Kinetic modeling of butane-2,3-diol dehydration over Nb 2 O 5 .nH 2 O

Abstract

This study employed Nb₂O₅.nH₂O for the dehydration of butane-2,3-diol, which could be derived from biomass or waste gas using a fermentation process. The experiments were conducted at a temperature ranging from 220 °C to 260 °C and a weight hourly space velocity of 0.01–0.05 min⁻¹. There are three main products that include methyl ethyl ketone, isobutyraldehyde, and butadiene. The yield of products increased with the reaction temperature. Rate data for the dehydration reaction were well represented by Langmuir–Hinshelwood kinetics with adsorption parameters in the rate equations, which assumed the formation of products was reversible with single-site reaction. The apparent activation energies for the dehydration reaction of methyl ethyl ketone, isobutyraldehyde, and butadiene obtained from the Arrhenius plot data were 19.5, 24.0, and 23.7 kJ mol⁻¹, respectively. The adsorption energies for butane-2,3-diol, methyl ethyl ketone, isobutyraldehyde, and butadiene were −182.4, −142.1, −136.1, and −105.6 kJ mol⁻¹, respectively.

Keywords

Butane-2,3-diol dehydration niobium pentoxide methyl ethyl ketone biomass

Introduction

Butane-2,3-diol (BDO) is produced by a variety of micro-organisms in a process known as butanediol fermentation.¹ BDO production via fermentation started from the Second World War for the production of rubber, plastics, lubricants, latex, and other products. However, the lack of economic competitiveness led to its unsuccessful application in industry. Recently, microbial BDO production has gained global attention as renewable biomass has emerged as a promising route for the development of a low carbon economy. BDO production without concern for the isomer product can produce BDO at a concentration of 150 g L⁻¹.² Lately, LanzaTech has developed novel technology that converts local, highly abundant waste and low-cost resources into sustainable, valuable commodities. The patented process uses an acetogenic microbe to convert waste gas containing carbon dioxide (from industrial sources like steel mills and processing plants), or syngas generated from any biomass resources (e.g. municipal solid waste, organic industrial waste, or agricultural waste), into fuels and chemicals including BDO.³ The catalytic dehydration of BDO has been investigated since the 1940s; however, the known methods for conversation of BDO suffer from several disadvantages, including corrosive reagents, radioactive catalysts, and undesirable products. Dehydration of BDO can be proceeded by different mechanisms depending upon the catalysts used,⁴ including Brønsted acid (alumina),⁵ silica-supported sodium phosphates,⁶γ-Al₂O₃,⁷ and ceria.⁸ As used herein, dehydration refers to a reaction that removes H₂O from an alcohol to form an alkene.⁹ A diol, such as BDO, may be partially dehydrated by removing one molecule of water, or completely dehydrated by removing two molecules of water.¹⁰ The biologically fermented butane-1,3-diol is usually present in an aqueous solution, so that water-tolerant solid acid catalysts are better for BDO dehydration. Niobic acid (hydrated niobium oxide, Nb₂O₅.nH₂O) has been reported to possess high acid strength¹¹ and to be a water-tolerant Lewis acid catalyst.¹² In addition, Nb₂O₅.nH₂O is inexpensive, thermally stable, and qualified as an industrial catalyst. The main purpose of this research was to acquire more insight into the dehydration of BDO over Nb₂O₅.nH₂O and the kinetics of BDO dehydration. A series of dehydration experiments was systematically carried out and used to formulate micro-kinetic models. The kinetic constants obtained from this research would be very useful to simulate bio-BDO dehydration on an industrial level.

Experimental

Chemicals

BDO (mixture of stereoisomers) was purchased from Tokyo Chemical Industry Co., Ltd. (Japan) Niobium oxide (HY-340) was provided by Companhia Brasileira de Metalurgia e Mineração (CBMM).

Blank experiment

Nb₂O₅.nH₂O (1.0 g) was loaded into the reactor, which was set at 523 K and purged with nitrogen at a rate of 100 mL min⁻¹. There were no chemicals detected by the gas chromatograph (GC). This indicated that interference from the reactor for these catalytic reactions could be ignored.

BDO dehydration experiments

The catalytic reactions were performed in a conventional continuous flow fixed-bed reactor made of stainless steel (ID = 8 mm) under atmospheric pressure. Prior to reaction, the catalyst sample (weight = 1.0 g) was treated in the reactor under N₂ flow (flow rate of 100 mL min⁻¹) at 300 °C for 2 h. The N₂ flow was controlled with mass-flow controllers. BDO was fed into the reactor via a micro pump at 3 mL h⁻¹ together with the N₂ flow. The reactor temperature was set between 220 °C and 260 °C (Figure 1).

Figure 1.

Experimental setup.

Product compositions were analyzed by an on-line GC (SRI 8610C) equipped with an MXT-1 column (non-polar phase, 60 m, ID 0.25 mm, film thickness 0.25 µm), equipped with a thermal conductivity detector (TCD), and flame ionization detector (FID) for the analysis of hydrocarbons and oxygenated chemicals, and quantified by injecting calibration standards to the GC system. The temperature of the tubing from the bottom of the reactor to the inlet of the GC was maintained at 230 °C to avoid the condensation of liquid products. The products were injected through the sample loop, which was controlled by a high-temperature 10-port valve. The oven was kept at 40 °C for 5 min, then raised to 120 °C at a ramp rate of 40 °C min⁻¹, and finally, raised to 250 °C at a rate of 20 °C min⁻¹ and held at this temperature for 10 min. To ensure the identification of products, gas chromatography–mass spectrometer (GC–MS) analyses were also carried out using an Agilent 7890A GC system equipped with an Agilent 5975C MS detector (Agilent, Santa Clara, California, USA) and HP-1 capillary column. The carbon selectivity and conversion were calculated by area normalization methods. Two repeat runs were performed under each reaction condition, and any two trials were generally within 5% of each other. The carbon balances closed with above 90% for all runs in this study.

Results and discussion

Effects of BDO concentration and reaction time

Nb₂O₅.nH₂O used in this study was commercially available. Nb₂O₅.nH₂O was found to exhibit a high acid strength (H₀ = −5.6), corresponding to the acid strength of 70% H₂SO₄, when calcined at relatively low temperatures (373–573 K). Nb₂O₅.nH₂O showed high acid strength on the surface in spite of containing water in an unusual solid acid. To test the selectivity of Nb₂O₅.nH₂O during the reaction, a different set of BDO concentration tests were carried out under the same conditions as before, except for the temperature, which was held at 250 °C.

Nb₂O₅.nH₂O showed excellent stability as a catalyst for BDO dehydration over 60 hours of study. The change in concentration of BDO did not affect the composition of the product (Figures 2 and 3). The selectivity of methyl ethyl ketone (MEK) was always maintained at 80%, and isobutyraldehyde (IBA) and butadiene (BD) were maintained around 10%.

Figure 2.

Effect of BDO concentration on selectivity.

Figure 3.

Effect of reaction time on selectivity.

Mechanism of BDO dehydration on Nb₂O₅.nH₂O

The dehydration reaction of BDO on Nb₂O₅.nH₂O is a very complex process. When Nb₂O₅.nH₂O is taken as the catalyst, there can be mainly three products.

The reaction routes to the products are summarized in Scheme 1. The reaction products over Nb₂O₅.nH₂O catalyst have not been discussed before. It could be observed that MEK and IBA were the primary products from the dehydration reaction of BDO. Acid sites on the surface of the Nb₂O₅.nH₂O¹³ could act as the active sites for converting BDO to IBA through a pinacol rearrangement.^14,15 3-Buten-2-ol was the dehydration product from the first step and this could be followed by further dehydration to the final product BD at high reaction temperature.

Scheme 1.

Mechanism of BDO dehydration on Nb₂O₅.nH₂O.

Effects of internal and external diffusion

Nb₂O₅.nH₂O used in this reaction was as porous solid particles. In order to induce the reaction, the reacting species were required to be diffused on the active surface of the catalyst. For this reason, the effect of the transport process became more and more significant. External diffusion could be excluded by changing the space velocity of the mixture at nitrogen flows of 20, 40, 60, 80, and 100 mL min⁻¹. The BDO vapor fraction in the feed was 0.02 mL min⁻¹, and the carrier flow was maximized (100 mL min⁻¹) in order to guarantee minimum external resistances. The results, presented in Figure 4, showed that the component of products did not depend on nitrogen flow when it was above 60 mL min⁻¹ and so external diffusion could be neglected.

Figure 4.

Effect of N₂ carrier flow on selectivity.

The internal mass transfer resistances were investigated by performing reactions with a 1.0 g load of catalyst at three particle size ranges: namely, 100–300, 300–500, and 500–700 μm. As suggested in Figure 5, the moderate particle size of 300–500 μm was chosen for the remainder of the work.

Figure 5.

Effect of particle size on selectivity.

Flow–temperature profiles

Concentration–time profiles on Nb₂O₅.nH₂O catalysts at different temperatures and initial substrate concentrations were obtained to capture the temporal evolution of various intermediate products and to establish acceptable carbon balance closure (Figure 6). Such knowledge is essential to the development of detailed and reliable kinetic models. To confirm the absence of external and intraparticle mass transfer limitations, components of the products were utilized.

Figure 6.

Catalytic results for the conversion of BDO over different temperatures and weight hourly space velocity.

Kinetic model

In this study, the Langmuir–Hinshelwood (LH) model was applied to derive the reaction rate equations. The reaction mechanism followed a classical scheme of reactant–adsorption/surface–reaction/product–desorption, and all the kinetic models included the effects of reactant and product concentrations. This involved adsorption of BDO, reaction between adsorbed reactant (BDO) and the catalyst’s active site (S), and finally, desorption of BD, IBA, MEK, and water from the catalyst surface. Hence, the surface reaction was assumed to be the rate-determining step. The surface reaction mechanism was described by the following equations

BDO + S \leftrightarrow BDO \cdot S, θ_{BDO} = K_{BDO} P_{BDO} θ_{V}

(1)

BDO \cdot S + S \begin{matrix} \to^{k_{MEK}^{+}} \\ \leftarrow_{k_{MEK}^{-}} \end{matrix} MEK \cdot S + S + W, r_{MEK} = k_{MEK}^{+} θ_{BDO} θ_{V} - k_{MEK}^{-} θ_{MEK} θ_{V}

(2)

BDO \cdot S + S \to^{k_{MEK}^{+}} MEK \cdot S + S + W, r_{MEK} = k_{MEK}^{+} θ_{BDO} θ_{V}

(3)

BDO \cdot S + S \begin{matrix} \to^{k_{IBA}^{+}} \\ \leftarrow_{k_{IBA}^{-}} \end{matrix} MEK \cdot S + S + W, r_{IBA} = k_{IBA}^{+} θ_{BDO} θ_{V} - k_{IBA}^{-} θ_{IBA} θ_{V}

(4)

BDO \cdot S + S \to^{k_{IBA}^{+}} IBA \cdot S + S + W, r_{IBA} = k_{IBA}^{+} θ_{BDO} θ_{V}

(5)

BDO \cdot S + S \begin{matrix} \to^{k_{BD}^{+}} \\ \leftarrow_{k_{BD}^{-}} \end{matrix} BD \cdot S + S + 2 W, r_{BD} = k_{BD}^{+} θ_{BDO} θ_{V} - k_{BD}^{-} θ_{IBA} θ_{V}

(6)

BDO \cdot S + S \to^{k_{IBA}^{+}} BD \cdot S + S + 2 W, r_{BD} = k_{BD}^{+} θ_{BDO} θ_{V}

(7)

MEK \cdot S \leftrightarrow MEK + S, θ_{MEK} = K_{MEK} P_{MEK} θ_{V}

(8)

IBA \cdot S \leftrightarrow IBA + S, θ_{IBA} = K_{IBA} P_{IBA} θ_{V}

(9)

BD \cdot S \leftrightarrow BD + S, θ_{BD} = K_{BD} P_{BD} θ_{V}

(10)

θ_{V} + θ_{BDO} + θ_{MEK} + θ_{IBA} + θ_{BD} = 1

(11)

θ_{V} = \frac{1}{1 + K_{EG}}

(12)

The value of the partial pressure of each species was calculated using

P_{i} = (\frac{N_{i}}{N_{total}}) P_{total}

(13)

Polymath 6.0¹⁶ was used to fit the experimental data. The main function of the Levenberg–Marquardt algorithm was to minimize the sum of square errors $(r^{2})$ created by the difference between the experimental rates $(r_{iexp})$ and the calculated rates $(r_{ical})$

r^{2} = 1 - \frac{\sum_{1}^{n} {(r_{iexp} - r_{ical})}^{2}}{\sum_{1}^{n} {(r_{iexp} - \hat{r})}^{2}}

(14)

Furthermore, the standard error (SE) was calculated by dividing the standard deviation $(σ)$ by the square root of the number of experiment trails (N) as shown in the equation (15)

SE = \frac{σ}{\sqrt{N}}

(15)

To construct an accurate kinetic model, the first step was to make reasonable assumptions for the proposed reactions (Table 1). There were two alternative assumptions for BDO dehydration. The first assumption, as shown in equations (3), (5), and (7), anticipated that the reaction was irreversible. The second assumption, as shown in equations (2), (4), and (6), supposed that the reaction was reversible. The objective of this research was to find out which assumption could generate the most suitable model for BDO dehydration. The results of the regression analysis using the Polymath program are illustrated in Tables 2 and 3.

Table 1.

Brief description of kinetic models.

No.	Assumptions	Equations
I	Reaction was irreversible	$r_{i} = \frac{k_{i}^{+} K_{BDO} P_{BDO}}{{(1 + K_{EQ})}^{2}}$
II	Reaction was reversible	$r_{i} = \frac{k_{i}^{+} K_{BDO} P_{BDO} - k_{i}^{-} K_{i} P_{i}}{{(1 + K_{EQ})}^{2}}$

Table 2.

Rate constants and activation energies of the surface reactions based on Model I.

Rate constant (×10⁻⁶)	493 K	503 K	513 K	523 K	533 K	$r^{2}$	E_a
$k_{MEK}^{+}$	6.2	7.1	8.5	9.3	10.8	0.76	30.164
$k_{IBA}^{+}$	4.7	5.3	6.7	7.4	8.9	0.68	35.18
$k_{BD}^{+}$	2.8	3.7	4.6	5.8	7.5	0.83	52.86
Adsorption constant (×10⁻³)	220°	230°	240°	250°	260°	$r^{2}$	$- Δ H$
$K_{BDO}$	10.3	14.3	27.9	279.1	687.1	0.85	247.1
$K_{MEK}$	45.1	87.9	117.8	457.1	789.2	0.87	165.4
$K_{IBA}$	38.7	60.7	70.5	400.7	650.9	0.81	163.8
$K_{BD}$	91.3	145.7	250.8	875.6	1660.3	0.78	160.6

Rate constant unit: m³ kg⁻¹ s⁻¹; adsorption equilibrium constants unit: m³ kmol⁻¹; E_a and $Δ H$ unit: kJ mol⁻¹.

Table 3.

Rate constants and activation energies of the surface reactions based on Model II.

Rate constant	493 K	503 K	513 K	523 K	533 K	$r^{2}$	E_a
$k_{MEK}^{+}$	7.3	8.5	10.5	13.8	14.8	0.95	41.5
$k_{MEK}^{-}$	1.3	1.4	1.6	1.8	1.9		22.0
$k_{IBA}^{+}$	4.8	3.7	5.1	6.7	7.8	0.96	43.8
$k_{IBA}^{-}$	1.2	1.4	1.5	1.7	1.8		19.8
$k_{BD}^{+}$	2.8	4.0	4.3	5.3	6.8	0.95	44.9
$k_{BD}^{-}$	1.1	1.2	1.3	1.5	1.6		21.2
Adsorption constant	493 K	503 K	513 K	523 K	533 K	$r^{2}$	$- Δ H$
$K_{BDO}$	11.8	53.9	98.3	239.7	357.0	0.97	182.4
$K_{MEK}$	47.9	85.7	178.8	377.3	589.2	0.93	142.1
$K_{IBA}$	41.3	64.7	174.3	295.7	433.3	0.96	136.1
$K_{BD}$	94.5	143.8	233.7	448.9	599.7	0.94	105.6

Rate constant unit: m³ kg^-1 s^-1; adsorption equilibrium constants unit: m³ kmol^-1; E_a and ΔH unit: kJ mol^-1.

The rate constants for each reaction at various temperatures and the activation energies, and adsorption enthalpies were calculated based on the Arrhenius and van’t Hoff equations. The relative quality of the fit of Models I and II, specifically the sum of squares of residuals, is compared in Tables 2 and 3. It is observed that, among the two models, Model II best represents the experimental data and trends. The estimated parameters of Model II, except r², exhibit a good fit for the concentration of all other species. The kinetic parameters obtained from fitting this model to experimental data are presented in Table 3. The E_a for BDO dehydration to form MEK is approximately 19.5 kJ mol⁻¹, which is smaller than the value for dehydration conversion of BDO on ZSM-5, which is 32.3 kJ mol⁻¹.¹⁷E_a for BDO dehydration to form IBA and BD are 24.0 and 23.7 kJ mol⁻¹, respectively. Both of these values are larger than the activation energy to form MEK. Furthermore, while the adsorption energies are comparable, and it can be shown that the adsorption energy of BDO is the largest, due to the presence of two hydroxyl groups in the structure of BDO. The adsorption energy of BD is 105.6 kJ mol⁻¹. It is necessary to point out that while Model II describes a reversible reaction on the catalyst surface, diffusion in the catalyst pellets is also important.

Conclusion

Nb₂O₅.nH₂O is an active and selective catalyst for dehydration of BDO and therefore can be used to dehydrate BDO produced through fermentation. More than 80% selectivity of BDO dehydration was produced on the catalyst Nb₂O₅.nH₂O, with BD and two other products detected. Mechanistic modeling of BDO conversion over Nb₂O₅.nH₂O catalyst at 220 °C–260 °C under an inert atmosphere revealed mechanistic insights into the dehydration of BDO. Detailed kinetic analysis illustrated that a model assuming a reversible reaction on the catalyst surface fitted temporal concentration–time data well. The apparent activation energies for the dehydration reaction of MEK, IBA, and BD obtained from the Arrhenius plot data were 19.5, 24.0, and 23.7 kJ mol⁻¹, respectively. The adsorption energies for BDO, MEK, IBA, and BD were −182.4, −142.1, −136.1, and −105.6 kJ mol⁻¹, respectively. These insights should be useful to influence the rational design of catalysts and processes for the selective BDO dehydration in general.

Footnotes

Appendix 1 Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (grant no. 21602206) and Zhejiang Provincial Natural Science Foundation of China (grant no. LY15B060008).

Author biographies

Gui-Lin Cheng is an Assistant Professor of Pharmaceutical Research Center of Public Platform, Zhejiang Chinese Medical University. She has a PhD in Chemistry from Zhejiang University.

Lin-Yan Wang is Experimenter of Pharmaceutical Research Center of Public Platform, Zhejiang Chinese Medical University. She has a Master’s degree.

Cheng-Jun Jiang is an Associate Professor in Zhejiang University of Science and Technology. She has a PhD in Chemcal Engineering from Zhejiang University.

References

Biddy

Davis

Humbird

et al . ACS Sustain Chem Eng 2016; 4: 3196–3211.

Huang

Ouyang

PK.

Biotechnol Adv 2011; 29: 351–364.

Köpke

Mihalcea

Liew

et al . Appl Environ Microb 2011; 77: 5467–5475.

Zheng

Wales

Heidlage

et al . J Catal 2015; 330: 222–237.

Jing

Katryniok

Araque

et al . Catal Sci Technol 2016; 6: 5830–5840.

Kim

Shin

Lee

et al . Appl Catal A Gen 2016; 511: 156–167.

Liu

Fabos

Taylor

et al . Chem Eur J 2016; 22: 12290–12294.

Sato

Takahashi

Sodesawa

et al . Catal Commun 2004; 5: 397–400.

Nash

Ramanathan

Ruddy

et al . Appl Catal A Gen 2016; 510: 110–124.

10.

Nikitina

Ivanova

. Chem Cat Chem 2016; 8: 1346–1353.

11.

Okuhara

Chem Rev 2002; 102: 3641–3666.

12.

Ali

Moromi

Touchy

et al . Chem Cat Chem 2016; 8: 891–894.

13.

Nakajima

Baba

Noma

et al . J Amer Chem Soc 2011; 133: 4224–4227.

14.

Duan

Sun

Yamada

et al . Catal Commun 2014; 48: 1–4.

15.

Zhang

et al . Green Chem 2012; 14: 3441–3450.

16.

Polymath 6.0. Available at www.polymath-software.com

17.

Multer

McGraw

Hohn

et al . Ind Eng Chem Res 2013; 52: 56–60.