Solvent Effects in the Homogeneous Catalytic Reduction of Propionaldehyde with Aluminium Isopropoxide Catalyst: New Insights from PFG NMR and NMR Relaxation Studies

Article abstract of DOI:10.1002/cphc.202000267

Solvent effects in homogeneous catalysis are known to affect catalytic activity. Whilst these effects are often described using qualitative features, such as Kamlet-Taft parameters, experimental tools able to quantify and reveal in more depth such effects have remained unexplored. In this work, PFG NMR diffusion and T₁ relaxation measurements have been carried out to probe solvent effects in the homogeneous catalytic reduction of propionaldehyde to 1-propanol in the presence of aluminium isopropoxide catalyst. Using data on diffusion coefficients it was possible to estimate trends in aggregation of different solvents. The results show that solvents with a high hydrogen-bond accepting ability, such as ethers, tend to form larger aggregates, which slow down the molecular dynamics of aldehyde molecules, as also suggested by T₁ measurements, and preventing their access to the catalytic sites, which results in the observed decrease of catalytic activity. Conversely, weakly interacting solvents, such as alkanes, do not lead to the formation of such aggregates, hence allowing easy access of the aldehyde molecules to the catalytic sites, resulting in higher catalytic activity. The work reported here is a clear example on how combining traditional catalyst screening in homogeneous catalysis with NMR diffusion and relaxation time measurements can lead to new physico-chemical insights into such systems by providing data able to quantify aggregation phenomena and molecular dynamics.

Full text of DOI:10.1002/cphc.202000267

Accepted Article
Title: Solvent effects in the homogeneous catalytic reduction of
propionaldehyde with aluminium isopropoxide catalyst: New
insights from PFG NMR and NMR relaxation studies
Authors: Atika Muhammad, Graziano Di Carmine, Luke Forster, and
Carmine D'Agostino
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemPhysChem 10.1002/cphc.202000267
Link to VoR: https://doi.org/10.1002/cphc.202000267
0
1/2020

ChemPhysChem
10.1002/cphc.202000267
Solvent effects in the homogeneous catalytic reduction of propionaldehyde with
aluminium isopropoxide catalyst: New insights from PFG NMR and NMR relaxation
studies
a
a
a
a
Atika Muhammad , Graziano Di Carmine , Luke Forster , Carmine D’Agostino *
aDepartment of Chemical Engineering and Analytical Science, The University of Manchester,
The Mill, Sackville Street, Manchester, M13 9PL, UK
Academic title:
Ms Atika Muhammad (PhD student)
Dr Graziano Di Carmine (Research Associate)
Mr Luke Forster (PhD student)
Dr Carmine D’Agostino (Lecturer)
*
Corresponding Author:
Dr Carmine D’Agostino
Email: carmine.dagostino@manchester.ac.uk
ORCID ID: 0000-0003-3391-8320
Abstract
Solvent effects in homogeneous catalysis are known to affect catalytic activity. Whilst these
effects are often described using qualitative features, such as Kamlet-Taft parameters,
experimental tools able to quantify and reveal in more depth such effects have remained
unexplored. In this work, NMR diffusion and T relaxation measurements have been carried
1
out to probe solvent effects in the homogeneous catalytic reduction of propionaldehyde to 1-
propanol in the presence of aluminium isopropoxide catalyst. Using data on diffusion
coefficients it was possible to estimate trends in aggregation of different solvents. The results
show that solvents with a high hydrogen bonding accepting ability, such as ethers, tend to form
larger aggregates, which slow down the molecular dynamics of aldehyde molecules, as also
suggested by T measurements, and preventing their access to the catalytic sites, which results
1
in the observed decrease of catalytic activity. Conversely, weakly interacting solvents, such as
alkanes, do not lead to the formation of such aggregates, hence allowing easy access of the
aldehyde molecules to the catalytic sites, resulting in higher catalytic activity. The work
reported here is a clear example on how combining traditional catalyst screening in
homogeneous catalysis with NMR diffusion and relaxation time measurements can lead to new
physico-chemical insights into such systems by providing data able to quantify aggregation
phenomena and molecular dynamics.
Keywords: NMR diffusion, T
1
relaxation, solvent effects, homogeneous catalysis, aggregation
1
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
1
. Introduction
The use of solvents is ubiquitous in chemical processes, including homogeneous and
heterogeneous catalytic processes. The primary role of a solvent is for dissolving and/or
diluting the reactive species, as well as increase the rate of mass and heat transfer. However,
solvent properties can strongly influence catalysis and interfere with the catalytic reaction due
to their ability to interact with both reactants and the catalyst. In heterogeneous catalysis,
solvents may influence a variety of phenomena that can ultimately affect catalytic activity,
[
1]
[2]
[3, 4]
including solvatochromic parameters, proton transfer, mass transport by diffusion
and
[
5, 6]
poisoning of active sites due to competitive adsorption with the reactants.
For example,
[
1]
Marchi et al. reported that, in addition to solvent-catalyst interactions, solvent polarity and
its ability for hydrogen bonding formation with the reactants can significantly affect the rate of
catalytic hydrogenation in the liquid-phase occurring inside heterogeneous catalysts. Similar
effects can also be observed in homogeneous catalysis, whereby solvent properties such as
polarity, polarizability, hydrogen-bond donating ability (proticity) and hydrogen-bond
[
7]
accepting ability (basicity) can interfere with the catalytic reactions. In this context, Kamlet
[
8]
and Taft defined three parameters, π*, α and β, which are able to quantify polarizability,
hydrogen-bond donating ability and hydrogen-bond accepting ability of a solvent, respectively.
Using such parameters it has been possible to elucidate solvent effects in a variety of
[9]
homogeneously catalysed reactions. For example, Chase et al. studied the isomerization of
allyl alcohol to propionaldehyde using an iron pentacarbonyl catalyst. While solvent
polarizability and proticity had a minor influence on the reaction, solvents with a high
hydrogen-bond accepting ability (β) were seen to inhibit the catalytic activity. A plausible
hypothesis given by the authors is that such solvents are able to coordinate with the catalytic
species, hence inhibiting the binding of the reactant. In another study by Wynne et al.[ the
influence of solvent on enantioselectivity of the cyclopropanation of styrene and methyl
phenyldiazoacetate was studied. Their work showed that the enantioselectivity depends on both
steric and electronic effects. It was observed that the best enantioselectivity was obtained in
solvents that do not show evidence of binding to the catalyst. Macquarrie and co-workers also
10]
[
11]
studied the solvent dependence on catalytic Knoevenagel reactions. They also found that
low polarity solvents significantly promote the reaction compared to the polar ones; however,
the reason for such solvent effects remained largely unknown and was not further investigated.
It is worth noting that solvents may also affect reaction rate of homogeneously catalysed
reactions by transition state stabilization.[ Such transition states are difficult to detect but
12]
[9]
their existence is often supported by density functional theory (DFT) calculations.
Whilst the Kamlet and Taft parameters provide a good general and qualitative description for
solvent effects, experimental measurements able to reveal physico-chemical insights at a
molecular level in solvent-based homogeneous catalytic processes have remained overlooked.
Computational tools can be used to explain solvent effects in catalysis.[ Experimental tools
are however essential and, among the potential techniques for experimental studies of solvent
effects in homogeneous catalysis, nuclear magnetic resonance (NMR) measurements of
diffusion coefficients represent a powerful, non-invasive method for studying interactions in
homogeneously catalysed chemical reactions. Pioneering work from Pregosin[ has shown
that through suitable analysis of diffusion data obtained by pulsed-field-gradient (PFG) NMR
measurements it is possible to understand ion pairing and aggregation in reactions catalysed by
13]
14]
2
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
transition metal salts. This methodology was recently used to study the effect of temperature,
type of anion and presence of additives onto ion pairing and aggregation of imidazolium-based
[15, 16]
ionic liquids as well as in deep eutectic solvents.
Similar studies on polyaromatic
hydrocarbon aggregation, in particular on the effect of attached functional groups on
aggregation, have recently been reported.[ Additional insights into solvents effects onto
molecular dynamics of homogeneous catalytic processes can potentially be obtained by studies
of NMR relaxation times. In particular, it is well known that the spin-lattice relaxation time,
17]
T , can be considered a finger print of molecular dynamics of liquids and has been often used
1
[18]
to unravel physical and chemical properties of complex fluids such as ionic liquids. Yet,
according to our knowledge, the use of NMR relaxation measurements for rationalising solvent
effects in homogeneous catalysis has not been reported.
In this work we have used PFG NMR in conjunction with T NMR relaxation measurements
1
to unravel the solvent effects in homogeneous catalysis, choosing as a model reaction system
the reduction of propionaldehyde to 1-propanol though hydrogen transfer in the presence of 2-
propanol as a sacrificial agent and utilising an aluminium isopropoxide catalyst, Al(O-i-Pr) .
3
[
19]
This type of reaction is well-known
and hence serves as a good model system for the
proposed NMR approach. Experimental NMR results are compared with data on catalytic
activity in different solvents and with the Kamlet-Taft parameters of the solvent used in order
to understand and rationalise the catalytic behaviour.
2
2
. Experimental
.1 Materials and chemicals
Aluminium isopropoxide (98%), 2-propanol (anhydrous, 99.5%), n-hexane (anhydrous, >99%)
,4-dioxane (anhydrous, 99.8%), diethyl ether (>98%) were obtained from Sigma Aldrich, UK.
1
Toluene (99.5%) and propionaldehyde (extra pure, SLR) were obtained from Fisher Scientific
UK, while cyclohexane (99%) was purchased from Acros.
2
.2 Reaction studies
The reduction of propionaldehyde to 1-propanol in the presence of 2-propanol as the sacrificial
agent and aluminium isopropoxide as the catalyst, Al(O-i-Pr) , was used as a model reaction.
3
The catalyst was dissolved in different solvents, namely, n-hexane, cyclohexane, toluene, 1,4-
dioxane and diethyl ether. The catalyst solution was prepared by dissolving 0.25 g of the
catalyst in 5 ml of the solvent of interest. The experiment was then carried out by adding 2 ml
of the catalyst solution to 1.4 mmol of propionaldehyde and 6.6 mmol of 2-propanol. An excess
of 2-propanol was needed to shift the equilibrium towards the desired product (1-propanol).
1
The reaction was followed by acquiring H NMR spectra of the reacting solution at regular
time intervals of 1 h for a period of 4 h using a benchtop Magritek Spinsolve 43 MHz NMR
spectrometer. The feasibility of using NMR to follow such reactions has previously been
[
20]
reported.
2
.3 NMR diffusion and T relaxation measurements
1
NMR measurements were conducted using 5 mm NMR tubes in the Magritek Spinsolve 43
MHz NMR spectrometer. Pulsed-field gradient (PFG) NMR diffusion measurements were
carried out using the PGSTE pulse sequence[ with an observation time Δ = 50 ms, a gradient
21]
3
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
-1
strength of 164 mT m , and a gradient pulse duration in the range 3 – 5 ms, with rectangular
gradient pulses, for which the diffusion coefficient value, D, can be obtained by fitting the
experimental data according to the equation:[
22]
2
2 2
퐸 ⁄퐸 = exp[−퐷훾 훿 ꢀ (∆ − 훿⁄3)]
(1)
(푔)
(0)
where 퐸(푔) and 퐸(0) are NMR echo signal intensities in the presence and absence of the
gradient respectively, ꢀ is the gradient strength of pulse duration length δ, 훾 is the
1
gyromagnetic ratio of the studied nuclei ( H), Δ is the observation time, i.e., the time delay
between the midpoints of the gradients, and D is the diffusion coefficient of the examined
species.
NMR T
1
relaxation measurements were carried out using an inversion recovery pulse sequence
[
23]
technique. More details on the pulse sequence and data analysis can be found elsewhere.
3
. Results and discussion
Results of catalytic activity in different solvents are reported in Figure 1, which shows the
aldehyde conversion profile as a function of time in different solvents. Data on final turnover
number (TON) and turnover frequency (TOF) are reported in Table S1 of the Supporting
Information. The selectivity to 1-propanol was in all cases close to 100%. It is clear that the
solvent has a dramatic effect on the conversion, with reactions carried out in diethyl ether and
1
,4-dioxane being significantly slower compared to toluene, and with aliphatic solvents such
as n-hexane and cyclohexane being the best solvents for the reaction.
n-hexane
cyclohexane
toluene
1
,4-dioxane
diethyl ether
Time [h]
Figure 1. Conversion profile of propionaldehyde in different solvents.
In order to understand how the solvent may affect reactivity, we start by looking at the Kamlet-
Taft parameters of the different solvents, reported in Table 1, which were taken from the
[
24, 25]
literature.
From a comparison of those parameters with the catalytic data, it is clear that
4
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
푁
푇
solvents with a high hydrogen bond accepting ability, (β), and high polarizability, (퐸 ), are
those producing detrimental effects on catalytic activity, whereas non-polar, weakly-
interacting solvents such as alkanes lead to much higher reactivity. Similar findings have been
reported for other homogeneously catalysed reactions in the liquid-phase. This first observation
suggests that solvents with a high hydrogen bonding accepting ability are inhibiting catalyst
activity, possibly by a coordination with the metal centre of the aluminium catalyst, hence
blocking the access of the aldehyde reactant to the active metal centre. This explanation has
[
9]
been reported in other homogeneous catalytic processes using similar type of catalysts.
Solvation effects could also occur by producing solvation shells of solvent molecules around
the reactant species, hence preventing the reactant from accessing such active sites.[ It is
reasonable to assume that low polarity or aprotic solvents produce less solvation effects in the
26]
1
reaction medium compared to polar ones. Such effects have previously been observed by H
NMR NOESY studies of binary mixtures of organic compounds and have been related to
[
27]
hydrogen accepting ability.
Table 1. Kamlet-Taft parameters for the solvents used in this study.
푵
Solvent
n-hexane
cyclohexane
toluene
α
β
π*
푬
푻
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.11
0.37
0.47
-0.04
0.000
0.535
0.553
0.27
0.009
0.006
0.099
0.207
0.210
1,4-dioxane
diethyl ether
Whilst some of the parameters in Table 1 give a good qualitative description of the catalyst
activity, they do not give an in-depth description of the physico-chemical phenomena occurring
in the liquid-phase reaction system. In order to further understand the effect of solvents in those
reactions and probe solvation effects, we carried out calculations on molecular aggregations.
In particular, the NMR diffusion data obtained from PFG NMR experiments can be used to
estimate aggregation; this can be done by calculating the “apparent” hydrodynamic radius of
the solvent molecules, r
H
, using the Stokes-Einstein relation as a starting point, according to:
푘 푇
퐵
푟퐻 = 6
(2)
휋휂ꢁ
where k
B
is the Boltzmann constant, T the absolute temperature, η the dynamic viscosity and D
the self-diffusion coefficient from PFG NMR measurements. A main drawback of this equation
is that for small molecules, such as the case of those studied here, this equation leads to errors
in predictions up to approximately 40%, hence it is a poor guide to small molecule diffusion,
as also reported by Evans et al. in recent reports.[28, 29] The main assumption of the standard
Stokes-Einstein model is that the size of the diffusing molecules is much larger than the size
of the molecules of the medium in which they diffuse, that is, diffusion occurs in a continuum
[
30]
medium. Gierer and Wirtz
have derived a modified version of Equation (2), which
5
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
introduces a friction coefficient and removes the assumption of continuous medium.
Accordingly, a new equation can be obtained:
푘 푇 ꢃꢄ푆
1
ꢇ
퐵
푟퐻 = 6
ꢂ
+ 1ꢆ
ꢈ
(3)
푆
휋휂ꢁ 2ꢄ
ꢅ
ꢇ
ꢅ
where r
interest, hence r
s
is the radius of the solvent, which in our case is also the hydrodynamic radius of
= r . It is important to note that the choice of the diffusion model will
s
H
determine the calculated value of the hydrodynamic radius. Hence, absolute values of this
parameter may vary according to the model used; however, assuming the systems being
compared are similar, which is the case of our work, the trend of r will not be affected by the
H
choice of the diffusion model.
Figure 2 shows the PFG NMR log attenuation plots of the solvent signal (see Figure S1 of the
Supporting Information for typical PFG NMR spectra and peak assignment) in mixture with
the catalyst and the aldehyde using different solvents. The values of D can be calculated as the
negative of the slope of the plots in Figure 2 and are reported in Table S2 of the Supporting
Information.
2
2
2
-2
-9
γ δ g (Δ-δ/3) [s m ] × 10
0
.00 0E +00
1
1
5.0 00 E. 5+ 08
1.00 1E +09
1.5 10 .E5 +09
2.0 02 E+09
nn - -hhe ex xaa nne e
cc yy cc l loo hh ee x xaa nn ee
t too l luu ee nn ee
1
1, ,44 - -dd i oi ox xaa nn ee
dd i ei et ht hy yl l ee t thh ee rr
0
0..11
0
0. .00 11
Figure 2. PFG NMR log attenuation plots of the solvent signal for different solvents in solvent/catalyst/aldehyde
mixtures. Lines are fittings to Equation (1).
Using values of diffusion coefficients, the hydrodynamic radius of the diffusing species can be
calculated. The knowledge of the hydrodynamic radius of the diffusing species (see Table S3
of the Supporting Information for numerical values) can be used to quantify the aggregation
0
number, N, which can be calculated as ꢉ = 푉 /푉 , with 푉 being the hydrodynamic volume
H
H
H
0
H
[31, 32]
of the actual diffusing species in solution, and 푉 the volume of the single molecule.
If
we assume spherical aggregates, a reasonable assumption for relatively small molecules, 푉H
can be calculated as the volume of a sphere with the hydrodynamic radius, 푟 , derived from
H
6
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
0
H
Equation (3). As for the values of 푉 , this represents the molecular volume, that is, the volume
of a single molecule, which for the solvents under investigation was taken from literature values
[33]
reported by Gavezzotti,
who used data on molecular geometry and atomic radii for
estimating the molecular volume.
Values of N for the different solvents in the solvent/aldehyde/catalyst system are reported in
Figure 3 and compared with values of N calculated for the pure solvents and for solvent/catalyst
systems, that is, without the aldehyde reactant.
5.00
4.50
4.00
3.50
3.00
2.50
2.00
1.50
1.00
0.50
0.00
Pp uu r re es so ol vl ev en tn t
Ss oo l lvv ee n nt /t c/ ca at at lay lsyt st
Ss oo l lvv ee n nt /t c/ ca at at lay lsyt /sa t l/da el dh ye dh ey de
n-Hexane Cyclohexane Toluene
n-hexane cyclohexane
toluene
1,4-Dioxane Diethyl ether
1,4-dioxane diethyl ether
Figure 3. Aggregation number, N, for different solvents in different conditions.
The results clearly show that for pure solvents and solvents in the presence of the catalyst the
values of N are similar for n-hexane, cyclohexane, toluene and 1,4-dioxane and slightly higher
for diethyl ether and in most cases are close to 1, that is, such species diffuse as single
molecules. When the catalyst is added to the solvent, the N values remain similar to the values
reported for the pure solvent in most cases, except for diethyl ether, where an increase is
observed. When the aldehyde is further added to the systems, the N values for n-hexane and
cyclohexane do not change significantly compared to the other cases, whereas a small but
significant increase is observed for toluene and a very sharp increase is observed for 1,4-
dioxane and particularly for diethyl ether, with values of the aggregation number close to 5.
We would like to stress that whilst absolute values of N are affected by the choice of the
diffusion model, the trend across the range of solvents is largely independent of this and hence
a relative comparison of N values is appropriate. These results strongly suggest that solvents
such as 1,4-dioxane and diethyl ether have the tendency to form large aggregates when the
aldehyde is present in solution.
Based on such results and on the values of the Kamlet-Taft parameters reported in Table 1, we
suggest that solvents with a high hydrogen bond accepting ability, such as 1,4-dioxane and
diethyl ether, tend to form larger aggregates in the presence of aldehyde, which could result in
solvation shells around the aldehyde molecules. The data reported here provide a direct
quantification of such phenomena. Such explanation is further supported by T
1
NMR relaxation
measurements performed on the aldehyde species within the different solvents. Typical T
1
inversion recovery data are depicted in Figure 4a, which shows the spectral evolution of
propionaldehyde in mixture with the n-hexane solvent as a function of the time delay used in
7
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
the NMR inversion recovery pulse sequence (see full spectrum of the mixture in Figure S2 of
the Supporting Information). Despite the small concentration of propionaldehyde in the
mixture, it is still possible to clearly detect the carbonyl resonance of the aldehyde at
approximately 9.5 ppm, hence measure the inversion recovery rates, which are shown in Figure
4
b for all the solvents studied. The obtained values of T
1
are reported in Table 2 together with
the solvent aggregation number, N.
(a)
(b)
1
0
.5
0
n
n-hexane
cyclohexane
toluene
-0.5
1
,4-dioxane
diethyl ether
10 9 8 7 6 5 4 3 2 1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10
-1
Chemical shift [ppm]
0
10
20
30
40
50
60
70
80
Time [s]
Figure 4. (a) Typical T
The NMR peak at approximately 9.5 ppm is that of the carbonyl proton of the aldehyde. (b) T
rates for the aldehyde in different solvents. Lines are fittings to the inversion recovery model.
1
inversion recovery spectra for propionaldehyde in mixture with n-hexane as the solvent.
inversion recovery
1
2
3
Table 2. T
relative error on T
1
values for aldehyde in different solvents and solvent aggregation number, N, in the mixture. The
measurements is in the range 2 – 3 %.
1
Solvent
n-hexane
cyclohexane
toluene
T
1
aldehyde [s]
20.0 ± 0.5
15.9 ± 0.4
14.3 ± 0.3
13.5 ± 0.2
11.1 ± 0.2
N [-]
1.46
1.40
1.70
2.86
4.49
1,4-dioxane
diethyl ether
It is possible to observe that generally for the solvents whereby solvent aggregation becomes
more significant, that is, N increases, the T value of the aldehyde shows a decreasing trend.
Smaller values of T
in bulk liquids are ascribed to slower molecular dynamics;[34] hence, this
1
1
suggests the formation of larger aggregates surrounding the aldehyde molecules and slowing
down their molecular dynamics.
It is now of interest to compare the results on solvent aggregation to the aldehyde catalytic
conversion obtained in each solvent. Such results are presented in Figure 5, which shows that
as solvent aggregation decreases, the catalytic conversion increases.
8
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
n-hexane
1
00
9
8
7
6
5
4
3
2
1
0
0
0
0
0
0
0
0
0
0
cyclohexane
toluene
1
,4-dioxane
diethyl ether
0
.00
1.00
2.00
3.00
4.00
5.00
N [-]
Figure 5. Catalytic conversion of aldehyde against solvent aggregation number, N, within the
solvent/aldehyde/catalyst mixture.
In light of these results, it is therefore possible that solvents with a high hydrogen bond
accepting ability, such as 1,4-dioxane and diethyl ether, tend to form larger aggregates (higher
N values) surrounding aldehyde molecules and slowing down their molecular dynamics
(
decrease in T ), hence blocking access to the catalytic sites, which in turn leads to a decreased
1
reactivity. On the contrary, non-polar solvents such as alkanes, with no ability to accept
hydrogen bonds do not lead to the formation of such aggregates, hence having no detrimental
influence on the aldehyde access to catalytic sites and resulting in much higher catalytic
activity. This explanation ties in well with what has been previously suggested on solvation of
reactants increasing the free energy barrier during reaction, which may result in a slower
reaction rate and a lower conversion.[ The experimental results reported here support this
explanation and highlight the use of NMR relaxation and diffusion methods as useful tools to
study the behaviour of homogeneous catalytic processes.
13]
We would like to highlight that structuring phenomena in the liquid phase are likely to involve
also the solubilised homogeneous catalyst. However, probing such effects is challenging due
to the low concentration of catalyst used, that is, it might be difficult to perform meaningful
NMR measurements to probe the catalyst signal. Possible alternatives could be the use of
deuterated solvents and high-field NMR spectrometers or using model systems with higher
concentration of catalyst.
4
. Conclusions
Solvent effects in the homogeneous catalytic reduction of propionaldehyde by hydrogen
transfer in the presence of aluminium isopropoxide catalyst have been investigated by reaction
studies combined with PFG NMR diffusion and NMR relaxation measurements aiming at
understanding solvent aggregation and interactions. The results show that solvents with a high
hydrogen bond accepting ability lead to a significant decrease of catalytic activity, which is
attributed to their ability to coordinate with the catalytic sites, hence blocking access to
aldehyde molecules. In addition, calculations of aggregation numbers show a significant
9
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
aggregation in such solvents, which are thought to hinder the mobility and access of the
aldehyde reactant to the catalytic sites, as suggested by T relaxation measurements. The work
1
proposed here shows the usefulness of NMR diffusion and relaxation time measurements in
gaining a deeper understanding of solvent effects in homogeneous catalysis.
Acknowledgements
We would like to acknowledge the Petroleum Technology Development Fund (PDTF) for
funding this research work. Graziano Di Carmine would also like to acknowledge the EPSRC,
grant no. EP/S019138/1, for funding his research activities.
References
[
[
1] N. M. Bertero, A. F. Trasarti, C. R. Apesteguia, A. J. Marchi, Appl. Catal., A 2011, 394, 228-238.
2] B. S. Akpa, C. D'Agostino, L. F. Gladden, K. Hindle, H. Manyar, J. McGregor, R. Li, M. Neurock, N.
Sinha, E. H. Stitt, D. Weber, J. A. Zeitler, D. W. Rooney, J. Catal. 2012, 289, 30-41.
3] M. A. Isaacs, N. Robinson, B. Barbero, L. J. Durndell, J. C. Manayil, Christopher M. A. Parlett, C.
D'Agostino, K. Wilson, A. F. Lee, J. Mater. Chem. A 2019, 7, 11814-11825.
4] L. Gilbert, C. Mercier, Solvent effects in heterogeneous catalysis : Application to the synthesis of
[
[
fine chemicals, Vol. 78 (Eds.: M. Guisnet, J. Barbier, J. Barrault, C. Bouchoule, D. Duprez, G. Pérot, C.
Montassier), Elsevier, 1993, pp.51-66.
[
5] C. D'Agostino, G. L. Brett, P. J. Miedziak, D. W. Knight, G. J. Hutchings, L. F. Gladden, M. D. Mantle,
Chem. Eur. J. 2012, 18, 14426-14433.
[
[
[
[
[
[
6] C. D’Agostino, R. D. Armstrong, G. J. Hutchings, L. F. Gladden, ACS Catal. 2018, 8, 7334-7339.
7] P. J. Dyson, P. G. Jessop, Catal. Sci. Technol. 2016, 6, 3302-3316.
8] M. J. Kamlet, J. L. M. Abboud, M. H. Abraham, R. W. Taft, J. Org. Chem. 1983, 48, 2877-2887.
9] H. M. Chase, T. J. McDonough, K. R. Overly, C. M. Laperle, J. Phys. Org. Chem. 2013, 26, 322-326.
10] D. C. Wynne, M. M. Olmstead, P. G. Jessop, J. Am. Chem. Soc. 2000, 122, 7638-7647.
11] D. J. Macquarrie, J. H. Clark, A. Lambert, J. E. G. Mdoe, A. Priest, React. Funct. Polym. 1997, 35,
1
53-158.
[
[
[
[
12] C. Reichardt, Solvents and solvent effects in organic chemistry, Wiley, 2004.
13] J. J. Varghese, S. H. Mushrif, React. Chem. Eng. 2019, 4, 165-206.
14] P. S. Pregosin, Prog. Nucl. Magn. Reson. Spectrosc. 2006, 49, 261–288.
15] C. D'Agostino, M. D. Mantle, C. L. Mullan, C. Hardacre, L. F. Gladden, ChemPhysChem 2018, 19,
1
081-1088.
16] C. D'Agostino, L. F. Gladden, M. D. Mantle, A. P. Abbott, E. I. Ahmed, A. Y. M. Al-Murshedi, R. C.
Harris, Phys. Chem. Chem. Phys. 2015, 17, 15297-15304.
17] D. Simionesie, G. O’Callaghan, R. Laurent, J. A. Preece, R. Evans, Z. J. Zhang, Ind. Eng. Chem. Res.
019, 58, 20505-20515.
[
[
2
[
[
1
18] K. Hayamizu, S. Tsuzuki, S. Seki, J. Phys. Chem. A 2008, 112, 12027–12036.
19] R. Cohen, C. R. Graves, S. T. Nguyen, J. M. L. Martin, M. A. Ratner, J. Am. Chem. Soc. 2004, 126,
4796-14803.
[
[
[
[
[
[
[
[
20] Y. Wu, C. D'Agostino, D. J. Holland, L. F. Gladden, Chem. Commun. 2014, 50, 14137-14140.
21] J. E. Tanner, J. Chem. Phys. 1970, 52, 2523-2526.
22] D. Sinnaeve, Concepts Magn. Reson., Part A 2012, 40A, 39-65.
23] E. Fukushima, S. W. Roeder, Experimental pulse NMR, Addison-Weslkey, Reading, US, 1981.
24] A. de Juan, G. Fonrodona, E. Casassas, Trends Anal. Chem. 1997, 16, 52-62.
25] P. G. Jessop, D. A. Jessop, D. Fu, L. Phan, Green Chem. 2012, 14, 1245-1259.
26] J. R. Pliego, J. Phys. Chem. B 2009, 113, 505-510.
27] A. Bagno, J. Phys. Org. Chem. 2002, 15, 790-795.
1
0
This article is protected by copyright. All rights reserved.

ChemPhysChem
10.1002/cphc.202000267
[
28] R. Evans, Z. Deng, A. K. Rogerson, A. S. McLachlan, J. J. Richards, M. Nilsson, G. A. Morris, Angew.
Chem. Int. Ed. 2013, 52, 3199-3202.
[
[
[
[
29] R. Evans, Prog. Nucl. Magn. Reson. Spectrosc. (doi.org/10.1016/j.pnmrs.2019.11.002) 2019.
30] A. Gierer, K. Wirtz, Z. Naturforsch. A 1953, 8, 532-538.
31] A. Macchioni, G. Ciancaleoni, C. Zuccaccia, D. Zuccaccia, Chem. Soc. Rev. 2008, 37, 479-489.
32] L. Rocchigiani, G. Bellachioma, G. Ciancaleoni, S. Crocchianti, A. Laganà, C. Zuccaccia, D. Zuccaccia,
A. Macchioni, ChemPhysChem 2010, 11, 3243-3254.
[
[
33] A. Gavezzotti, J. Am. Chem. Soc. 1983, 105, 5220-5225.
34] N. Bloembergen, E. M. Purcell, R. V. Pound, Phys. Rev. 1948, 73, 679-746.
1
1
This article is protected by copyright. All rights reserved.