+ All Categories
Home > Documents > 1-s2.0-S138151480400094X-main.pdf

1-s2.0-S138151480400094X-main.pdf

Date post: 03-Apr-2018
Category:
Upload: er-mayur-patil
View: 212 times
Download: 0 times
Share this document with a friend

of 14

Transcript
  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    1/14

    Kinetics of esterification of lactic acid with ethanol

    catalyzed by cation-exchange resins

    Yang Zhang, Li Ma, Jichu Yang *

    Department of Chemical Engineering, Tsinghua University, Beijing 100084, Peoples Republic of China

    Received 17 March 2004; accepted 26 April 2004Available online 25 June 2004

    Abstract

    The esterification of lactic acid with ethanol was carried out in the presence of five different cation-exchange resins.

    The effect of catalyst type, catalyst loading, and temperature on reaction kinetics was evaluated. In order to study which

    components had the strongest adsorption strength on the resin surface, two simplified mechanisms based on Langmuir

    Hinselwood model were compared by correlating the experimental data. FTIR method was used to verify the ratio-

    nality of the mechanism. Nonideality of the liquid phase was taken into account by using activities, which were

    predicted by UNIFAC method instead of concentrations. The thermal stability and mechanical strength of the resin

    catalysts were tested by SEM. 2004 Elsevier B.V. All rights reserved.

    Keywords: Lactic acid; Esterification; Ethanol; Kinetics; Cation-exchange resin

    1. Introduction

    Ethyl lactate is an important organic ester,

    which is biodegradable and can be used as food

    additive, perfumery, flavor chemicals and solvent,which can dissolve acetic acid cellulose and many

    resins [1]. Furthermore, the esterification of lactic

    acid with ethanol is a step in the purification of

    lactic acid by reactive distillation [2,3].

    The conventional way to produce ethyl lactate

    is the esterification of lactic acid with ethanol

    catalyzed by sulphuric acid. Since this kind of

    homogeneous catalyst may cause a lot of prob-

    lems, many heterogeneous solid catalysts were

    used in the reaction such as ion-exchange resin,

    clays and clay supported heteropoly acids [4].Among these catalysts, cation-exchange resin is a

    perfect substitute which has many advantages such

    as: (a) corrosion problems could be avoided and it

    is easier to dispose of the waste liquor from the

    reaction mixture; (b) continuous operations in

    columns are possible; (c) the catalyst can be easily

    removed from the reaction products by decanta-

    tion or filtration; (d) the purity of the products is

    higher since side reactions can be eliminated or are

    less significant [5,6].

    * Corresponding author. Tel.: +86-10-62-788568/785514; fax:

    +86-10-62-770304.

    E-mail address: [email protected] (J. Yang).

    1381-5148/$ - see front matter 2004 Elsevier B.V. All rights reserved.

    doi:10.1016/j.reactfunctpolym.2004.04.003

    Reactive & Functional Polymers 61 (2004) 101114

    www.elsevier.com/locate/react

    REACTIVE

    &

    FUNCTIONAL

    POLYMERS

    http://mail%20to:%[email protected]/http://mail%20to:%[email protected]/
  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    2/14

    The industrial production of esters by esterifi-

    cation of acid and alcohol was first carried out in a

    continuous stirred tank reactor (CSTR) and later

    in a catalytic distillation column over cation-ex-

    change resins. Nowadays, some investigations

    have focused on the water-permeable membrane

    reactors applied to liquid-phase reversible reac-

    tions [1,7,8]. The conversion of the esterification of

    lactic acid with ethanol exceeded the equilibrium

    limit remarkably with the aid of vapor-permeation

    according to their researches.In order to optimize the design of a CSTR, a

    reactive distillation column or a membrane reac-

    tor, it is necessary to have some information on the

    reaction kinetics. Although there have been many

    researches about the esterification of lactic acid

    with different alcohols over cation-exchange resins

    [49], very few reports concern the synthesis of

    ethyl lactate with heterogeneous catalysts. In this

    paper, the emphasis of our work was to use

    LangmuirHinselwood (LH) model [10] in a study

    of the esterification kinetics over different cation-

    exchange resins. Two simplified LH mechanisms

    were compared in order to find which one de-

    scribed the reaction kinetics better. Fourier trans-

    form infrared spectroscopic (FTIR) analysis was

    also used to verify the rationality of the mecha-

    nism. Scanning electron microscopy (SEM) was

    introduced to evaluate the mechanical strength of

    the resins and their thermal stability.

    2. Materials and catalysts

    Ethanol (purity >99.7 wt%) was purchased

    from Yili fine chemical Co., Beijing. Ethyl

    LL-lactate was purchased from SigmaAldrich.LL-

    lactic acid (80 wt%) was obtained from PURAC

    Biochem, Netherlands. The catalysts used in the

    experiments were commercial strong-acid cation-

    exchange resins. Their physical properties are

    listed in Table 1. Before use, the fresh catalysts

    Nomenclature

    aiactivity of i (mol/l)

    ci concentration of i at the surface of the

    catalyst (mol/l)

    DA molecular diffusion coefficient

    De effective diffusion coefficient

    EA apparent activation energy (kJ mol1)

    DH enthalpy change (kJ mol1)Keq reaction equilibrium constant

    k reaction rate constant (mol g1 min1)k0 pre-exponential factor (mol g1 min1)k0i adsorption coefficient at initial temper-

    ature

    ki adsorption coefficientM molar mass of the solvent

    MRD mean relative deviation

    N number of data points

    nLA;0 initial molar amount of lactic acid (mol)

    R gas constant (J mol1 K1)r reaction rate (mol g1 min1)r0 radius of catalyst particle

    S vacant site on catalyst surface

    SRS sum of residual squares

    Ttemperature (K)

    t time (min)

    W dry catalyst weight (g)

    X conversion

    VA molar volume at boiling point

    Greek symbols

    c activity coefficient

    e porosity

    s tortuosity

    / thiele modulus

    U0 association factorl viscosity (Pa s)

    Subscripts

    calc calculated values

    exp experimental values

    LA lactic acid

    E ethanol

    EL ethyl lactate

    W water

    102 Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    3/14

    were dried at 90 C over 12 h to remove moisture

    completely after being washed with pure water,

    ethanol, hydrochloric acid and pure water

    sequentially.

    3. Kinetics experimental apparatus and procedure

    3.1. Apparatus

    The esterification reactor consisted of a three-

    necked flask of 250 ml capacity fitted with a con-

    denser, a sampling device and a thermometer. The

    temperature was controlled by a thermostating

    bath, which ensured a temperature constancy of

    0.2 C in the reactor. A magnetic stirrer was usedto mix the reactants, and the frequency was about

    450 rpm.

    3.2. Procedure

    Lactic acid and cation-exchange resins were

    charged into the reactor, and then heated to the

    desired temperature. Finally, ethanol was added.

    This was taken as zero time for a run. The initial

    molar ratio of lactic and ethanol was 1:3, and the

    total volume of the reactant was 196.77 ml. About

    1 ml of liquid sample was withdrawn from the

    reactor at regular intervals for gas chromatogra-

    phy (GC) analysis. In a typical run, about 10

    samples were taken from the system.

    4. GC analysis

    GC (Shimadzu GC-9AM) was used to analyze

    the sample, which was equipped with a flame

    ionization detector (FID). The sample size for GC

    was 0.2 ll. The injection port and detector tem-

    peratures were set to be 240 and 250 C separately.

    A capillary column (25 m 0.5 mm, SE-30) wasused. The column temperature was programmed

    to rise from an initial value of 100125 C at

    2.5 C/min, and then held constant at 125 Cfor an additional 4 min. High purity helium gas

    (99.999%) was used as a carrier gas. The flow rate

    of the carrier gas was 40 ml/min.

    5. FTIR and SEM

    The catalyst samples, which had been used in

    the kinetics experiments, were tested in FTIR

    analysis. Before test, they were treated under 100

    C with different desorption times from 0 to 8 h inan oven, and then sealed in plastic bags.

    The FTIR analysis was carried out at room

    temperature using NICOLET 560 equipment. The

    samples were ground to fine powders using an

    agate mortar immediately prior to analysis. KBr

    was used as the embedding medium to make every

    sample tablets containing around 1.3 mg of ion-

    exchange resin powder [11,12].

    A SEM equipment (KYKY-2800, KYKY

    Technology Development Ltd.) was used to

    Table 1

    Physical properties of the strong-acid cation-exchange resins

    Property Amberlyst-15 (Mp) D001 (Mp) D002 (Mp) NKC (Mp) 002 (G)

    Shape Bead Bead Bead Bead BeadSize (mm) 0.5P90% 0.321.25P 95% 0.41.25P 95% 0.41. 25P 95% 0.51.25P95%

    Internal surface

    area (m2/g)

    50 N/A 3540 77 N/A

    Weight capacity

    (mEq/g)

    4.7 P 4.35 P 4.8 P 4.7 P 5.0

    Temperature

    stability (C)

    120 100 120 100 180

    Manufacturer Rohm and Haas Co.,

    USA

    Shandong Dongda

    Chemical Industry

    Co., PRC

    Jiangyin Organic

    Chemical Plant,

    PRC

    Chemical Plant of

    Nankai Univ.,

    PRC

    Jiangyin Organic

    Chemical Plant,

    PRC

    Mp Macroporous; G Gel; N/A Not available.

    Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114 103

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    4/14

    evaluate the mechanical strength and the thermal

    stability of the catalysts, and the acceleration

    voltage is 20 kV.

    6. Results and discussion

    6.1. External and internal diffusion

    In a heterogeneous catalytic reaction, there are

    several processes that influence the rate of reac-

    tion, which are external and internal diffusion of

    reactants, adsorption, surface reaction, desorp-

    tion, internal and external diffusion of reaction

    products. In order to study the intrinsic kinetics atthe catalyst surface, external and internal diffusion

    should not be the rate limiting steps [8]. According

    to Sanzs work [9], if the speed of stirring was

    between 300 and 700 rpm for the similar cation-

    exchange resins, the influence of external diffusion

    could be neglected. Therefore, the stirring speed

    was set around 450 rpm in all further experiments

    to ensure the absence of external mass transfer

    resistance.

    To evaluate the effect of internal diffusion on

    the cation-exchange resins, Eq. (1) was used [13].

    / r20k

    9De; 1

    where r0 and De denote the radius of catalyst

    particle and the effective diffusion coefficient re-

    spectively. k is the reaction rate constant, and / is

    the thiele modulus. If the calculated value of/ was

    smaller than 1, the internal diffusion could be

    neglected.

    The effective diffusion coefficient was defined as

    follows:

    De DAes

    ; 2

    where DA is the molecular diffusion coefficient. s is

    the particle tortuosity and e is the porosity. For

    most resin catalysts, the values of e=s are between0.25 and 0.50 [14]. In the calculation, the value of

    e=s was taken 0.50. The molecular diffusion coef-ficient for liquid phase diffusion can be evaluated

    from the WilkeChang equation (Eq. (3)) [15].

    DA 7:4 108 ffiffiffiffiffiffiffiffiffiffiffiffiU0MTp

    VA0:6l: 3

    In Eq. (3), U0 and M denote the associationfactor and the molar mass of the solvent. VA is the

    molar volume at boiling point and l is the vis-

    cosity of the solution. For solvent mixtures, the

    association molar mass can be calculated byffiffiffiffiffiffiffiffiffiU0M

    p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPxiU0iMip . Le Bas method was used in

    order to calculate VA [15].

    With the increase of temperature, the viscosity

    of the reaction solution will decrease while the

    diffusion coefficient will increase. As the result, if

    the influence of internal diffusion could be ne-

    glected at lower temperature, the influence at

    higher temperature could also be neglected ac-

    cording to Eq. (1). At 25 C, through experiment,

    the viscosity of the equilibrium reaction solution

    was 2.3963 103 Pa s. Then the calculation re-sults of Eqs. (1) and (2) were / 0:02 andDe 2:2 106 m2 s1, which indicated that theeffect of internal diffusion on the reaction rate

    could be ignored reasonably. Based on the dis-

    cussion above, the experimental kinetic results

    could be considered to reflect the intrinsic kinetics

    of the esterification reaction catalyzed by cation-

    exchange resins.

    6.2. Assumption of the model and parameters

    estimation

    The heterogeneous reaction has been described

    with many models such as LangmuirHinselwood,

    EleyRideal (ER) and pseudo-homogeneous

    model [6,10]. Among them, LH model was found

    to be more appropriate for this kind of esterifica-

    tion reaction [9,14,16]. The reaction mechanism

    can be described as follows:

    LA S () LA S

    E S () E S

    E S LA S () EL S W S

    EL S () EL S

    W S () W S

    104 Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    5/14

    Therefore, the LH model could be written as:

    r nLA;0W

    dX

    dt k

    aLAaE aELaWKeq1 kWaW kEaE kLAaLA kELaEL2

    ;

    ki ciSaicS

    ; Keq aEL aWaLA aE

    eq

    ;

    i Water, Ethanol, Lactic acid, Ethyl lactate4

    In the equations, nLA;0 is the initial molar con-

    centration of lactic acid and W is the catalyst load-

    ing, k represents the reaction rate constant and ki

    represents the adsorption coefficient. cS and ciSdenote the concentration of vacant site on catalyst

    surface and the concentration of component i at the

    catalyst surface respectively. ai is the activity for

    each component andKeq is the reaction equilibrium

    constant. The activities were calculated by the

    UNIFAC method [15]. The splittingof the groups is

    shown in Table 2. The volume and area parameters

    of the groups and their interaction parameters were

    taken from the book of Fredenslund et al. [16].

    Since the denominator of Eq. (4) is a multi-

    component complex, the parameters cannot beregressed correctly from the experimental results.

    It was assumed that the adsorption of the mole-

    cules was competitive on the same active site, and

    only those that had the strongest adsorption were

    taken into account in the simplified mechanisms.

    There were some different opinions about which

    components had the strongest adsorption on the

    catalyst surface [810,14,17]. Two mechanisms

    were tested in our work to find which one was

    more reliable. Mechanism A assumes that ethanol

    and water adsorbed much stronger than other

    components in the esterification solution, so the

    adsorption of ethyl lactate and lactic acid wasneglected. We get

    r kaLAaE aELaWKeq

    1 kWaW kEaE2: 5

    In contrast to mechanism A, it was assumed

    that lactic acid and water had the strongest ad-

    sorption in mechanism B (Eq. (6)).

    r kaLAaE aELaWKeq

    1 kWaW kLAaLA2: 6

    In mechanisms A or B, there are three param-eters instead of five (Eq. (4)) to be estimated at a

    constant reaction temperature. In mechanism A,

    they are k, kW and kE. While for mechanism B, the

    corresponding parameters are k, kW and kLA. A

    two-stage optimization procedure was adopted for

    parameter evaluation. Firstly, the regression of

    these parameters was carried out by minimizing

    the sum of residual squares (SRS) between the

    experimental and calculated reaction rates. Then,

    numerical integration method was used for inte-

    grating the calculated reaction rates with previ-ously determined parameters to get the

    conversions. The calculated conversion values

    were then compared with experimental values

    through the mean relative deviation (MRD).

    SRS XN

    rexp rcalc2; 7

    MRD 1N

    XN

    Xcalc XexpXexp

    ! 100%: 8

    Table 2UNIFAC group identification of the components

    Molecule Group identification vij Volume parameter Rj Area parameter Qj

    Group name Main Secondary

    Ethyl Lactate CH3 1 1 1 0.9011 0.8480

    CH3OHCH 5 15 1 1.8780 1.660

    CH2COO 11 26 1 1.6764 1.420

    Water H2O 7 20 1 0.92 1.40

    Ethanol CH3CH2OH 5 17 1 2.1055 1.972

    Lactic acid COOH 17 40 1 1.3013 1.224

    CH3OHCH 5 15 1 1.8780 1.660

    Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114 105

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    6/14

    Regression results are summarized in Table 3,

    which included the values of the regression pa-

    rameters, SRS and MRD at 353 K. The differences

    of the two mechanisms were compared through

    the values of SRS and MRD. From the regression

    results, the rate equation for mechanism A de-

    scribed more accurately the experimental data

    than mechanism B. Therefore, the former was

    more reliable for this kind of esterification reac-

    tion. Similar mechanisms were applied in Sanz and

    Wei Songs work [9,17]. In order to verify this

    conclusion, FTIR analysis was used to test the

    component absorbing strength on the catalyst

    surfaces as discussed below.

    6.3. Desorption experiment and FTIR analysis

    In the model assumption section, through the

    comparison of two different mechanisms, it was

    concluded that ethanol and water absorbed

    stronger than other components in the reaction

    solution. In order to examine the previous results,

    FTIR analysis was introduced. The used resins

    (002) were treated at high temperature in order to

    determine the desorption information which was

    inverse to the reaction of adsorption. The resinsamples were withdrawn from the esterification

    reactor, and then treated at 100 C in an oven for

    different period of time (0, 0.5 and 8 h). In Fig. 1,

    the overall spectra of three samples are presented

    (wave numbers from 400 to 4000 cm1), in whichany component changes on the surfaces of the

    catalysts could be observed. There is a broad band

    in the range from 3200 to 3600 cm1, which is dueto the symmetric and asymmetric stretching of the

    OH groups [18]. Fig. 1 shows that the intensity of

    Table 3

    Values of parameters for the kinetic equations based on the mechanisms A and B

    Mechanism Parameters 002 NKC

    Value SRS MRD (%) Value SRS MRD (%)

    k (molg1 min1) 1.704 1.701A kW 4.973 1.83 109 4.27 5.073 2.10 109 4.16

    kE 2.093 2.245

    k (molg1 min1) 1.650 1.449B kW 5.016 5.88 109 8.61 5.510 9.06 _109 7.30

    kLA 1.554 2.010

    Fig. 1. Absorbance FTIR spectrum of 002 at different de-sorption times: 0 h (), 0.5 h (- - - - -), 8 h ( ), under 100 C,wave number region: 4004000 cm1.

    Fig. 2. Absorbance FTIR spectrum of 002 at different de-

    sorption times: 0 h (), 0.5 h (- - - - -), 8 h ( ), under 100 C,wave number region: 15001800 cm1.

    106 Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    7/14

    the band decreases with desorption time, which

    means that the extent of the desorption reaction

    took place on the catalyst surface can be measured

    by FTIR.

    Fig. 2 shows two important bands, which are

    located at 15001800 cm1. One is at around 1640cm1, and the other is located at 1740 cm1. Theformer band is attributed to the bending vibration

    of water and the latter to the stretching vibration of

    the C@O [18]. It can be seen that the absorbance

    of the band at 1640 cm1 decreases as a function ofdesorption time, which indicates that water held a

    large proportion of active sites on the catalyst

    surfaces. The intensity of the band at 1740 cm1 isrelatively much smaller in comparison with the

    band at 1640 cm1. However, the intensity of thestretching vibration of the C@O group is always

    very strong and sharp, which can be testified by the

    Fig. 3. Absorbance FTIR spectrum of lactic acid, wave number

    region: 4004000 cm1.

    Fig. 4. Absorbance FTIR spectrum of ethyl lactate, wave

    number region: 4004000 cm1.

    Fig. 6. Conversion versus time for the esterification with NKC

    at different temperatures, 333 K ., 343 K d, 353 K N,361 K j. The continuous lines represent the results of the LHmodel. All the reactions were carried out with an initial molar

    ratio 3:1 (ethanol:lactic acid). The catalyst loading was 4%

    (w/w) in all experiments.

    Fig. 5. Conversion versus time for the esterification with 002 atdifferent temperatures, 333 K ., 343 K d, 353 K N, 361K j. The continuous lines represent the results of the LHmodel. All the reactions were carried out with an initial molar

    ratio 3:1 (ethanol:lactic acid). The catalyst loading was 4%

    (w/w) in all experiments.

    Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114 107

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    8/14

    Table 5

    Regression results of NKC at different reaction temperatures

    Temperature (K) k (molg1 min1) kW kE SRS MRD (%)

    333 0.578 5.921 3.922 3.99 1010 3.94343 1.101 5.507 2.945 2.58 109 4.05361 2.505 5.501 2.902 3.60 10

    10

    2.33The results at 353 K are shown in Table 3.

    Table 4

    Regression results of 002 at different reaction temperatures

    Temperature (K) k (molg1 min1) kW kE SRS MRD (%)

    333 0.762 5.999 3.254 8.01 1010 4.58343 1.015 5.475 2.601 2.84 1010 4.79361 2.555 5.500 2.900 7.23 1010 2.68

    The results at 353 K are shown in Table 3.

    Fig. 7. (a) Arrhenius plot of esterification catalyzed with 002. The continuous line represents the result of linear regression. (b) vant

    Hoff plot of adsorption coefficient of water catalyzed with 002. The continuous line represents the result of linear regression. (c) vant

    Hoff plot of adsorption coefficient of ethanol catalyzed with 002. The continuous line represents the result of linear regression.

    108 Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    9/14

    FTIR spectra of lactic acid and ethyl lactate in

    Figs. 3 and 4. Thus it was concluded that lactic acid

    and ethyl lactate were not absorbed strongly in

    comparison with water on the resin surfaces.

    Through the FTIR experiments, it is proved

    that water absorbs much stronger on the resin

    surfaces than lactic acid and ethyl lactate do. Al-

    though the adsorption intensity of ethanol couldnot be confirmed in this way, mechanism B is

    Fig. 8. (a) Arrhenius plot of esterification catalyzed with NKC. The continuous line represents the result of linear regression. (b) vant

    Hoff plot of adsorption coefficient of water catalyzed with NKC. The continuous line represents the result of linear regression. (c) vant

    Hoff plot of adsorption coefficient of ethanol catalyzed with NKC. The continuous line represents the result of linear regression.

    Table 6

    Regression results ofDH for 002 and NKC

    002 NKC

    Water Ethanol Water Ethanol

    DH (kJ/mol) )6.42 )21.33 )7.47 )26.97

    Fig. 9. Conversion versus time for the esterification with 002.

    The catalyst loading was 2% (w/w) j, 4% N and 6% drespectively. All the reactions were carried out with an initial

    molar ratio 3:1 (ethanol:lactic acid), at temperature 343 K.

    Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114 109

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    10/14

    surely unreasonable since the adsorption of lactic

    acid is thought as one of the strongest.

    6.4. Esterification of lactic acid with ethanol

    6.4.1. Effect of reaction temperature

    The effect of temperature on the esterification

    reaction was studied over a temperature range

    from 333 to 361 K, under atmospheric pressure,

    with 002 and NKC as catalysts. The boiling point

    of this mixture is about 361 K. As mentioned

    above, the stirring speed was set to around 450

    rpm. So the influence of external and internal

    diffusion could be neglected under the given con-

    ditions. The experimental and regression resultsare displayed in Figs. 5 and 6. From the figures,

    the reaction rate increases with increasing tem-

    perature. The same trend was described in the re-

    action of lactic acid with methanol by Seo and

    Fig. 10. Conversion versus time for the esterification with fivedifferent cation-exchange resins. The symbols: 002 j, Am-berlyst-15 N, D001 d, D002 ., NKC r. All the re-actions were carried out with an initial molar ratio 3:1

    (ethanol:lactic acid), at temperature 353 K. The catalyst loading

    was 4% (w/w) in all experiments.

    Fig. 11. SEM photomicrographs of the surface of the resins: (a) fresh catalyst, (b) after esterification experiment and (c) reused in the

    catalytic distillation column. Catalyst: 002. Magnification: 150.

    110 Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    11/14

    Sanz et al. [6,9], and in the reaction of isopropanol

    with lactic acid [4]. The reaction constant, ad-

    sorption coefficients, SRS and MRD are provided

    in Tables 4 and 5. In both tables, the values of SRSwere smaller than 3.0 109 and MRD weresmaller than 5%. The adsorption coefficient of

    water is greater than that of ethanol, which sug-

    gests that water adsorbs more strongly than etha-

    nol on the active sites of the resins.

    6.4.2. Activation energy and adsorption coefficient

    Figs. 7 and 8 show the Arrhenius plots for the

    esterification reaction with 002 and NKC as cata-

    lysts at different temperatures. Arrhenius relation

    was introduced to describe the change of the re-

    action rate constant with temperature. The linear

    correlation coefficients were higher than 0.995 for

    both catalysts. From the slopes of the straight lines

    in Figs. 7(a) and 8(a), the activation energies can be

    calculated by Eq. (9), which are 51.58 kJ/mol for

    002 and 52.26 kJ/mol for NKC. The high values of

    the activation energy indicate that the reaction is

    kinetically controlled. From Figs. 7(b) and (c), it

    can be deduced that the variation of adsorptioncoefficient can be described by vant Hoff law (Eq.

    (10)) [8] in the temperature range from 333 to 353

    K, but there was a big deviation at 361 K (the

    boiling point of the reaction mixture). Since a lot of

    bubbles appeared in the liquid reaction mixture at

    the boiling point, the concentrations ai of com-ponents in the liquid phase, mainly ethanol and

    water, became lower because of their higher vola-

    tility. Meanwhile, as the surface reaction was the

    rate-determining step, the concentration of ab-

    sorbed components on catalyst surface was con-

    sidered unchanged. In Eq. (4), ciS and cS did notchange, while ai got smaller. So the regression value

    ofki was higher than that estimated by vant Hoff

    law. Figs. 7 and 8 show that the adsorption of

    water and ethanol on cation-exchange resin surface

    Fig. 12. SEM photomicrographs of the surface of the resins: (a) fresh catalyst, (b) after esterification experiment and (c) reused in the

    catalytic distillation column. Catalyst: 002. Magnification: 5000.

    Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114 111

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    12/14

    is an exothermic process, while the whole esterifi-

    cation reaction is an endothermic reaction. The

    regression results of DH for both catalysts were

    listed in Table 6.

    lnk lnk0 EART

    ; 9

    lnki lnk0i DHi

    RT: 10

    6.4.3. Effect of catalyst loading

    Fig. 9 shows the effect of catalyst loading on re-

    action rates. It was observed that the equilibrium

    constant nearly did not change with the increase of

    catalyst loading. The time required to reach theequilibrium was reduced as the catalyst loading in-

    creased. The reason was that the more catalysts loa-

    ded, the more active sites were available for reaction.

    6.4.4. Effect of catalyst type

    Different types of cation-exchange resins were

    used to assess their efficiency in the esterification

    reaction. Fig. 10 shows the plots of conversion oflactic acid against time for the various catalysts. It

    can be concluded that the catalytic activity

    increased in the order D002 < D001

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    13/14

    more than 40 h at around 363 K in the lactic acid

    ester catalytic distillation process [19].

    Figs. 11 and 12 show the SEM photographs of

    002 under magnification of 150 and 5000. In these

    figures, the surface of 002 was very smooth like

    most gel resins. Figs. 12(b) and (c) show that after

    being used in the esterification and catalytic dis-

    tillation experiments, there were few changes on

    the gel surfaces. Some small defects were observed

    on their surfaces, which were probably caused byimpurities in the column or by a small amount of

    poly-lactic acid formed in the experiments, just like

    what happened on other catalysts after being used

    for a long time.

    In the same way, SEM micrographs of NKC

    are shown in Figs. 13 and 14. The surface had

    obvious differences in comparison with 002. The

    pore diameters on the surfaces of fresh NKC

    ranged from 50 to 1000 nm. The surfaces of NKC

    had much more cracks and mass loss than 002

    after the same intensity of stirring according to

    Figs. 13(b) and 14(b). In Fig. 14(c), the pore sizes

    of NKC became larger than the fresh catalysts. It

    was due to the swelling effect of reaction liquid

    under high temperature in the catalytic distillation

    column. All these indicated that 002 had better

    mechanical strength than NKC.

    For industrial applications, mechanical strength

    is an important factor in evaluation of catalyst.

    Taking lactic acid ester production for example,cation-exchange resins were used in a CSTR or a

    catalytic distillation column. Under the stirring

    force, the resins may be destroyed if they do not

    have enough mechanical strength. And that would

    cause a lot of problems in further separation and

    purification processes. In a reactive distillation

    column, the destroyed resin fragments would plug

    the packing section and lead to much pressure

    drop in it. Taking this factor into account, 002 is a

    better choice for industrial applications.

    Fig. 14. SEM photomicrographs of the surface of the resins: (a) fresh catalyst, (b) after esterification experiment and (c) reused in the

    catalytic distillation column. Catalyst: NKC. Magnification: 5000.

    Y. Zhang et al. / Reactive & Functional Polymers 61 (2004) 101114 113

  • 7/28/2019 1-s2.0-S138151480400094X-main.pdf

    14/14

    7. Conclusions

    Esterification of lactic acid with ethanol over

    different kinds of cation-exchange resins was stud-ied in this paper. The order of catalytic activity was

    found to be: D002


Recommended