s, 994 m-s cm-1
Ethyltriphenylphosphonium 1974 w, 1908 w, 1821 w, 1777 w, 1665 w, 1591 m, 1485 m-s, 1439 s, 1420 m-s,
bromide 1390 m, 1333 m, 1313 m, 1110 s, 995 m-s cm1
Ethyltriphenylphosphonium 1978 w, 1907 w, 1824 w, 1780 w, 1672 w, 1594 m-s, 1490 m-s, 1442 s, 1425 m-s,
iodide 1389 m, 1317 m, 1113 s, 998 s, cm-1
Allyltriphenylphosphonium 1979 w, 1903 w, 1821 w, 1778 w, 1639 m, 1613 m-s, 1592 m, 1487 m-s, 1439 s,
chloride 1422 m-s, 1390 m, 1334 w-m, 1313 m-s, 1109 s, 997 m-s, 960 m cm-1
Allyltriphenylphosphonium 1981 w, 1906 w, 1820 w, 1778 w, 1616 w, 1593 m, 1487 m, 1442 s, 1423 m-s,
bromide 1394 m, 1335 w-m, 1315 w-m, 1113 s, 998 m-s, 990 sh cm-I
Allyltriphenylphosphonium 1979 w, 1899 w, 1820 w, 1776 w, 1635 w, 1611 w, 1590 m, 1483 m, 1438 s,
iodide 1420 m-s, 1391 m, 1333 w-m, 1312 w-m, 1108 s, 994 m-s, 986 sh cm-1
n-Propyltriphenylphosphonium 1978 w, 1912 w, 1822 w, 1781 w, 1673 w, 1590 m-s, 1484 m-s, 1463 m, 1438 s,
bromide 1334 m, 1313 m, 1110 s, 1076 m-s, 994 m-s cm-1
alkyltriphenylphosphonium salts may be identified by careful analysis of the
The methyltriphenylphosphonium salts all appear to have a characteristic
strong absorption at 895-900 cm-1 which is missing in the ethyl, allyl and propyl
The allyltriphenylphosphonium salts have the double bond stretching frequency
at approximately 1635-1640 cm through which they may be identified.
The ethyl and propyltriphenylphosphonium salts have no special absorption
bands by which they may be identified.
The infrared spectra were taken on the Beckman 1R 10 as 0. 25 M solutions
in chloroform in a 0.175 mm cell. The reported range is from 2000 to 800 cm-1
in chloroform in a 0.175 mm cell. The reported range is from 2000 to 800 cm
The purpose of this research was to investigate factors influencing the
stereochemistry of the Wittig reaction. Based on this work the following con-
clusions may be drawn: (1) There is no observable solvent effect on the stereo-
chemistry of the reaction under the experimental conditions used, although these
observations are contrary to the results reported by other investigations;
(2) Reaction temperature has no effect on the stereochemistry of the reaction
although it has a considerable effect on the yield of product; (3) Even though the
reaction is extremely fast at low temperature, reaction time does not affect the
stereochemistry of the reaction; (4) The relative concentration of reactants,
contrary to the reports of other investigators, has no effect on the stereochemistry
of the reaction, although by proper choice of concentrations, high yields of
products can be obtained; (5) The nature of the substituents, particularly those
on the ylide, has a profound effect on the stereochemical outcome of the reaction;
(6) There appears to be either a cation or an anion effect influencing both the
ylide and the stereochemistry of the system studied, however, the nature of this
effect is too complex to explain with the existing data; and (7) No 1,4-addition of
the methylene ylide occurred with crotonaldehyde. A proposal to explain the
observations concerned with substituent effects is offered.
LIST OF REFERENCES
1. G. Wittig and G. Geissler, Ann., 580, 44 (1953).
2. W. von E. Doering and A. K. Hoffman, J. Am. Chem. Soc., 77, 521 (1955).
3. H. H. Jaffe, J. Phys. Chem., 58, 185 (1954).
4. A. W. Johnson and R. B. LaCount, Chem. & Ind. (London), 1959, 52.
5. G. Wittig and U. Schollkopf, Ber., 87, 1318 (1954).
6. G. Wittig, H. Weizmann and M. Schlosser, Ber., 94, 676 (1961).
7. S. Fliszar, R. F. Hudson and G. Salvadori, Helv. Chim. Acta., 46 1580
8. A. J. Speziale and D. E. Bissing, J. Am. Chem. Soc., 85, 3878 (1963).
9. H. 0. House and G. H. Rasmusson, J. Org. Chem., 26, 4278 (1961).
10. L. D. Bergelson and M. M. Shemyakin, Tetrahedron, 1, 149 (1963).
11. S. Trippett, Pure & Appl. Chem., 9, 255 (1964).
12. R. Ketcham, D. Jambotkar and L. Martinelli, J. Org. Chem., 27, 4666
13. A. J. Speziale and K. W. Ratts, J. Org. Chem., 28, 465 (1963).
14. P. C. Wailes, Chem. & Ind. (London), 1958, 1086.
15. E. Truscheit, K. Eiter, A. Butenandt and E. Hecker, Ger. 1, 138, 037,
C.A. 58, 66941(1963).
16. V. F. Kucherov, B. G. Kovalev, G. A. Kogan and L. A. Yanovskaya,
DokladyAkad. Nauk S. S. S. R., 138 1115 (1961), C.A. 55, 24560i(1961)
17. L. D. Bergelson, V. A. Vaver and M. M. Shemyakin, Izvest. Akad. Nauk
S.S.S.R., Odtel. Khim. Nauk, 1961, 729, C.A. 55, 22196c (1961).
18. L. D. Bergelson and M. M. Shemyakin, Pure & Appl. Chem., 9, 271 (1964).
19. L. D. Bergelson, V. A. Vavor, L. J. Barsukov and M. M. Shemyakin,
Tetrahedron Letters, 38, 2669 (1964).
20. H. 0. House, V. K. Jones, G. A. Frank, J. Org. Chem., 29 3327 (1964).
21. V. G. Drefahl, D. Lorenz and G. Schnitt, J. Prakt. Chem., 23, 143 (1964).
22. C. F. Hauser, T. W. Brooks, M. L. Miles, M. A. Raymond and G. B.
Butler, J. Org. Chem., 28, 372 (1963).
23. G. B. Butler and C. F. Hauser, Unpublished results.
24. H. S. Gutowsky and C. H. Holm, J. Chem. Phys., 25, 1228 (1956).
25. H. E. Zaugg, J. Am. Chem. Soc., 83 837 (1961).
26. M. Schlosser and K. F. Christmann, Angew. Chem. Inter. Ed., 3, 636
27. D. Seyferth, J. K. Heoren and W. B. Hughes, Jr., J. Am. Chem. Soc., 34
28. G. W. Fenton and C. K. Ingold, J. Chem. Soc., 1929, 2342a.
29. A. J. Speziale and K. W. Ratts, J. Am. Chem. Soc., 84, 854 (1962).
30. H. C. Brown and H. L. Berneis, J. Am. Chem. Soc., 75, 10 (1953).
31. D. E. Bissing and A. J. Speziale, J. Am. Chem. Soc., 87, 2783 (1965).
32. L. D. Bergelson and M. M. Shemyakin, Private communication to H. O.
House et al., J. Org. Chem., 29, 3327 (1964) Footnote (8).
33. H. H. Inhoffen, K. Bruckner, G. F. Domagk and H. Erdmann, Ber., 88,
34. J. P. Freeman, Chem. & Ind., 1959, 1254.
35. F. Bohlmann, Ber., 89 2191 (1956).
36. The Merck Index of Chemicals and Drugs, 7th Ed., Rahway, New Jersey,
Merck& Co., Inc., 1960, p. 1023.
37. Ibid., p. 1063.
38. Ibid., p. 1052.
39. Ibid., p. 352.
0. :-' p. 37
6. R. L. Le-Isi:.;er and J G.Traynham, J. Am. Chem. Soc., 70, 2S13 (1948).
47. H. Grayson and P. T. Keough, J. Am. Chem. Soc., S2, 3919 (1960).
4S. A. .:ichaelis ar.d H. von Soclen, Ann., 229, 295 (1SS5).
49. G. Wittig and D. Wittenberg, Ann. 006, 1 (1957).
50. Ii. J. Bastniann and B. Arnason, Ber., 95, 1513 (1962).
51. L. D. Borgeison, V. A. Vaver, L. J. Barsukov and M. M. Shemyakin,
Izv. Akad. Nauk S. S. S. R., Otd. Khim. Nauk (English Translation) 1963,
52. S. W. Fer-is, Handbook of Hydrocarbons, Academic Press, Inc., New York,
Y., 1955, p. 21-23.
53. L. J. Bellamy, The Infrared Soectra of Complex Molecules, John Wiley &
Sons, Inc., New York, N. Y., 1955, p. 45-48.
54. R. S. Rasmiussen and R. R. Bar.atain, J. Chem. Phys., 15, 131 (1947).
55. C. W. Kamienski and D. L. Esmay, J. Org. Chem., 25, 115 (1960).
Jerome Thomas Kresse was born December 29, 1931, at Buffalo, New York.
He obtained his elementary and secondary education in Buffalo graduating from
Buffalo Technical High School in 1949. He attended the University of Buffalo from
1949 until 1953. After spending two years in the United States Army he worked
for 18 months as a technician with the Silicone's Division of Union Carbide Corpora-
tion in Tonawanda, New York. In April 1957 he entered Michigan State University
and received the degree of Bachelor of Science in Chemistry in June 1958. He
entered the Graduate School of the University of Florida in September 1958.
The author is a member of Alpha Chi Sigma, Professional Chemistry
Fraternity and the American Chemical Society. He is married to the former
Joan Margaret Schmid and is the father of two children, Jennifer Ann and
This dissertation was prepared under the direction of the chairman
of the candidate's supervisory committee and has been approved by all members
of that committee. It was submitted to the Dean of the College of Arts and
Sciences and to the Graduate Council, and was approved as partial fulfillment
of the requirements for the degree of Doctor of Philosophy.
December 18, 1965
Dean, College of Arts and Sciences
Dean, Graduate School
^Y/jy -^^ <
UNIVERSITY OF FLORIDA
3 1262 08553 3825