Adipic acid production via metal-free selective hydrogenolysis of biomass-derived tetrahydrofuran-2,5-dicarboxylic acid

Article abstract of DOI:10.1021/acscatal.7b01753

Biomass-derived furans offer sustainable routes to adipic acid (AA), a key chemical in Nylon-6,6 synthesis. In this work, we show that tetrahydrofuran-2,5-dicarboxylic acid (THFDCA) is a viable precursor for AA production, achieving up to 89% yield in a metal-free system containing HI and molecular H₂ in a propionic acid solvent at 160 °C. Reactivity studies demonstrate that the interplay between HI, H₂, and the solvent is essential for effective THFDCA ring opening. By measuring the reaction orders of HI and molecular H₂ and calculating an acid?base equilibrium constant in a nonaqueous solvent, we show that HI plays a multifaceted role in the reaction by acting both as a proton source and an iodide source to selectively cleave C?O bonds without overhydrogenation of carboxylic acid groups. Using reactivity studies, kinetic measurements, and first-principles computational insights, we demonstrate that metal-free activation of molecular H₂ plays a key role in the reaction, following HI-mediated cleavage of the etheric C?O bond in THFDCA.

Full text of DOI:10.1021/acscatal.7b01753

Subscriber access provided by UNIVERSITY OF ADELAIDE LIBRARIES
Article
Adipic Acid Production via Metal-Free Selective Hydrogenolysis
of Biomass-Derived Tetrahydrofuran-2,5-dicarboxylic Acid
Matthew J. Gilkey, Alexander V. Mironenko, Dionisios G. Vlachos, and Bingjun Xu
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.7b01753 • Publication Date (Web): 16 Aug 2017
Downloaded from http://pubs.acs.org on August 16, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street
N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 41
ACS Catalysis
1
2
3
4
5
6
7
Adipic Acid Production via Metal-Free Selective Hydrogenolysis of Biomass-
Derived Tetrahydrofuran-2,5-Dicarboxylic Acid
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Matthew J. Gilkey, Alexander V. Mironenko, Dionisios G. Vlachos*, Bingjun Xu*
Catalysis Center for Energy Innovation, Department of Chemical & Biomolecular Engineering,
University of Delaware, Newark, DE 19716
Emails: vlachos@udel.edu, bxu@udel.edu
Keywords: biomass conversion, hydrodeoxygenation, metalꢀfree catalysis, ring opening, adipic acid,
tetrahydrofuranꢀ2,5ꢀdicarboxylic acid, solvent effects
1
ACS Paragon Plus Environment

ACS Catalysis
Page 2 of 41
1
2
3
4
5
Abstract
Biomassꢀderived furans offer sustainable routes to adipic acid (AA), a key chemical in Nylon 6,6
synthesis. In this work, we show that tetrahydrofuranꢀ2,5ꢀdicarboxylic acid (THFDCA) is a viable
precursor for AA production, achieving up to 89% yield in a metalꢀfree system containing HI and
molecular H2 in a propionic acid solvent at 160 °C. Reactivity studies demonstrate that the interplay
between HI, H2, and the solvent is essential for effective THFDCA ring opening. By measuring the
reaction orders of HI and molecular H2 and calculating an acidꢀbase equilibrium constant in a nonꢀ
aqueous solvent, we show that HI plays a multifaceted role in the reaction by acting both as a proton
source and an iodide source to selectively cleave CꢀO bonds without overhydrogenation of carboxylic
acid groups. Using reactivity studies, kinetic measurements, and firstꢀprinciples computational insights,
we demonstrate that metalꢀfree activation of molecular H2 plays a key role in the reaction, following HIꢀ
mediated cleavage of the etheric CꢀO bond in THFDCA.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
ACS Paragon Plus Environment

Page 3 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
1. Introduction
Adipic acid (AA) is an essential commodity chemical used in the commercial production of
Nylon 66 and polyurethanes.1, AA has been produced almost exclusively from petroleumꢀderived
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
cyclohexane in an inefficient oxidation process which releases N2O, a potent greenhouse gas (Scheme 1).
In this process, cyclohexane is first oxidized to create KA oil, a mixture of cyclohexanone and
cyclohexanol, over Co catalysts (cobalt(II) naphthenate) using air as an oxidant. To maintain high
selectivity in this reaction (70 ꢀ 90%), the reaction is typically conducted at low conversions (3 to 8%),
which necessitates extensive feed recycling and inflated capital costs. AA is then obtained by oxidizing
KA oil under harsh conditions using nitric acid, with the production of undesired N2O.2 Tightening
regulations on greenhouse gas emissions make the development of renewable and environmentally benign
processes for AA production increasingly desirable.
Scheme 1. Commercial AA production from cyclohexane for use as a precursor for 6,6ꢀnylon.2
Biomassꢀderived 5ꢀhydroxymethylfurfural (HMF) has the potential to replace cyclohexane for
renewable AA production, owing to HMF’s structure similarity to AA.3 Furthermore, HMF production
from biomass has significantly matured with advances in both glucose isomerization to fructose and
fructose dehydration to HMF,4ꢀ6 thereby paving the way to key commodity chemicals, such as AA. The
conversion of HMF to AA requires the oxidation of terminal oxygenꢀcontaining groups into carboxyls
and the hydrogenolysis/hydrogenation of the furan ring via two possible pathways (Scheme 2): (1) ring
opening and hydrogenation of HMF to 1,6ꢀhexanediol (1,6ꢀHDL), followed by oxidation to AA, or (2)
oxidation of HMF to 2,5ꢀfurandicarboxylic acid (FDCA), followed by ring opening and hydrogenolysis to
3
ACS Paragon Plus Environment

ACS Catalysis
Page 4 of 41
1
2
AA. The former pathway poses selectivity challenges both in the ring opening step to 1,6ꢀHDL7ꢀ9 and in
the oxidation of 1,6ꢀHDL.10ꢀ12 Several authors have reported that metal/metal oxide catalysts, e.g., Irꢀ
ReOx or RhꢀReOx, can selectively convert 2,5ꢀ(bishydroxymethyl)tetrahydrofuran (BHTHF), a
hydrogenation product of HMF, to 1,2,6ꢀhexanetriol (1,2,6ꢀHTL) with high selectivity;9, 13 however,
removing the secondary alcohol in 1,2,6ꢀHTL to make 1,6ꢀHDL is a bottleneck to high conversion and
selectivity.9 Moreover, total oxidation of two terminal alcohol groups in 1,6ꢀHDL to carboxylic acid
groups remains challenging.10ꢀ12 In contrast, the oxidation of HMF to FDCA is quite facile, where yields
of >99% can be achieved over Au and Pt catalysts.14ꢀ18 While AA production through FDCA is a
promising route, significant work is needed to understand and perform selective FDCA ring opening, e.g.,
selectively cleaving CꢀO bonds within the ring, rather than hydrogenating the terminal carboxylic acid
groups.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Proposed pathway to AA from celluloseꢀ and glucoseꢀderived HMF, via (1) via ring opening followed
by oxidation (denoted by blue arrows), or (2) oxidation followed by ring opening (denoted by orange arrows).
HMF = 5ꢀhydroxymethylfurfural, BHMTHF = 2,5ꢀ(bishydroxymethyl)tetrahydrofuran, 1,6ꢀHDL = 1,6ꢀ
hexanediol, FDCA = 2,5ꢀfurandicarboxylic acid, THFDCA = tetrahydrofuranꢀ2,5ꢀdicarboxylic acid, AA = adipic
acid.
Boussie et al. proposed that conversion of FDCA can occur through sequential hydrogenation of
the furan ring to tetrahydrofuranꢀ2,5ꢀdicarboxylic acid (THFDCA) and ring opening and hydrogenolysis
of THFDCA to AA.19 In the first step, Boussie et al. conducted the hydrogenation from FDCA in acetic
acid solvent over noble metal catalysts, such as Pd, at high H2 pressure (710 psi). In the second step, they
performed the hydrogenolysis in a similar manner with the addition of 0.3 M hydriodic acid (HI) to a
solution THFDCA in acetic acid. An AA yield of >99% was claimed. In this reaction scheme, the
hydrogenation of FDCA to THFDCA is quite facile, occurring at temperatures as low as 40 °C over
Pd/Al2O3 in a toluene solvent; at such low temperatures, the carboxylic acid groups remained intact.20
4
ACS Paragon Plus Environment

Page 5 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
Despite reported high yields of AA from THFDCA, little is known about the underlying reaction
mechanism. In particular, the coexistence of HI, H2, and the organic acid solvent obscures their respective
roles. To this end, we seek molecular level understanding of the ring opening chemistry in order to
rationally design this process with less corrosive materials to effectively ringꢀopen THFDCA to AA.
Herein, we present a detailed experimental study, complemented with computational insights, on
the ring opening of THFDCA to AA. Reactivity studies show that no metal is needed to facilitate
hydrogenolysis of THFDCA in the presence of HI. Instead, HI and H2 in propionic acid solvent work
synergistically to cleave CꢀO bonds in the THF ring and form the aliphatic diacid, where both solvent
choice and proton source are critical. Detailed kinetic measurements show that both H2 and HI are key in
effective ring opening, where H2 and HI have ~1st and ~2nd order dependences on the AA formation rate.
By combining insights from theory, reactivity data, and kinetic insights using model compounds, we
propose that THFDCA ring opening proceeds by HIꢀmediated ring opening, followed by metalꢀfree
activation of molecular H2 to form AA.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2. Methods and Materials
2.1 Reactivity evaluations
Reactions were conducted in 50 mL stainless steel pressure vessels (Parr Instruments) equipped
with a stainless steel thermocouple and Teflonꢀcoated magnetic stir bar. A typical reaction mixture was
made by adding the 1 wt% of THFDCA (Carbosynth Ltd., 98%) to a 15 mL mixture of 0.3 M HI (57 wt%
in water, Sigma Aldrich) in propionic acid (99%, Sigma Aldrich). Propionic acid, rather than acetic acid,
was used as a solvent due to severe overlap between acetic acid and AA peaks in the high performance
liquid chromatography (HPLC). Due to the corrosive nature of the reaction mixture, glass inserts were
used in the reaction vessel as well as Teflonꢀcoated thermocouples. After collecting an initial sample, the
mixture was then added to the glass insert, placed in the reactor vessel, and sealed. The reactor headspace
was purged 3 times with N2 and 3 times with H2 before pressurizing to the desired reaction pressure. The
reactor was subsequently placed over a magnetic stirrer and heated to the desired temperature, regulated
5
ACS Paragon Plus Environment

ACS Catalysis
Page 6 of 41
1
2
3
4
5
6
7
8
9
by a PIDꢀcontrolled ceramic band heater. After the desired reaction time, the reaction was quenched by
placing the reactor in an ice bath. Once the reaction mixture was cooled, a sample was collected and
stored until analysis in both HPLC and HPLC with on line mass spectrometry (LCꢀMS). All chemicals
were used as received without further purification.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2.2 Analytical techniques
2.2.1 High performance liquid chromatography (HPLC) analysis
Liquid samples, diluted 1:10 with subꢀmicron filtered water (Thermo Fisher), were analyzed in
both HPLC (Alliance Waters System e2695) and LCꢀMS (Agilent 6120), both equipped with a BioꢀRad
Aminex HPXꢀ87H (300 mm × 7.8 mm) column. The HPLC method was optimized, using 0.50 mL/min of
mobile phase (5 mM H2SO4) with a column temperature of 50 °C. For LCꢀMS, a mobile phase of 1%
formic acid was used instead of 5 mM H2SO4 for compatibility with the mass spectrometer. Conversions
and yields were calculated using the following equations:
ꢀꢁ = ꢂ
ꢃ,ꢆꢇꢈꢉꢊ ∙ 100% (1)
ꢋ = ꢂꢍ,ꢆꢇꢈꢉꢊ ∙ 100%
(2)
ꢅꢂ
ꢃ,ꢄ
ꢌ
ꢂ
ꢂ
ꢃ,ꢄ
ꢃ,ꢄ
where Xi and Yj correspond to conversion of reactant i and yield of product j, respectively, Ci,0 is the initial
concentration of the reactant i (mol Lꢀ1), Ci,final is the final concentration of the reactant i (mol Lꢀ1), and
Cj,final is the final concentration of a product j (mol Lꢀ1). Concentrations were measured by integrating
peaks HPLC chromatograms which were calibrated based on prepared solutions of known concentration.
2.2.2 Inductively-coupled plasma mass spectrometry (ICP-MS)
Quantification of trace metals in the solution was conducted using an inductivelyꢀcoupled plasma
mass spectrometer (ICPꢀMS). Trace metal standards containing 100 µg/mL Au, Pt, Pd, Ir, Ru, Rh, Cr, Fe,
and Ni in 5% HCl, were obtained from HighꢀPurity Standards. These solutions were diluted in 2% HCl to
solutions of varying known concentrations for accurate calibration. ICPꢀMS samples of stock HI were
prepared by diluting stock HI in 2% HCl by a factor of 100. ICPꢀMS samples from postꢀreaction mixtures
were obtained by reacting 0.3 M HI (no THFDCA) in 15 mL propionic acid under 500 psi H2 pressure for
2 h at 160 °C. Because organic compounds are incompatible with the ICPꢀMS, the solvent was removed
by allowing the postꢀreaction solution to stir via magnetic stir bar over a hot plate set to ~100 °C in a
6
ACS Paragon Plus Environment

Page 7 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
ventilation hood. After about 4 h, the solvent was completely removed. The remaining solid was washed
with water and again left to evaporate at ~40 °C to ensure that all organic matter was removed; this was
repeated 3 times. After the final washing and evaporation step, 15 mL of water was added and stirred
vigorously to dissolve all the remaining solid. This mixture was then diluted 1:10 in 2% HCl and injected
into the ICPꢀMS for quantitative analysis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2.3 Scanning electron microscopy with X-ray microanalysis
Scanning electron microscopy (SEM) with Xꢀray microanalysis (SEMꢀEDX) was conducted on a
Joule 7400 instrument equipped with Inca Xꢀray microanalysis. Samples were prepared by evaporating
the solvent, vented to a ventilation hood. The solid was collected and placed on carbon tape prior to SEMꢀ
EDX analysis. EDX was conducted with an electron energy of 25 keV.
2.4 Kinetic modeling of tetrahydrofuran-2,5-dicarboxylic acid (to adipic acid) and malic
acid (to succinic acid) hydrogenolysis
Kinetic models were created to describe timeꢀdependent data of the ring opening of THFDCA to
AA as well as the hydrogenolysis of malic acid (MA), a model compound, to succinic acid (SA) in the
presence of HI and H2 in propionic acid solvent. In describing MA hydrogenolysis, we compared two
reaction networks: a twoꢀstep dehydrationꢀhydrogenation mechanism via fumaric acid (FA), referred to as
Model A, and a combination of dehydrationꢀhydrogenation and a direct CꢀO bond hydrogenolysis to form
SA in a single step (model B; Scheme S1). Both models contained a reversible esterification reaction
between MA and the propionic acid solvent to form 2ꢀ(1ꢀoxopropoxy)butanedioic acid (BEA). THFDCA
hydrogenolysis network, referred to as Model C, included three reactions: ring opening of THFDCA to 2ꢀ
hydroxyadipic acid (HAA), hydrogenolysis of HAA to AA, and esterification of HAA with the solvent to
form 2ꢀ(1ꢀoxopropoxy)hexanedioic acid (PEA; Scheme S1). No iodineꢀcontaining species were included
in models, since they were not observed in any timeꢀresolved experiment. Regression/estimation of rate
constants was conducted in MATLAB’s Optimization Toolbox using the lsqnonlin function. Full model
details, including derivations and equations, can be found in the Supplementary Information (Section S1).
7
ACS Paragon Plus Environment

ACS Catalysis
Page 8 of 41
1
2
3
4
2.5 Ab initio reaction energetics
Gibbs free energies of THFDCAꢀpertinent reactions and HI acid dissociation constants were
calculated using the Gaussian 09 software package21. Selected basis sets were obtained from the EMSL
Basis Set Library22, 23. Due to both neutral and charged species involved, we employed a composite
computational scheme inspired by Shields and coꢀworkers24, 25, in which gas phase energies and solvation
energies are calculated at different levels of theory. The method was proven to be successful for accurate
pKa estimates of acids.25 Solvation contributions to Gibbs free energies of acid deprotonation and
THFDCA ringꢀopening reactions were essential to explain the role of acid nature and the solvent mixture
composition on AA yields. All molecules were preꢀoptimized in the gas phase at the restricted Hartreeꢀ
Fock (RHF) level of theory, using the 3ꢀ21G basis set,26, 27 Vibrational frequencies were calculated using
density functional theory,28, 29 M06ꢀ2X exchangeꢀcorrelation functional,30 and the LANL2DZ basis set,31ꢀ
33 following the geometry optimization at the same level of theory. We employed the Def2ꢀTZVPD basis
set34 for accurate structure and electronic energy calculations, as it contains diffuse basis functions,
crucial for description of anions. Gibbs free energies of solvation were computed as an energy difference
between the molecule in the gas phase and in the implicit solvent, calculated at the RHF/6ꢀ311G35ꢀ39 level
of theory with geometry optimizations. No geometry optimization in the solvent was performed for
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
cationic species (H3O+, protonated THFDCA), due to convergence difficulties. The implicit solvent was
41
modeled using the conductorꢀlike polarizable continuum model (CPCM).40,
A solute cavity was
specified using the United Atom Topological Model applied on radii optimized for the HF/6ꢀ31G(d) level
of theory (UAHF). We employed the 6ꢀ311G instead of 6ꢀ31G(d) basis set, used in the original CPCM
parameterization,40 as the former was available for all chemical elements utilized in this study, including
iodine. We used acetic acid (HAc) as a surrogate for propionic acid, due to availability of HAc
experimental data42, 43 for method benchmarking, which is justified by the similar AA yields achieved in
HAc and propionic acid solvents. Coꢀsolvent (H2O) effects on the HAc dielectric constant are neglected
in this study, unless noted otherwise.
8
ACS Paragon Plus Environment

Page 9 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
In order to validate the composite method for our solvent and the basis set choice, in Table S1, we
compare the computed acetic acid selfꢀdissociation constant (K = [H2Ac+][ Acꢀ], corresponding to 2HAc
ꢀ H2Ac+ + Acꢀ) with its experimental value (1.2×10ꢀ19 at 105.7 °C42). The CPCMꢀUAHF/RHF/6ꢀ311G
method predicts the autoprotolysis constant to be 2.6×10ꢀ20, which is less a factor of 5 smaller than the
experimental value. For comparison, a commonly employed semiꢀempirical SMD implicit solvation
model44 at the M06ꢀ2X/Def2ꢀQZVPD level of theory predicts it to be 8 orders of magnitude lower
(6.2×10ꢀ27).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In order to calculate the HI acid dissociation constant in an organic acid (glacial acetic acid;
ꢎꢏꢐꢑ,ꢐꢒꢓ), we developed a computational scheme that requires the following properties as an input: HAc
experimental selfꢀdissociation constant ꢎꢏꢐꢒꢓ,ꢐꢒꢓ, HI and HAc experimental acid dissociation constants
in water (ꢎꢏꢐꢑ,ꢔ, ꢎꢏꢐꢒꢓ,ꢔ), and calculated solvation Gibbs free energies of HI, Iꢀ, HAc, and Acꢀ species.
The scheme benefits from cancellation of errors and eliminates H3O+ (as well as H2Ac+), known to be
poorly described by the implicit solvent model. In Scheme 3 we depict two thermodynamic cycles
outlined in Section S3. HI acid dissociation constants are calculated according to Equations (1a) and (1b);
the notation corresponds to Scheme 3.
Scheme 3.Thermodynamic cycles used to derive the HI pKa expression. (a) HI acid dissociation in water
(w). (b) HI dissociation in acetic acid (HAc). pK values are related to Gibbs free energies ∆ꢕ of the
corresponding processes as ꢎꢏ = ∆ꢕ/2.303ꢖꢗ, where R is the gas constant, T is temperature.
9
ACS Paragon Plus Environment

ACS Catalysis
Page 10 of 41
1
2
3
4
5
6
7
8
9
(3a)
(3b)
ꢎꢏꢐꢑ,ꢐꢒꢓ = ꢎꢏꢐꢒꢓ,ꢐꢒꢓ + ꢘꢎꢏꢐꢑ,ꢔ − ꢎꢏꢐꢒꢓ,ꢔꢙ + ∆ꢎꢏꢚꢛꢜꢝ,ꢑꢅ − ∆ꢎꢏꢚꢛꢜꢝ,ꢐꢑ − ∆ꢎꢏꢚꢛꢜꢝ,ꢒꢓꢅ
+ ∆ꢎꢏꢚꢛꢜꢝ,ꢐꢒꢓ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
∆ꢎꢏꢚꢛꢜꢝ,ꢁ = ꢎꢏꢚꢛꢜꢝ,ꢁ/ꢐꢒꢓ − ꢎꢏꢚꢛꢜꢝ,ꢁ/ꢔ
We benchmarked the pKa computational scheme against experimental pKa of HCl, HBr, H2SO4,
and HClO4 in glacial acetic acid; the data are reported in Table S2. Errors in predicted pKa values are less
than 1.1 pKa unit for HCl, HBr, and H2SO4. Success of the implicit solvation scheme indicates that the
previously hypothesized ion pair formation after acid dissociation in glacial acetic acid43 does not occur to
a significant extent.
To calculate the equilibrium constant of the H2Oꢀassisted HI dissociation in acetic acid containing
water impurity, we first obtain an estimate of the H3O+ solvation energy, known to be poorly described by
the implicit solvent model. We exploit the Br/I similarity and use the experimental HBr dissociation
constant in the HAc solvent with water impurity,43 together with a range of experimental and computed
properties (see Section S4, Tables S3 and S4). We find the Gibbs free energy of H3O+ solvation in acetic
acid to be equal to ꢀ113.84 kcal/mol; the corresponding value in water, obtained using a similar method, is
ꢀ114.06 kcal/mol, i.e., the model predicts similar H3O+ solvation energies in H2O and HAc+H2O impurity.
For comparison, the experimental H3O+ solvation energy in water is ꢀ103.4 kcal/mol.45 We note that two
most successful methods for pKa calculations (CPCM46 and COSMOꢀRS47) systematically overestimate
ion solvation energies, leading to H3O+ solvation energy overestimation, with a minor effect on pKa
results. We calculate the HI pKa in water to be equal to ꢀ9.70, in excellent agreement with the
experimental value (ꢀ10), validating this approach.
In order to calculate an equilibrium constant of H3O+ + THFDCA ꢀH2O + THFDCA·H+ over a
range of water concentrations, we utilize the experimentally measured dielectric constant dependence on
the HAc/H2O mixture composition (Figure S1),48 as an input to the CPCM model. Although absolute
10
ACS Paragon Plus Environment

Page 11 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
H3O+ (and probably THFDCA·H+) solvation energy cannot be accurately calculated by this method,
successful application of the Born equation to describe pKa variations in water/alcohol mixtures49
indicates that compositionꢀdependent acid dissociation trends can be well described by the implicit
solvent model (Section S5, Tables S5ꢀS7).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3. Results and Discussion
Table 1. Reactivity data from THFDCA at 160 °C
Conversion
Acid
Catalyst PH2 (psi)
Solvent
Time (h)
YAA (%)
(%)
1
2
Propionic
acid
0.3 M HI
0.3 M HI
None
Pd/C
None
None
None
None
None
None
None
None
None
None
500
500
500
None
500
500
500
500
500
500
500
500
2
2
2
2
2
2
2
2
2
2
2
2
99
89
82
0
Propionic
acid
99
0
3
Propionic
acid
None
4
Propionic
acid
0.3 M HI
0
0
5
0.3 M HI
Water
3
0
6
0.3 M HI
2ꢀpropanol
<1
24
32
16
11
12
24
<1
14
3
7
Propionic
acid
0.3 M HBr
0.3 M HCl
0.3 M H2SO4
0.3 M HClO4
8
Propionic
acid
9
Propionic
acid
<1
<1
<1
<1
10
Propionic
acid
11 0.3 M H2SO4 + 0.3
M LiI
Propionic
acid
12
0.3 M HClO4 +
0.3 M LiI
Propionic
acid
13 0.3 M H2SO4 + 0.3
M LiI + 1.8 M
H2O
Propionic
acid
None
500
2
9
<1
Conditions: 1 wt% THFDCA in 15 mL propionic acid.
11
ACS Paragon Plus Environment

ACS Catalysis
Page 12 of 41
1
2
3
4
5
3.1 Metal-free ring opening in organic acid solvents
6
7
Though metals or metal oxides are typically necessary to catalyze hydrogenolysis reactions,
8
9
effective ring opening of THFDCA to AA occurs facilely in the absence of a metal catalyst when HI and
H2 are employed in organic acid solvents. In the presence of Pd/C, 500 psi H2, and 0.3 M HI in propionic
acid, full conversion of THFDCA was achieved after reaction at 160 °C, forming ~82% yield of AA,
consistent with a recent patent.19 Surprisingly, full conversion and slightly higher AA yield (89%) was
reached in the absence of Pd/C under otherwise identical conditions, indicating that hydrogenation or
hydrogenolysis reactions occur in the presence of HI and H2 without Pdꢀmediated hydrogen activation
(Table 1, entries 1ꢀ2). Control experiments indicate that ring opening is only effective when all three of
HI, H2, and propionic acid are employed simultaneously. Removing one of the compounds results in
complete loss in activity at 160 °C (Table 1, entries 3ꢀ5), which raises several questions: (1) the role of
HI, (2) the nature of H2 activation, and (3) the role of the organic acid solvent. Below we address these
questions via systematic reactivity, computational, and kinetic modeling studies.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
3.2 Effect of hydriodic acid on THFDCA ring opening
3.2.1. Experimental assessment
HI plays a unique role in THFDCA ring opening, which likely stems from both HI’s strong
acidity and iodide’s direct participation in the reaction. To access the role of HI’s acidity, we
systematically explored the impact of alternative sources of proton and iodide on THFDCA conversion.
The effect of proton was investigated by replacing stock HI (pKa ~ ꢀ10; all pKa values are defined in an
aqueous solution unless noted otherwise, 57 wt% HI in water)50 with other strong acids, e.g., 70 wt%
HClO4 in water (pKa ~ ꢀ10) and 99.99 wt% H2SO4 (pKa ~ ꢀ3 for the first proton, ~2 for the second
proton),50 in the absence of iodide. Doing so resulted in a stark decrease (99% vs. 11% and 16%,
respectively) in conversion and negligible AA yield (Figure 1a) under identical conditions as compared
with HI, suggesting that the acid strength alone cannot explain the effectiveness of HI in the ring opening
of THFDCA. Using other hydrohalic acids, e.g., HBr (pKa ~ ꢀ9) and HCl (pKa ~ ꢀ7),50 in place of HI
12
ACS Paragon Plus Environment

Page 13 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
again resulted in a significant loss in activity when compared with HI (Figure 1a, 99% conversion vs.
24% and 32%, respectively), where the AA yield decreases in the sequence of HI > HBr > HCl (Figure
1a). This relationship parallels the acid strength of each of these acids, suggesting that protonation of
THFDCA could play a key role in ring opening. Additionally, the sequence also parallels the halides’
nucleophilicity in polar protic solvents, where iodides are the strongest nucleophiles.51 Increasing HI
concentration positively impacts the AA formation rate (Figure 1b). Furthermore, analysis of the initial
AA formation rate (at THFDCA conversions <15%, see Section S6 and Table S8) at 100 °C versus HI
concentration results in a slope of 1.6 ± 0.1, suggesting that the reaction order with respect to HI is close
to 2nd order. This is consistent with the reaction orders previously proposed for ether bond cleavage by
hydrogen halides,52, 53 where HI was used to cleave the CꢀO bond in various alkyl ethers to an alcohol and
an iodoalkane. No detectable conversion of THFDCA was observed in the absence of HI at 100 °C after 4
h, demonstrating that HI’s participation is critical.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Just as HI’s acid strength cannot fully explain its high activity, the iodide component alone cannot
account for HI’s performance. When a solution of 0.3 M lithium iodide (LiI) in propionic acid was used
as a substitute for 0.3 M HI in propionic acid at 160 °C, THFDCA remained unreactive, confirming the
importance of protons in driving the ring opening reaction. Adding a strong acid to a mixture of
THFDCA and LiI in propionic acid still cannot reproduce the conversion and yield obtained with stock
HI. This is evidenced by reactions with 0.3 M LiI and 0.3 M H2SO4 or HClO4 in propionic acid, forming
no appreciable AA (<1%) at <25% THFDCA conversion in both cases under otherwise identical
conditions (Table 1, entries 11 and 12). Adding iodide and proton components separately, e.g., H2SO4 and
LiI, may not be representative of adding HI, because the stock HI solution contains 43% water by weight.
Thus, a 0.3 M HI solution in propionic acid intrinsically results in a water concentration of 1.8 M in the
final reaction mixture. To test the role of water, 0.48 g water was added to a propionic acid mixture
containing 1 wt% THFDCA, 0.3 M LiI, and 0.3 M H2SO4 to mimic the water concentration when HI is
added; again, no appreciable AA formation was observed as compared to the case with HI (Table 1, entry
13). Although employing HBr in place of HI results in a decrease in AA yield (14% vs. 87%, Table 1),
13
ACS Paragon Plus Environment

ACS Catalysis
Page 14 of 41
1
2
3
4
5
6
7
8
9
the production of AA from HBr and HI suggests that the halide component is critical, where either the
nucleophilicity of the halide, the strong acidity, or both, could explain the performance of hydrohalic
acids over other mineral acids. These experiments suggest that a physical mixture of an iodide source and
a proton source (aside from the solvent) is not equivalent to HI. This could be due to the degree of
dissociation of mineral acids, which is dependent on the nature of the solvent (Leveling effect).54, 55
Computations, discussed next, shed light on the acid dissociation trend in an organic acid solvent.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Effect of acid at 160 °C after 2 h on the THFDCA conversion and AA yield (a) and effect of
varying HI concentration at 100 °C after 4 h on the AA production rate (b). Conditions: 1 wt% THFDCA
in 0.3 M HI under 500 psi H2.
3.2.2. Theoretical modeling of dissociation constants
To calculate HI’s acid dissociation constant in an organic acid (glacial acetic acid; ꢎꢏꢐꢑ,ꢐꢒꢓ), we
used the computational scheme outlined in the section 2.5. The calculated HI pKa value for the HI + HAc
ꢀ H2Ac+ + Iꢀ process is 4.65, indicating that HI is a weak acid, and H2Ac+ ions are unlikely to participate
in the THFDCA ringꢀopening chemistry due to their negligible concentrations.
14
ACS Paragon Plus Environment

Page 15 of 41
ACS Catalysis
1
2
3
4
As the HI stock solution contains ~7 mol% of water, we propose that HI dissociation is primarily
5
6
H2Oꢀassisted, corresponding to HI + H2O ꢀ H3O+ + Iꢀ
where all species are solvated by the organic acid.
,
7
8
9
The H2Oꢀassisted HI dissociation constant, defined as Ka = [H3O+][Iꢀ]/[HI][H2O], is calculated to be equal
to 440, or pKa = ꢀ2.6, at 298.15 K, significantly greater than the HAcꢀmediated dissociation constant,
reported above. For comparison, experimental H2Oꢀassisted Ka of HClO4, HBr, and H2SO4 in acetic acid
are 27.8, 14.1, and 4, respectively.43 Larger Ka’s in the presence of water are consistent with the
significant increase in conductances of HCl and H2SO4 in acetic acid solutions upon water addition,56 due
to their increased dissociation extent. We note, however, that the reported absolute values of calculated
and experimental Ka’s cannot be used to calculate the acidꢀdependent concentration of catalytic protons in
the THFDCA reaction system, since they correspond to much lower acid and water concentrations in HAc
(5·10ꢀ3 M and 0.067 M, respectively43). As to the relative magnitudes of Ka’s, the comparison is
complicated by ion pairs, potentially formed in solvents with low dielectric constants (acetic acid,
ꢞ =6.25; propionic acid, ꢞ =3.44), due to incomplete screening of electrostatic interactions. The
indicatorꢀbased Ka measurement method43 is not sensitive to ion pair formation, and thus the reported Ka’s
may not reflect the true availability of protons for catalysis. However, we ruled out ion pair formation in
H2SO4, HCl, and HBr solutions in glacial HAc by reproducing their conductanceꢀbased experimental pKa
values within 1 unit, as we assumed nonꢀinteracting ions in the implicit solvent model; the reported
dissociation constants (as well as the HI dissociation constant by analogy with HBr) most likely reflect
their true relative acid strengths. HClO4 dissociation constant is an exception: conductance studies
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ꢀ
indicated the formation of H2Ac+·ClO4 ion pairs in the HClO4/HAc solution,56 consistent with inability of
the implicit solvent method to reproduce the HClO4 dissociation constant in glacial HAc (Table S2). In
addition, ion pair formation leads to only a weak increase in the HClO4/HAc solution conductivity upon
water addition,56 i.e., H2O does not promote HClO4 dissociation as efficiently as it does for other acids.
We conclude that in the THFDCA reaction system, acid dissociation constants follow the trend
HI>HBr>H2SO4>HClO4; a similar trend holds for their corresponding activity in AA production, even
when LiI is present. The correspondence between acid strengths and AA yields is consistent with the
15
ACS Paragon Plus Environment

ACS Catalysis
Page 16 of 41
1
2
3
4
5
6
Brønsted acidꢀcatalyzed ringꢀopening mechanism driven by HI dissociation, in which catalytically active
protons are bound to water molecules.
7
8
9
3.4 Effect of hydrogen and THFDCA on ring opening activity
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
THFDCA conversion and AA yield grow linearly with H2 pressure in the presence of 0.3 M HI in
propionic acid solvent (Figure 2a), suggesting that hydrogen plays an important part in AA’s formation.
When no hydrogen is used (500 psi N2), THFDCA remains unreacted in the temperature range of 100 to
160 °C after 4 h of reaction. At 115 °C, the formation rate of AA increases linearly with increasing H2
pressure in the range of 50 to 500 psi at low THFDCA conversions (<20%, see Section S6 and Tables S9
and S10). Plotting ln(rAA) versus ln(PH2) results in a roughly linear relationship with a slope of ~0.9 ± 0.1
(Figure 2a), which is close to 1st order with respect to H2. By a similar procedure, a firstꢀorder dependence
in THFDCA concentration is observed (a slope of 0.9 ± 0.1, Figure 2b). The strong correlation between
AA yield and H2 pressure in the presence of HI (Figure 2a) suggests that H2 actively participates in the
reaction in the absence of any metal catalyst. To confirm that hydrogen is activated in the absence of
metal catalysts, two possibilities need to be ruled out: (1) trace metal contaminations, introduced in situ or
from commercially available chemicals, dissociate molecular H2 to perform the reaction or (2) HI
mediates the necessary hydrogenolysis reactions by assisting in H2 activation.
16
ACS Paragon Plus Environment

Page 17 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. Effect of each reactant on AA formation rate: (a) effect of H2 (50, 100, 200, 300, and 500 psi)
using 1 wt% THFDCA in propionic acid and (b) effect of THFDCA concentration (0.1, 0.5, 1, 2 wt%)
under 500 psi H2. Conditions: THFDCA in a mixture of 0.2 M HI in propionic acid at 115 °C, t = 4 h.
Elemental analysis of stock HI via ICPꢀMS revealed no significant transition metal species. This
is evidenced by comparing a diluted stock HI mixture from SigmaꢀAldrich (0.1%) with a high purity
standard, containing the most common transition metals capable of activating molecular H2, including Pt,
Pd, Ir, Ru, Rh, Ni, Cu, Fe, and Cr. For each element tested, levels remained below the detection limit of
the instrument, thereby ruling out the possibility that stock HI directly introduces metal contaminants to
the reaction mixture.
Elemental analysis of postꢀreaction mixtures showed evidence of the metal species, such as Fe,
Cr, and Ni, the presence of which could facilitate hydrogen activation. Samples were prepared by
removing the solvent from a postꢀreaction mixture of THFDCA, 0.3 M HI, and 500 psi H2 in propionic
acid at 160 °C for 2 h. Removal of solvent was conducted via controlled evaporation at 100 °C until only
solid species remained. Then, the resulting solids were dissolved in pure H2O, diluted, and analyzed by
ICPꢀMS. Elemental analysis of solids in SEMꢀEDX revealed that both Fe and Cr were present in the dried
17
ACS Paragon Plus Environment

ACS Catalysis
Page 18 of 41
1
2
3
4
5
6
7
8
9
sample with a mass ratio of ~4:1. A similar ratio was observed by ICPꢀMS, where Fe and Cr was detected
at concentrations of 255 ppm and 63 ppm Cr, respectively (ratio ~4:1). Ni was also detected via ICPꢀMS
with a concentration of 202 ppm, where Cu, Mn, and Co levels remained < 20 ppm. The Fe:Cr ratio
strongly suggests that these metal species originate from the stainless steel of the inner walls of the
reaction vessel. Although glass inserts were used in the stainless steel reactors, vaporꢀphase HI and/or
propionic acid likely condense onto the inner walls of the top of the reactor due to their volatility and
leach metals into the reaction mixture. This provides a source of metal contamination in the reaction
system, which could provide metal sites for H2 activation. To test this possibility, mixtures containing
~0.05 M Fe, Ni and Cr salts (major metal components detected by ICPꢀMS), e.g., FeBr3, Ni(NO3)2, and
CrCl3·6H2O, and 1 wt% THFDCA in propionic acid were employed in reactions under 500 psi H2 (no
iodide source). We used metal salt concentrations significantly larger (0.05 M) than the concentration of
Fe detected by ICPꢀMS (255 ppm or ~0.005 M) to verify whether these metals play an active role in the
chemistry. After reaction with no iodide source, minimal activity was observed at 115 °C after 4 h
(Figure 3, red bars). However, when 0.3 M LiI was added to CrCl3·6H2O, a significant increase in AA
formation rate was observed as compared to that when no iodide was used (Figure 3, green bars). Iron
salts, e.g., FeBr3, do not show enhanced activity when iodide was added (Figure 3, green bars).
Furthermore, increasing the CrCl3·6H2O concentration from 0.05 to 0.15 M, results in increased AA
formation rate, suggesting that CrCl3 promotes the ring opening chemistry in the presence of LiI. Despite
the promising synergy displayed by the combination of Cr and iodide, the observed activity cannot match
that when 0.3 M HI is used. Insignificant role of Cr, Fe, and Ni in hydrogen activation is further
supported by experiments, where CrCl3·6H2O, FeBr3, or CrCl3 were added to reaction mixtures
containing 0.3 M HI (blue bars, Figure 3), in which no detectable increase in AA rate was achieved within
the experimental error. To confirm that it is not synergy between Cr3+ and proton, rather than iodide,
CrCl3·6H2O was added to 0.3 M H2SO4 resulting in only minimal AA yield (Figure 3, orange bars). In
summary, our control experiments show that the metal contaminants are unlikely to play any major role in
H2 activation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
18
ACS Paragon Plus Environment

Page 19 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. AA formation rate in the presence of Cr, Ni, and Fe salts alone (red bars), 0.3 M HI (blue
bars), 0.3 M LiI (green bars), and 0.3 M H2SO4 (orange bars). Three concentrations (200, 500, and 800
mg) of CrCl3·6H2O were plotted from bottom to top in the presence of LiI. Conditions: 1 wt% THFDCA
in 15 mL propionic acid at 115 °C for 4 h, msalt = 200 mg which corresponds to ~0.05 M of the metal.
3.4 Impact of solvent on the cleavage of THFDCA’s ether bond
3.4.1 Experimental assessment
The solvent plays a profound role in THFDCA ring opening. Among the tested solvents, only
organic acid solvents, i.e., acetic acid and propionic acid, are viable candidates for ring opening of
THFDCA. Water or 2ꢀpropanol are poor solvents, each resulting in <1% AA yield under otherwise
identical conditions, in stark contrast to the >80% AA yields obtained in organic acids (Figure 4a). To
understand the role of water in the reaction, the concentration of water in propionic acid was
systematically varied in the reactant mixture. Because HI is employed in its aqueous solution (57% HI by
weight), adding HI inherently introduces water to the reaction mixture. Water concentration in the
reaction mixture is captured by the bottom axis in Figure 4b, and the volume of water added in excess to
that from HI is shown in the top axis. The water concentration from the stock HI solution is 1.1 M. As the
water concentration varies between ~1.1 to 2.7 M (0 to 0.4 mL of added H2O), the THFDCA conversion
19
ACS Paragon Plus Environment

ACS Catalysis
Page 20 of 41
1
2
3
4
5
6
7
8
9
and the AA yield are not significantly affected (~39ꢀ45% and 30ꢀ36%, respectively). This indicates that
the results from the HI concentration dependence experiments (Figure 1b) are unlikely influenced by the
varying water concentration in any significant manner. However, further increasing water concentration
to 8.5 M resulted in a significant decrease in THFDCA conversion and AA yield. Once water
concentration exceeds 8.5 M, AA formation becomes too slow to observe over the 4 h reaction time.
Thus, at high concentrations, water plays a strongly inhibitive role in ring opening.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Solvent effects on AA yield: (a) conversion and yield in each solvent (0.3 M HI) at 160 °C for
2 h; (b) effect of water concentration in propionic acid at 115 °C for 4 h (0.2 M HI). Conditions: 1 wt%
THFDCA under 500 psi H2.
20
ACS Paragon Plus Environment

Page 21 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
Water’s ability to suppress AA production is likely associated with interference of the H2O (the
strongest base in the system) with the acidꢀcatalyzed THFDCA ringꢀopening mechanism. HI has been
shown to catalyze the cleavage of ether bonds effectively by addition of proton and iodide across the CꢀO
bond to form an iodoalkane and an alcohol, e.g., converting diethyl ether to iodoethane and ethanol
(Scheme 4a and 4b).52, 53 The ether bond breaking proceeds by protonation of the oxygen atom to form an
oxonium cation (Scheme 4c),53 followed by halide addition and release of an alcohol fragment via SN1 or
SN2ꢀtype reactions (Scheme 4c, pathways (1) and (2), respectively), with SN2 being the dominant one for
ethers with adjacent primary or secondary alkyl groups.53 At high HI concentrations the alcohol product
undergoes subsequent conversion to an iodoalkane by nucleophilic substitution (Scheme 4b).57 A similar
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 4. Ether bond scission mechanisms showing (a) general ether bond scission by HI, (b) diethyl ether bond
scission by HI, (c) SN1 (1) vs. SN2 (2) pathways for ether bond scission of 2ꢀmethoxypropane, (d) ether bond
scission of THFDCA by HI, (e) inhibition of ring opening via water’s basicity. THFDCA = tetrahydrofuranꢀ2,5ꢀ
dicarboxylic acid, IHA = 2ꢀiodoꢀ5ꢀhydroxyhexanedioic acid.
21
ACS Paragon Plus Environment

ACS Catalysis
Page 22 of 41
1
2
3
4
5
6
7
8
oxoniumꢀmediated CꢀO bond breaking mechanism in THFDCA is consistent with the acid strength
dependence of the AA formation rate (Figure 1a). THFDCA protonation leads to oxonium, referred to as
THFDCA·H+, that is likely to be strongly acidic, with pKa values similar to that of dimethyl ether (ꢀ3.8 in
9
58
aqueous H2SO4 ) or tetrahydrofuran (ꢀ2.1).27 Subsequent iodideꢀmediated ring opening is proposed to
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
form 2ꢀiodoꢀ5ꢀhydroxyhexanedioic acid (IHA), which is not detected in the products, likely due to the
fast subsequent hydrogenolysis step of the CꢀI bond to form HAA (vide infra; Scheme 4d). Due to strong
THFDCA·H+ acidity, at high water concentration protonation of THFDCA could be suppressed in favor
of hydronium formation, because water acts as the strongest base (Scheme 4e). Below, we quantify the
relative basicity of THFDCA and water in an organic acid solvent using first principles calculations.
3.4.2. Theoretical insights into water’s role in ring opening of THFDCA
By employing the pKa difference between protonated dimethyl ether and hydronium ion in water
as a reference (∆ꢎꢏꢟ = 2.1, calculated from dimethyl ether pKa = ꢀ3.858 and hydronium pKa = ꢀ1.7), we
compute the THFDCA protonation reaction (H3O+ + THFDCA ꢀH2O + THFDCA⋅H+) to be exoergonic
22
Figure 5. Experimental adipic acid yield and predicted protonated THFDCA molar fraction as a function of
ACS Paragon Plus Environment
water molar fraction. Highlighted green and blue areas indicate ranges of H2O molar fractions that
correspond to exoergonic and endoergonic THFDCA protonation, respectively.

Page 23 of 41
ACS Catalysis
1
2
o
3
4
(∆ꢕ = −1.04 kcal/mol at 25 C) at ꢠꢐ ꢢ = 0.07, becoming endoergonic above ꢠꢐ ꢢ~0.2, with ∆ꢕ
ꢡ
ꢡ
5
6
reaching 1.47 kcal/mol at ꢠꢐ ꢢ = 0.42 (see Section S5 for details). The combined effect of increasing
ꢡ
7
8
9
H2O concentration and the increasing solvent dielectric constant leads to the equilibrium concentration of
ꢅꢣ.ꢤꢥꢦ
, similar to the AA yield decay at ꢠꢐ ꢢ > 0.16 (~ꢠꢐꢅꢣꢢ.ꢥꢥꢧ
,
ꢡ
ꢡ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
the protonated THFDCA that decays as ꢠ
ꢐ ꢢ
ꢡ
Figure 5). Since the THFDCA ringꢀopening rate is expected to be first order in protonated THFDCA, we
propose that the large negative reaction order in water for AA production for ꢠꢐ ꢢ>0.20 is consistent with
ꢡ
THFDCA ringꢀopening becoming the rateꢀlimiting step. In other words, the presence of excess water
hinders furan ringꢀopening because of thermodynamics (due to LeChatelier’s principle along with a more
subtle change in dielectric constant). The plateau in the AA yield and a nearꢀ1stꢀorder reaction rate in H2
at low H2O concentrations (Figure 2b and 5) coincide with the change in the sign of the free energy and
indicate that ringꢀopening is not limiting the overall reaction rate at low water fractions.
3.5 Kinetic insights into C-OH bond scission by HI and H2
Time dependent data from THFDCA in 0.2 M HI at 100 °C reveals one critical ringꢀopened
intermediate (Figure 6a) – 2ꢀ(1ꢀoxypropoxy)ꢀhexanedioic acid (PEA), an ester formed between propionic
acid and 2ꢀhydroxyhexanedioic acid (HAA, Figure 6b). This species was identified by LCMS, where
masses of 217 through 218 amu were observed in the negative scan (Figure S2a). To verify the
esterification pathway, experiments were conducted using acetic acid instead of propionic acid, where 2ꢀ
acetoxyhexanedioic acid (AE) was expected to form. AE was indeed detected by observation of masses of
202 through 203 amu (Figure S2b). The existence of an ester as an intermediate suggests that upon ring
opening of THFDCA, esterification of the OH group with the organic acid solvent occurs. The doubly
esterification compound, i.e., 2,4ꢀdi(1ꢀoxoalkoxy)ꢀhexanedioic acid, was not detected, suggesting that
2,4ꢀdihydroxyhexanedioic acid is not a major intermediate. Thus, the ring opening of THFDCA is
unlikely to proceed via the hydrolysis of the protonated THFDCA. In Scheme 4d, we propose that
THFDCA ring opening occurs by HI addition across the etheric CꢀO bond to form an iodineꢀ and
hydroxylꢀcontaining species (IHA). In the timeꢀevolved series, however, neither IHA nor other iodineꢀ
23
ACS Paragon Plus Environment

ACS Catalysis
Page 24 of 41
1
2
3
4
5
6
7
8
9
containing species is observed, suggesting that these species are easily converted under reaction
conditions. IHA likely undergoes facile CꢀI bond hydrogenolysis to HAA in the presence of HI, H2, and
the organic acid solvent, and a high concentration of propionic acid as a solvent accelerates esterification.
The final hydrogenolysis step to convert HAA/PEA to AA, however, cannot be elucidated from the timeꢀ
dependent THFDCA data alone.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Timeꢀdependent data of THFDCA ring opening, where only 2ꢀ(1ꢀoxopropoxy)ꢀhexanedioic acid
(PEA) is observed as an intermediate. Conditions: 1 wt% THFDCA in 0.2 M HI and propionic acid at 100 °C
under 500 psi H2. Solid traces represent the best model fit, taking into account all observed products and one
undetected intermediate, 2ꢀhydroxyhexanedioic acid. THFDCA = tetrahydrofuranꢀ2,5ꢀdicarboxylic acid, HAA
= 2ꢀhydroxyhexanedioic acid, PEA = 2ꢀ(1ꢀoxopropoxy)hexanedioic acid, AA = adipic acid.
Table 2. Reactivity data from malic acid (MA) or fumaric acid (FA) at 160 °C in propionic acid to
succinic acid (SA).
24
ACS Paragon Plus Environment

Page 25 of 41
ACS Catalysis
1
2
3
4
Reactant
CHI
PH2 (psi)
Conversion
YSA
YFA
5
6
7
8
9
1
2
3
4
5
6
MA
MA
MA
FA
0.3 M
0.3 M
0 M
500
0
96%
84%
56%
99%
99%
10%
57%
46%
0%
0%
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
10%
500
500
0
21%
0.3 M
0.3 M
0 M
99%
99%
0%
ꢀ
ꢀ
ꢀ
FA
FA
500
Conditions: 1 wt% reactant in 15 mL propionic acid with or without HI under 500 psi N2 or H2 at 160 °C
for 2 h. MA = malic acid, FA = fumaric acid, SA = succinic acid.
To understand the pathway for CꢀO bond scission in HAA and PEA to yield AA, malic acid
(MA) was employed as a model compound, due to its structural similarity to HAA and commercial
availability, to probe the interplay between H2 and HI in removing the hydroxyl/ester group. There are
three potential reaction pathways (Scheme 5): (1) dehydration of the CꢀOH group to fumaric acid (FA),
(2) hydrogenolysis of the CꢀO bond to succinic acid (SA), and (3) esterification of the hydroxyl group to
2ꢀ(1ꢀoxopropoxy)ꢀbutanedioic acid (PBA). Moreover, FA could undergo hydrogenation to SA (Scheme 5,
pathway 4). MA hydrogenolysis at 160 °C in propionic acid (0.3 M HI, 500 psi H2) occurred with >99%
conversion and ~90% selectivity to SA after 2 h of reaction (Table 2, entry 1). When H2 was replaced
with 500 psi N2 under otherwise identical conditions, MA conversion dropped to 84%, and both SA and
FA were observed with yields of 46% and 10%, respectively (Table 2, entry 2). This suggests that in the
absence of H2, several pathways are at play, including protonꢀmediated dehydration from MA to FA
either via protons from HI or propionic acid. The ability of propionic acid to perform dehydration
reactions to FA was confirmed by a control experiment where no HI was added to the solution, and 21%
yield to FA was observed after using MA as a starting material under 500 psi H2 (Table 2, entry 3). No
25
ACS Paragon Plus Environment

ACS Catalysis
Page 26 of 41
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 5. Possible pathways from malic acid (MA) including (1) dehydration to fumaric acid (FA), (2)
hydrogenolysis to succinic acid (SA), and (3) esterification with the solvent to 2ꢀ(1ꢀoxopropoxy)ꢀbutanedioic
acid (BEA).
SA was observed in the absence of HI, suggesting that H2 cannot participate in the reaction without HI.
The formation of SA in the absence of H2 indicates that HI alone is able to either (1) hydrogenate FA or
(2) perform direct hydrogenolysis of MA. To test the first possibility, control experiments were performed
using FA as a starting material. After reaction at 160 °C with 0.3 M HI, quantitative yield of SA from FA
was seen in both the presence and absence of H2 (Table 2, entries 4 and 5). FA is unreactive without HI
(Table 2, entry 6), providing further evidence of H2’s inability to directly participate in the reaction
without HI. The formation of SA from FA in the presence of HI suggests that HI can indeed facilitate
hydrogenation of the C=C bond, via one of two pathways: (1) HI is added across the C=C bond to make
2ꢀiodobutanedioic acid followed by a displacement reaction with HI to form SA and I2; and (2) HI first
decomposes into I2 and H2, which is then used to hydrogenate the C=C bond.59 In contrast, HI cannot
drive the cascade reaction from THFDCA to AA in the absence of H2. Although H2 is not required for
MA hydrogenolysis, enhanced conversion of MA was observed when HI and external H2 were coꢀfed in
comparison with that when only HI was employed (see below). This, in combination with the THFDCA
conversion in the presence of both HI and H2, lead to the hypothesis that H2 participates in the reaction,
possibly through HIꢀmediated activation.
26
ACS Paragon Plus Environment

Page 27 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. Time dependent data in MA conversion in propionic acid at 500 psi N2 and 0. 2 M HI (a), in FA
conversion in propionic acid at 500 psi H2 and 0.2 M HI (b), and MA conversion in propionic acid at 500 psi H2
and 0.2 M HI (c). The dotted line in (c) denotes the use of model A (k2A derived from data in panel a followed by
quasiꢀsteady state approximation for C=C hydrogenation), showing a poor fit to the experimental data.
Conditions: 1 wt% MA or FA in 15 mL mixture of 0.2 M HI in propionic acid at 100 °C.
Timeꢀresolved MA reduction at 100 °C suggests that direct CꢀO bond scission occurs only when
both H2 and HI are coꢀfed. At 100 °C in 0.2 M HI and 500 psi N2, only dehydration to FA occurs over the
course of 6 h, where the majority of MA is converted to 2ꢀ(1ꢀoxopropoxy)ꢀbutanedioic acid (BEA) via
reversible esterification with solvent (Figure 7a). At 160 °C (t = 2 h) under otherwise identical
conditions, SA was observed with 46% yield from MA (Table 2). SA was not observed from MA at 100
°C, suggesting that hydrogenation of FA in the presence of HI alone is slow. However, when FA was
used as a starting material in the presence of HI and H2 at 100 °C, FA was quickly converted to SA in the
first 1 h of reaction (~99% SA yield, Figure 7b). Thus, the addition of molecular H2 to HI in propionic
acid significantly enhances hydrogenation compared to the case where only N2 was present. This is
further evidenced by timeꢀevolved data when MA was used as a starting material while employing 500
psi H2 in place of N2. In this case, FA was not observed as a reaction intermediate (Figure 7c). Instead,
SA accumulated at a faster rate than did FA under 500 psi N2. Below, we use the timeꢀevolved data in
Figure 7 to estimate rate constants and provide quantitative assessment of reaction pathways involved in
CꢀOH bond hydrogenolysis.
27
ACS Paragon Plus Environment

ACS Catalysis
Page 28 of 41
1
2
3
4
5
6
7
8
9
Kinetic analysis of timeꢀevolved results of MA and FA hydrogenolysis/hydrogenation suggests
that direct hydrogenolysis is the dominant mechanism in the removal of CꢀOH bonds by the combination
of HI and H2, although dehydration and hydrogenation reactions can still occur. As explained above, two
mechanisms could explain the conversion of MA to SA: (1) dehydrationꢀhydrogenation and/or (2) direct
hydrogenolysis (Scheme 5). To determine relative contributions of the two possible reaction pathways,
rate expressions were derived (see Supplementary Information) by incorporating experimentally
determined reaction order dependences with respect to all reactants and assuming equilibrium between
MA and BEA (Scheme 6). Additionally, the models do not include halogenation/dehalogenation reactions
because Iꢀcontaining species are not observed and are instead lumped into apparent rate constants. Using
nonlinear least squares regression (see details in Section 2.4) for MA reduction in the absence of H2
(Figure 7a), the rate constant for dehydration (k2A) was determined to be 0.052 L molꢀ1 hꢀ1 with r2A equal
to 6.50 × 10ꢀ4 mol Lꢀ1 hꢀ1, where r3A is 0 because SA formation is too slow without H2 at 100 °C. In
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
contrast, in the presence of both HI and H2 at 100 °C, FA hydrogenation proceeds too fast to accurately
determine a rate constant at different conversions, which is evidenced by the 75% conversion of FA at t =
0 h (~10 minute heatꢀup time, Figure 7b). Thus, in the presence of both HI and H2, it is reasonable to
apply quasiꢀsteady state approximation with FA being a reactive intermediate, i.e., cFA ~ 0 and rFA ~ 0,
when MA is used as a starting material, because r3A >> r2A. This is consistent with the timeꢀevolved data
from MA in the presence of HI and H2, where no FA was observed. If the dehydrationꢀhydrogenation
pathway is the only mechanism through which MA is converted to SA, employment of k2A as the
dehydration rate constant along with the use of a quasiꢀsteady state approximation for FA formation rate
should accurately simulate the SA yield over time in the presence of HI and H2 (Figure 7c). However,
doing so results in a poor model fit for SA yield (Figure 7c, dotted lines), suggesting that this model does
not provide an accurate description of the reaction network. In light of this, a second model was employed
(Model B, Scheme 6), taking into consideration both a dehydrationꢀhydrogenation pathway (k2B and k3B),
as well as a direct hydrogenolysis pathway (k4B). By using the previously determined dehydration rate
28
ACS Paragon Plus Environment

Page 29 of 41
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 6. MA reaction schemes. Model A (a): dehydration from MA to FA and hydrogenation from FA
to SA; model B (b): direct hydrogenolysis from MA to SA. MA = malic acid, FA = fumaric acid, SA =
succinic acid, BEA = 2ꢀ(1ꢀoxopropoxy)butanedioic acid.
constant (k2A = k2B) along with the quasiꢀsteady state approximation for FA, k4B was determined to be 0.25
L3 molꢀ3 hꢀ1 with r4B equal to 1.30 × 10ꢀ5 mol Lꢀ1 hꢀ1. The rate of direct hydrogenolysis (r4B) is roughly 1.5
times greater than the rate of dehydration (r2B), suggesting the direct hydrogenolysis pathway is the
dominant pathway in removing the CꢀOH bonds in the presence of both HI and H2, where ~60% of the
SA formation rate can be attributed to direct hydrogenolysis. This analysis provides quantitative evidence
of H2’s direct participation in CꢀO bond scission in the absence of any metal catalyst.
The structural similarity between the C4 model system and ringꢀopened intermediate from
THFDCA suggests that kinetic insights from MA hydrogenolysis could be used to evaluate kinetic
parameters from timeꢀresolved data in the ring opening of THFDCA. A twoꢀstep THFDCAꢀtoꢀAA
pathway was implemented in the kinetic model: (1) HIꢀmediated ether bond cleavage and hydrogenation
steps to produce HAA from THFDCA (Figure 6b), described by a “lumped” rate constant k1C, and (2) the
hydrogenolysis of HAA to AA, described by the rate constant k3C. The equilibrium constant K2C describes
the equilibrium between HAA and PEA. Based on the kinetic insights from MA hydrogenolysis, the
dehydration/hydrogenation mechanism was not considered. In addition, since no THFDCA conversion
was observed when only HI was employed (500 psi N2), an experimentally determined dehydration rate
constant for HAA could not be obtained. By nonlinear leastꢀsquares regression, k1C, k2C, K2C, and k3C were
evaluated to be 0.80 L3 molꢀ3 hꢀ1, 0.51 L molꢀ1 hꢀ1, 0.42, and 31 L3 molꢀ3 hꢀ1, respectively. Because k1C <<
29
ACS Paragon Plus Environment