2,3-Butanediol dehydration catalyzed by silica-supported sodium phosphates

Article abstract of DOI:10.1016/j.apcata.2015.11.043

Catalytic gas phase dehydration of 2,3 butanediol (BDO) was studied in the fixed bed reactor over silica-supported sodium phosphates for a wide range of Na/P ratios (Na/P = 0-3) and variable content of phosphates. Major products were 1,3-butadiene (BD) and methyl ethyl ketone (MEK) for long contact times. 3-Buten-2-ol (3B2OL) and MEK are major products for the short contact times. Achieved yields of elimination products (BD + 3B2OL) were over 60% per pass at Na/P = 1.8-1.9 which is the optimum combination of acidic and basic components for the long term production of BD or 3B2OL from BDO. Minor products detected at short contact times (2,3-epoxybutane, 2,3-butanedione and acetoin) imply that hydrogen transfer and dehydrative epoxidation are minor pathways. Mass transport limitations are more apparent for BDO → 3B2OL → BD route (elimination) than for BDO → MEK route (rearrangement). This leads to the lower selectivity of elimination products at long contact time and/or insufficient flow rate of the carrier gas. 2-butanol dehydration using the same catalysts shows considerable selectivity to 1-butene (1-BT) providing evidence for E_1cB elimination mechanism that requires concerted action of acid-base pairs. NH₃ and CO₂ TPD studies confirmed existence of weak acid and basic sites for the sodium phosphates supported on silica. High affinity adsorption of pyridine (Py) and benzoic acid (BA) shows maximum of acid/base ratio at Na/P～1.5-2. According to XRD and ³¹P MAS NMR data the catalysts providing the best performance are composed of pyrophosphates and tripolyphosphates finely dispersed on the surface of silica. XPS shows considerable migration of surface sodium cations into the bulk when Na/P ratio exceeds 1.4.

Full text of DOI:10.1016/j.apcata.2015.11.043

Applied Catalysis A: General 511 (2016) 156–167
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
2,3-Butanediol dehydration catalyzed by silica-supported sodium
phosphates
Wooyoung Kim, Wookyun Shin, Kyoung Jun Lee, Hyohak Song, Hyung Soon Kim,
Doyoung Seung, Igor N. Filimonov∗
GS Caltex Corporation, Research Center, 104-4 Moonji-dong, Yuseong-gu, Daejeon 305-380, Republic of Korea
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 30 August 2015
Received in revised form 9 November 2015
Accepted 24 November 2015
Available online 2 December 2015
Catalytic gas phase dehydration of 2,3 butanediol (BDO) was studied in the fixed bed reactor over
silica-supported sodium phosphates for a wide range of Na/P ratios (Na/P = 0–3) and variable content
of phosphates. Major products were 1,3-butadiene (BD) and methyl ethyl ketone (MEK) for long contact
times. 3-Buten-2-ol (3B2OL) and MEK are major products for the short contact times. Achieved yields
of elimination products (BD + 3B2OL) were over 60% per pass at Na/P = 1.8–1.9 which is the optimum
combination of acidic and basic components for the long term production of BD or 3B2OL from BDO.
Minor products detected at short contact times (2,3-epoxybutane, 2,3-butanedione and acetoin) imply
that hydrogen transfer and dehydrative epoxidation are minor pathways.
Keywords:
2,3-Butanediol dehydration
Silica-supported sodium phosphates
1,3-Butadiene
Methyl ethyl ketone
3-Buten-2-ol
Mass transport limitations are more apparent for BDO → 3B2OL → BD route (elimination) than for
BDO → MEK route (rearrangement). This leads to the lower selectivity of elimination products at long
contact time and/or insufficient flow rate of the carrier gas.
Acid-base site density
2-butanol dehydration using the same catalysts shows considerable selectivity to 1-butene (1-BT)
providing evidence for E1cB elimination mechanism that requires concerted action of acid-base pairs.
NH3 and CO2 TPD studies confirmed existence of weak acid and basic sites for the sodium phosphates
supported on silica. High affinity adsorption of pyridine (Py) and benzoic acid (BA) shows maximum of
acid/base ratio at Na/P∼1.5–2.
According to XRD and 31P MAS NMR data the catalysts providing the best performance are composed of
pyrophosphates and tripolyphosphates finely dispersed on the surface of silica. XPS shows considerable
migration of surface sodium cations into the bulk when Na/P ratio exceeds 1.4.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
ciency of fermentation [1] and high titers of BDO in microbial broth
[2]. BDO can be further dehydrated to MEK (over 1.3 million met-
Renewable sources are becoming important alternatives to
classical petroleum based methods for the production of general
platform chemicals and value-added products. Among a variety of
possible fermentation strategies formation of 2,3-butanediol (BDO)
from the carbohydrate feedstock is attractive in terms of high effi-
ric tons of current market size) and BD (over 10 million metric
tons—the size of the current market).
Over 50 years of research on BDO dehydration were recently
reviewed by Makshina et al. [3] where one can find a good survey
of the work performed on the dehydration of C4-diols. For the sake
of a very brief introduction we would like to emphasize that BDO
dehydration is just one of the typical examples of the dehydration of
vicinal diols. Usually the dehydration of vic-diols almost inevitably
leads to the pinacol-pinacolone rearrangement explicitly discussed
almost in every organic chemistry textbook. All the experimental
evidence retrieved by Makshina et al. [3] shows that under condi-
tions of acidic catalysis mostly MEK is formed indicating preferable
rearrangement route. In the case of bifunctional acid-base cata-
lysts 1,2-elimination becomes more pronounced. The most cited
example of BDO dehydration is a process conducted using ThO2
Abbreviations: Actn, acetoin (3-hydroxybutanone); BA, benzoic acid; BD, 1,3-
butadiene; BDO, 2,3-butanediol; BDOn, 2,3-butanedione; 1-BT, 1-butene; 2-BT,
2-butene; 3B2OL, 3-buten-2-ol; CA, crotyl alcohol (cis-, trans-); 2M2P1OL, 2-methyl-
2-propen-1-ol (isobutenyl alcohol); IBA, isobutyraldehyde (2-methylpropanal);
MEK, methyl ethyl ketone (2-butanone); Py, pyridine; 1.82Na P/SiO2 9.4, silica sup-
ported sodium phosphate having ratio Na/P 1.82 (mol/mol) and 9.4 wt.% of sodium
phosphate phases; E/R, elimination/rearrangement ratio: (SBD + S3B2OL)/(SIBA + SMEK).
∗
Corresponding author. Tel.: +82 42 866 1634; fax: +82 42 866 1649.
E-mail addresses: FLM@gscaltex.co.kr, infilimonov@gmail.com (I.N. Filimonov).
http://dx.doi.org/10.1016/j.apcata.2015.11.043
0926-860X/© 2015 Elsevier B.V. All rights reserved.

W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
157
catalyst that under the appropriate conditions produces mostly BD
[4,5]. ThO2 has both acid and base centers [6] and provides 40–80%
selectivity to 1-butene (1-BT) in 2-butanol dehydration [7] that is
interpreted as a manifestation of E1cB mechanism requiring con-
certed acid-base action.
Important achievements in BDO dehydration to BD and 3B2OL
were accomplished by the research group of professor Sato using
rare earth oxides as catalysts [8–11]. The reaction was performed
using hydrogen as a carrier gas and 275–425 ◦C temperature range.
The catalysts provided stable performance for 5 h time on stream.
The highest yield of BD (88% at 411 ◦C) was reported for Sc2O3 [11]
while at the lower temperature (325 ◦C) about 85% yield to 3B2OL
was achieved [10]. These data are excellent indeed but rather short
time on stream reported (5 h) and the necessity of hydrogen usage
as a carrier gas may lead to certain complications for the scaled-up
production.
Another interesting recent research by Kim et al. [12] on alterna-
tive pathways for vic-diol dehydration shows that for metal doped
silicas considerable amount of 2,3-epoxybutane may be obtained
from BDO at 400 ◦C using the mixture of nitrogen and water vapor
as a carrier gas. Cs/SiO2 was the most active catalyst for this process.
The basic component, indeed, plays important role in the clas-
sic acid-catalyzed pinacol transformations—i.e., when a conjugate
base was added to the homogeneous acid catalyst the amount of
elimination products (e.g., 2,3-dimethyl-3-butene-2-ol) increased
[13].
second route is a sequential 1,2-elimination of water leading to
3B2OL after the first dehydration and BD after the second one.
Assuming the mechanism of elimination reactions [16] 3B2OL
and BD should arise due to the same chemistry of E1cB-type
eliminations, while MEK and IBA may be formed by the stan-
dard acid-catalyzed E2 elimination. Thus the ratio of selectivities
(SBD + S3B2OL)/(SIBA + SMEK) will characterize the relative rates of the
two major pathways of BDO dehydration. For a brevity and conve-
nience we will term it elimination/rearrangement ratio (E/R).
In this preliminary publication we would like to discuss the case
of catalyzed vapor phase BDO dehydration over silica-supported
sodium phosphate catalysts with a special emphasis on the varia-
tion of Na/P ratio and its effect on the selectivity to major products.
Wider class of catalysts and supports will be discussed in the next
publications.
To our knowledge, a limited number of publications deals with
the alkali metal phosphates supported on silica. Lion et al. char-
acterized acidity of silica-supported sodium phosphate catalysts
by FTIR using adsorption of pyridine (Py) and piperidine as probe
molecules [17]. Gunter et al. [18], Tam et al. [19,20] characterized
silica-supported sodium phosphate in the lactic acid conversion.
Zhang et al. [21] characterized conversion of methyl lactate cat-
alyzed by silica-supported sodium phosphates and interactions of
NaH2PO4 and H3PO4 with silica using 31P MAS NMR and DFT cal-
culations [22].
Thus taking into account theoretic background [14] and classic
review on the dehydration of diols by Bartók and Molnár [15] it is
possible to describe BDO dehydration as a set of two parallel routes
(Scheme 1). The first route represents traditional rearrangement
mechanism via 1,2- hydride shift leading to MEK or via 1,2-methyl
shift leading to isobutyraldehyde (IBA). Since hydride shift is faster
than methyl shift MEK dominates over IBA in the products. The
2. Experimental
2.1. Catalyst preparation
In this study Davisil type silicas were used: grade
636 (36–60 mesh); grade 635 (60–200 mesh), grade 633
Scheme 1. Major (solid arrows) and minor (dashed arrows) pathways of BDO dehydration.

158
W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
Table 1
Composition and textural properties of catalysts.
Sample name
NaOH:H3PO4 (mol:mol)
Na/P, ICPa
% wt.
Specific surface area, (m2/g)
Pore volume (ml/g)
SiO2
–
–
442
279
419
388
387
311
330
336
363
362
396
388
398
384
224
448
371
338
273
0.90
0.58
0.82
0.74
0.73
0.70
0.70
0.70
0.74
0.72
0.73
0.68
0.72
0.67
0.65
0.79
0.64
0.63
0.51
Na/SiO2 5.8b
NaOH only
H3PO4 only
0.50
1.00
1.50
1.67
1.80
1.82
1.84
1.89
2.00
2.25
2.50
3.00
1.82
1.82
1.82
570
5.79
9.78
9.78
9.78
9.78
9.78
9.78
9.43
9.78
9.78
9.78
9.78
9.78
9.78
4.95
17.24
23.81
29.41
P/SiO2 9.8c
0.08
0.81
1.42
2.05
2.39
2.49
2.33
2.30
2.29
2.55
2.76
2.91
3.36
2.08
2.57
2.61
2.57
0.5Na P/SiO2 9.8
1Na P/SiO2 9.8
1.5Na P/SiO2 9.8
1.67Na P/SiO2 9.8
1.8Na P/SiO2 9.8
1.82Na P/SiO2 9.4
1.84Na P/SiO2 9.8
1.89Na P/SiO2 9.8
2Na P/SiO2 9.8
2.25Na P/SiO2 9.8
2.5Na P/SiO2 9.8
3Na P/SiO2 9.8
1.82Na P/SiO2
5
1.82Na P/SiO2 17.2
1.82Na P/SiO2 23.8
1.82Na P/SiO2 29.4
1.82
a
–Analysis of water soluble material (catalyst washed by water).
–5.79% wt. Na2O.
–9.78% wt. P2O5.
b
c
(200–425 mesh) obtained from Sigma–Aldrich. Before the
impregnation step silica gel was precalcined at 500 ◦C for 3 h.
Calculated volumes of 5 mol/L sodium hydroxide solution (Sam-
chun Chemicals) and orthophosphoric acid (Sigma–Adrich) diluted
to 5 mol/L were mixed in the required proportion. Deionized water
was added to maintain the volume required for the incipient wet-
ness preparation of catalysts.
Vacuum impregnation method according to Lu et al. [23] was
employed. Before the impregnation silica gel was evacuated at
<10−2 Torr, 350 ◦C for 3 h. After cooling down to an ambient tem-
perature under vacuo sodium phosphate solution was added to the
gel and left for 1 h. Then the wet solid was evacuated at 90 ◦C for
2–3 h. Dried solid was then calcined in air at 500 ◦C for 3 h (heating
rate 2 ◦C/min).
The reactor was held at a constant temperature in the 5-zone
oven regulated by individual thermocouples. Before each experi-
ment the catalyst was pretreated in N2 flow (100 ml/min) at 400 ◦C
for 1 h (heating rate 10 ◦C/min).
The effluent gas was analyzed by an online GC (Agilent 6890N)
using an FID detector and Restek RTX-VRX capillary column (60 m).
For the identification purposes liquid products were collected and
analyzed by an offline GC/MSD (Agilent 5973) using the same cap-
illary column. Typical carbon material balance was within 5% for
kinetic regime conditions. Carbon balance deteriorates for condi-
tions when mass transport limitations become evident (e.g., low
carrier gas flow rates). This situation will be discussed later in
the text and corresponding mass balance values will be presented
whenever is necessary.
In the typical catalytic run the catalyst was kept for 1 h at 400 ◦C
under the flow of inert gas, then BDO feed was introduced. After
reaching the steady state (usually within 10–20 h of time on stream)
the sampling was continued for additional 20 h to ensure catalyst
stability and/or notice any deactivation effects. In general typical
catalysts that do not possess high basicity show excellent stability
and within a single run several hundred hours of time on stream
may be achieved without deactivation. More data on the stability
of the catalysts are presented in the Supplementary materials.
2.2. Reactor setup and catalytic test
Catalysts (0.03–1.0 g) were mixed with 3 g of silica sand (Sam-
chun Chemicals) and packed between quartz wool plugs in the
center of 50 cm long tubular reactor. Both stainless steel tubular
reactors (1/4ꢀꢀ and 1/2ꢀꢀ) as well as quartz ones (3/8ꢀꢀ) were tested
in blank runs. In the experiments when long contact times were
required 8–12 g of catalyst were placed into 3/8ꢀꢀ quartz reactor
without dilution by quartz sand. Material of reactor was of minor
importance compared to the void volume effects in the blank runs.
Glass beads of appropriate diameter filling all void space of the reac-
tor were used to suppress non-specific reactions. Acetoin (Actn)
is a characteristic product of non-specific processes in the blank
runs and overall conversion of BDO in blanks did not exceed 1–4%
for 1/4ꢀꢀ reactor under the typical test conditions. When long con-
tact time conditions were used Actn was absent and in this case
non-specific reactions are negligible.
Liquid BDO (97% Sigma–Aldrich) was fed by the syringe pump
(0.01–1 ml/h) into the empty 5 l volume kept at 180 ◦C (depending
on BDO partial pressure) to eliminate concentration fluctuations
(caused by the syringe pump delivery and droplet evaporation into
the gas phase). Nitrogen carrier gas was used to maintain desired
partial pressure of BDO. Except where noted otherwise N2 flow was
fixed to 100 ml/min. All communications were maintained at 180 ◦C
under the typical reaction conditions. 1 atm system pressure was
kept constant in all the experiments.
2.3. Catalyst characterization
Water soluble sodium and phosphate content was determined
by suspending 0.4 g of catalyst in 80 ml of water, shaking the mix-
ture for 30 min, filtration and analysis of Na/P content in the filtrate
using ICP-AES. More details on this procedure may be found in
Supplementary materials.
Surface area and porosity of all the samples was determined
by N2 adsorption-desorption isotherms at 77 K using ASAP 2020 or
Tristar II apparatuses (Micromeritics). The samples were evacuated
at 350 ◦C overnight before the measurements.
Acid-base site density and distribution was determined by
AutoChem II (Micromeritics) by temperature programmed desorp-
tion (TPD) of ammonia and carbon dioxide. About 0.3 g catalyst
were pretreated in helium flow (50 ml/min, 550 ◦C), cooled down
to 100 ◦C and exposed to 50 ml/min flow of pure CO2 or 15% v/v
NH3/He flows to saturate base or acid sites. Physically adsorbed
species were removed by He flow (100 ◦C, 3 h) and then TPD trace

W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
159
A
B
Fig. 1. (A) Dependence of specific surface areas on the Na/P ratio of silica-supported sodium phosphates. (B) Dependence of pore volume on the Na/P ratio of silica-supported
sodium phosphates.
was recorded at 10 ◦C/min heating rate and 50 ml/min He flow. Des-
orbed ammonia was trapped by 0.05 mol/l solution of H2SO4 and
desorbed carbon dioxide was trapped by 0.05 mol/l NaOH solution.
Total amounts of desorbed species were checked by titration.
In addition to TPD acid-base site density was evaluated by
Tanabe-Yamaguchi [24] titration method in a spectrophotometric
variant described by e.g., Bautista et al. [25] using freshly pre-
pared solutions of pyridine (Py) and benzoic acid (BA) in anhydrous
cyclohexane (all the chemicals from Sigma–Aldrich). Analysis of
titration data revealed considerable amount of weak adsorption
centers both for Py and BA thus instead of one-point measurement
we used 6 point isotherm measurement and treated our data as an
adsorption of probe molecules onto a limited number of high affin-
ity sites and excessive number of low affinity non-specific sites. In
these experiments about 0.1 g of the sample placed into each of the
six 20 ml capped vials, 10 ml of Py and BA solutions in cyclohexane
(2.5 × 10−3–2.5 × 10−2 mol/l) were added and the adsorption pro-
ceeded for 5 h under a vigorous stirring at room temperature. The
amount of adsorbed Py and BA was determined spectrophotomet-
rically at the wavelength of maximum absorption within the range
of Lambert-Beer law. More details on the determination of acid-
base sites by Py and BA adsorption may be found in Supplementary
materials.
202.5 MHz. The magic angle spin speed used for 31P spectral record-
ing was 6 and 8 kHz to distinguish isotropic resonances from the
side bands.
Sodium phosphate dibasic dodecahydrate (Na2HPO4·12H2O)
was used as a standard for isotropic shift positions.
3. Results and discussion
Textural properties and sample compositions are presented in
Table 1. All the catalysts have high surface area and good porosity.
One notable feature of the textural properties is a somewhat
strange dependence of specific surface area on the chemical com-
position of phosphate phase (Fig. 1A). Sigmoid shape of the curve
gives evidence of the fast rise of specific surface within the nar-
row range of Na/P ratios when 1.8 < Na/P < 2. This fast increase of
the surface area with Na/P ratio is contrary to the expected alkali-
induced textural shrinkage of silica gel leading to the substantial
decrease of silica surface area [26]. May be the basicity of solutions
is moderate or the preparation time is too short for a sufficient
shrinkage of the silica gel under the current conditions. (Indeed,
we noticed considerable loss of specific surface area only during
the hydrothermal ageing of the silica support in sodium phosphate
solutions). Plausible explanation of this increase of the surface area
may be related to the hypothetical increase of the aggregation of
sodium phosphate phase with the increase of NaOH/H3PO4 ratio.
This aggregation may result in the opening of certain amount of
blocked pores. Indeed, the dependence of pore volume on the Na/P
ratio within certain error level supports this idea (Fig. 1B). Unfor-
tunately, very poor crystallinity of supported sodium phosphates
cannot give us substantial evidence on this interpretation.
Diffractograms of finely powdered samples were collected on
Xpert Pro (PANAlitical) using CuKa radiation at 45 kV power and
40 mA current.
X-ray photoelectron spectroscopy (XPS) was carried out on a
KRATOS AXIS electron spectrometer equipped with Al K␣ radi-
ation. The binding energies were corrected using C1s as an
internal standard (284.5 eV). The peaks were fitted by mixed
Gaussian–Lorentzian functions with subtraction of Shirley type
background.
XRD results (see lower) show that within 5–17 wt.% of loading
no diffraction peaks are observed suggesting almost a monolayer
coverage within this range. Assuming simple geometric calculation
this gives a rough estimate of 0.50 nm2 per each PO4 tetrahedron.
31P magic angle spinning (CP/MAS) nuclear magnetic resonance
(NMR) spectra of the samples were recorded on a Bruker AVANCE
II 500 WB (500 MHz) spectrometer operating at a frequency of

160
W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
Scheme 2. Possible pathways for the formation of 2M2P1OL and CA.
More about the coverage by phosphate species is presented in the
XRD section.
Hydrogen transfer reactions leading to hydrogenation and
dehydrogenation were observed earlier for the concentrated
non-supported phosphoric acid [27] and thus the formation of
dehydrogenation products seems to be possible. Another evident
indication of hydrogen transfer reactions is a formation of small
amounts of 1- and 2-butenes (BT) observed for all the catalysts.
Probably BT is formed as a result of hydrogenation of BD.
Dehydrogenation usually accompanies dehydration when basic
sites are present [28]. Moreover it is possible to find statements that
dehydrogenation reactions may be a good indication of E1cB mech-
anism [29]. And indeed Table 2 shows that more basic (Na/P > 1)
catalysts exhibit dehydrogenation activity, especially for a more
basic case when Na/P > 2. It is important to note again that the
amount of dehydrogenation products decreases virtually to zero
when long contact times are used combined with a lower quality
of the mass balance.
Since surface sodium phosphate species seem to be responsible
for the catalysis and they may be easily removed by water wash
then we used the latter procedure to analyze sodium and phos-
phate content of water soluble material (Table 1). There is linear
relationship between Na/P ratios used in the synthesis and Na/P
ratios determined by ICP (see Supplementary Materials). But clearly
the ratios determined by ICP are higher than synthetic ratios. Prob-
ably certain part of phosphate species cannot be removed by the
water extraction due to the stronger binding to silica.
3.1. Catalytic properties
3.1.1. Product overview
Catalytic data for the studied catalysts are accumulated in
Table 2. Typical behavior of the catalyst during the time on stream
for long and short contact times is presented in Fig. 2A–C. A majority
of the sodium phosphate containing catalysts has good conversion
of BDO and shows high selectivity to the elimination products (BD,
3B2OL) and lower selectivity to the rearrangement products (IBA,
MEK).
Silica support itself as well as basic silica promoted by NaOH
only (Na/SiO2) have poor catalytic activity and show poor car-
bon balance. Phosphoric acid supported on silica has high activity
and higher selectivity to MEK formation. It is important to note
that more acidic catalysts like silica supported phosphoric acid
(P/SiO2) and 1Na P/SiO2 (Na/P = 1) have relatively high selectiv-
ity to butanedione (BDOn) and Actn indicating considerable level
of dehydrogenation for the experiments with short contact times.
When the same experiments were repeated at a long contact time
dehydrogenation products were practically absent (Table 2 and
Fig. 2A). At the same time certain drop in the carbon balance was
observed (within 10–15%) indicating that probably dehydrogena-
tion products were consumed for a coke build-up.
Thus assuming BDOn and Actn formation we have to add one
more route to BDO transformation–dehydrogenation (Scheme 1).
These products can be observed only for a short contact time due
to secondary unspecified processes at long contact times.
Another characteristic minor products discovered during the
short contact time conversion are two butenols: 2-methyl-2-
propen-1-ol (2M2P1OL, isobutenyl alcohol) and cis-, trans-crotyl
alcohol (CA, 2-butenol), see Table 2 and Fig. 2C. These two alcohols
are observed mostly for the catalysts having ratio 1 < Na/P < 2 and
the yields of both alcohols do not exceed 6%. Following the stan-
dard carbocation view on pinacol rearrangement [13], formation of
these types of products is not surprising and Scheme 2 explains pos-
sible pathways of their formation. Taking into account that these
alcohols have been formed via elimination route then we also cal-
culated E/R ratio using these products as well. Since these alcohols
are minor products then E/R ratio does not change much when
considering CA and 2M2P1OL formation. Vanishing of these prod-
ucts during the long contact time seems also to be related to the
secondary reactions leading to the coke build-up.

W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
161
Fig. 3. Elimination-rearrangement ratio E/R upon the variation of contact time
and carrier gas flow rate: (a) W/F = 0.90 (0.18 g catalyst 1.82Na P/SiO2 17.2,
0.2 ml/h BDO, 100 ml/min N2, 400 ◦C); (b) W/F = 30.3 h (5.98 g catalyst
1.82Na P/SiO2 17.2, 0.2 ml/h BDO, 100 ml/min N2, 400 ◦C); (c) W/F = 31.5 h
(6.21 g catalyst 1.82Na P/SiO2 17.2, 0.2 ml/h BDO, 20 ml/min N2, 400 ◦C); (d)
W/F = 6.2 h (6.14 g catalyst 1.82Na P/SiO2 17.2, 1 ml/h BDO, 100 ml/min N2, 400 ◦C).
Finally, we also detected small amounts of cis-, trans-2,3-epoxy
butanes (BTO, see Fig. 2C and Table 2) that clearly result from
the dehydrative epoxidation already discovered for the basic alkali
metal-promoted silicas [12]. Thus we have to add one more path-
way to Scheme 1. According to Table 2 more basic samples provide
higher selectivity to epoxides. The epoxides also vanish during the
operation at a long contact time indicating their possible consump-
tion for the coke build-up.
3.1.2. Catalyst performance and mass-transfer limitations
In this study we will focus mostly on the main products of
reaction: BD, 3B2OL, IBA, MEK providing 80–90% of all the prod-
ucts and resulting from the elimination and rearrangement routes
(Scheme 1).
Typical evolution of the products and conversion of BDO dur-
ing the catalytic run is presented in Fig. 2A–C. Long contact time
provides quite simple product pattern showing only IBA, MEK and
BD as major constituents of the effluent at 100% conversion level of
BDO (Fig. 2A) having steady selectivity within ca. several hundred
hours of the catalytic run. E/R ratio reaches steady value after 10 h
of time on stream (Fig. 3b) in this case.
It is necessary to emphasize that selectivities of the main prod-
ucts depend on the contact time and carrier gas flow rates (Fig. 3).
For relatively high flow rates of the carrier gas (100 ml/min) and
small rate of BDO feed E/R ratio is relatively high and stable within
10–50 h time on stream (Fig. 3b). When a flow rate of the carrier is
fixed to low values (e.g., 10–30 ml/min) keeping the other parame-
ters same E/R ratio declines much faster (Fig. 3c), the plateau of high
E/R values diminishes to 8–14 h period of time on stream. Under the
low feed of a carrier the quality of the mass balance becomes worse
(it closes within 85–90%). Used catalyst in the latter case becomes
darker indicating faster coking. Upon the increase of BDO feed rate
from 0.2 ml/h to 1 ml/h (decreasing the contact time) and keep-
ing high (100 ml/min) carrier gas flow rate we see again relatively
stable but lower E/R ratio (Fig. 3d).
Fig. 2. (A) Typical evolution of selectivities to major products and conversion of
BDO for W/F = 30.3 (5.98 g catalyst 1.82Na P/SiO2 17.2, 0.2 ml/h BDO, 100 ml/min N2,
400 ◦C). (B) Typical evolution of selectivities to major products and conversion of
BDO for W/F = 0.90 (0.18 g catalyst 1.82Na P/SiO2 17.2, 0.2 ml/h BDO, 100 ml/min N2,
400 ◦C). (C) Typical evolution of selectivities to minor products for W/F = 0.90 (0.18 g
catalyst 1.82Na P/SiO2 17.2, 0.2 ml/h BDO, 100 ml/min N2, 400 ◦C).
In the case of short contact times the evolution of products and
their pattern changes substantially (Fig. 2B–C): 3B2OL becomes

162
W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
Fig. 4. ꢁ–Dependence of BDO conversion on the time on stream; ꢀ–dependence
of 3B2OL conversion on time on stream (0.1 g 1.82Na P/SiO2 9.4, 0.2 ml/h BDO or
3B2OL, 100 ml/min N2, 400 ◦C).
Fig. 5. First-order rate equation fitting the data of BDO conversion X at steady state
conditions and variable contact times (W/F) for the catalyst 1.82Na P/SiO2 9.4 at
400 ◦C (insert—exploded view for W/F = 0-1 h).
dominating within the first 30 h time on stream. MEK and IBA
selectivities do not change much and minor products appear in the
stream. We see steady decrease of BD selectivity with simultaneous
increase of 3B2OL selectivity. The E/R ratio reaches steady values
after ca. 20 h time on stream in this case (Fig. 3a) and after 35 h time
on stream begins to decline.
Taking into account all these features (dependence of the selec-
tivity on a contact time and carrier gas flow rate) clearly we have
a situation of mass-transfer limitations in the heterogeneous cat-
alytic process. Obviously, we have to find parameters of a kinetic
regime when a catalytic reaction itself becomes a limiting step of
the overall process.
In the ideal case when a catalyst does not deactivate and is in
its steady state this is relatively easy to realize. Our case clearly is
not ideal: there is certain level of deactivation that becomes more
evident when low flow rates of the carrier gas are used (Fig. 3c).
For the case of a relatively slow deactivating catalyst we can find
certain time frames when the catalyst is almost in a steady state (i.e.,
changes of the conversion-selectivity are small). We can see such
regions in Fig. 3 when plateaus in selectivity patterns are clearly
observed.
Another important consideration can be found in the mecha-
nism of BDO dehydration itself—Scheme 1. Actually we have two
distinct sequential steps: BDO → 3B2OL and 3B2OL → BD. Each of
these steps has its own kinetics, deactivation features etc. And
indeed, as we can see in Fig. 4 the conversion is stable for BDO
dehydration within 30 h time on stream while the same catalyst
under the same conditions deactivates relatively fast in the case of
3B2OL dehydration. Therefore, the catalyst may be quite stable and
have good performance in BDO dehydration step and may be not
really good in 3B2OL dehydration. Taking into account that changes
in 3B2OL/BD selectivity in Fig. 2B reflect deactivation of the cata-
lyst in 3B2OL → BD step, while the catalyst is relatively stable in
BDO → 3B2OL step, then within a certain accuracy we may be con-
fident about the possibility to determine a kinetic regime for BDO
dehydration step. Since for the most of the catalysts conversion of
BDO is quite stable within 30 h time on stream and usually after
10–20 h time on stream we get a plateau in selectivities then we
can use these parameters as markers for the steady state within a
certain approximation level, of course.
In order to find out experimental conditions of catalytic conver-
sions related to the kinetic regime and characterize mass-transfer
limitations a full set of experiments with respect to the mass,
feed rate, particle size, and carrier gas flow rate was performed.
Koros-Novak [30] type of approach to find out kinetic regime was
applied (mass of the catalysts varied from 0.032 g to 1 g and the
feed of BDO varied from 0.2 to 1 ml/h, 100 ml/min N2 flow rate).
The catalytic data were accumulated for 20–30 h to ensure steady
conversions and selectivities. The data are presented in Fig. 5 and
give an idea what region of possible experimental parameters may
be referred to kinetic regime, since good linearity is observed within
W/F = 0.1–1.0 h.
Quite important issue in the catalytic dehydration of BDO is an
external mass-transfer taking into account evident effects of the
carrier gas flow rate. Direct evaluation of this effect is presented
in Fig. 6. Clearly IBA and MEK selectivity is not susceptible to the
variation of N2 flow rate. At the same time BD and 3B2OL are sup-
pressed when a carrier gas flow rate decreases, especially when the
flow rate is less than 40 ml/min. Another very important feature of
the low carrier gas flow rates is a depreciation of the carbon balance
that almost linearly decreases upon the decrease of the carrier gas
flow rate from 80 ml/min to 10 ml/min. We did not make attempts
of a more detailed study of this effect, but our observations indicate
that at high concentration of 3B2OL and epoxides quick accumu-
lation of the hard, probably polymeric substance occurs in the
tube communications after the reactor. This implies that at higher
concentrations of the products polymerization reaction between
unsaturated hydrocarbons and epoxides occurs which leads to the
deterioration of the mass balance.
Taking all the data relevant to the mass transfer into account, the
most part of the catalytic runs was conducted using 0.15–0.22 g of
catalyst at 0.2 ml/h liquid BDO feed rate using 100 ml/min nitro-
gen carrier gas rate. At these conditions there is no appreciable
effect of mass-transfer limitations and steady conversions of BDO
are achieved.

W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
163
Fig. 8. ꢁ—1-BT/2-BT ratio of selectivities of 2-butanol dehydration and
ꢂ—conversion of 2-butanol over 1.82Na P/SiO2 9.4 catalyst (0.2 ml/h 2-butanol feed
rate, 0.17 g catalyst, 100 ml/min N2 flow rate).
Fig. 6. Dependence of selectivities on the carrier N2 flow rate:
ꢀ—SBD + 3B2OLꢁ—SIBA + MEK; ꢁ—quality of the mass-balance.
3.1.3. Na/P ratio and selectivity to main products
a maximum around Na/P = 1.8–1.9 suggests optimum combination
of acid and base sites required by E1cB mechanism for 3B2OL and
BD formation.
Clearly different sets of products that are obtained at short and
long contact time nicely correspond to the sequential steps of BDO
dehydration in the elimination route (Scheme 1). Complete con-
version of BDO into BD and MEK is achieved at long contact times
permitting both steps of dehydration to occur. Dominance of 3B2OL
in the products of BDO dehydration at short contact times indicates
that only one dehydration step is realized.
Assuming the effects of a conjugated base in the pinacol trans-
formation for the homogeneous catalysts [13] there should be a
certain opportunity in the tuning of E/R ratio by the changing the
acid-base balance of the catalyst. This may be easily achieved by a
simple variation of Na/P ratio during the preparation of the cata-
lysts.
The dominance of acidic sites (e.g., in the case of supported phos-
phoric acid) provides mostly MEK formation with a low selectivity
to 3B2OL + BD. For more basic samples having Na/P = 2.5–3 ratio we
observed relatively fast deactivation, poor material balance, sup-
pressed selectivity to all target products (MEK, BD, 3B2OL) and
considerable increase of by-product formation. Due to a relatively
fast deactivation it is impossible to get reliable catalytic data for the
basic samples.
3.1.4. 2-Butanol dehydration
Dehydration of 2-BuOH represents a good test for the E1cB mech-
anism since for this mechanism we expect formation of Hofmann
orientated product 1-butene (1-BT) [31]. In this experiment only
1- and 2-butenes were observed (Fig. 8). 1-BT/2-BT ratio increases
from 0.5 to 0.7 when increasing the temperature from 300 to 450 ◦C.
This is indicative of a higher activation barrier required for E1cB
type of elimination in this case. Formation of sufficient amounts
of Zaitzev product 2-BT may indicate that E2 elimination cannot
be ruled out. Rather interesting feature of a maximum of 2-BuOH
conversion observed at 400 ◦C clearly indicates a transition of the
limiting steps of the catalytic reaction due to the change of adsorp-
tion equilibria with the temperature.
The effect of Na/P ratio used in a catalyst preparation procedure
on the E/R ratio is presented in Fig. 7A. Volcano-shape curve with
A
B
3.2. XRD
Samples loaded with 5–17% wt. of sodium phosphates did not
reveal any diffraction pattern except the diffuse line typical to
amorphous silica. Starting from 24% wt. very weak diffraction peaks
appear superimposed on the diffuse line of support. An increase of
the sodium phosphate content improves the intensity of diffraction
peaks (Fig. 9)
In order to get better understanding of the formation and variety
of phases non-supported phosphate was prepared the same way
but omitting the addition of a silica support. In this case we obtain
the sample with a good crystallinity and low specific surface (ca.
0.5 m2/g). Phase analysis of the XRD pattern indicates the formation
of a sodium pyrophosphate phase Na4P2O7 with a smaller contribu-
tion of sodium tripolyphosphate Na5P3O10 and unidentified minor
peaks probably belonging to higher polyphosphates. Assuming the
Fig. 7. (A) Dependence of E/R ratio in catalytic tests on the Na/P ratio of sil-
ica-supported sodium phosphates at short contact times (W/F = 0.82-0.89 h). (B)
Dependence of the ratio of irreversibly adsorbed Py/BA on the Na/P ratio of silica-
supported sodium phosphates.

164
W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
Fig. 10. 31P MAS NMR of silica supported 1.82Na P/SiO2 9.8 (a) and
1.82Na P/SiO2 17.8; (b) pure sodium pyrophosphate phase (c); non-supported
sodium phosphate powders with compositions Na/P = 1.92 (d) and Na/P = 1.82 (e);
pure sodium tripolyphosphate phase (f). All the samples were calcined at 500 ◦C.
Fig. 9. XRD of non-supported sodium phosphate powder (Na/P = 1.82) calcined
at 500 ◦C (a);1.82Na P/SiO2 23.8 (b); and 1.82Na P/SiO2 29.4 (c) samples. Aster-
isks: + Na5P3O10, * Na4P2O7.
way of the preparation and a chemical composition of the initial
mixture used for the preparation of catalysts the formation of these
phases is quite expected.
Assuming the dominance of pyrophosphates and using an ele-
mentary cell geometry of Na4P2O7, from the simple geometric
considerations we can expect the highest density in the monolayer
coverage to be about 0.34 nm2 per each P2O7 unit. This is in line
with the estimate of Gunter et al. [18] of 0.17 nm2 per each PO4
tetrahedron.
The onset of the formation of diffraction patterns permits more
realistic estimate of the highest density at a monolayer coverage.
If phosphate groups are in monolayers in the amorphous samples
then this gives about 0.50 nm2 per each PO4 tetrahedron. Recent
studies of KH2PO4/SiO2 by Georgelin et al. [32] provide estimate
of ca. 1.20 nm2 per each PO4 unit using the same approach. The
difference in the estimates may be explained by the larger diameter
of K+ cations and differences in the thermal pretreatment of the
samples.
3.3. MAS NMR
31P MAS NMR data are presented in Fig. 10. In order to make eas-
Fig. 11. Dependence of surface Na/P ratio determined by XPS on Na/P ratio used for
the preparation of silica-supported catalysts.
ier assignments of peaks, pure sodium tripolyphosphate Na5P3O10
,
pure sodium pyrophosphate Na4P2O7 as well as non-supported
samples derived from different NaOH:H3PO4 compositions (pre-
calcined at 500 ◦C) were also analyzed. Pure Na5P3O10 exhibits
resonances at −1.4 and 2.3 ppm while pure Na4P2O7 exhibits some-
what broad line centered at −0.33 ppm obviously comprised of two
or more unresolved lines.
phates also may be present in smaller quantities assuming the
line broadness of supported samples and very weak peaks around
2 ppm.
Comparing the spectra of the individual phases and non-
supported mixtures of phases it is clear that non-supported
samples with compositions Na/P = 1.8–1.9 exhibit superposition of
lines typical to pyrophosphates and lines of smaller intensity typi-
cal to tripolyphosphates.
Silica supported samples with the same Na/P ratio show broader
and less intense lines in the same region as the lines from pure
pyrophosphates. Therefore on the surface of silica sodium phos-
phate is mostly in the state of pyrophosphates interacting with the
surface of the support that causes line broadening. Tripolyphos-
3.4. XPS
XPS analysis of the surface Na:P composition is presented in
Fig. 11. There are several notable features in the surface compo-
sition regarding sodium and phosphorus: first, surface Na/P ratio
determined by XPS does not follow chemical composition during
the synthesis. Second—“concave” shape of the curve of the XPS
ratio versus chemical ratio Na/P. Third—noticeable depletion of the
surface sodium when NaOH/H3PO4 > 1.4.

W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
165
Table 2
Catalytic properties.
Sample name
W/F (h)
XBDO (%)
Elimination
Rearrangement
E/R
Hydrogen transfer
Epoxidation
SBD (%)
S3B2OL (%)
S2M2P1OL + CA (%)
SIBA (%)
SMEK (%)
SActn + BDOn (%)
SBT (%)
SBTO (%)
a
SiO2
0.53
1.13
0.89
0.89
27.66
0.87
0.86
0.90
0.89
0.88
0.87
0.88
0.88
0.82
0.82
0.87
0.87
0.90
0.17
6.22
30.29
31.48
0.18
0.22
73.7
85.9
100
80.6
100
0.3
0.2
0.5
1.6
0
8.8
0
0.1
0
0
0.4
0
0
3.1
2.8
4.4
4.0
4.3
2.4
2.4
1.9
0.7
1.9
3.2
5.1
3.9
0
1.9
0.8
14.1
15.7
11.8
5.6
2.9
1.7
2.6
1.6
1.8
1.4
1.6
1.5
1.3
2.6
1.8
1.5
2.1
2.8
1.6
2.3
1.3
2.0
13.8
16.4
45.6
44.9
49.2
32.8
21.5
18.9
19.4
19.3
19.5
19.1
21.7
35.5
30.5
33.1
20.1
17.4
18.0
25.3
22.1
22.5
16.0
16.2
0.06
0.10
0.38
0.32
0.35
1.37
2.46
2.43
2.94
3.00
3.03
3.27
2.58
0.81
0.64
0.83
2.70
3.83
3.01
2.22
2.80
1.92
3.45
3.21
6.3
7.1
20.5
7.1
0
1.3
0.6
1.5
1.7
2.9
0.7
1.8
2.0
1.1
1.2
1.2
0.9
1.0
1.2
1.0
4.2
0.5
0.6
0.7
1.3
1.3
5.2
0.4
0.6
1.5
2.0
0
0.6
0
b
Na/SiO2
P/SiO2 9.8c
22.9
10.5
21.4
52.3
13.5
11.3
8.9
10.1
9.2
9.6
5.5
3.6
2.7
2.9
6.7
13.5
5.9
60.2
66.4
47.4
5.6
P/SiO2 9.8d
P/SiO2 9.8
1Na P/SiO2 9.8
1.5Na P/SiO2 9.8
1.67Na P/SiO2 9.8
1.8Na P/SiO2 9.8
1.82Na P/SiO2 9.4
1.84Na P/SiO2 9.8
1.89Na P/SiO2 9.8
2Na P/SiO2 9.8
2.25Na P/SiO2 9.8
2.5Na P/SiO2 9.8
3Na P/SiO2 9.8
100
0.2
7.4
3.9
27.1
5.8
4.2
2.9
8.4
5.4
7.2
6.3
13.0
2.5
1.9
4.3
0
0
79.3
96.0
95.0
96.4
92.7
92.4
90.4
70.8
70.3
80.4
90.6
94.6
90.8
100
43.4
39.8
51.4
48.5
51.6
53.0
52.3
25.3
16.9
24.8
49.2
53.7
51.0
0
4.1
3.7
3.8
4.4
4.7
4.5
6.1
11.4
10.0
8.5
5.1
4.4
4.7
0
0
0
4.8
4.3
1.82Na P/SiO2
5
1.82Na P/SiO2 17.2
1.82Na P/SiO2 17.2
1.82Na P/SiO2 17.2
1.82Na P/SiO2 17.2
1.82Na P/SiO2 17.2e
1.82Na P/SiO2 23.8
1.82Na P/SiO2 29.4
100
100
83.2
84.7
0
0
51.2
47.2
0
0.3
0.3
3.3
2.5
0.2
2.9
3.4
8.0
The data were accumulated for at least 30 h of the time on stream. Unless otherwise stated, the values in the Table are based on averaged values during 20–30 h (i.e., 10 h
total averaging) period of the time on stream.
a
–Poor mass-balance (∼45%).
b
–Poor mass-balance (∼39%).
c
–After 15 h time on stream.
–After 90 h time on stream.
–20 ml/min N2 flow.
d
e
All these features can be rationalized as a competition of phos-
phates and silanols of silica support for Na+ cations. Existence
of silanol groups renders weak acidic properties to silica surface
[33–35], thus the following reactions may occur:
Main desorption peaks have maxima around 150–170 ◦C char-
acteristic of very weakly bound NH3 species. Probably this peak
corresponds to physically bound ammonia since the intensity of
this peak is dependent on the time of helium purge after ammonia
adsorption. Medium-weakly bound ammonia is represented by the
peak around 200–250 ◦C. The peak area of the desorbed ammonia
decreases in a monotonous way with the increase of a sodium con-
tent of the samples i.e., the more NaOH is added the less acidic is
the surface according to ammonia TPD for the 1.8 < Na/P < 2.
SiOH ↔ SiO− + H+
(1)
(2)
(3)
SiOH + Na+ ↔ SiONa + H+
NaxH3−xPO4 + H+ ↔ Na+ + Nax−1H3−x+1PO4
Since phosphate groups are localized mostly on the sur-
face while silanols are localized both in the bulk and surface
then migration of Na+ cations into the bulk at high tempera-
tures is expected. Taking into account that “practical” silanol pKa
is around 6.5–7.2 while for H3PO4 pKa1 = 2.12, pKa2 = 7.21, and
pKa3 = 12.67 we can expect that sodium may migrate to silica
when NaOH/H3PO4 > 1—the curve in Fig. 11 confirms this. Assum-
ing the XRD results it is better to consider pyrophosphoric acid
and its dissociation constants (pKa1 = 0.85, pKa2 = 1.96, pKa3 = 6.68
and pKa4 = 9.39). According to the latter constants again we can
expect that sodium migration from phosphates to silanols will start
at Na/P > 1 as well.
3.5.2. Carbon dioxide TPD
CO2 desorption curves of the samples having ratios 1.8 < Na/P < 2
are presented in Fig. 12B.
Overall peak areas are increasing with the decrease of the
sodium content. Both physically bound CO2 (maximum desorp-
tion temperature∼150–170 ◦C) and medium-weakly adsorbed CO2
peaks are increasing with the decrease of NaOH/H3PO4 ratio mean-
ing that the apparent number of basic sites is increasing with the
decrease of NaOH added. This is quite contrary to the expectations
based on the general chemistry. Comparison with CO2 desorption
from a pure silica shows that sodium phosphate modified silicas
have anyhow considerably more basic centers.
3.5. Acid-base properties
3.5.3. Na/P ratio and a site density
Since BDO dehydration is described by E1cB mechanism it is
obvious that a characterization of the acid and base properties is
important.
The total amount of CO2 and NH3 desorbed with respect to the
wider NaOH:H3PO4 range is presented in Fig. 13. For the molar
ratios 1 ≤ NaOH/H3PO4 ≤ 3 both basic and acidic sites qualitatively
follow the same pattern. Clearly we see three distinct regions of
acid/base density dependence on Na/P ratio. For the 1 < Na/P < 1.8 a
monotonous increase of both basic and acidic sites occurs. For the
narrow region 1.8 < Na/P < 2 we see a dramatic change of the trend
with almost a vertical decrease of the acidic and basic site density.
For the Na/P > 2 the trend is reverted and we see almost a parabolic
increase of the acidic and basic site densities.
3.5.1. Ammonia TPD
Assuming the chemical nature of silica and sodium phosphates
we can expect absence of strong acid and basic centers for the silica-
supported sodium phosphate. Typical ammonia TPD curves for the
samples with compositions 1.8 ≤ NaOH/H3PO4 ≤ 2 are presented in
Fig. 12A.

166
W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
A
Fig. 13. Dependence of NH3 (solid symbols) and CO2 (empty symbols) total desorp-
tion on the Na/P ratio of silica-supported sodium phosphates.
B
Fig. 12. (A) Ammonia TPD curves for the catalysts loaded with 9.8% wt. phos-
phates having variable Na/P ratios; (a) silica support; (b) Na/P = 2; (c) Na/P = 1.89;
(d) Na/P = 1.84; (e) Na/P = 1.82; (f) Na/P = 1.80. (B) Carbon dioxide TPD curves for
the catalysts loaded with 9.8% wt. phosphates having variable Na/P ratios; (a) silica
support; (b) Na/P = 2; (c) Na/P = 1.89; (d) Na/P = 1.84; (e) Na/P = 1.82; (f) Na/P = 1.80.
Fig. 14. Dependence of high affinity site density determined by Py (solid symbols)
and BA (empty symbols) adsorption on the Na/P ratio of silica-supported sodium
phosphates.
Such a fast drop in the acid-base site density when 1.8 < Na/P < 2
is really striking. It is interesting to compare Fig. 1A and Fig. 13
where we can see that a fast change in the surface area and a
fast change in the acidity happens within the same narrow inter-
val 1.8 < Na/P < 2. If the coverage of silica by phosphates is not a
monolayer but certain aggregation into e.g., platelets occurs when
1.8 < Na/P < 2 then it is possible to explain the drop of acidity-
basicity by the increase of the number of inaccessible sites since
sodium phosphates are non-porous and probe molecules cannot
penetrate into the bulk. The increase of a surface area also may be
explained by this effect assuming that certain amount of pores may
become accessible due to the agglomeration of sodium phosphate
(see Fig. 1B).
and Na/P > 2 is related to (i) increase of accessible Na+ cations, acting
as Lewis centers localized on the surface of catalysts, (ii) increase of
negatively charged O␦− species in phosphate tetrahedra (resulted
from NaOH and H3PO4 neutralization), acting as base centers.
Though ammonia and carbon dioxide TPD is a standard method
for the characterization of acid-base site densities and strength in
heterogeneous catalysts nevertheless for very weak acid and base
sites of sodium phosphate it is somewhat controversial method.
This happens since it is very difficult to discriminate adsorption of
probe molecules on weak acid-base sites from physically adsorbed
NH3 or CO2.
Decrease of ammonia TPD peaks for the samples of the compa-
rable composition ratio 0.83 < Na/P < 2 was reported by Zhang et al.
[21] though they used different TPD protocol with a smaller time
for the removal of loosely bound ammonia than in our study.
The increase of both basic and acidic site densities upon the
increase of NaOH/H3PO4 ratio, evident for the regions 0 < Na/P < 1.8
We also tried to evaluate acid and base site densities by mea-
suring pyridine (Py) and benzoic acid (BA) adsorption isotherms in
the liquid cyclohexane. The results are presented in Fig. 14. Com-
parable trends in Figs. 13 and 14 suggests that both methods are
in the qualitative agreement. BA titration indicates much steeper
increase of basic sites for Na/P > 2 than CO2 TPD.

W. Kim et al. / Applied Catalysis A: General 511 (2016) 156–167
167
The ratio of the acid/base site densities calculated for Py and
Appendix A. Supplementary data
BA adsorption shows a clear maximum at 1.5 < Na/P < 2 (Fig. 7B).
Comparing Fig. 7A and B it is clear that certain optimum in the
acid/base balance corresponds to the maximum of E/R ratio. This is
in turn supports the idea of E1cB mechanism for the BD formation
requiring conjugated acid-base pairs and thus certain optimum in
the Na/P ratio.
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.apcata.2015.11.
043.
References
The optimum acid:base density balance determined by Py/BA
probe molecules is about 1.6–1.7 according to Fig. 7B. This is not
1:1 ratio that is expected from the mechanistic considerations and
relates to the fact that probe molecules are binding to all accessible
basic and acidic sites, not only to those that are directly involved in
the catalytic action.
[1] C. Ma, A. Wang, J. Qin, L. Li, X. Ai, T. Jiang, H. Tang, P. Xu, Appl. Microbiol.
Biotechnol. 82 (2009) 49–57.
[2] J.-H. Cho, C. Rathnasingh, H. Song, B.-W. Chung, H. Lee, D. Seung, Bioprocess
Biosyst. Eng. 35 (2012) 1081–1088.
[3] E.V. Makshina, M. Dusselier, W. Janssens, J. Degreve, P.A. Jacobs, B.F. Sels,
Chem. Soc. Rev. 43 (2014) 7917–7953.
[4] M.E. Winfield, J. Counc. Sci. Ind. Res. 18 (1945) 412–423.
[5] M.E. Winfield, Aust. J. Chem. 3 (1950) 290–305.
[6] A. Auroux, A. Gervasini, J. Phys. Chem. 94 (1990) 6371–6379.
[7] K. Tanabe, H. Hattori, Y. Ono, New Solid Acids and Bases: Their Catalytic
Properties, Elsevier Science, 1990.
[8] H. Duan, D. Sun, Y. Yamada, S. Sato, Catal. Commun. 48 (2014) 1–4.
[9] H. Duan, Y. Yamada, S. Sato, Appl. Catal. A 487 (2014) 226–233.
[10] H. Duan, Y. Yamada, S. Sato, Chem. Lett. 43 (2014) 1773–1775.
[11] H. Duan, Y. Yamada, S. Sato, Appl. Catal. A 491 (2015) 163–169.
[12] T.Y. Kim, J. Baek, C.K. Song, Y.S. Yun, D.S. Park, W. Kim, J.W. Han, J. Yi, J. Catal.
323 (2015) 85–99.
4. Conclusions
Silica-supported sodium phosphates demonstrated high stabil-
ity and excellent performance in the gas phase dehydration of
BDO. Conversions can be easily tuned up to 100%, while overall
selectivity to BD + MEK being over 80% and the ratio of elimination-
rearrangement products E/R = 2–2.8. For short contact times 3B2OL
becomes dominating product indicating stepwise dehydration of
BDO while E/R ratio increases up to 3–3.8. Short contact times also
revealed some minor products (epoxides, butanedione, acetoin)
that provide evidence for additional pathways (hydrogen trans-
fer, epoxidation). Minor products vanish at long contact times and
this may happen due to the unspecified polymerization processes
leading to the build-up of coke.
[13] M. De Lezaeta, W. Sattar, P. Svoronos, S. Karimi, G. Subramaniam, Tetrahedron
Lett. 43 (2002) 9307–9309.
[14] K. Nakamura, Y. Osamura, J. Am. Chem. Soc. 115 (1993) 9112–9120.
[15] M. Bartók, Á. Molnár, Ethers Crown Ethers Hydroxyl Groups and Their
Sulphur Analogs, John Wiley & Sons, Ltd., 1980, 1981, pp. 721–760.
[16] R.A. Biggs, W.W. Ogilvie, Contemporary Organic Synthesis II, in: P. Knochel
(Ed.), Second ed., Elsevier, Amsterdam, 2014, pp. 802–841.
[17] M. Lion, M. Maache, J.C. Lavalley, G. Ramis, G. Busca, P.F. Rossi, V. Lorenzelli,
Mol. Spectrosc. Mol. Struct. 1989, Proc. XIXth Eur. Congr. Mol. Spectrosc., 218
417–422 (1990).
Fine tuning of selectivities may be achieved by choosing an
appropriate Na/P ratio for the preparation of sodium phosphate.
The highest yields of elimination products (BD, 3B2OL) are achieved
for the catalysts having ratios 1.8 < Na/P < 2. Within this range phos-
phate structures are represented by pyrophosphates with a smaller
amount of tripolyphosphate according to 31P MAS NMR and XRD.
According to XPS sodium cations migrate from the surface to the
bulk of silica probably due to the competition of acidic silanols for
sodium when Na/P > 1.
[18] G.C. Gunter, R. Craciun, M.S. Tam, J.E. Jackson, D.J. Miller, J. Catal. 164 (1996)
207–219.
[19] M.S. Tam, G.C. Gunter, R. Craciun, D.J. Miller, J.E. Jackson, Ind. Eng. Chem. Res.
36 (1997) 3505–3512.
[20] M.S. Tam, R. Craciun, D.J. Miller, J.E. Jackson, Ind. Eng. Chem. Res. 37 (1998)
2360–2366.
[21] Z. Zhang, Y. Qu, S. Wang, J. Wang, Ind. Eng. Chem. Res. 48 (2009) 9083–9089.
[22] Zhi-qiang Zhang, Y. Qu, S. Wang, J. Wang, Chin. J. Chem. Phys. 22 (2009)
315–321.
[23] S. Lu, D. Wang, S.P. Jiang, Y. Xiang, J. Lu, J. Zeng, Adv. Mater. 22 (2010) 971–976.
[24] K. Tanabe, T. Yamaguchi, J. Res. Inst. Catal. Hokkaido Univ. 11 (1964) 178–184.
[25] F.M. Bautista, J.M. Campelo, A. Garcıı´a, D. Luna, J.M. Marinas, R.A. Quirós, A.A.
Romero, Appl. Catal. A. 243 (2003) 93–107.
[26] J. Goworek, A. Borówka, R. Kusak, Colloids Surf. Physicochem. Eng. Asp. 157
(1999) 127–135.
[27] S.N. Vereshchagin, K.P. Dugaev, N.P. Kirik, N.N. Shishkina, A.G. Anshits, Catal.
Today 24 (1995) 349–355.
[28] M.A. Aramendıı´a, V. Borau, C. Jiménez, J.M. Marinas, A. Porras, F.J. Urbano, J.
According to the expected chemistry of the elimination mech-
anisms formation of 3B2OL and BD may be interpreted in terms of
E1cB mechanism requiring conjugated acid-base pairs. Dehydration
of 2-butanol catalyzed by the same catalysts leads to the formation
of 1-BT (Hofmann product) that is a good marker of E1cB mecha-
nism.
Existence of weak acid-base pairs is revealed by CO2/NH3 TPD
but in this case there is a problem of discriminating sites of proper
strength relevant to the catalytic action. The number of high affin-
ity sites binding to pyridine and benzoic acid calculated from the
adsorption isotherms gives somewhat better support for the idea
of the optimum balance of acid-base sites realized in the interval
of 1.8 < Na/P < 2 of the catalyst composition.
Catal. 161 (1996) 829–838.
[29] T.T.N. Nguyen, V. Ruaux, L. Massin, C. Lorentz, P. Afanasiev, F. Maugé, V.
Bellière-Baca, P. Rey, J.M.M. Millet, Appl. Catal. B Environ. 166–167 (2015)
432–444.
[30] R.M. Koros, E.J. Nowak, Chem. Eng. Sci. 22 (1967) 470.
[31] Y. Ono, H. Hattori, Solid Base Catalysis, Springer, Berlin, Heidelberg, 2012.
[32] T. Georgelin, M. Jaber, T. Onfroy, A.-A. Hargrove, F. Costa-Torro, J.-F. Lambert, J.
Phys. Chem. C 117 (2013) 12579–12590.
[33] E. Papirer, Adsorption on Silica Surfaces, Taylor & Francis, 2000.
[34] K.K. Unger, Porous Silica, Elsevier Science, 1979.
[35] J. Nawrocki, J. Chromatogr. A 779 (1997) 29–71.
Acknowledgements
This work was supported by the Industrial Strategic Technology
Development Program (No. 10050407) funded by the Ministry of
Trade, Industry and Energy (MOTIE, Korea).