Liquid-assisted grinding and ion pairing regulates percentage conversion and diastereoselectivity of the Wittig reaction under mechanochemical conditions

Article abstract of DOI:10.3762/bjoc.14.57

Mechanochemistry is maturing as a discipline and continuing to grow, so it is important to continue understanding the rules governing the system. In a mechanochemical reaction, the reactants are added into a vessel along with one or more grinding balls and the vessel is shaken at high speeds to facilitate a chemical reaction. The dielectric constant of the solvent used in liquid-assisted grinding (LAG) and properly chosen counter-ion pairing increases the percentage conversion of stilbenes in a mechanochemical Wittig reaction. Utilizing stepwise addition/evaporation of ethanol in liquid-assisted grinding also allows for the tuning of the diastereoselectivity in the Wittig reaction.

Full text of DOI:10.3762/bjoc.14.57

Liquid-assisted grinding and ion pairing regulates percentage
conversion and diastereoselectivity of the Wittig reaction
under mechanochemical conditions
Kendra Leahy Denlinger, Lianna Ortiz-Trankina, Preston Carr, Kingsley Benson,
Daniel C. Waddell and James Mack*
Full Research Paper
Open Access
Address:
Beilstein J. Org. Chem. 2018, 14, 688–696.
Department of Chemistry, University of Cincinnati, PO Box 210172,
Cincinnati, OH 45221-0172, USA
doi:10.3762/bjoc.14.57
Received: 20 December 2017
Accepted: 06 March 2018
Published: 23 March 2018
Email:
James Mack* - james.mack@uc.edu
* Corresponding author
This article is part of the Thematic Series "Mechanochemistry".
Guest Editor: J. G. Hernández
Keywords:
green chemistry; high-speed ball milling; HSBM; LAG; liquid-assisted
grinding; Wittig
© 2018 Denlinger et al.; licensee Beilstein-Institut.
License and terms: see end of document.
Abstract
Mechanochemistry is maturing as a discipline and continuing to grow, so it is important to continue understanding the rules
governing the system. In a mechanochemical reaction, the reactants are added into a vessel along with one or more grinding balls
and the vessel is shaken at high speeds to facilitate a chemical reaction. The dielectric constant of the solvent used in liquid-assisted
grinding (LAG) and properly chosen counter-ion pairing increases the percentage conversion of stilbenes in a mechanochemical
Wittig reaction. Utilizing stepwise addition/evaporation of ethanol in liquid-assisted grinding also allows for the tuning of the dia-
stereoselectivity in the Wittig reaction.
Introduction
Mechanochemistry is maturing as a discipline and continuing to tions for the synthesis of olefins [19-23]. Aside from its synthe-
develop and grow [1-16]. Thus it is important to continue tic utility, its unique reaction mechanism (shown in Figure 1)
studying and understanding the rules governing the system. and inherent diastereoselectivity has led to a vast amount of
Under mechanochemical conditions, the reactants are added intrigue by the chemical community [24-26].
into a vessel along with one or more grinding balls, and the
vessel is shaken at high speeds to generate the product. Several Our research group has continued the study of the Wittig reac-
years ago, Balema and Percharsky first demonstrated the tion under mechanochemical conditions with the use of a func-
success of the Wittig reaction under mechanochemical condi- tionalized polymer resin. During these studies, we discovered a
tions [17,18]. The Wittig reaction is one of the most useful reac- few exciting differences between our results and the ones ob-
688

Beilstein J. Org. Chem. 2018, 14, 688–696.
Figure 1: Solution-based Wittig reaction mechanism.
tained under traditional solution-based conditions. First, we ob- carbonate were ball-milled in a stainless steel vial with two
served that using functionalized resins allowed us to isolate the LAG solvents at opposite ends of the dielectric spectrum, as
desired product in an easy and environmentally benign manner. well as a control without any solvent (Table 1).
Second, we observed that the incorporation of liquid-assisted
grinding (LAG) increased the rate of the reaction in compari- In general, we noticed that more polar solvents (high dielectric
son to completely solvent-free conditions. Finally, we observed constants) favour Z selectivity and a higher overall conversion,
that there was an effect of the dielectric constant of the solvent whereas the use of less polar or no solvent (lower dielectric
used in LAG on the stereochemistry of the product [27]. Al- constants) favour E selectivity and a lower overall conversion.
though previously we were able to generate high yields of
Wittig products under liquid-assisted grinding conditions, we As shown in Table 1, a side product (benzyl benzoate) is also
did not truly understand the influence of the reaction medium generated during the reaction. 1H NMR spectroscopy was used
on the reaction. Therefore, we were interested in better under- to determine the E:Z ratio of the product as well as the percent-
standing these observations, with the goal of increasing the age of the side product formation. Figure 2 shows the particular
overall conversion and having better control over the stereose- peaks for each compound integrated to determine product
lectivity of the product.
ratios.
Results and Discussion
Liquid-assisted grinding
It is important to note that under traditional solution-based
conditions, benzyl benzoate has never been reported as a prod-
To focus the study, benzaldehyde, benzyl bromide, polymer- uct in the Wittig reaction. Under mechanochemical conditions,
supported triphenylphosphine (PS-C6H4-PPh2) and potassium the side-product formation appears to be hindered when
Table 1: LAG solvent effect on the mechanochemical Wittig reaction.
LAG solvent
dielectric constant [17]
% conversion to stilbene
E:Z ratio
stilbene/side product ratio
none
–
30
25
95
67:33
61:39
40:60
1/0.83
1/0.44
1/0.03
toluene
ethanol
2.38
24.5
689

Beilstein J. Org. Chem. 2018, 14, 688–696.
Figure 2: 1H NMR spectra of stilbene mixture (a) and benzyl benzoate (b).
utilizing a LAG solvent with a high dielectric constant. These side product, which involves the addition of benzaldehyde to
results were used to probe the ability to tune the Wittig reaction the traditional betaine intermediate of the Wittig reaction. This
under mechanochemical conditions.
addition step occurs before the rotation and formation of the
oxaphosphetane (Scheme 1). In the presence of very small
To determine the origin of benzyl benzoate, we performed a amounts of solvent (LAG) or no solvent at all, the concentra-
number of control reactions to determine if all reactants are tion of reactants is very high. This could cause the additional
necessary to form the side product (Table 2). Benzyl bromide benzaldehyde to be close enough to the intermediate to react
was absent in each control reaction, and a “√” in the table indi- before the rotation occurs. To bolster further this argument, the
cates reactants that were present in each control trial.
highest amount of side product is observed in the absence of
solvent, i.e., at highest reactant concentration (Table 1).
Interestingly, it was found that all four components of the reac-
tion (benzyl bromide, benzaldehyde, base, and PS-C6H4-PPh2) Further, our results show that the E selectivity and benzyl
are necessary for the production of benzyl benzoate. Based on benzoate formation are observed together (when the dielectric
this evidence, we propose a mechanism for the formation of this constant of the LAG solvent is low). Therefore, if the benzyl
690

Beilstein J. Org. Chem. 2018, 14, 688–696.
Table 2: Control reactions to determine the origin of the side product benzyl benzoate.
control trial
benzaldehyde
K2CO3
PS-C6H4-PPh2
result
1
2
3
4
√
√
√
√
No reaction
No reaction
No reaction
No reaction
√
√
√
√
Scheme 1: Possible mechanism of benzyl benzoate formation.
benzoate formation follows the path presented in Scheme 1, benzaldehyde (Scheme 3) to increase the concentration of reac-
perhaps the mechanism of the E selectivity is similar. The inter- tants. At the same time, benzaldehyde can be considered a LAG
mediate (Scheme 1) allows for a reaction pathway involving the solvent with a high dielectric constant (benzaldehyde has a
formation of a six-membered ring instead of the traditional dielectric constant of 17.8 [28]). The increased concentration
four-membered oxaphosphetane ring of the Wittig reaction. should favour E selectivity, but the high dielectric constant
This six-membered ring may account for the higher E selec- should favour Z selectivity.
tivity due to the preference for larger groups to be in equatorial
positions in cyclohexane rings (Scheme 2).
Compared to the original reaction (E:Z ratio 67:33, Table 1,
entry 1), the reaction with excess benzaldehyde resulted in an
The E selectivity is driven by both the high concentrations of increase in Z selectivity with an E:Z ratio of 52:48. Therefore, it
reactants and a low dielectric constant of the LAG solvent (or can be concluded that, if a LAG solvent is present, its dielectric
no solvent). To evaluate which might be playing a more critical constant will be the determining factor in diastereoselectivity,
role in the selectivity, the reaction was run with an excess of consistent with our previous observations.
Scheme 2: A possible mechanistic explanation for the E selectivity.
691

Beilstein J. Org. Chem. 2018, 14, 688–696.
Scheme 3: Ball-milled Wittig reaction using excess benzaldehyde.
Counter-ion pairing
However, this pathway is shut down under solvent-free
We further hypothesized that, if the benzyl benzoate is formed conditions, making everything in the system a contact
through a six-membered ring intermediate to give (E)-stilbene ion pair. The traditional solution-based mechanism of the Wittig
as the major product, then the same rationale could be used in reaction proceeds via a four-membered oxaphosphetane inter-
the case of our solvent-free conditions. In solution, ions are sep- mediate. However, by incorporating the halide anion
arated and stabilized by solvent molecules. Mechanistically we and the alkali metal cation into the mechanism, a six-mem-
envision ions to start out as contact ion pairs, then solvent sepa- bered ring, similar to that proposed in Scheme 2, would result
rated ion pairs (i.e., loose ion pairs) followed by free ion pairs. (Figure 3).
Figure 3: Comparison of solution based Wittig reaction (a) with polymer-supported mechanochemical Wittig reaction (b).
692

Beilstein J. Org. Chem. 2018, 14, 688–696.
Table 3: Counter-ion partnerships.
Jones–Dole viscosity B
coefficient [31]
Ion
Pearson HSAB concept
Ionic radius (pm)
Li+
Na+
K+
hard
hard
0.150
0.086
76
102
138
152
167
181
196
hard
−0.007
−0.030
−0.045
−0.007
−0.032
Rb+
Cs+
Cl-
borderline
soft
hard
Br-
borderline
Based on this hypothesis, in addition to the oxygen and phos- polymer-supported Wittig reaction with various carbonate salts
phorous forming a stable bond, the alkali metal and halide must and alkyl halides (Table 4).
form a stable contact ion pair as well. These interactions and the
formation of this bond may have a large influence under Pairing of a hard acid (Li+ or Na+) with a moderately soft base
mechanochemical conditions because there is not a solvent (Br−) leads to no or poor conversion to stilbene products. Con-
reservoir to accept the dispersion of these ions. Using Pearson’s versely, the best conversion resulted when combining Cs+ (soft
hard and soft acid and base (HSAB) theory [29] and the acid) with Br− (borderline soft base). The benzyl benzoate side-
Jones–Dole viscosity B coefficient [30] (Table 3), we can product observed in trials 1, 2, and 5 indicates that the
predict which alkali metal and halide pairs would be most carbonate bases deprotonated the phosphonium salt to
favourable. For example, bromide is a borderline soft anion, so form the ylide which then subsequently added to the
based on the proposed mechanism we would expect more prod- benzaldehyde. However, the oxygen anion could not
uct to form if the counter ion is Cs+(soft) than if it was Li+ bind to the phosphorus cation to produce the stilbene
(hard). Using the Jones–Dole viscosity B coefficient, we could product, presumably due to the mismatched counter
also predict that Cs+ and Br− would be a good pair, since their ion pair. After identifying the caesium–bromine
values are similar. We would expect Rb+ to pair well with Br− pair was optimal for the conversion, caesium carbonate
for the same reason. To test this idea and to understand the was the base of choice for the described stepwise study with
effect of the interaction, we conducted the solvent-free ethanol.
Table 4: Counter-ion partnerships in the solvent-free mechanochemical Wittig reaction.
conversion to
stilbene
conversion to
benzyl benzoate
trial
cation (M+)
anion (X−)
E:Z ratio
1
2
3
4
5
6
7
8
Li
Na
K
Br
Br
Br
Br
Cl
Cl
Cl
Cl
–
0%
0%
6%
–
29%
45%
9%
67:33
78:22
–
30%
72%
0%
Cs
Li
10%
13%
24%
28%
Na
K
72:28
69:31
74:26
11%
37%
36%
Cs
693

Beilstein J. Org. Chem. 2018, 14, 688–696.
Tuning liquid-assisted grinding with ethanol
the reaction stepwise could influence our ability to select for
Because using ethanol (high dielectric constant) as the LAG both percent conversion and diastereoselectivity. Using a step-
solvent afforded the highest conversion to stilbene and the least wise reaction approach with ethanol as the LAG solvent (no
amount of benzyl benzoate, we began our study on the yield and work-up performed between the steps), a 98% conversion to
diastereoselectivity using this solvent. First, we were interested stilbene was observed with an E:Z ratio of 43:57 (Scheme 4).
in the influence of ethanol on the mechanochemical reaction of
the alkyl halide and the polymer-supported triphenylphosphine. As can be seen from the scheme, the reaction proceeded with
For this purpose, PS-C6H4-PPh2 and 4-nitrobenzyl bromide high conversion and Z selectivity, which was ascribed to the
were ball-milled for two hours with and without ethanol as the high dielectric constant of ethanol. However, if ethanol was
LAG solvent. Afterwards, the reaction mixture was filtered with allowed to evaporate from the vial before the addition of
ethanol to determine the amount of 4-nitrobenzyl bromide in benzaldehyde, thus creating a non-LAG condition, then E selec-
solution: the higher the amount recovered means that less tivity should be favoured for the Wittig reaction. Indeed, it
4-nitrobenzyl bromide reacted and is bound to the polymer and turned out that a 98% conversion to the product along with an
thus less production of the desired phosphonium salt. As ex- E:Z ratio of 78:22 (Scheme 5) occurred.
pected, only 8.5% unreacted 4-nitrobenzyl bromide was recov-
ered when ethanol was used as a LAG solvent, demonstrating As shown, the dielectric constant of the solvent used in LAG
that ethanol is an effective LAG solvent for the production of can affect both the percent conversion of the reaction as well as
the phosphonium salt (Table 5).
the diastereoselectivity. By running the reaction stepwise, we
can tune the reaction to proceed with high percent conversion
while changing the diastereoselectivity of the product.
Table 5: How much 4-nitrobenzyl bromide adds to the polymer-sup-
ported triphenylphosphine, PS-C6H4-PPh2?
Conclusion
Both a high dielectric constant of the solvent used in liquid-
assisted grinding (LAG) and proper ion pairing were found to
increase the percent conversion to stilbenes under
mechanochemical conditions. Choosing appropriate ion pairs
when LAG is utilized in the system also allowed tuning the dia-
stereoselectivity. Specifically, this selectivity could be achieved
by combining the Cs+/Br− pair with the LAG solvent as
follows: if one millilitre of ethanol was present in both reaction
steps a higher Z selectivity was obtained. If one millilitre of
ethanol was present only in the first step of the experiment, a
higher E selectivity was obtained. The high concentration of
reactants under mechanochemical conditions allows for unique
ethanol as LAG solvent?
percent mass recoverya
no
79.5 %
8.5%
yes (1 mL)
aAverage of two trials.
Because the formation of the phosphonium salt is the first step and potentially selective reactions that may not be achievable by
of the Wittig reaction, the question arose whether performing traditional synthetic means. Further studies on the influence of
Scheme 4: Stepwise ball-milled Wittig reaction with ethanol as the LAG solvent.
694

Beilstein J. Org. Chem. 2018, 14, 688–696.
Scheme 5: Stepwise ball-milled Wittig reaction with ethanol evaporation between the steps.
ORCID® iDs
Daniel C. Waddell - https://orcid.org/0000-0001-7318-7687
HSAB theory and the Jones–Dole viscosity B coefficient under
mechanochemical conditions are ongoing.
References
Experimental
1. Do, J.-L.; Friščić, T. ACS Cent. Sci. 2017, 3, 13–19.
NMR spectra were obtained using a Bruker Avance 400 MHz
spectrometer. Deuterated chloroform was obtained from
Cambridge Isotope Laboratories Inc., Addover, MA, and used
without further purification. Triphenylphosphine-functionali-
zed polystyrene, 2% cross-linked with divinylbenzene (PS-
C6H4-PPh2) was obtained from Biotage® and used without
further purification. Benzaldehyde was obtained from Sigma
Aldrich and used without further purification. Alkyl halides and
carbonate bases were obtained from Fisher Scientific and used
without further purification.
doi:10.1021/acscentsci.6b00277
2. James, S.; Adams, C.; Bolm, C.; Braga, D.; Collier, P.; Friščić, T.;
Grepioni, F.; Harris, K.; Hyett, G.; Jones, W.; Krebs, A.; Mack, J.;
Maini, L.; Orpen, A.; Parkin, I.; Shearouse, W.; Steed, J.; Waddell, D.
Chem. Soc. Rev. 2012, 41, 413–447. doi:10.1039/C1CS15171A
3. Wang, G.-W. Chem. Soc. Rev. 2013, 42, 7668–7700.
doi:10.1039/c3cs35526h
4. Takacs, L. Chem. Soc. Rev. 2013, 42, 7649–7659.
doi:10.1039/c2cs35442j
5. Boldyreva, E. Chem. Soc. Rev. 2013, 42, 7719–7738.
doi:10.1039/c3cs60052a
6. Strukil, V. Beilstein J. Org. Chem. 2017, 13, 1828–1849.
doi:10.3762/bjoc.13.178
Mechanochemical Wittig reaction
7. Metro, T.-X. M.; Martinez, J.; Lamaty, F. ACS Sustainable Chem. Eng.
2017, 5, 9599–9602. doi:10.1021/acssuschemeng.7b03260
8. Do, J.-L.; Friščić, T. Synlett 2017, 28, 2066–2092.
doi:10.1055/s-0036-1590854
To a customized stainless steel vial (3.0 mL volume) was added
1 mmol (500 mg) of PS-C6H4-PPh2, 0.998 mmol alkyl halide,
0.58 mmol aldehyde, and 1.3 mmol carbonate base. This mix-
ture was ball-milled for 16 h. For liquid-assisted grinding exper-
iments, also 1 mL solvent was added. For stepwise reactions,
PS-C6H4-PPh2, the alkyl halide, and the LAG solvent were ball-
milled for 2 h. Afterwards, the aldehyde and carbonate bases
were added, and the reaction mixture was ball-milled for further
16 h. A 4.8 mm (3/16”) stainless steel ball was added to the vial
for all steps. The vial was shaken at 18 Hz in a Spex8000M
Mixer/Mill. To work up the reaction after the reaction was com-
plete, 2 mL of ethyl acetate were added to the vial, and the vial
was returned to the mill for 5 min. The resulting mixture was
gravity filtered with ethyl acetate. The solvent was removed
under reduced pressure. 1H NMR spectroscopy was performed
to assess the composition of the filtrate.
9. Achar, T. K.; Bose, A.; Mal, P. Beilstein J. Org. Chem. 2017, 13,
1907–1931. doi:10.3762/bjoc.13.186
10.Zhao, S.; Li, Y.; Liu, C.; Zhao, Y. Tetrahedron Lett. 2018, 59, 317–324.
doi:10.1016/j.tetlet.2017.12.021
11.Tan, D.; Friščić, T. Eur. J. Org. Chem. 2018, 2018, 18–33.
doi:10.1002/ejoc.201700961
12.Strukil, V. Synlett 2018, in press. doi:10.1055/s-0036-1591868
13.Intasa-ard, S. G.; Imwiset, K.; Bureekaew, S.; Ogawa, M. Dalton Trans.
2018, 47, 2896–2916. doi:10.1039/c7dt03736h
14.Hernandez, J. G. Chem. – Eur. J. 2017, 23, 17157–17165.
doi:10.1002/chem.201703605
15.Geciauskaite, A. A.; Garcia, F. Beilstein J. Org. Chem. 2017, 13,
2068–2077. doi:10.3762/bjoc.13.204
16.Andre, V.; Quaresma, S.; Ferreira da Silva, J. L.; Duarte, M. T.
Beilstein J. Org. Chem. 2017, 13, 2416–2427. doi:10.3762/bjoc.13.239
17.Balema, V.; Wiench, J.; Pruski, M.; Pecharsky, V. J. Am. Chem. Soc.
2002, 124, 6244–6245. doi:10.1021/ja017908p
Acknowledgements
We are thankful for financial support for this research from the
National Science Foundation, CHE-1465110. We also would
like to thank Dr. Travis Pollard for helpful discussions.
18.Balema, V.; Wiench, J.; Pruski, M.; Pecharsky, V. Chem. Commun.
2002, 724–725. doi:10.1039/b111515d
19.Cristau, H.-J. Chem. Rev. 1994, 94, 1299–1313.
doi:10.1021/cr00029a006
695

Beilstein J. Org. Chem. 2018, 14, 688–696.
20.Murphy, P.; Lee, S. J. Chem. Soc., Perkin Trans. 1 1999, 3049–3066.
doi:10.1039/a803560a
21.Hajos, G.; Nagy, I. Curr. Org. Chem. 2008, 12, 39–58.
doi:10.2174/138527208783330082
22.Fairlamb, I. ChemSusChem 2009, 2, 1021–1024.
doi:10.1002/cssc.200900208
23.Marsden, S. Nat. Chem. 2009, 1, 685–687. doi:10.1038/nchem.458
24.Schlosser, M. Top. Stereochem. 1970, 5, 1–30.
doi:10.1002/9780470147146.ch1
25.Vedejs, E.; Peterson, M. Top. Stereochem. 1994, 21, 1–157.
doi:10.1002/9780470147306.ch1
26.Vedejs, E.; Peterson, M. Adv. Carbanion Chem. 1996, 2, 1–85.
27.Shearouse, W.; Mack, J. Green Chem. 2012, 14, 2771–2775.
doi:10.1039/c2gc35669d
28.El-Anwnr, I. M.; Foad, I. Z. S. A. A. J. Mater. Sci. Technol. 1995, 11,
222–228.
29.Pearson, R. G. J. Am. Chem. Soc. 1963, 85, 3533–3539.
doi:10.1021/ja00905a001
30.Jones, G.; Dole, M. J. Am. Chem. Soc. 1929, 51, 2950–2964.
doi:10.1021/ja01385a012
31.Jenkins, H. D. B.; Marcus, Y. Chem. Rev. 1995, 95, 2695–2724.
doi:10.1021/cr00040a004
License and Terms
This is an Open Access article under the terms of the
Creative Commons Attribution License
(http://creativecommons.org/licenses/by/4.0), which
permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
The license is subject to the Beilstein Journal of Organic
Chemistry terms and conditions:
(https://www.beilstein-journals.org/bjoc)
The definitive version of this article is the electronic one
which can be found at:
doi:10.3762/bjoc.14.57
696