You are on page 1of 10

Chemical Engineering Journal 375 (2019) 122037

Contents lists available at ScienceDirect

Chemical Engineering Journal


journal homepage: www.elsevier.com/locate/cej

On the kinetics of multiphase etherification of glycerol with isobutene T


⁎ ⁎
Jingjun Liu , Zhiguang Zhang, Peng Zhang, Bolun Yang
Department of Chemical Engineering, Xi’an Jiaotong University, Xi’an 710049, PR China

H I GH L IG H T S

• The phase separation of reactants was considered in kinetic modeling.


• The initial rate for each etherification step was measured separately.
• Adsorption of reactants competed with each other remarkably.

A R T I C LE I N FO A B S T R A C T

Keywords: The etherification of the hydroxyl groups of glycerol with isobutene successively forms mono-ethers (ME), di-
Glycerol etherification ethers (DE) and tri-ether (TE) of glycerol, which are desired fuel additives. The initial rates of each step in the
Glycerol ether consecutive etherification were measured independently under various concentrations and temperatures. The
Isobutene initial rate of ME synthesis from glycerol and isobutene was independent of the initial composition due to the
Kinetics
invariant concentrations of reactants in the reaction phase. Adding ME to the glycerol-isobutene reaction was
observed to accelerate the glycerol conversion by enhancing the solvation of isobutene in glycerol. Rate ex-
pressions in the form of power-law with parameters regressed from initial rates measured separately for each
etherification step were found to be unable to predict the kinetics of complex reaction where all reaction steps
coexisted together. This signifies significant interaction of reactions and competitive adsorption of reactants.
Glycerol etherification with isobutene under high conversions where all reactions and reactants exist simulta-
neously was carried out to obtain kinetic data containing the interactions between different reactions. Thereafter
a model considering phase separation and competitive adsorption was proposed and correlated with the initial
rates and kinetic data under high conversions. The model was capable to describe all observed kinetics.

1. Introduction isomers of ME and DE are treated as one species respectively. DE and TE


are the desired fuel additives. This etherification process can be roughly
A huge amount of glycerol is produced in the biodiesel production divided into three successive steps, i.e., ME synthesis from G and IB, DE
in recent years. Efforts have been made to find out new outlets for the synthesis from ME and IB, and TE synthesis from DE and IB.
bio-renewable glycerol. On the other hand, glycerol is a versatile mo- Most relative works devoted to developing active and selective
lecule which can be transferred into a lot of valuable products [1–6]. catalysts for glycerol etherification [7,11–23]. Our previous work [10]
tert-Butyl ethers of glycerol obtained by etherifying with isobutene were studied the mass transfer properties and phase equilibrium behavior in
proved to be excellent fuel (gasoline, diesel, and biodiesel) additives, glycerol etherification with isobutene. It shows that the reactants se-
which are able to reduce the emissions of particle matter and green- parated into two liquid phases and the reaction rates were limited by
house gas of fuel combustion [7,8] and improve the low-temperature low concentration of isobutene in the reaction phase, where the solid
behavior of biodiesel [9]. catalyst distributed. Along with the catalytic process, there is complex
As presented in Fig. 1, glycerol etherification with isobutene forms multiphase mass transfer and phase equilibration in the reaction, in-
mono-tert-butyl ethers of glycerol (ME), di-tert-butyl ethers of glycerol teracting with the catalytic reaction. The formed mono ethers of gly-
(DE) and tri-tert-butyl ether of glycerol (TE) consecutively. The di- cerol were thought to accelerate the reaction by enhancing the solva-
merization of isobutene is normally observed as a side reaction under tion of isobutene in the polar reaction phase.
high conversions, which forms dimers of isobutene (DIB). Usually, As for the kinetics of this reaction, most studies [24–27] ignored the


Corresponding authors.
E-mail addresses: jingjun_liu@mail.xjtu.edu.cn (J. Liu), blunyang@mail.xjtu.edu.cn (B. Yang).

https://doi.org/10.1016/j.cej.2019.122037
Received 19 March 2019; Received in revised form 19 June 2019; Accepted 21 June 2019
Available online 22 June 2019
1385-8947/ © 2019 Elsevier B.V. All rights reserved.
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

Nomenclature T time, s
T absolute temperature, K
c concentration, mol/L X molar fraction
Deff effective diffusivity, m2∙s−1 γ activity coefficient
Ea activation energy, kJ/mol ϕ fraction of the glycerol rich phase in the biphasic mixture
Eq equilibrium state
I component index Abbreviations
Ka adsorption equilibrium constant
k 0,i pre-exponential factor, L(molH + ) - 1s - 1 or cal calculated value
L2mol−1 (molH + ) - 1s - 1 DE di-tert-butyl ethers of glycerol
k reaction rate constant, L(molH + ) - 1s - 1 or exp experimental value
2 −1 + - 1 - 1
L mol (molH ) s G glycerol
mcat amount of fed catalyst, kg GP glycerol phase
n molar number of component, mol IB isobutene
NWP Weisz-Prater number IBP isobutene phase
QE ion-exchange capacity, mol H+/kg ME mono-tert-butyl ethers of glycerol
R universal gas constant, J∙mol−1∙K−1 TE tri-tert-butyl ether of glycerol
Rp radius of catalyst particle, m TPDF the tangent plane distance function
r reaction rate, mol∙s−1∙(mol H+)−1
r 0
initial reaction rate, mol∙s−1∙(mol H+)−1

Fig. 1. Reaction scheme of glycerol etherification with isobutene [10].

phase separation of the reaction mixture. Behr and Obendorf [24] taking place.
studied the kinetics for this reaction at 343 K and 363 K. They corre- All former kinetic studies employed batch stirred reactors, where
lated reaction rates with the overall concentrations of reactants using glycerol and isobutene were fed and the etherification process took
rate expressions in the form of power-law. We built a complete model place consecutively. The rates of the three steps were measured si-
for glycerol etherification based on Eley–Rideal mechanism [26]. Ac- multaneously and then analyzed mathematically. Only the feed of the
tivities of components calculated from the overall concentration were first step (G and IB) of the consecutive reaction was varied, while the
used in the rate expressions. Singh et al. [27] modeled the kinetics of reactants for the subsequent steps were formed in the reaction. As a
glycerol etherification with isobutene produced in situ from the dehy- consequence, the kinetics for the last two steps were studied under
dration of tert-butyl alcohol. Again, a power-law model based on the limited ranges of reactant concentrations, determined by the former
overall concentrations of reactants was used. To avoid phase separa- steps. However, in all proposed processes to produce glycerol ethers,
tion, Klepáčová et al. [25] employed dioxane as a solvent in the kinetic continuous stirred reactors with part of the products being cycled back
study for glycerol etherification. A pseudo-homogeneous model with to the inlet were used [24,28–30]. The three etherification steps occur
power-law rate expressions for all independent components was for- parallelly as well as consecutively. The concentrations of reactants,
mulated. But only part of the kinetic parameters (mostly for the for- especially the glycerol ethers, could exceed the applicable scope of the
mation of primary products) was solved, due to the complexity of the existing kinetic models. For the consecutive reaction, it is also necessary
model. to find out the slowest step which needs to be strengthened, to improve
The available kinetic models for glycerol etherification with iso- the efficiency of the whole process through the catalyst and reactor
butene have defects in both mechanism interpretation and practical design.
application. The reaction rates are determined by the concentrations of The present work engages to investigate the kinetics of the multi-
reactants in the reaction phase. The phase separation of reactants has a phase etherification. The single etherification steps in the consecutive
significant effect on the reaction kinetics and should be taken into ac- reaction were studied independently, as well as in one pot, to explore
count since the compositions of each phase are largely different from their kinetics under a wide range of concentrations. The effect of ME on
one another and the overall compositions. The kinetic models ignoring the glycerol etherification was verified experimentally. The obtained
phase separation are just phenomenological correlations, application of kinetic data were then analyzed by taking the phase separation of re-
which is thus limited to the corresponding experimental conditions. actants into consideration. This kinetic study is expected to provide new
These models can not reveal the intrinsic kinetics of reactions actually insights into the catalyst development, reactor design, process design

2
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

and intensification for glycerol etherification. an ice bath (about 273 K) immediately to bring down the temperature
and stop the reaction timely. The pressure was then released slowly
2. Experimental section under the low temperature. It can be expected that almost all reactants
(glycerol, glycerol ethers, and DIB in some cases) except isobutene re-
2.1. Chemicals and catalyst tained in the reactor as their boiling points are much higher than 273 K
[9]. The remained reactants formed a miscible liquid mixture, which
Isobutene (> 99.99%, liquid, 1 MPa) was obtained from Shandong was centrifuged to remove catalyst. A liquid sample was taken and
Wusheng Gas Technology Co., Ltd, China. Glycerol (> 99.0%), n-bu- added into a volumetric flask containing a certain amount of n-butanol
tanol (> 99.5%) and ethanol (> 99.7%) were purchased from Tianjin (internal standard chemical). The sample was then diluted to the cali-
Fuyu Fine Chemical Co., Ltd, China. The glycerol ethers were synthe- brated volume using ethanol and analyzed by a gas chromatograph
sized by etherifying glycerol with isobutene, and then separated to get (GC9790Ⅱ, Zhejiang Fuli Analytical Instrument Co., Ltd) equipped with
ME, DE and TE by extraction and vacuum distillation (see the a DB-wax capillary column (length 30 m, film thickness 0.25 μm and
Supplementary Material) [26,31]. The purities of the synthesized ethers diameter 0.32 mm) and an FID detector. Glycerol, glycerol ethers, and
(> 99.5%) were verified by gas chromatograph (GC) before use. DIB after the reaction were quantified using the internal standard
A strong acidic ion-exchange resin, NKC-9, purchased from the method. The amount of isobutene was calculated through material
Chemical Plant of Nankai University was used in the kinetic experi- balance.
ments. The resin consists of a styrene–divinylbenzene base and sulfonic The reaction rate was defined as:
acid active sites. NKC-9 particles were washed with ethanol to remove
−dni
possible impurities and then dried at 373 K. The particles were then − ri =
mcat QE dt (1)
ground and sieved into fine powder (diameter ≤ 0.05 mm) in order to
exclude internal diffusion [26,32]. The ion-exchange capacity of the where ni is the number of moles of component i, QE the ion-exchange
NKC-9 powder is 4.7 mol H+/kg, verified by titration after cation ex- capacity (4.7 mol H+/kg) of the catalyst and mcat the amount of fed
change with 1 M sodium chloride. The specific surface area and pore catalyst. In the reactions performed to measure the initial rates, the
volume of the powder are 7.25 m2/g and 0.08 cm3/g, determined by conversions were controlled within 8%-15% by manipulating the
nitrogen adsorption–desorption (see the Supplementary Material). The quantity of catalyst and reaction time. The initial rate was calculated as
powder was dried at 373 K under vacuum overnight before use. the average rate in the low conversion region, i.e.
− ri0 = −Δni /(mcat QE Δt ) .
2.2. Experimental apparatus and procedure Weisz-Prater criterion (Weisz-Prater number) [33,34] was used to
evaluate the effect of internal diffusion on each reaction:
For simplicity and as usual in the tracking literature, isomers of ME,
rRp2
DE and DIB are lumped as one species respectively. The etherification NWP =
system thus can be divided into three main consecutive steps, i.e., the cs Deff (2)
reaction of G with IB, ME with IB, and DE with IB. Three categories of In the above equation, r is the reaction rate, Rp the radius of the
reactions were carried out. The first category measured the initial rates catalyst particle, cs the concentration of reactant on the catalyst surface,
of the three etherification steps separately under a variety of compo- and Deff the effective diffusivity. The number was evaluated at condi-
sitions. The initial molar ratio of isobutene to glycerol (or glycerol tions where the reaction was the fastest, presenting the most stringent
ethers) was varied from 0.4 to 9. In the second category, initial rates conditions for the calculation. The conditions were 373 K and the cor-
were measured for G + IB with the addition of different portions of ME. responding initial concentrations resulting in the highest rate. The ab-
In the third category, the reactions were fed with isobutene and glycerol sence of significant internal diffusion limitation can be assured when
(initial molar ratio fixed to 4:1) and were performed for different in- the calculated NWP is less than 0.3. The calculation details are presented
tervals, to obtain concentration evolution data. All reactions were in the Supplementary Material.
performed in batch model in a 20 ml stainless-steel batch reactor (see
Scheme S1 in the Supplementary Material). The reaction temperatures
3. Results and discussion
were 343 K, 353 K, 363 K, and 373 K, controlled within a deviation
of ± 0.3 K. Conditions of reactions performed were tabulated as Table
Based on experimental results, the following kinetic scheme con-
S1 in the Supplementary Material. Some reactions were repeated 3–4 sidering seven reactions with six components (or lumps) was proposed:
times.
k1
In a typical run, the catalyst was loaded to the bottom of the reactor G + IB → ME (R1)
cell carefully to avoid spraying on the side wall. Then the weighted
k2
amount of glycerol or glycerol ethers, which were in liquid form under ME → G + IB (R2)
room temperature and pressure, were added to the reactor, which was
k3
then purged with nitrogen twice after closing the lid. The pressure of ME + IB → DE (R3)
the reactor was raised to 0.5 MPa by filling nitrogen, under which the k4
isobutene was pumped into the reactor in liquid form and weighted. DE → ME + IB (R4)
The pressure was increased to 3 MPa by filling nitrogen to ensure a k5
liquid phase reaction. In these operations, the reactor was kept upright DE + IB → TE (R5)
and shaking was avoided to lessen the mixing of the catalyst with re- k6
actants. Afterward, the reactor was put into a big thermal jacket where TE → DE + IB (R6)
the temperature was maintained at the reaction temperature, to heat up k7
2IB → DIB (R7)
the reactants without stirring. It took 6–10 min for the reactants to
reach the reaction temperature. It’s verified that the reaction during the In experiments, we failed to prepare enough TE to study its de-
short heating process was ignorable thanks to the limited contact of composition. Thus, only the initial rates of reaction R1-R5 were mea-
reactants with the catalyst. Magnetic stirring was then switched on and sured by feeding G + IB, ME + IB, and DE + IB to the batch reactions
the instant was considered as the start of the reaction. The stirring of the first category. Reaction temperatures and concentrations of re-
speed was kept at 1400 rpm to eliminate external diffusion [26]. actants were varied in order to obtain necessary kinetic data and to
At the end of each batch reaction, the reactor cell was quenched in understand their effects on the reaction rate and selectivity. These

3
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

kinetic data were obtained without the limitation of external diffusion In the reaction fed with ME and isobutene, reaction R(2) and R(3)
by using strong stirring in the experiments, proved in our former papers take place at the same time. The relative magnitudes of rates reveal the
[10,26]. The potential effect of internal diffusion was tested by Weisz- selectivity to each reaction under the experimental conditions. At low
Prater criterion [33,34]. As presented in Table 1, the Weisz-Prater ME concentration, the DE synthesis rate from ME is higher than the rate
numbers for all reactions are much less than 0.3. The low value of the of ME decomposition. With the increasing of ME fraction in the reaction
estimated Weisz-Prater number supports the assumption of the absence mixture, the ME decomposition rate gradually surpasses the DE for-
of internal diffusion limitation. Thus, without the limitation of internal mation.
and external diffusion, the bulk concentration of the reaction phase can In the third set of experiments, DE and isobutene were fed to the
be treated as the concentration on the catalyst surface. reactor. The only hydroxyl group in DE molecule further etherified with
Corresponding to the three categories of kinetic reactions, three sets isobutene to form TE (reaction R(5)). The two tert-butyl ether groups in
of kinetic data were obtained, i.e., 1) the initial rates of each ether- DE also broke to form isobutene and ME (reaction R(4)). The reactants
ification step measured separately; 2) initial rates of glycerol conversion in this system also formed a single homogeneous liquid phase.
with the addition of ME; and 3) the concentration evolution data for The rates of TE formation and DE decomposition under different
glycerol etherification with isobutene. They are discussed in turn to concentrations of reactants and temperatures are presented in Figs. 5
understand the kinetics of this reaction in the following sections. and 6. The two reactions behaved similarly with those in the ether-
ification of ME with isobutene. The highest rates of TE formation were
3.1. Initial rate of each etherification step obtained at a nearly equal concentration of DE and isobutene. In-
creasing or decreasing the fraction of DE or isobutene would lower the
Initial rates for the reaction of glycerol with isobutene were mea- TE synthesis rate. By contrast, the decomposition rate of DE to ME in-
sured with different concentrations at 363 K. The initial rates are in- creases with the concentration of DE, shown in Fig. 6. The decom-
dependent of the reactant composition in the investigated range, as position rate of DE to ME is much higher than the TE synthesis rate,
shown in Fig. 2. It seems to be very strange at first glance. But it is under most of the reactant compositions. This means that DE prefers to
indeed the case when the intrinsic of the process is fully understood. decompose to ME rather than react with isobutene to form TE. In terms
When the reaction was fed with glycerol and isobutene, the mixture of the temperature dependence of the reaction rate, for the TE synthesis
of reactant separates into two liquid phases (under pressure) due to the from DE and isobutene, the only safe conclusion to be drawn is that the
component immiscibility. As we proposed in previous work [10], the rates under different temperatures are very close. The DE decomposi-
hydrophilic catalyst powder (NKC-9) only appeared in the glycerol rich tion rates rise normally with increasing temperature.
phase, which was thus the reaction phase. It was proved that the li- Comparing the initial rates of each etherification step, the rate of the
quid–liquid-solid mass transfer proceeds quickly and has ignorable in- last step, from DE to TE, is one magnitude lower than other steps. This
fluence on the transferring of reactants to the catalyst surface. The re- is consistent with the low yield of TE under high conversions. All other
actant concentrations on the external surface of the catalyst approach reactions have initial rates of the same magnitude. Although the re-
the liquid–liquid phase equilibrium composition of the reaction phase actants separated into two phases, which would normally lower the
[10]. At the initial stage of the G-IB reaction, the reaction mixture is a concentration of one reactant, the initial rate of ME synthesis from G
binary liquid–liquid system, as no or very small amount of product is and IB is just a bit smaller than the DE synthesis. This reveals the high
formed. The component concentrations of the two phases reach equi- reactivity of glycerol. Enhancing the mutual solution of glycerol with
librium and do not alter with the changing of glycerol/isobutene ratio, isobutene should be an effective way to strengthen the glycerol con-
for the binary system. The concentration of isobutene in the glycerol version.
phase, i.e. the reaction phase, is the solubility of isobutene in glycerol,
which is a constant value under a certain temperature. The glycerol
concentration in the glycerol phase is in the same situation. In short, the 3.2. Effect of ME addition on the G-IB reaction
concentrations of reactants in the reaction phase are the same for dif-
ferent initial compositions, in the investigated range. As a result, the In our previous work, we proposed that ME boosts the solvation of
reaction rates are the same under different (overall) concentrations at isobutene in glycerol, and thus accelerates the rate of glycerol conver-
the same temperature. This phenomenon was also verified at 353 K, sion [10]. A set of experiments were carried out to investigate the in-
with the corresponding results shown in Fig. 2. The independence of fluence of ME addition on the initial rate of G-IB etherification and thus
initial rates from reactant concentration also supports the proposal that to verify the proposal. The measured initial rates of glycerol conversion
the catalyst powder only distributed in one of the two liquid phases, with different portions of ME added are shown as hollow circles in
most likely in the glycerol phase due to the hydrophilic property of the Fig. 7. The initial molar ratio of G/IB was fixed to 1:4 when varying the
catalyst surface. Otherwise, the reaction rate would alter with the amount of ME. With increasingly adding of ME to the reaction, the
changing of the initial composition of reactants. glycerol conversion rate increases gradually. The highest glycerol
When the reaction was fed with ME and isobutene, the hydroxyl conversion rate obtained with the addition of ME doubles the rate
groups in ME molecule could further etherify with isobutene to form DE without ME addition. The experimental result confirms that adding a
(reaction R(3)). The ether bond in ME molecule could also break to certain amount of ME can indeed promote the reaction of glycerol with
form glycerol and isobutene (reaction R(2)). The measured rates for isobutene. This effect will be further discussed in the following kinetic
these two reactions under different temperatures and reactant compo- analysis.
sitions are demonstrated using different markers in Fig. 3 and Fig. 4.
Without doubt, the higher the temperature, the higher the reaction
rates, for both of the two reactions. In the ME + IB reaction system, the Table 1
reactants form a homogeneous single liquid phase. The initial rates of Results of Weisz-Prater test.
the two reactions show normal trends. In the DE synthesis from ME and Reactants Reaction Weisz-Prater Number
isobutene, the rate increases with increasing ME concentration at first,
and then decreases after reaching a maximum value, as shown in Fig. 3. G + IB G + IB → ME 6.71 × 10−3
As a bimolecular reaction, the largest rate is achieved at equal molar ME + IB ME + IB → DE 6.21 × 10−5
ME → G + IB 3.61 × 10−4
concentrations of the two reactants. For the decomposition of ME, the
DE + IB DE + IB → TE 1.61 × 10−6
initial rate increases with the increasing of ME concentration, presented DE → ME + IB 8.84 × 10−5
in Fig. 4.

4
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

Fig. 2. Initial rate of ME formation from G and IB as a function of glycerol Fig. 5. Initial rate of TE formation from DE and isobutene as a function of DE
molar fractions. molar fraction under different temperatures.

Fig. 3. Initial rate of DE formation from ME and isobutene as a function of ME Fig. 6. Initial rate of ME formation from DE decomposition as a function of DE
molar fraction under different temperatures. molar fraction (with isobutene co-feed) under different temperatures.

Fig. 4. Initial rate of G formation from ME decomposition as a function of ME


molar fraction (with isobutene co-feed) under different temperatures. Fig. 7. Effects of ME addition on the initial rate of glycerol etherification with
isobutene at 353 K.

5
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

3.3. Kinetic analysis beginning of the reaction homogenized to a single phase during the
reaction under some conditions.
When the reaction mixture separates into two liquid phases, the With the method to deal with phase separation in kinetic modeling,
phase separation has to be taken into consideration in kinetic analysis. our first attempt was to correlate the initial rates of each etherification
This is one of the motivations of the present work. The phase separation step using simple power-law rate expressions. The reaction kinetics of
appears when the reaction is fed with glycerol and isobutene, even with each step can be well interpreted by first-order power-law expressions
the addition of ME, in both the initial and long period of the reaction. (see Figure S2). Then those rate expressions for each step were gathered
To deal with the kinetics of the three-phase reaction, the reaction phase up to predict the reaction rate of G-IB-ME reaction and the composition
where the catalyst immersed should be determined. And then the of the reaction mixture under long reaction time. This methodology
concentrations of reactants on the catalyst surface need to be obtained, soon found failure in the prediction of the glycerol conversion rate in G-
which can then be correlated with rate expressions. Following the IB-ME reaction. Although the simple model was unsuccessful, it pro-
proposal of our previous work [10], the catalyst was assumed to im- vided the direction to improve the kinetic model and important en-
merse in the more polar glycerol rich phase (GP), and the glycerol phase lightenment, which are demonstrated briefly in the following discus-
was thus treated as the reaction phase. This presumption is also sup- sion.
ported by the kinetics of G-IB reaction, as discussed in section 3.1. The rate expressions in the power-law form were proposed to de-
Detailed mass transfer simulations show that the transferring of re- scribe the initial rates of ME synthesis (R1) and decomposition (R2):
actants between the liquid–liquid-solid phases proceeds more quickly
rG → ME = kGM cG cIB (8)
than the reaction, making the external diffusion not a restriction [10].
The Weisz-Prater test reveals that the internal diffusion limitation is rME → G = kMG cME (9)
negligible. These imply that the concentrations of reactants on the
catalyst surface are the bulk composition of the reaction phase (glycerol In modeling the rate of ME formation from glycerol, the reaction
phase), which is also in phase equilibrium with the other phase (named rate constant kGM was obtained by equation solving, since only one
isobutene phase, IBP). Consequently, the equilibrium compositions of average rate can be gained for one temperature from experiments. With
the two phases need to be calculated. the concentration of the glycerol phase, the rate constant was solved as
To ensure the phase separation was treated rigorously and capture kGM = 1.08 × 10−2 L2mol−1 (molH + ) - 1s - 1 at 353 K. Fitting Eq. (9) with
the possible change of phase state during the reaction, the tangent plane the measured initial rates of ME decomposition at 353 K obtained the
distance function (TPDF) was used to determine whether a reaction rate constant kMG = 3.87 × 10−2 L(molH + ) - 1s - 1. The comparison of
mixture separated into two phases or not [35,36]. The phase stability experimental and calculated initial rates (by Eq. (9)) was illustrated as
analysis using TPDF was detailed in the Supplementary Material. When Figure S2 in the Supplementary Material.
a reaction mixture separated into two phases, their equilibrium com- In the reaction fed with G + ME + IB, ME decomposition (reaction
positions were calculated by solving the phase equilibrium equations: R(2)) and its formation (reaction R(1)) take place simultaneously. With
the estimated rate constants for these two reactions, it is desirable to
γi,IBP IBP GP GP
Eq x i, Eq = γi, Eq x i, Eq (3) simulate the glycerol conversion rate in the G-IB-ME reaction. Thus, the
concentration in the reaction phase was calculated first. Although the
in which γ and x are activity coefficient and molar fraction. The su-
added ME partially homogenizes glycerol and isobutene, the G-IB-ME
perscripts IBP and GP indicate isobutene rich phase and glycerol rich
reaction mixture still splits into two phases in the investigated com-
phase respectively. The subscript Eq denotes equilibrium state. The
positions ranges, according to the phase equilibrium computation. The
activity coefficient was calculated using the NRTL model [37] with
calculated concentrations of the glycerol phase with different portions
parameters adapted from literature [24,29,38]. The mass balance
of ME at 353 K are shown in Fig. 8. The concentration of isobutene
equations of the two-phase mixture are:
increases significantly with the adding of ME, which benefits the gly-
ϕx iGP IBP
, Eq + (1 − ϕ ) x i, Eq = x i
m
(4) cerol conversion. Besides promoting the solvation of IB into glycerol,
ME itself also trends to distribute in the hydrophilic glycerol phase,
where ϕ is the fraction of the glycerol rich phase in the mixture, and x im which facilitates its decomposition. Thus, excessive addition of ME
the overall composition of the reaction mixture. The following con- should be avoided to intensify glycerol conversion. The concentration
straints also hold: of glycerol in the glycerol phase declines quickly with the adding of ME,
NC
∑ xiGP
, Eq = 1
i=1 (5)
NC
∑ xiIBP
, Eq = 1
i=1 (6)

0 ⩽ ϕ, x iGP IBP
, Eq , x i, Eq ⩽ 1 (7)
Equations (3)–(7) were solved to get the equilibrium compositions
GP IBP
of each phase, x i,Eq and x i,Eq . The equilibrium concentration of glycerol
GP GP
rich phase ci,Eq was then calculated from x i,Eq and can be used in kinetic
analysis.
The concentration on the catalyst surface, which is equal to the
concentration of the reaction phase without the limitation of diffusion,
was used in the rate expressions. The concentration of the glycerol
phase was used when phase separation occurred. In the following dis-
cussion, ci is used in rate expressions to present the concentration of the
reaction phase, without further adding phase (GP) or equilibrium state
(Eq) label to the symbol particularly for phase separation cases. This
symbol simplification also aimed at a uniform notation, especially in Fig. 8. Calculated concentration of the glycerol phase in the G-IB-ME ether-
the cases when the reaction mixture separating into two phases at the ification upon feeding at 353 K.

6
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

as the result of dilution and its migration into the isobutene phase. The Table 2
concentrations of components in the isobutene phase upon feeding Estimated values for kinetic parameters at the reference temperature of 357 K
were illustrated in Figure S3 in the Supplementary Material. and the errors corresponding to the 95% confidence interval.
The concentration of isobutene in the reaction phase increases with Rate constant k 0, i (L(molH + ) - 1s - 1 or L2mol−1(molH + ) - 1s - 1) Eai (kJ/mol)
the addition of ME as expected, and this provides the possibility of the
increase of glycerol conversion rate. In G-IB-ME reaction, the net rate of R1 – k1 7.36 ± 0.05 82.0 ± 2.1
glycerol conversion equals to ME synthesis rate minus ME decomposi- R2 – k2 0.62 ± 0.17 97.7 ± 16.6
R3 – k3 2.43 ± 0.44 89.1 ± 17.1
tion rate:
R4 – k4 0.55 ± 0.21 65.3 ± 15.1
R5 – k5 0.49 ± 0.21 35.0 ± 9.0
− rG = rG → ME − rME → G = kGM cG cIB − kMG cME (10)
R6 –k6′ 0.03 ± 0.01 39.3 ± 18.5
R7 – k7 0.01 ± 0.00 20.0 ± 0.5
The concentrations of glycerol phase (Fig. 8) and the obtained va-
lues of kGM and kMG were used to calculate the rate of glycerol con-
version by Equation (10). The calculated rate, shown as the dashed line Table 3
in Fig. 7, decreases quickly with the addition of ME, emerging an op- Estimated values for adsorption parameters at the reference temperature of
posite trend with the experimental results. This indicates that the rate 357 K and the errors corresponding to the 95% confidence interval.
expressions and the corresponding parameters proposed for the single
Components Δa Hi (kJ/mol) Δa Si (J/mol/K)
reaction step are unable to predict the rate of complex reaction. The ME
decomposition in the presence of G is slower than in the single reaction ME −43.4 ± 14.6 −10.8 ± 2.6
(fed with ME and IB). This may be because the adsorption of ME was DE –23.7 ± 13.5 −15.6 ± 6.0
weakened by competitive adsorption and thus the decomposition of ME IB −44.7 ± 0.5 −15.7 ± 0.4
G −2.5 ± 0.0 −8.7 ± 0.1
was suppressed in the presence of glycerol. The deviation between
experiments and rates calculated by Eq. (10) reflects remarkable in-
teraction among the reactants and reactions. temperatures). As a consecutive reaction, the concentrations of ME, DE
Kinetic study for every single step discovers the intrinsic properties and TE increase in sequence. The accompanying dimerization of iso-
of each etherification step, without the influence of other components butene is not severe in the investigated period. While the reaction was
and reactions. However, as a consecutive reaction, glycerol etherifica- fed with glycerol and isobutene, all reactions (R1-R7) and reactants
tion with isobutene cannot be controlled within one step. The industrial coexisted in the process. This provides effective data accounting for the
reactors normally operate under high conversion, where all reactants interactions of different reactions and the influences of reactants. These
and reactions coexist in one pot at the same time. Hence, the applicable kinetic data under long reaction period, together with the initial rates of
kinetic model for the reactor and process design should present the single step were then used to regress the kinetic model.
interactions among reactants and reactions by considering the adsorp- All components were thought to be adsorbed on the catalyst surface,
tion of components. To this end, the glycerol etherification with iso- described by the Langmuir isotherm. By assuming the surface reaction
butene was carried out under higher conversions (compared with initial between adsorbed species as the rate determining step, the rate ex-
rate) at different reaction times and temperatures. As an example, the pressions for reaction R1-R7 considering phase separation were for-
scatters in Fig. 9 show the concentration evolutions of reactants at mulated as (see the Supplementary Material for derivation):
363 K (see the Supplementary Material for data under other

Fig. 9. The evolution of component concentration under long reaction period at 363 K.

7
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

Fig. 10. Parity plots for component concentrations under high conversions.

k1 KaG KaIB cG cIB rME = r1 − r2 − r3 + r4 (20)


r1 =
(1 + ∑i Kai ci )2 (11) rDE = r3 − r4 − r5 + r6 (21)
k2 KaME cME rTE = r5 − r6 (22)
r2 =
1 + ∑i Kai ci (12)
rDIB = r7 (23)
k Ka Ka c c
r3 = 3 ME IB ME 2IB rIB = −r1 + r2 − r3 + r4 − r5 + r6 − 2r7 (24)
(1 + ∑i Kai ci ) (13)
rG = −r1 + r2 (25)
k 4 KaDE cDE
r4 =
1 + ∑i Kai ci (14) The kinetic model above was fitted to kinetic data under high
conversions and the initial rates of single etherification steps by mini-
k5 KaDE KaIB cDE cIB mizing the quadratic sum of scaled deviations between experimental
r5 =
(1 + ∑i Kai ci )2 (15) and calculated values:
0 0
k6 KaTE cTE k6′ cTE rcal , ijk − rexp, ijk ccal, ijk − cexp, ijk
r6 = = f= ∑∑∑ [ 0
]2 + ∑∑∑ [ max(cexp, ij )
]2
1 + ∑i Kai ci 1 + ∑i Kai ci (16) i j k
max(rexp , ij ) i j k (26)
)2
k 7 (KaIB cIB In the above equation, the subscripts i, j and k indicate temperature,
r7 =
(1 + ∑i Kai ci )2 (17) component and measured data point respectively. The subscripts exp
and cal denote experimental value and calculated value. The first term
It’s prudent to emphasize that the reactant concentrations in the of the objective function f indicates the deviation of the model with the
above equations are the concentrations in the reaction phase, i.e. the 0
measured initial rates (data points in Figs. 2–6). The initial rates rcal , ijk
concentrations in the glycerol phase if phase separation emerges, were calculated by Eqs. (20)–(25) with the initial concentration of the
otherwise the overall concentration. The adsorption constants for re- reaction phase. The second term presents the deviation of the model
actants were estimated as well as the rate constants since there is no with the experimental data under high conversion, expressed as con-
reliable source for these parameters. In modeling, it’s found that the centration evolution data of components (Fig. 9). The rate expressions
adsorption equilibrium constants for TE and DIB were one to two (Eqs. (20)–(25)) in combining with the definition of reaction rate
magnitudes smaller than for other components. The adsorption terms (Equation (1)) were integrated numerically to obtain concentrations of
for TE and DIB were thus omitted in the nominator of these equations, different instants ccal, ijk under high conversions. It’s prudent to note
i.e., i = ME , DE , IB, G . The temperature dependence of adsorption and again that the concentrations of reaction phase were used in Eqs.
rate constants is described by: (20)–(25) and phase equilibrium calculation was performed to obtain
Eai 1 1 these concentrations when phase separation occurred. The objective
ki = k 0, i exp[− ( − )], i = 1, 2, ...,7
R T Tr (18) function enables fitting the model with the initial rates and con-
centration evolution data simultaneously. With the temperature de-
Δa Hi 1 1 Δ S
Kai = exp[− ( − ) + a i ], i = ME , DE , IB, G pendence of rate constants and adsorption constants described by Eqs.
R T Tr R (19)
(18), (19), the activation energies and adsorption parameters were es-
The reference temperature Tr is specified as 357 K. timated by fitting the model with kinetic data under all temperatures.
The rates for each species are: The estimated values for the kinetic and adsorption parameters

8
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

together with their 95% confidence intervals are shown in Tables 2 and constant concentrations of reactants in the reaction phase, for such a
3. It should be noted that the pre-exponential factor and activation binary liquid–liquid system. When the reaction was fed with glycerol
energy for R6 actually involve the effect of adsorption (k6′ = k6 KaTE ). ethers (ME or DE) and isobutene, further etherification of the hydroxyl
The obtained values for parameters are in the normal ranges. The group of the ethers with isobutene and decomposition of glycerol ethers
confidence interval for the estimated activation energy of R6 is rela- took place simultaneously. It was experimentally verified that adding
tively large, due to the less effective data points compared with other ME to the G-IB reaction can accelerate the glycerol conversion.
reactions. The initial rate for TE decomposition (R6) was not measured, The rate expressions in the power-law form and parameters re-
and R6 only occurs when significant TE accumulated in the reaction gressed from the initial rates of single etherification steps failed to
under high conversion. The estimated adsorption parameters for gly- predict the kinetics of complex reaction which was affected by other
cerol and isobutene accord with the values estimated for their homologs reactions and components. That implies the necessity of considering
from adsorption experiments on resin [39]. The adsorption enthalpy for component adsorption in kinetic modeling. To get effective data ac-
glycerol is smaller than for glycerol ethers and isobutene. The calcu- counting for the interaction of reactions and effect of components, the
lated adsorption equilibrium constant (by Equation (19)) for glycerol is glycerol etherification with isobutene was carried out under high con-
larger than for ME, supporting the presumption that the ME adsorption versions, where all reactions and reactants coexisted. Then those re-
was hindered by glycerol in the G-IB-ME reaction. The values of action data under high conversions and the initial rates were correlated
k5 KaDE KaIB under different temperatures are close, due to the opposite with a model considering the component adsorption and phase se-
trends of rate constant (k5 ) and adsorption equilibrium constant paration to solve the model parameters. The proposed model is capable
(KaDE KaIB ) with the change of temperature. This explains the close of interpreting the observed kinetics of the complex reaction.
initial rates observed for TE formation from DE and IB under different
temperatures (Fig. 5). Acknowledgments
The concentrations calculated by the kinetic model under high
conversions are plotted as solid lines in Fig. 9, which tightly capture the This work was supported by the National Natural Science
evolution trends of the measured concentrations. The corresponding Foundation of China (grant number 21706202), China Postdoctoral
concentrations in the glycerol phase and isobutene phase are also de- Science Foundation (grant number 2017M623179), and Natural
monstrated as dashed and dot-dashed lines. During the reaction, the Science Basic Research Plan in Shaanxi Province of China (grant
reaction mixture changes from two liquid phases to one homogeneous number. 2018JQ2059). J Liu was also supported by the International
liquid phase. The concentrations change suddenly instead of smoothly Postdoctoral Exchange Fellowship Program by the Office of the China
because the homogenization process does not pass the plaint point of Postdoctoral Council (grant number 20180066).
the corresponding phase diagram. The calculated evolution data of
overall concentrations, and concentrations in the two separate phases Appendix A. Supplementary data
under other temperatures were illustrated in Figures S4-S6 in the
Supplementary Material. Fig. 10 shows the parity plots of experimental Supplementary data to this article can be found online at https://
and calculated (overall) concentrations under all temperatures. The doi.org/10.1016/j.cej.2019.122037.
relatively large deviation for DIB concentrations between measured and
predicted values is attributed to its low magnitude and lack of effective References
data, as DIB only appears at the last half the reaction. The initial rates of
each etherification step calculated using estimated parameters are il- [1] M.R. Monteiro, C.L. Kugelmeier, R.S. Pinheiro, M.O. Batalha, A. da Silva César,
lustrated in the corresponding figures (Figs. 2–6). The good accordance Glycerol from biodiesel production: technological paths for sustainability, Renew
Sust. Energ Rev. 88 (2018) 109–122.
of the calculated values with the measured values signifies that the [2] A. Rodrigues, J.C. Bordado, R.G.d. Santos, Upgrading the glycerol from biodiesel
proposed model is capable of describing the kinetic data accurately. As production as a source of energy carriers and chemicals—a technological review for
a test for the application of the proposed model, it was employed to three chemical pathways, Energies 10 (2017) 1817.
[3] S. Veluturla, N. Archna, D. Subba Rao, N. Hezil, I.S. Indraja, S. Spoorthi, Catalytic
predict the initial rates of glycerol conversion in the G-IB-ME reaction. valorization of raw glycerol derived from biodiesel: a review, Biofuels 9 (2016)
The model provides reasonably good prediction to the observed initial 305–314.
rates, as depicted by the solid line in Fig. 7. [4] S. Bagheri, N.M. Julkapli, W.A. Yehye, Catalytic conversion of biodiesel derived raw
glycerol to value added products, Renew. Sust. Energ. Rev. 41 (2015) 113–127.
We also tried estimating the parameters of the model (Equations [5] H.W. Tan, A.R. Abdul Aziz, M.K. Aroua, Glycerol production and its applications as
(11) - (25)) only using the kinetic data under high conversions. The a raw material: a review, Renew. Sust. Energ. Rev. 27 (2013) 118–127.
regression agreed well with the experimental observations. But the [6] C.A.G. Quispe, C.J.R. Coronado, J.A. Carvalho Jr, Glycerol: production, consump-
tion, prices, characterization and new trends in combustion, Renew. Sust. Energ.
model with the obtained parameters cannot reproduce the initial rates
Rev. 27 (2013) 475–493.
of each etherification step (Figs. 3–6), except for R1. In the reaction fed [7] C. Beatrice, G. Di Blasio, M. Lazzaro, C. Cannilla, G. Bonura, F. Frusteri,
with glycerol and isobutene to obtain kinetic data under high conver- F. Asdrubali, G. Baldinelli, A. Presciutti, F. Fantozzi, G. Bidini, P. Bartocci,
sions, the reactant concentrations for R2 - R5 just varied in narrow Technologies for energetic exploitation of biodiesel chain derived glycerol: oxy-
fuels production by catalytic conversion, Appl. Energ. 102 (2013) 63–71.
ranges, as pointed out in the Introduction section. Using the model [8] C. Beatrice, G. Di Blasio, C. Guido, C. Cannilla, G. Bonura, F. Frusteri, Mixture of
parameters, regressed from kinetic data under high conversions alone, glycerol ethers as diesel bio-derivable oxy-fuel: Impact on combustion and emis-
to predict the initial rates thus actually extrapolated the model to wide sions of an automotive engine combustion system, Appl. Energ. 132 (2014)
236–247.
ranges of reactant concentrations. However, the extrapolation failed for [9] J.A. Melero, G. Vicente, G. Morales, M. Paniagua, J. Bustamante, Oxygenated
this reaction system. This trial confirmed the necessity of combining compounds derived from glycerol for biodiesel formulation: influence on EN 14214
both initial rates of each step and kinetic data under high conversions in quality parameters, Fuel 89 (2010) 2011–2018.
[10] J. Liu, B. Yang, Liquid-liquid-solid mass transfer and phase behavior of hetero-
the kinetic modeling for this consecutive reaction. geneous etherification of glycerol with isobutene, AIChE J. 64 (2018) 2526–2535.
[11] L. Frusteri, C. Cannilla, G. Bonura, A.L. Chuvilin, S. Perathoner, G. Centi, F. Frusteri,
4. Conclusion Carbon microspheres preparation, graphitization and surface functionalization for
glycerol etherification, Catal. Today 277 (2016) 68–77.
[12] Ö.D. Bozkurt, F.M. Tunç, N. Bağlar, S. Çelebi, İ.D. Günbaş, A. Uzun, Alternative fuel
The kinetics of consecutive glycerol etherification with isobutene additives from glycerol by etherification with isobutene: structure–performance
catalyzed by NKC-9 were studied stepwise. For the synthesis of ME from relationships in solid catalysts, Fuel Process. Technol. 138 (2015) 780–804.
[13] J. Zhou, Y. Wang, X. Guo, J. Mao, S. Zhang, Etherification of glycerol with iso-
glycerol and isobutene, the reactants separate into two phases which
butene on sulfonated graphene: reaction and separation, Green Chem. 16 (2014)
are in equilibrium in the reaction. The measured initial rates for ME 4669–4679.
synthesis do not change with the variation of G/IB ratios, due to the [14] M.D. González, P. Salagre, R. Mokaya, Y. Cesteros, Tuning the acidic and textural

9
J. Liu, et al. Chemical Engineering Journal 375 (2019) 122037

properties of ordered mesoporous silicas for their application as catalysts in the [26] J. Liu, B. Yang, C. Yi, Kinetic study of glycerol etherification with isobutene, Ind.
etherification of glycerol with isobutene, Catal. Today 227 (2014) 171–178. Eng. Chem. Res. 52 (2013) 3742–3751.
[15] M.D. González, P. Salagre, M. Linares, R. García, D. Serrano, Y. Cesteros, Effect of [27] J. Singh, J. Kumar, M.S. Negi, D. Bangwal, S. Kaul, M.O. Garg, Kinetics and mod-
hierarchical porosity and fluorination on the catalytic properties of zeolite beta for eling study on etherification of glycerol using isobutylene by in situ production
glycerol etherification, Appl. Catal. A 473 (2014) 75–82. from tert-butyl alcohol, Ind. Eng. Chem. Res. 54 (2015) 5213–5219.
[16] C. Cannilla, G. Bonura, L. Frusteri, F. Frusteri, Catalytic production of oxygenated [28] J.K. Cheng, C.-L. Lee, Y.-T. Jhuang, J.D.C. Ward, I-Lung Design and control of the
additives by glycerol etherification, Cent. Eur. J. Chem. 12 (2014) 1248–1254. glycerol tertiary butyl ethers process for the utilization of a renewable resource, Ind.
[17] W. Zhao, C. Yi, B. Yang, J. Hu, X. Huang, Etherification of glycerol and isobutylene Eng. Chem. Res. 50 (2011) 12706–12716.
catalyzed over rare earth modified Hβ-zeolite, Fuel Process. Technol. 112 (2013) [29] M. Di Serio, L. Casale, R. Tesser, E. Santacesaria, New process for the production of
70–75. glycerol tert-butyl ethers, Energ Fuel 24 (2010) 4668–4672.
[18] A. Turan, M. Hrivnák, K. Klepáčová, A. Kaszonyi, D. Mravec, Catalytic etherification [30] J. Liu, P. Daoutidis, B. Yang, Process design and optimization for etherification of
of bioglycerol with C4 fraction, Appl. Catal. A 468 (2013) 313–321. glycerol with isobutene, Chem. Eng. Sci. 144 (2016) 326–335.
[19] M.D. González, P. Salagre, E. Taboada, J. Llorca, E. Molins, Y. Cesteros, Sulfonic [31] M.E. Jamroz, M. Jarosz, J. Witowska-Jarosz, E. Bednarek, W. Tecza, M.H. Jamroz,
acid-functionalized aerogels as high resistant to deactivation catalysts for the J.C. Dobrowolski, J. Kijenski, Mono-, di-, and tri-tert-butyl ethers of glycerol. A
etherification of glycerol with isobutene, Appl. Catal. B 136–137 (2013) 287–293. molecular spectroscopic study, Spectrochim Acta, Part A 67 (2007) 980–988.
[20] M.D. González, P. Salagre, E. Taboada, J. Llorca, Y. Cesteros, Microwave-assisted [32] H.J. Lee, D. Seung, K.S. Jung, H. Kim, I.N. Filimonov, Etherification of glycerol by
synthesis of sulfonic acid-functionalized microporous materials for the catalytic isobutylene: tuning the product composition, Appl. Catal. A 390 (2010) 235–244.
etherification of glycerol with isobutene, Green Chem. 15 (2013) 2230–2239. [33] M.A. Vannice, W.H. Joyce, Kinetics of catalytic reactions, Springer, 2005.
[21] F. Frusteri, C. Cannilla, G. Bonura, L. Spadaro, A. Mezzapica, C. Beatrice, G. Di [34] P.B. Weisz, C. Prater, Interpretation of measurements in experimental catalysis, Adv
Blasio, C. Guido, Glycerol ethers production and engine performance with diesel/ Catal, Elsevier, 1954, pp. 143–196.
ethers blend, Top. Catal. 56 (2013) 378–383. [35] M.L. Michelsen, The isothermal flash problem. Part I. Stability, Fluid Phase Equilibr.
[22] M.D. González, Y. Cesteros, J. Llorca, P. Salagre, Boosted selectivity toward high 9 (1982) 1–19.
glycerol tertiary butyl ethers by microwave-assisted sulfonic acid-functionalization [36] L.E. Baker, A.C. Pierce, K.D. Luks, Gibbs energy analysis of phase equilibria, Soc.
of SBA-15 and beta zeolite, J. Catal. 290 (2012) 202–209. Petrol. Eng. J. 22 (1982) 731–742.
[23] F. Frusteri, L. Frusteri, C. Cannilla, G. Bonura, Catalytic etherification of glycerol to [37] H. Renon, J.M. Prausnitz, Local compositions in thermodynamic excess functions
produce biofuels over novel spherical silica supported Hyflon(R) catalysts, for liquid mixtures, AIChE J. 14 (1968) 135–144.
Bioresour. Technol. 118 (2012) 350–358. [38] J. Liu, Y. Yuan, Y. Pan, Z. Huang, B. Yang, Liquid–liquid equilibrium for systems of
[24] A. Behr, L. Obendorf, Development of a process for the acid-catalyzed etherification glycerol and glycerol tert-butyl ethers, Fluid Phase Equilibr. 365 (2014) 50–57.
of gycerine and isobutene forming glycerine tertiary butyl ethers, Eng. Life Sci. 2 [39] R. Soto, N. Oktar, C. Fité, E. Ramírez, R. Bringué, J. Tejero, Adsorption of C1–C4
(2002) 185–189. alcohols, C4–C5 isoolefins, and their corresponding ethers over Amberlyst™35,
[25] K. Klepáčová, D. Mravec, A. Kaszonyi, M. Bajus, Etherification of glycerol and Chem. Eng. Technol. 40 (2017) 889–899.
ethylene glycol by isobutylene, Appl. Catal. A 328 (2007) 1–13.

10

You might also like