Mechanistic Insight to C-C Bond Formation and Predictive Models for Cascade Reactions among Alcohols on Ca- and Sr-Hydroxyapatites

Article abstract of DOI:10.1021/acscatal.6b00556

Biomass-derived light alcohols (e.g., ethanol) may be upgraded via C-C bond formation to form larger alcohols and chemicals. The mechanism for coupling reactions among alcohols (i.e., Guerbet chemistry) is still debated, and the factors that determine the rates of subsequent, and inevitable, reactions among coupling products, and thus the product distributions, are not well understood. Here, the interpretation of the formation rates of products, in situ spectroscopy of surface intermediates, and evidence from isotope labeling experiments are combined to clarify the mechanism of the ethanol coupling over hydroxyapatite (HAP) catalysts. Initial C-C bonds are created by aldol condensation of acetaldehyde, derived in situ from ethanol, and involves a kinetically relevant deprotonation step to form the reactive enolate. In situ infrared spectra show that the coverage of ethanol-derived species far exceeds that of reactive aldehyde intermediates, which is consistent with C-C formation rates that inversely depend on ethanol pressure. Unsaturated aldehyde products are sequentially hydrogenated by the Meerwein-Ponndorf-Verley (MPV) reaction (i.e., C=O bond hydrogenation) and surface-mediated H-transfer (i.e., C=C bond hydrogenation). The MPV reaction simultaneously supplies reactive acetaldehyde needed for the coupling reaction by dehydrogenating ethanol. The rates of self- and cross-coupling reactions among C₂-C₄ alcohols are similar as are the values of the apparent activation enthalpies, which shows that self- and cross-coupling rates depend weakly on the structure of the reactants on HAP catalysts, with few exceptions. The carbon number distribution of the products from ethanol coupling closely matches predictions from an adapted step-growth model. Together, these findings show the mechanism for C-C bond formation between alcohol reactants on HAP catalysts and provide guidance for the production of higher carbon number species from alcohol coupling reactions.

Full text of DOI:10.1021/acscatal.6b00556

Subscriber access provided by McMaster University Library
Article
Mechanistic Insight to C–C Bond Formation and Predictive Models for
Cascade Reactions among Alcohols on Ca- and Sr-Hydroxyapatites
Takahiko Moteki, and David W. Flaherty
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.6b00556 • Publication Date (Web): 04 May 2016
Downloaded from http://pubs.acs.org on May 7, 2016
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 71
ACS Catalysis
1
2
3
4
5
6
Mechanistic Insight to C–C Bond Formation and Predictive Models for
Cascade Reactions among Alcohols on Ca- and Sr-Hydroxyapatites
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
Takahiko Moteki†,‡ and David W. Flaherty*,†,‡
†Department of Chemical and Biomolecular Engineering, University of Illinois Urbanaꢀ
Champaign, Urbana, IL 61801
‡Energy Biosciences Institute, University of Illinois UrbanaꢀChampaign, Urbana, IL 61801
*Corresponding Author
Phone: (217) 244ꢀ2816
Email: dwflhrty@illinois.edu
1
ACS Paragon Plus Environment

ACS Catalysis
Page 2 of 71
1
2
3
4
Abstract
5
6
7
8
9
Biomassꢀderived light alcohols (e.g., ethanol) may be upgraded via C‒C bond formation to form
larger alcohols and chemicals. The mechanism for coupling reactions among alcohols (i.e.,
Guerbet chemistry) is still debated, and the factors that determine the rates of subsequent, and
inevitable, reactions among coupling products, and thus the product distributions, are not well
understood. Here, the interpretation of the formation rates of products, in situ spectroscopy of
surface intermediates, and evidence from isotope labeling experiments are combined to clarify
the mechanism of the ethanol coupling over hydroxyapatite (HAP) catalysts. Initial C‒C bonds
are created by aldol condensation of acetaldehyde, derived in situ from ethanol, and involves a
kinetically relevant deprotonation step to form the reactive enolate. In situ infrared spectra show
that the coverage of ethanolꢀderived species far exceeds that of reactive aldehyde intermediates,
which is consistent with C–C formation rates that inversely depend on ethanol pressure.
Unsaturated aldehyde products are sequentially hydrogenated by the Meerwein–Ponndorf–
Verley (MPV) reaction (i.e., C=O bond hydrogenation) and surface mediated Hꢀtransfer (i.e.,
C=C bond hydrogenation). The MPV reaction simultaneously supplies reactive acetaldehyde
needed for the coupling reaction by dehydrogenating ethanol. Rate of selfꢀ and crossꢀcoupling
reactions among C2‒C4 alcohols are similar as are the values of the apparent activation enthalpies,
which shows that selfꢀ and crossꢀcoupling rates depend weakly on the structure of the reactants
on HAP catalysts, with few exceptions. The carbon number distribution of the products from
ethanol coupling closely match predictions from an adapted stepꢀgrowth model. Together, these
findings show the mechanism for CꢀC bond formation between alcohol reactant on HAP
catalysts and provide guidance for the production of higher carbon number species from alcohol
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
coupling reactions.
2
ACS Paragon Plus Environment

Page 3 of 71
ACS Catalysis
1
2
3
4
5
6
7
Key words
8
9
alcohol coupling, hydroxyapatite, Guerbet reaction, step growth, Meerwein-Ponndorf-
Verley, aldol condensation, cascade reaction
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
ACS Paragon Plus Environment

ACS Catalysis
Page 4 of 71
1
2
3
4
1. Introduction
5
6
7
8
9
Biomass has the potential to provide renewable and carbon neutral feedstocks for production
of fuels and chemicals.1ꢀ4 Many forms of biomass (e.g., sugarcane, corn, or straw) can be readily
converted into ethanol, a platform chemical, by fermenting sugars (e.g., glucose, fructose,
sucrose).4ꢀ6 In turn, ethanol can be transformed to more valuable and higher carbon number (Cn)
products by C‒C bond formation reactions.7ꢀ11 Reactive species derived from ethanol form 1ꢀ
butanol, larger alcohols, and alkenes over heterogeneous catalysts that expose both acid and base
sites (i.e., metal oxides,12ꢀ15 mixedꢀmetal oxides,16ꢀ20 and hydroxyapatites21ꢀ28).7,8 Both
stoichiometric hydroxyapatite (HAP, Ca10(PO4)6(OH)2) and modified HAP have received a
significant amount of attention for ethanol conversion,7,8,21ꢀ29 because HAP produces 1ꢀbutanol
with reasonable selectivity (~ 75%) at 10–15% conversion22,23 and possesses acid and base
properties that can be tuned by changing the identity of the metal (e.g., Ca or Sr) and the metal to
phosphorous molar ratio (M/P) of the material.21ꢀ24,30
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
Following the formation of 1ꢀbutanol, C‒C bond forming reactions continue and create
heavier products (C≥6).21,22 The relative rates of selfꢀ and crossꢀcondensation reactions
determines the Cn and isomer distribution of the product,22 and thus, their value for particular
applications. For example, 1ꢀbutanol is desirable as a fuel additive, midꢀchain (C8‒C12) alcohols
are useful for jet fuels, and predominately linear C12‒C18 alcohols can be used for diesel fuels.31ꢀ
33
Ideally, the selectivity towards a given category of alcohols would be high, because the
fractionation of longꢀchain alcohols is energy intensive. Clarifying the network of reactions
present during ethanol conversion to higher alcohols over heterogeneous catalysts would help to
show how, and if, the product distribution may be controlled.
4
ACS Paragon Plus Environment

Page 5 of 71
ACS Catalysis
1
2
The details of the mechanism,14,15,19,20,22ꢀ28 active intermediate,20,24,29,34 and nature of the
active sites25,26,29,35ꢀ37 for forming the first C‒C bonds between ethanolꢀderived intermediates on
oxide catalysts is an active area of research, as summarized in recent reviews.7,8 Yet, the
mechanisms and networks of continuous secondary coupling reactions that produce C≥6 products
have received less attention. C‒C bonds form over heterogeneous catalysts via one of three
proposed pathways;7,8 direct condensation between two alcohols,15,27,28 condensation between
aldehydes and alcohols,14 or a more complex series of reactions that involve aldol condensation
between aldehydes,21ꢀ26 and the relative contributions of these pathways likely depend on both
the reaction conditions and the identity of the catalyst. The latter pathway consists of a sequence
of steps that is thought to include dehydrogenation of alcohols to aldehydes, aldol addition of
two aldehydes, dehydration of the aldol, and subsequent hydrogenation steps to form aldehydes
or alcohols.7,8,38 Insight into the factors that ultimately determine the Cn and structural
distributions of C≥6 product alcohols requires that we understand also the mechanism for C‒C
bond formation between alcohol derivatives and the identity of the reactive species that
participate in each elementary step.
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
Here, we describe the network of reactions that convert ethanol into C4‒C12 alcohols and
aldehydes over Caꢀ and SrꢀHAP, and show how the rates of the parallel coupling pathways
depend on intrinsic properties of the reactive intermediates and the catalytic surfaces.
Comparisons of reaction rates for C‒C bond formation measured as functions of ethanol (1.2‒6.2
kPa), acetaldehyde (0.07‒0.26 kPa), and hydrogen (20‒100 kPa) pressures implicate
acetaldehyde as a reactive intermediate. C‒C bonds form by steps involving kinetically relevant
deprotonation of acetaldehyde and subsequent aldol condensation. Isotope labeling experiments
show that direct Hꢀatom transfer between ethanol and formed aldehydes via the Meerweinꢀ
5
ACS Paragon Plus Environment

ACS Catalysis
Page 6 of 71
1
2
3
4
5
6
7
8
9
PonndorfꢀVerley (MPV) reaction concurrently saturates C=O bonds to form heavier alcohols and
produces acetaldehyde, however, surface mediated Hꢀatom transfer pathways hydrogenate C=C
bonds. Independent rate measurements for selfꢀcoupling reactions of C2, C3, and C4 alcohols as a
function of reactant conversion (0‒10%) show that pseudoꢀfirst order rate constants vary only by
a factor of two between ethanol and larger alcohols. In addition, rate ratios for crossꢀcoupling
reaction pathways that involve the deprotonation of either C2 or C4 aldehydes to form either 1ꢀ
hexanol or 2ꢀethylꢀbutanol, respectively, remain constant and close to unity (~ 0.8) from 548ꢀ598
K. These results, and expectations that further increases in alcohol chain length will have
diminishing effects on reactivity, suggest that apparent rate constants and activation enthalpies
for all simultaneous selfꢀ and crossꢀcoupling pathways are similar. Moreover, measurements of
ethanol consumption rates and quantitative analysis of product distributions among C4ꢀC12
alcohols show that the reaction network of all condensation pathways match predictions of an
adapted stepꢀgrowth polymerization model.39,40 The degree of branching in these species
increases monotonically with conversion, because reactions between two C>2 aldehydes create
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
products with one more Cꢀatom than the reactants. Overall, these findings show that a simple
model provides useful predictions for several properties of the final product distribution.
However, this also indicates that such reaction networks are not suitable for selectively creating
product alcohols with narrow distributions of Cn or branching characteristics.
2. Materials and Methods
2.1. Preparation of Ca- and Sr-HAP Catalysts
6
ACS Paragon Plus Environment

Page 7 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
The CaꢀHAP catalyst with a Ca/P ratio of 1.67 was purchased from Acros Organics (Lot:
A0333466), while other Caꢀ and SrꢀHAP catalysts with different metal to phosphorous ratios
(M/P) were prepared by precipitation methods described previously.21ꢀ24,30 Briefly, Caꢀ and Srꢀ
HAP catalysts were prepared from solutions created by combining deionized water (17.8 Mꢁ)
with (NH4)2HPO4 (SigmaꢀAldrich, 99.0%) and either Ca(NO3)2·4H2O (SigmaꢀAldrich, 99.0%)
or Sr(NO3)2 (SigmaꢀAldrich, 99.0%). The M/P of the resulting catalysts were controlled over the
range of 1.61‒1.74 by changing the ratio of the metal and phosphorous precursors in the
synthesis solutions. Specifically, 300 cm3 of an aqueous solution containing either
Ca(NO3)2·4H2O or Sr(NO3)2 at a concentration of 0.667‒0.8 M was added to 300 cm3 of aqueous
0.4 M (NH4)2HPO4. Aqueous NH4OH (28‒30% as NH3, Macron Fine Chemicals) was added to
the solution to achieve an initial pH of 10. The mixed synthesis solution was then heated to 353
K and stirred for 12 h to complete the precipitation of the HAP product. The precipitated
products were filtered and washed with 1 L of deionized water. Subsequently, the recovered
sample was dried in a stagnant air at 373 K for 12 h in a preheated oven. The dried powders were
pelletized, crushed, and sieved to obtain HAP agglomerations with diameters of 230‒560 ꢂm.
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. Characterization of HAP Materials
Powder Xꢀray diffraction (XRD) patterns were collected on a Siemens Bruker D5000 using
CuKα radiation (λ = 0.1542 nm, 40 kV, 30 mA) between 5° and 60° (2θ) with 0.05° steps and a
scanning speed of 2° min−1. The XRD patterns of all five materials are shown in the Supporting
Information (Figure S1) and all detected peaks match only those for the known structure of
hydroxyapatite (Table 1), which confirms that only Caꢀ and SrꢀHAPs phases formed without
structural impurities, such as brushite.30 Scanning electron microscope (SEM) images were
obtained on a Hitachi S4700 with an accelerating voltage of 3‒10 keV. Prior to SEM
7
ACS Paragon Plus Environment

ACS Catalysis
Page 8 of 71
1
2
3
4
5
6
7
8
9
measurements, the samples were lightly sputtered with Pt/Au in an Ar atmosphere (EMITECH
K575). The morphologies of the HAP samples were spherical and rodꢀlike for Caꢀ and SrꢀHAP
samples, respectively (Figure S2). These morphologies differ from the plateꢀlike CaꢀHAP
synthesized in solutions with pH values of 9 or less,21,22,30 and this indicates that our synthesis
directly forms HAP due to the higher pH value of 10 and does not involve an intermediate
brushite structure.30 The specific surface area of each sample was determined using a single point
N2 adsorption measurement at 77.3 K at a relative pressure (P/P0) of 0.31 by Micrometrics. Prior
to N2 adsorption, the HAP samples were degassed at 673 K in vacuum for 4 h. These
physisorption measurements show that the synthesized Caꢀ and SrꢀHAP materials possess
specific surface areas of 60‒70 and 35‒40 m2g−1, respectively (Table 1).
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 chemical composition of each sample was measured by inductively coupled plasmaꢀmass
spectrometry (ICPꢀMS; PerkinElmer, ELAN DRCꢀe) to determine the bulk molar ratios of Ca to
P (Ca/P) and Sr to P (Sr/P). All samples fell within the range of M/P values of (1.61‒1.74), and
bulk M/P values changed monotonically with that of the synthesis solution. It should be noted
that the M/P ratio of surface is correlated with, but not equal to, that of bulk.22 Additional
measurements are, however, necessary to determine how the chemical function and number of
base sites on the HAP surface change with M/P ratios and metal identity.30,36 Therefore, the
number and functional strength of base sites on HAP surfaces was determined by temperatureꢀ
programmed desorption (TPD) spectra of CO2 using a customꢀmade system equipped with a
quadrupole mass spectrometer (ThermoStar, Pfeiffer). During TPD measurements, 0.1 g of HAP
was loaded into a tubular glass reactor and heated at 5 K minꢀ1 to 773 K in flowing He (25 cm3
minꢀ1, SJ Smith, Ultra High Purity) and held at 773 K for 1 h, with the intent to desorb all H2O
and other impurities adsorbed on the surface. Subsequently, the sample was cooled to 323 K in
8
ACS Paragon Plus Environment

Page 9 of 71
ACS Catalysis
1
2
stagnant He, after which, the catalyst was exposed to a flowing stream of 50 % CO2/He (50 cm3
minꢀ1, SJ Smith, 99.995%) for 1 h to saturate the surface with CO2. Next the reactor and catalyst
bed was flushed with He (25 cm3 minꢀ1) for 3 h at 323 K to remove weakly adsorbed CO2. The
TPD spectra was then obtained by heating the sample at a rate of 5 K minꢀ1 to 873 K in a 25 cm3
minꢀ1 He flow while monitoring massꢀtoꢀcharge ratios of 44, 18, and 4 using the mass
spectrometer to detect the partial pressures of CO2, H2O, and He respectively. The absolute
amount of CO2 that desorbed from each sample was calculated using the known flow rate of the
system and the calibrated sensitivity of the mass spectrometer for CO2. The resulting CO2 TPD
spectra (Figure S3) show distinct desorption peaks at ~360 K and ~420 K on CaꢀHAP, which
correspond to distinct adsorption sites that can be qualitatively described as weaker and stronger
base sites, respectively. The integrals of the TPD peaks increase with increasing M/P ratios,
which suggest that HAP catalysts with higher M/P ratio have a larger number of surface base
sites. These results are consistent with expectations that larger Ca content will result in the
increase of base sites,21ꢀ23,30 possibly OH− and O2− moieties.35,36
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
2.3. In Situ Transmission Fourier Transform Infrared Spectroscopy
The identity and coverage of surface intermediates formed during ethanol coupling reactions on
HAP catalysts was probed using in situ transmission Fourier transform infrared (FTIR)
spectroscopy using a customꢀmade transmission cell, described previously.41 Catalyst wafers
(50–100 mg) were pressed into a selfꢀsupporting wafer and placed within the stainlessꢀsteel cell
equipped with CaF2 windows and connected to a gas manifold, which enabled thermal treatments
and adsorptionꢀdesorption experiments to be carried out in situ. The temperature of the cell was
controlled with a customꢀmade metal heating assembly,41 and the temperature of the wafer was
measured with a Kꢀtype thermocouple located within the cell. Ethanol was fed via a syringe
9
ACS Paragon Plus Environment

ACS Catalysis
Page 10 of 71
1
2
3
4
5
6
7
8
9
pump (Legato 100, KD Scientific) and was vaporized inside the gas transfer lines, which were
heated to 323 K using electrical heating tape. Wafers were first pretreated at 673 K for 1 h under
helium flow (100 cm3 minꢀ1), then cooled to 573 K prior to introducing ethanol (4.6 kPa C2H5OH,
96 kPa H2, Ca/P =1.67). In order to minimize the spectral contributions from aromatic surface
residues that form during the reaction, the ethanol flow was stopped after 30 min and the cell was
flushed with He prior to taking a background scan. Ethanol was then reꢀintroduced into the cell
and the in situ sample spectra were obtained after waiting 30 min for the system to reach steadyꢀ
state. Spectra were recorded by the spectrometer (Bruker, TENSOR 37), and each spectra was
coꢀadded from 32 scans obtained at a resolution of 4 cmꢀ1.
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
2.4. Steady-State Catalytic C–C Bond Formation and 1-Butanol Formation Rates and
Product Selectivities
Alcohol coupling reactions were carried out using a Uꢀshaped, tubular glass reactor with a
packedꢀbed configuration. The reactor was placed in a verticallyꢀaligned temperatureꢀcontrolled
tubular furnace (National Element, Inc., FAꢀ120). The temperature of the catalyst bed was
measured with a Kꢀtype thermocouple coaxiallyꢀaligned with the glass reactor and in direct
contact with the catalyst bed (0.05–0.2 g) and was controlled by a PID controller (WATLOW,
EZꢀZONE) connected to a variable transformer. The HAP catalysts were treated in situ by
heating at 10 K minꢀ1 to 773 K and holding for 2 h under He flow (20 cm3 minꢀ1) to remove
adsorbed water and other molecules (e.g., CO2) before measuring catalytic rates. Flow rates of
He (SJ Smith, Ultra High Purity) and H2 (SJ Smith, 99.999%) were set by mass flow controllers
(MFC, 601 series, Porter). Liquid phase reactants (ethanol, Decon's Pure Ethanol 200 Proof,
100%; acetaldehyde, Fluka, 99.5%; 1ꢀpropanol, Fisher Scientific, certified grade; 1ꢀbutanol,
Fisher Scientific, 99.9%; and crotonaldehyde, Aldrich, 99%, predominantly trans) were fed via a
10
ACS Paragon Plus Environment

Page 11 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
syringe pump (Legato 110, KD Scientific) and were vaporized inside the 1/4" stainless steel
transfer lines, which were heated to > 473 K using electrical heating tape. Volatile reactants (e.g.,
acetaldehyde) were fed as 10–30 vol % mixture in ethanol. The flow rates of all reactants were
varied in order to control both the partial pressure and surface residence time of the reactants
during experiments. The total pressure of the system was maintained at 101 kPa by coꢀfeeding
H2 with the other reactants as needed. H2 gas was used as a carrier gas instead of conventional
inert gas (i.e., He) to maintain higher CꢀC bond formation rates, because H2 slows the rate of
carbon deposition on the catalyst as shown by comparisons between carbon balances and
thermogravimetric analysis of spent catalysts (Figure S4). Before measuring the change in CꢀC
bond formation rates as a function of reactant pressures or temperature, the system was allowed
to reach steadyꢀstate at 5 kPa ethanol and 96 kPa H2, which took approximately 6ꢀ8 hours
(Figure S4a).
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 identity and concentrations of reactants and products in the reactor effluent stream were
measured using a gasꢀchromatograph (GC, 6850, Agilent) equipped with a capillary column
(DBꢀWAX, 30 m × 0.25 mm × 0.25 µm) and a flame ionization detector (FID). The retention
time for each component was determined using standard chemicals (SigmaꢀAldrich, C2‒C8 nꢀ
alcohols and C2‒C4 nꢀaldehydes), and the molecular speciation of these and heavier products was
confirmed using a gasꢀchromatograph massꢀspectrometer (GCꢀMS, QP2010 Ultra, Shimadzu).
Rates were measured at <15% conversion of the limiting reactant to identify primary reactions
and to ensure that depletion of reactants across the bed had a minimal influence on measured
rates. Massꢀaveraged rates for each reaction pathway are calculated as a number of the reaction
event per unit time normalized by the initial moles of reactants and the weight of catalyst. The
number of the C‒C bond formation events is calculated by multiplying the moles of formed C2i
11
ACS Paragon Plus Environment

ACS Catalysis
Page 12 of 71
1
2
3
4
5
6
7
8
9
species (i ≥ 2) by iꢀ1. For example, C6, C8, and larger C2i products are formed using 2, 3, and i−1
ethanol molecules, respectively. C‒C bond formation rates are reported as a number of C‒C
bond formation event per second per grams of catalyst ((mole) (g catalyst s)ꢀ1). Selectivities are
reported based on moles of carbon of the products (Cꢀ%). Measured rates were corrected to
account for catalyst deactivation by measuring rates at a standard set of conditions multiple times
within each experiment. After initial rapid deactivation over a period of 6 hours, the CꢀC bond
formation rates decrease slowly over time (Figure S4a). The change in the number of active sites
as a function of time on stream was determined by linear interpolation of relative site counts
estimated from rates measured every 10‒12 hours at identical conditions.
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
2.5. Catalytic Hydrogenation of Butyraldehyde, Crotyl Alcohol, and Crotonaldehyde to 1-
Butanol
Hydrogenation reactions of butyraldehyde, crotyl alcohol, and crotonaldehyde were carried out
using the reactor and pretreatment protocols used to study C–C bond formation reactions (section
2.4). The liquid phase reactants (2ꢀpropanolꢀd8, Aldrich, 99.5 atom % D; butyraldehyde, Fluka,
99.0%; crotyl alcohol, Aldrich, 96%, mixture of cis and trans; crotonaldehyde, Aldrich, 99%,
predominantly trans) were fed into flowing H2 within reactor using a syringe pump. 2ꢀPropanolꢀ
d8 was fed to the reactor in a 20 mol % mixture in tertꢀbutanol (Fischer Scientific, certified
grade). The reactor effluent was collected by bubbling through an ethanol solution at the outlet of
the reactor, and the mass spectra of all product species were collected using a gasꢀchromatograph
massꢀspectrometer (GCꢀMS, QP2010 Ultra, Shimadzu). The shift in the mass spectrum of the 1ꢀ
butanol obtained by catalytic reaction with the mixture of 2ꢀpropanolꢀd8 and tertꢀbutanol was
compared with that of purely hydrogen containing 1ꢀbutanol and crotyl alcohol.
12
ACS Paragon Plus Environment

Page 13 of 71
ACS Catalysis
1
2
3
4
5
6
7
3. Results and Discussion
8
9
3.1.1. Evidence for Aldehyde as Reactive Intermediates
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
42
Figure 1a shows that the steadyꢀstate C–C bond formation rate (rC–C
)
on CaꢀHAP is
approximately 0.1 µmol gcatal sꢀ1 (4.8 kPa C2H5OH, 96 kPa H2, 1.67 Ca/P, 548 K), and rC–C
increases by more than an order of magnitude when 0.26 kPa C2H4O is coꢀfed to the reactor.
This significant change suggests that acetaldehyde is an active intermediate in C–C bond
formation, which produces C4 and larger enals, aldehydes, and alcohols. The products contain a
broad distribution of species (Table 2), which at 8.2% conversion include C4 alcohols and
aldehydes (~86 Cꢀ%, 1ꢀbutanol, butyraldehyde, crotyl alcohol, and crotonaldehyde), C6–C8
alcohols and aldehydes (~8 Cꢀ%), C2–C6 alkenes (~4 Cꢀ%), and other species (~2 Cꢀ%) when
only ethanol is fed to the reactor (4.8 kPa C2H5OH, 96 kPa H2, 1.67 Ca/P, 548 K). The C4
products predominantly consist of 1ꢀbutanol and crotyl alcohol while crotonaldehyde and
butyraldehyde are present only in trace quantities (Table 2), which suggests that surface reaction
pathways hydrogenate C=O bonds more readily than C=C bonds. The large number of products
shows that the system involves a complex sequence of intermediate steps that form C–C bonds.
The significant fraction of C≥6 products suggest that these reactions occur in cascades that add
significant amounts of C2ꢀintermediates to primary and secondary products even at low
conversions (<10%) of acetaldehyde and ethanol. Figure 1b shows the product selectivities for
C4 products as a function of the total C2 conversion at higher acetaldehyde and lower ethanol
pressures (1 kPa C2H5OH, 0.35 kPa C2H4O, 100 kPa H2, 1.67 Ca/P, 548 K). Selectivities
extrapolated to zero conversion show that crotonaldehyde, which is the product from the aldol
ꢀ1
13
ACS Paragon Plus Environment

ACS Catalysis
Page 14 of 71
1
2
3
4
5
6
7
8
9
condensation of acetaldehyde, is the sole primary product and that more saturated C4 species
(e.g., butyraldehyde, crotyl alcohol, and 1ꢀbutanol) form as secondary products following
hydrogenation. These results (Figure 1 and Table 2) strongly suggest that C–C bond formation
occurs by aldol condensation of two acetaldehyde molecules over stoichiometric CaꢀHAP
catalyst at these conditions. Moreover, the change in product selectivities with conversion
(Figure 1b) are not consistent with other mechanisms proposed such as direct condensation
between two ethanol molecules15,27,28 or between ethanol and acetaldehyde,14,15,26 because such
mechanisms are reported to produce 1ꢀbutanol, butyraldehyde, and crotyl alcohol as primary
products. The elementary steps and mechanism for CꢀC bond formation on HAP are further
described below.
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
3.1.2. C–C Bond Formation and Hydrogenation
Figure 2 shows that C–C bond formation rates increase in proportion to the acetaldehyde
pressure (Figure 2a, rC–C ~ [C2H4O]1.0; 4.8 kPa C2H5OH, 96 kPa H2), decrease with the ethanol
pressure (Figure 2a, rC–C ~ [C2H5OH]−0.3±0.2; 0.1 kPa C2H4O, 96 kPa H2), and do not depend on
the hydrogen pressure (Figure 2b, rC–C ~ [H2]0; 5 kPa C2H5OH) over CaꢀHAP (1.67 Ca/P, 548 K).
These observations (Figure 2) are consistent with previous measurements, suggesting a
kineticallyꢀrelevant step that involves a reaction of acetaldehyde and active site inhibition by
ethanol.23,26 Neither gaseous H2 or surface intermediates derived from H2 (i.e., chemisorbed Hꢀ
atoms (H*)) appear to play a role in any kinetically relevant steps that couple acetaldehyde
surface intermediates.23,24
14
ACS Paragon Plus Environment

Page 15 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
The products contain high ratios of alcohols to aldehydes for each carbon backbone observed
(Table 2), which suggests that rapid hydrogen transfer saturates unsaturated products (e.g.,
crotonaldehyde) prior to their desorption from HAP surfaces. Hydrogen transfer between
alcohols and aldehydes on oxide surfaces may occur by one of two mechanisms7,8,43 which
include direct intermolecular hydride transfer, also known as the Meerwein–Ponndorf–Verley
(MPV) reaction (Scheme S1a)23,26,44 or surface mediated hydrogenation (Scheme S1b).20 Surface
mediated hydrogenation, which is common on metals,20 involves activation and transfer of Hꢀ
atoms from both gas phase H2 and alcohols. However, our results (Figure 2b) and others23,24
show that gaseous H2 has no effect on reaction rates or product selectivity, which suggest that H2
is not an effective Hꢀatom donor during reactions on HAP surfaces. Recent reports have
suggested that aldol condensation products (e.g., crotonaldehyde, butyraldehyde) hydrogenate by
Hꢀatom transfer from surface ethanol molecules following the MPV mechanism over
HAP,7,8,23,26 yet have not provided direct evidence for this pathway. It is difficult to determine
whether hydrogen transfer occurs by surface mediated routes or by the MPV mechanism using
rate measurements alone, however, isotope labeling techniques45 can differentiate between these
pathways.
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 shows the mass spectrum of 1ꢀbutanol formed after hydrogenation of butyraldehyde or
crotyl alcohol with a mixture of 2ꢀpropanolꢀd8 and tertꢀbutanol (0.5 kPa C4H8O or C4H7OH, 0.6
kPa C3D7OD, 2.4 kPa tertꢀC4H9OH, 97 kPa H2, 1.67 Ca/P, 573 K). Under these conditions with
the excess of tertꢀbutanol, the HAP surface mediates fast H/D exchange between the –OD and –
OH groups in the alcohols (2ꢀpropanolꢀd8 and tertꢀbutanol, respectively), which causes the
majority of the 2ꢀpropanol molecules to become the C3D7OH isotopologue. The ratio of Hꢀ to Dꢀ
atoms on the CaꢀHAP surface is assumed to be approximately four, because the zero point
15
ACS Paragon Plus Environment

ACS Catalysis
Page 16 of 71
1
2
3
4
5
6
7
8
9
energy differences between the OꢀH and OꢀD bonds in the organic reactants and those between
the OꢀH and OꢀD bonds of surface hydroxyls are similar46 and consequently these Hꢀ to Dꢀatom
ratio should match that of –OH to –OD groups in the reactant mixture of perhydrogenated tertꢀ
butanol (2.4 kPa) and perdeuterated 2ꢀpropanol (0.6 kPa). Figure 3a shows that the 1ꢀbutanol
produced by hydrogenating butyraldehyde (Figure 3a, red) has a molecular weight 1 amu greater
orthan that of perhydrogenated 1ꢀbutanol (C4H9OH, Figure 3a, black), and this comparison
suggests that 2ꢀpropanolꢀd7 (CD3CDOHCD3) formed in situ acts as the hydrogen donor and
directly transfers the βꢀD and the αꢀH to the Cꢀatom and the Oꢀatoms, respectively, of the C=O
group in butyraldehyde to produce C4H8DOH by the MPV reaction (Scheme S1a). If hydrogen
transfer occurred by the surface mediated process (Scheme S1b), butyraldehyde would be
hydrogenated with surface H/Dꢀatoms, among which the majority are H*ꢀatoms under these
conditions (2.4 kPa tertꢀC4H9OH, 0.6 kPa C3D7OD). Consequently, the product 1ꢀbutanol would
be primarily perhydrogenated and would have a mass spectrum indistinguishable from that for
C4H9OH. On the other hand, Figure 3b shows that the 1ꢀbutanol produced by hydrogenating
crotyl alcohol (Figure 3b, blue) has a mass spectrum almost identical to that of perhydrogenated
1ꢀbutanol (Figure 3b, black), which shows that C=C bond hydrogenation proceeds via surface
mediated Hꢀatom transfer (Scheme S1b). The comparisons between the spectra (Figure 3) would
give compelling evidence that, within this reaction network, the hydrogenation of C=O bonds
occurs by Hꢀtransfer from alcohol reactants via the MPV reaction mechanism, which is broadly
thought to transfer Hꢀatoms in a concerted manner.43 These data, however, would be consistent
also with a sequential transfer of Hꢀatoms such as direct Hꢀatom transfer from the αꢀcarbon of
the surface alkoxide to that of the aldehyde via a MPVꢀlike intermediate44 and subsequent Hꢀ
atom transfer to the Oꢀatoms of the aldehyde via a surface mediated pathway. On the other hand,
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
16
ACS Paragon Plus Environment

Page 17 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
hydrogenation of C=C bonds occurs by transfer of two Hꢀatoms formed on the catalyst surface
via surface mediated hydrogenation. Isomerization reaction between butyraldehyde and crotyl
alcohol are much slower than these hydrogenation reactions, because we observed nearly perfect
1 or 0 amu MS shifts suggesting that each hydrogenation pathway does not contribute
simultaneously.
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
These findings are consistent also with the outcome of similar isotope labeling experiments that
probed pathways for Hꢀtransfer to crotonaldehyde. The hydrogenation of crotonaldehyde to
crotyl alcohol and that of crotonaldehyde to 1ꢀbutanol (Figure S5) both show a 1 amu shift
relative to the perhydrogenated products. These results, together with those for hydrogenation of
butyraldehyde (Figure 3a) and crotyl alcohol (Figure 3b), suggest that among all unsaturated
species in this system, the MPV reactions is the predominant mechanism to hydrogenate C=O
bonds and surface mediated Hꢀtransfer hydrogenates C=C bonds. In addition, the high ratios of
alcohols to aldehydes (i.e., [C4H9OH]/[C4H8O] > 100; [C4H7OH]/[C4H6O] > 10) for each carbon
backbone observed (Table 2) suggests that the MPV reaction (i.e., C=O bonds hydrogenation)
occurs more readily than surface mediated hydrogenation (C=C bonds hydrogenation). These
observations strongly suggest that enals formed by aldol condensation reactions first form
unsaturated alcohols by the MPV reaction and subsequently hydrogenate to form alcohols by
surface mediated Hꢀtransfer. At steadyꢀstate, the MPV process also dehydrogenates ethanol to
form acetaldehyde, which is the active intermediate for C–C bond formation, as discussed in the
following section.
3.1.3. Elementary Steps and Rate Equation of Initial Ethanol Coupling Reaction
17
ACS Paragon Plus Environment

ACS Catalysis
Page 18 of 71
1
2
3
4
5
6
7
8
9
Scheme 1 shows a sequence of elementary steps that accurately describes the effects of
acetaldehyde, ethanol, and hydrogen pressures on C–C bond formation rates (Figure 2) and
accounts for Hꢀatom transfer between alcohol and aldehyde species by the MPV mechanism on
HAP catalysts (Figure 3). These steps involve Lewis acid sites (*A; Caꢀ or Srꢀatoms) and base
sites (*B; Oꢀ or OHꢀmoieties) that are known to exist on HAP surfaces based on FTIR
studies26,35,36 and the absence of spectral features indicative of Brønsted acid sites (i.e., formation
of pyridinium ions upon adsorption of pyridine).37,47
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 1 includes ethanol adsorption on acid sites (1.1) and dehydrogenation to form ethoxide
(1.2),29,48,49 and then further dehydrogenation to form acetaldehyde (1.3). Acetaldehyde may
desorb (1.4) or deprotonate at the αꢀcarbon to the carbonyl by reaction with an adjacent base site
to form a carbanion (i.e., enolate) (1.5). The enolate undergoes nucleophilic attack onto an
acetaldehyde molecule bound at a vicinal acid site to form a C–C bond (1.6). An aldol forms by
subsequent protonation (1.7) and dehydrates to form the unsaturated aldehyde (i.e.,
crotonaldehyde) (1.8). Crotonaldehyde may hydrogenate to form crotyl alcohol by the MPV
reaction (1.9), and then hydrogenate further by surface mediated Hꢀtransfer to form 1ꢀbutanol
(1.10). All C4 products (i.e., crotonaldehyde, crotyl alcohol, and 1ꢀbutanol) may desorb in
reversible steps (1.11–1.13). Concomitantly, the MPV reaction that transfers hydrogen from
ethanol to crotonaldehyde also produces acetaldehyde needed for the reaction (1.9). Finally, Hꢀ
atoms bound to base sites may recombinatively desorb to form H2 (1.14).7,8 Adsorption of
ethanol (1.1), desorption of acetaldehyde (1.4), and desorption of C4 alcohols (1.12–1.13) are
assumed to be quasiꢀequilibrated (QE), whereas deprotonation of the aldehyde (1.5) to form the
carbanion is kinetically relevant.26 Step (1.14) is not quasiꢀequilibrated and further the rate of
18
ACS Paragon Plus Environment

Page 19 of 71
ACS Catalysis
1
2
3
4
5
6
formation of surface Hꢀatoms by dissociative adsorption of gas phase H2 must be slow, because
C–C bond formation rates do not depend on the H2 pressure (Figure 2b).
7
8
9
Following Scheme 1, C–C bond formation rates are proportional to the number of adsorbed
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
aldehydes ([C2H4O*A]) and the probability of finding an unoccupied, adjacent base site ([*B]):
ꢅ
ꢋ ꢅ
ꢋ
ꢁ ꢇ ꢉ∗ ∙ ∗
ꢆ
ꢈ
ꢊ
ꢌ
ꢀꢁꢂꢁ = ꢃꢄ
(1)
ꢅ
ꢋ
ꢍ
ꢌ
Here, k5 is the rate constant for deprotonating acetaldehyde, and [C2H4O*A], [*B], [LB] are the
numbers of acetaldehyde surface intermediates, unoccupied base sites, and the total number of
reactive base sites, respectively. It should be noted that the number of base sites, [LB], also reflect
the number of acid sites, [LA], because the acid and base sites in Eq. (1) should be an adjacent
pair (see supporting information for further details) as proposed in the previous literature, in
3ꢀ
which CaO/PO4 pairs is shown to catalyze the aldol condensation.26 Equation 1 takes a new
form after accounting for quasiꢀequilibrated (QE) acetaldehyde desorption (Scheme 1, step 1.4).
ꢅ
ꢋ ꢅ ꢋ ꢅ ꢋ
ꢁ ꢇ ꢉ ∙ ∗ ∙ ∗
ꢌ
ꢆ
ꢈ
ꢊ
ꢀꢁꢂꢁ = ꢃꢄꢎꢏ
(2)
ꢅ
ꢋ
ꢍ
ꢌ
Here, K4 is the equilibrium constant for desorption of acetaldehyde (Scheme 1, steps 1.4),
[C2H4O] is the gasꢀphase concentration of acetaldehyde, and [*A] is the number of unoccupied
acid sites. The total numbers of available acid sites ([LA]) and those of base sites ([LB]) are equal
to the sum of all likely surface intermediates at each site:
ꢗ
ꢅ ꢋ ꢅ ꢋ
ꢐꢑ = ∗ꢑ
∑
∑
ꢘ
+
ꢘꢒꢓꢔ,ꢕ ∗ꢑꢖ + ꢙꢒꢓꢏ,ꢕ ∗ꢑꢖ
(3)
ꢅ
ꢋ
ꢅ ꢋ ꢋ
ꢅ
ꢐꢚ = ∗ꢚ + ꢛ ∗ꢚ
(4)
19
ACS Paragon Plus Environment

ACS Catalysis
Page 20 of 71
1
2
3
4
ꢗ
where ꢒꢓꢔ,ꢕ ∗ꢑꢖ and ꢙꢒꢓꢏ,ꢕ ∗ꢑꢖ represent the total number of all possible C2 (e.g., ethanol,
∑
∑
ꢘ
ꢘ
5
6
7
8
9
ethoxide, acetaldehyde) and C4 (e.g., 1ꢀbutanol, butyraldehyde, etc.) surface species, respectively,
and [H*B] is the number of Hꢀatoms on base sites. The number of C≥6 oxygenates and other
species (e.g., alkanes and ethers) are assumed to be much smaller than C2–C4 intermediates, and
therefore negligible, because the concentrations of C4 products are much higher than that of C≥6
products or side products (Table 2). The combination of Eqs. (2–4) gives the following rate
expression for C–C bond formation:
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
ꢅ
ꢋ ꢅ ꢋ ꢅ ꢋ
ꢁ ꢇ ꢉ ∙ ∗ ∙ ∗
ꢌ
ꢜ
ꢝꢞꢝ
ꢆ
ꢈ
ꢊ
= ꢃꢄꢎꢏ ( ∗ ꢟꢠ
(5)
ꢢ
ꢣ
ꢤ
ꢅ
ꢋ
∑
∑
ꢅ
ꢋ ꢅ ꢋ
ꢒꢁ ꢖ)( ∗ ꢟ ꢇ∗
)
ꢌ
ꢅꢍ ꢋ
ꢒꢁ ꢖꢟꢠ
ꢊ
ꢊ
ꢌ
ꢡ
ꢆ,ꢡ
ꢡ
ꢈ,ꢡ
ꢣ
Measured C–C bond formation rates show a linear dependence on acetaldehyde (Figure 2a) and
negative dependence on ethanol (rC–C ~ [C2H5OH]−0.3±0.2; Figure 2a), which is consistent with Eq.
(5) when ethanolꢀderived species occupy a significant fraction of the acid sites. Thus, the
mechanistic interpretation of C–C bond formation rates imply that surface species derived from
gaseous ethanol (e.g., ethoxide or ethanol) are among the most abundant reactive intermediates
(MARI) on acid sites. The identity of these species was determined by analysis of in situ FTIR
spectra, described below.
Figure 4 shows in situ FTIR spectra (3100–1500 cm−1 region) of the surface species that form on
CaꢀHAP under ethanol flow (5 kPa C2H5OH, 96 kPa H2, 573 K, 1.67 Ca/P). Figure 4a spectra of
the surface species that form feeding ethanol for 2 – 50 minutes, and a control measurement
taken under same condition with an empty cell (Figure 4a, bottom) shows that the IR absorption
features from gasꢀphase species are negligible. Absorption features at 3000–2800 and 1600–1500
cm−1 correspond to alkane v(C–H) and aromatic v(C=C) modes, respectively.50 A complete
absence of features between 1800–1600 cm−1 indicates that the number of carbonyl groups
20
ACS Paragon Plus Environment

Page 21 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
among the surface species is negligible, and therefore, that the surface coverage of aldehydes is
insignificant. Removal of ethanol from the inlet to the FTIR cell causes the intensity of the
features at 3000–2800 cm−1 region to drop by more than 50% within 100 seconds, however, the
intensity of the features at 1600–1500 cm−1 remain constant (Figure S6), which suggests that
both weakly adsorbed reactive species and strongly adsorbed aromatic species coexist on the
surface. Figure 4b shows a spectrum of only the reactive surface intermediates obtained by
subtracting a spectrum taken of a used catalyst after removing ethanol from the feed. The main
four bands at 2972, 2934, 2905, and 2877 cm−1 (Figure 4b) are attributed to the vas(CH3),
vas(CH2), vs(CH3), and vs(CH2) modes, respectively, of adsorbed ethanol (or ethoxide) on Caꢀ
HAP.29,48,49 Together these observations (Figures 4 and S6) suggest that absorbed ethanol (or
ethoxide) are a MARI on acid sites. This conclusion is consistent with steadyꢀstate isotopic
transient kinetic analysis (SSITKA) by Davis, which showed that the number of surface
intermediates on CaꢀHAP that desorb to form ethanol (5 µmol mꢀ2) is larger than those that form
1ꢀbutanol (1.1 µmol mꢀ2) and much larger than those that form acetaldehyde (0.082 µmol mꢀ2) at
operating conditions (8.1 kPa C2H5OH, 123 kPa He, 613 K, 75 cm3 minꢀ1)29 similar to those used
here.
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
These results, together with the negative dependence of C–C bond formation rates on ethanol
pressure (Figure 2a), suggest that ethanol inhibits C–C bond formation by displacing
3−
acetaldehyde from the pairs of neighboring acid and base sites (e.g., adjacent CaO and PO4
moieties)26 needed for kinetically relevant acetaldehyde deprotonation (Scheme 1, step 1.5).”
Surface Hꢀatoms will be also formed via O‒H rapture of adsorbed ethanol along with ethoxide
formation (Scheme 1, step 1.2), and these species will increase with the pressure of ethanol.
21
ACS Paragon Plus Environment

ACS Catalysis
Page 22 of 71
1
2
3
4
5
6
Therefore, even if this step is ineligible in Eq.5, both will contribute to the inverse dependence of
C‒C bond formation rates on ethanol pressure.
7
8
9
Collectively, these assumptions regarding surface coverages cause Eq. 5 to take the form:
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
ꢅ
ꢋ
ꢁ ꢇ ꢉ
ꢆ ꢈ
ꢜ
ꢝꢞꢝ
= ꢃꢄꢎꢏ ꢘꢟꢠ ꢁ ꢇ ꢉꢇ
(6)
ꢅ
ꢆ ꢥ
ꢋ
ꢅꢍ ꢋ
ꢊ
ꢣ
which is consistent with the measured dependencies of rC–C on reactant pressures (Figure 2) and
with the in situ FTIR spectra (Figures 4 and S6). Again, the term of number of acid site, [LA], in
the left hand side represents the number of acidꢀbase site pairs. It appears that ethanol, and by
extension other alcohols, couple via aldol condensation of aldehydes formed by dehydrogenation
of the reactant alcohol by direct transfer of Hꢀatoms to the product aldehyde via the MPV
reaction. This sequence of steps likely involves kinetically relevant deprotonation of the
aldehyde reactants on a surface predominantly covered by ethanol (or ethoxide) species.
These results are consistent with the widely accepted mechanism for the “Guerbet Reaction” of
ethanol, which involves aldol condensation for the C–C bond formation reaction between
aldehydes,7,8,22ꢀ26,29 but provide new evidence for the pathways for Hꢀtransfer, the coverage of
reactive species, and the identity of the kinetically relevant step. Scheme 2 summarizes the
important reaction boundaries within this system. The kinetic and spectroscopic data suggest that
enolate formation is the kinetically relevant step for C–C bond formation in the presence of
reactant mixtures containing acetaldehyde and ethanol at 548 K on CaꢀHAP. However, this
conclusion will not necessarily apply for the rate of formation of 1ꢀbutanol or for reactant
mixtures containing only ethanol (e.g., ethanol Guerbet reaction), because hydrogen transfer
steps that dehydrogenate ethanol and saturate product enals have rates comparable to that for Cꢀ
C bond formation. The following sections show how these findings can be used to explain, and
22
ACS Paragon Plus Environment

Page 23 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
predict, the complex product distributions that result from sequences of selfꢀ and crossꢀcoupling
reactions that occur at longer residence time and at high conversions of ethanol over HAP
catalysts.
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. Mechanistic Implications of Cn and Branching Distribution of Products
3.2.1. Effects of Reactant Chain Length on Rate Constants for Coupling C2-C4 Reactants
1ꢀButanol and also heavier (C≥6) products are formed by aldol condensation reactions among C≥2
aldehydes and subsequent hydrogen transfer by a combination of the MPV and surface mediated
Hꢀtransfer mechanisms. The C≥6 products must form by sequential reactions, in which the
primary products from aldol condensations of the acetaldehyde condense with other aldehydes
(C≥2) to create secondary products in a cascade of reactions analogous to the primary reaction
shown in Scheme 2.21,22 Consequently, the Cn distribution of the final products reflects the rates
at which a given product aldehyde (e.g., C4H8O) reacts with any of the other aldehydes present
(i.e., C2–C12).
The rate expression, Eq. (6), can be generalized to describe the C–C bond formation reaction
among heavier aldehydes (CmH2mO) with the rate constant for deprotonation ( ꢃꢄꢦ ) and the
equilibrium constant for desorption ( ꢎꢏ′). The formation of the same CꢀC bond, between the
same reactant aldehydes, can be restated with respect to how it depends on the observed
concentration of the reactant alcohol and an apparent rate constant (ꢃꢙꢂꢗ) that describes the
sequence of adsorption, dehydrogenation, and deprotonation:
ꢅ
ꢋ
ꢅ
ꢁ ꢇ
ꢤ ꢆꢤꢧꢣ
ꢋ
ꢉꢇ
ꢜ
ꢁ
ꢇ
ꢉ
ꢝꢞꢝ
ꢤ
ꢆꢤ
= ꢃꢄ′ꢎꢏ′ ꢘꢟꢠ ꢁ ꢇ ꢉꢇꢋ = ꢃꢙꢂꢗ
(7)
ꢅ
ꢅ
ꢋ
ꢘꢟꢠ ꢁ ꢇ ꢉꢇ
ꢅꢍ ꢋ
ꢊ
ꢣ
ꢆ
ꢥ
ꢣ
ꢆ ꢥ
23
ACS Paragon Plus Environment

ACS Catalysis
Page 24 of 71
1
2
3
4
5
6
7
8
9
ꢅ
ꢋ
where ꢓꢙꢛꢔꢙꢨ and ꢅꢓꢙꢛꢔꢙꢟꢘꢨꢛꢋ are gasꢀphase concentration of Cm aldehydes and alcohols,
respectively. While this restated form of the rate equation does not precisely describe the full set
of intervening elementary steps, it does provide a conceptual bridge from the fundamental
mechanism for C–C bond formation to a functional prediction for the distribution of products
formed. The final Cn distribution of the alcohol products depends on the relative values of ꢃꢙꢂꢗ
for all combinations of m (identity of the carbanion) and n (the aldehyde which is attacked by the
carbanion), which may depend on the size (i.e., Cn) of the reactants.
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 5 shows that the rate constants of C–C bond formation, calculated from Eq. 7 for selfꢀ
coupling (ꢃꢔꢂꢔ , ꢃꢩꢂꢩ , ꢃꢏꢂꢏ ) and crossꢀcoupling (ꢃꢔꢂꢏ , ꢃꢏꢂꢔ ) of C2, C3, and C4 aldehydes
increase by a factor of two from ꢃꢔꢂꢔ to ꢃꢩꢂꢩ, after which all value of ꢃꢙꢂꢗ are similar among
combinations of reactants with Cn greater than two (5 kPa alcohol, 96 kPa H2, 573 K, 1.67 Ca/P
These comparisons (Figure 5) are consistent with previous work that showed empirically fit rate
constants for formation of C4, C6, and C8 alcohols to be similar within an order of magnitude
during ethanol conversion on CaꢀHAP.21 This indicates that differences between the formation
rates of C4–C8 species from condensation reactions of C2–C4 aldehydes and alcohols depends
primarily on the concentration of the reactants and only weakly on the exact identity or structure
of the aldehyde intermediates. These findings (Figure 5), together with expectations that further
increases in chain length will have a diminished effect on reactivity,51 suggest that the values of
kmꢀn will be similar for nearly all reactants with three or more Cꢀatoms.
The underlying factors that determine the degree of branching within the products can also be
explained by fundamental arguments tied to the mechanism for C–C bond formation. An aldol
condensation reaction between two distinct aldehydes (e.g., acetaldehyde and butyraldehyde)
forms one of two potential products, and the structure of the product depends on which of the
24
ACS Paragon Plus Environment

Page 25 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
aldehydes deprotonates prior to nucleophilic attack. Scheme 3 describes the details of the
proposed C–C bond formation mechanism between different two aldehyde surface intermediates
(possessing alkyl groups R1 and R2). Specifically, the deprotonation of aldehyde A with the R1
group will give a different structural isomers comparing to that of product formed by
deprotonation of aldehyde B, containing the R2 group. Linear products (i.e., nꢀaldehydes and
alcohols) form only when the carbanion forms from acetaldehyde and attacks another normal
aldehyde. In contrast, deprotonation of C≥4 aldehydes ultimately form branched products (i.e.,
isoꢀaldehydes and alcohols). Deprotonation energies for aldehydes reflect inductive and resonant
effects that depend on the structure of pendant alkyl groups to the carbonyl function,51 and such
changes may influence also the barriers for catalytic reactions that involve deprotonation.
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
ꢜ
ꢫ,ꢬ
Figure 6 shows that the ratio of formation rates of linear to branched C6 alcohols ( ꢪ = ꢜ
,
ꢫ,ꢌ
where ꢀꢭ,ꢍ and ꢀꢭ,ꢚ are the formation rates of 1ꢀhexanol and 2ꢀethylꢀbutanol, respectively)
changes weakly as a function of inverse temperature on CaꢀHAP (5 kPa C2H5OH, 0.1 kPa
C2H4O, 0.1 kPa C4H8O, 96 kPa H2, 1.67 Ca/P).52 When these alcohols are the main products and
the amount of other enals or aldehydes are negligible, the ratio of the formation rates of the C6
products can be restated using the form of the general rate expression (Eq. 7) such that:
ꢅ
ꢅ
ꢋ
ꢋ
ꢁ ꢇ ꢉ
ꢪ = ꢮꢮ
(8)
ꢆꢞꢈ
ꢈꢞꢆ
ꢆ ꢈ
ꢁ ꢇ ꢉ
ꢈ
ꢯ
Thus, ꢪ depends only on the ratio of the rate constants in Eq. 8 when the concentration of C2 and
C4 aldehydes are equal. Values of ꢪ are near unity (Figure 6) at all temperatures, indicating that
the rate constants for these two pathways are similar, which is consistent with comparisons of kmꢀ
n values for reactions beginning with alcohols (Figure 5).
25
ACS Paragon Plus Environment

ACS Catalysis
Page 26 of 71
1
2
3
4
The change in the value of ꢪ as a function of temperature reflects the activation enthalpy
5
6
7
(∆ꢛꢁ ) for each reaction pathway when the rate ratio (Eq. 8) is restated in the form proposed by
ꢢ
Eyring:51
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
ꢂ∆ꢇ
)
ꢝ
ꢈ
ꢪ = ꢰ ∙ ꢱꢲꢳ ꢴꢂ(∆ꢇ
ꢷ ꢅꢅ
(9)
ꢋ
ꢋ
ꢁ ꢇ ꢉ
ꢝ
ꢆ
ꢈ
ꢆ
ꢵꢶ
ꢁ ꢇ ꢉ
ꢈ ꢯ
where R and T are the universal gas constant and the absolute temperature, ∆ꢛꢁ is the apparent
ꢢ
activation enthalpy for the pathways that deprotonate each Cn aldehyde species, and A is a preꢀ
factor related to entropic changes, respectively. Therefore, the change in the γ with respect to
inverse temperature (Figure 6) reflects the difference between the activation enthalpies for
pathways involving deprotonation of acetaldehyde or butyraldehyde (∆∆ꢛ = ∆ꢛꢁ − ∆ꢛꢁ ).
ꢆ
ꢈ
The calculated ∆∆ꢛ value (−11 ± 4.9 kJ molꢀ1) shows that the apparent barrier for acetaldehyde
deprotonation is slightly lower than that for butyraldehyde on CaꢀHAP, which agrees with the
smaller pKa value of acetaldehyde (14.5) than that of butyraldehyde (15.74).53
Overall, rate measurements (Figures 5 and 6) show that apparent rate constants for selfꢀ and
crossꢀcoupling reactions among C2, C3, and C4 species do not vary significantly with Cn of
aldehyde or alcohols reactants. Moreover, values of ∆ꢛꢁ depend weakly on the structure of the
ꢢ
aldehyde that is deprotonated. These trends may be extended to larger Cn species, which would
suggest that the specific values of kmꢀn (Figure 5) are nearly identical for all alcohols with Cn
greater than three. This finding suggests that the final Cn distribution of the products may be
predicted without individually measuring rate constants for all possible reaction pathways. In the
final section, we show that these product distributions can be predicted for any given extent of
conversion using a simple statistical model that exploits the fact that kmꢀn values are weakly
dependent on the exact structure of the reactant aldehydes and alcohols.
26
ACS Paragon Plus Environment

Page 27 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
3.3. Cascades of C-C Bond Formation Reaction and Product Cn Distribution
3.3.1. The Factors to Continue the C–C Bond Formation Reaction
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
Table 2 shows that significant fractions of C≥6 product alcohols form even at low ethanol
conversions on CaꢀHAP, and the cascade of coupling reactions responsible for these subsequent
steps decreases the yield to 1ꢀbutanol. In practice, aldol condensation reactions will continue
until all reactive groups for coupling (i.e., –C=O and –COH) are eliminated either by
condensation reactions to form larger oxygenates or by dehydration (the major side reaction) of
the alcohols to form hydrocarbons. Consequently, the formation of larger alcohol products
requires much greater selectivities for alcohol dehydrogenation (Scheme 1, steps 1.2–1.3, and
1.9) and aldol condensation (Scheme 1, steps 1.5–1.8) than those for alcohol dehydration.
The selectivity towards larger alcohols depends on both the reaction temperature and the M/P of
the sample. Figure 7a shows that the selectivity to larger alcohols and aldehydes is significantly
greater at 573 K than at 623 K at all ethanol conversions over CaꢀHAP (5 kPa C2H5OH, 96 kPa
H2, 1.67 Ca/P). Specifically, the selectivity towards alcohols and aldehydes is greater than 95%
at an ethanol conversion of 55% (Figure 7a). The greater selectivities towards alcohols and
aldehydes at 573 K than at 623 K suggest that dehydration has greater apparent activation
enthalpies than those for C–C bond formation. Figure 7b shows that the product selectivity
depends strongly on the Ca/P ratio of CaꢀHAP materials (1.61, 1.67, and 1.74 Ca/P) at all ethanol
conversions (5 kPa C2H5OH, 96 kPa H2, 623 K), because the Ca/P ratio largely controls the acidꢀ
base character of the surface.22,23,30 The HAP catalysts with higher base site density, which
correlate to higher Ca/P ratio (Table 1), shows higher selectivities to alcohols and aldehydes.
27
ACS Paragon Plus Environment

ACS Catalysis
Page 28 of 71
1
2
3
4
5
6
7
8
9
These results, together with Figure S3 and previous discussions, show that lower reaction
temperature and use of HAP catalysts with higher densities of base sites and stronger base sites
increase selectivities towards CꢀC bond formation and decrease that for dehydration. Thus, larger
(C≥6) products are obtained at high reactant conversions.
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
3.3.2. Model Describing the Cn and Branching Distribution of Ethanol Guerbet Products
Figure 8 shows that the selectivity towards larger Cn products (e.g., C4–C12) increases with
ethanol conversion due to secondary selfꢀ and crossꢀcondensation reactions among C≥2 aldehydes
over CaꢀHAP (5 kPa C2H5OH, 96 kPa H2, 1.67 Ca/P, 573 K) (similar plots for SrꢀHAP (1.71
Sr/P) are shown in Figure S7). These larger products include a variety of saturated linear and
branched C4–C12 alcohols (C≥14 alcohols are negligible at these conversion ranges) with a
maximum of two 3Cꢀatoms. More than 20 different isomers of C4–C12 alcohols were observed at
ethanol conversion of 20% over CaꢀHAP, and Figure S8 lists all observed alcohols. The
intervening steps that form C≥6 products are analogous to those shown in Scheme 2, which
involves dehydrogenation by the MPV reaction, aldol condensation via kinetically relevant
deprotonation of an aldehyde, and sequential MPV and surface mediated hydrogenation steps to
form the product alcohols. Below, we show that the resulting distribution of Cn and the mixture
of nꢀ and isoꢀalcohols can be explained by extending the discussions related to the formation of
the first C–C bond.
To a first approximation, the condensation reactions within this reaction network can be
categorized as either monomerꢀmonomer coupling (i.e., condensation between C2 aldehydes, also
initiation), monomerꢀoligomer coupling (i.e., condensation between C2 and C≥4 aldehydes), or
28
ACS Paragon Plus Environment

Page 29 of 71
ACS Catalysis
1
2
3
4
5
6
7
8
9
oligomerꢀoligomer coupling (i.e., condensation between C≥4 aldehydes). An average apparent
rate constant can be defined to describe each category (i.e., km–m, km–o, and ko–o, respectively).
Large differences in the values of these rate constants provide different product Cn distributions
that evolve in distinct ways as a function of the conversion of the monomer (acetaldehyde) and
that of the reactive functional group (i.e., –C=O/–COH). For example, when km–o is much larger
than km–m or ko–o, the products form by a chainꢀgrowth mechanism and the product Cn
distribution matches the AndersonꢀSchulzꢀFlory model, as seen in FischerꢀTropsch chemistry.54
Consequently, comparisons of these apparent rate constants can give useful predictions for the
products that will form at a given extent of conversion of the reactive species within networks of
ethanolꢀderived species.
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
For these reaction network, the progress towards completion for condensation reactions is
described by the fractional consumption of the initial number of reactive groups (i.e., –C=O
derived from –C–OH) and is not wellꢀdescribed by the ethanol conversion. The fraction of
reactive groups consumed can be defined as α:39,40
ꢸ = (
(10)
∑
)
ꢹ
ꢹ ꢂ
ꢺ
ꢡꢻꢆ ꢆꢡ
ꢹ
ꢺ
where N0 is the initial number of reactive groups present in the feed (i.e., equal to the number of
ethanol molecules) and N2i is the number of reactive groups remaining among all the molecules
with carbon number of C2i after a given extent of reaction. Thus, the value of α varies from zero
at 0% ethanol conversion to unity when all alcohols and aldehydes, of all Cn, are consumed by
condensation or termination (i.e., dehydration) reactions.
Figure 9 shows that the value of α increases on CaꢀHAP and SrꢀHAP as function of ethanol
conversion, controlled by varying the residence time (5 kPa C2H5OH, 96 kPa H2, 573 K). The
29
ACS Paragon Plus Environment