Sei sulla pagina 1di 11
Separation and Purification Technology 145 (2015) 39–49 Contents lists available at ScienceDirect Separation and

Contents lists available at ScienceDirect

Separation and Purification Technology

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

journal homepage: www.elsevier.com/locate/seppur Experimental and theoretical study on the adsorptive drying

Experimental and theoretical study on the adsorptive drying of primary alcohols in a fixed bed adsorber

Bernd Burrichter a , , Christoph Pasel a , Michael Luckas a , Dieter Bathen a , b

a Chair of Thermal Process Engineering, University of Duisburg-Essen, Lotharstraße 1, D-47057 Duisburg, Germany b Institute for Energy and Environmental Technology e.V. (IUTA), Bliersheimer Str. 60, D-47229 Duisburg, Germany

e.V. (IUTA), Bliersheimer Str. 60, D-47229 Duisburg, Germany article info Article history: Received 23 September 2014

article info

Article history:

Received 23 September 2014 Received in revised form 6 February 2015 Accepted 21 February 2015 Available online 6 March 2015

Keywords:

Adsorptive drying Adsorption isotherm Fixed bed adsorber Zeolite Modeling Alcohols

abstract

In this paper systematic investigations on the adsorptive drying of primary alcohols from methanol to 1-

hexanol on 3A and 4A zeolites are presented. The focus of the work is laid on the superdrying of these

polar solvents in the ppm w -region. In addition to equilibrium data gained in shaker bottle experiments information regarding the dynamic drying behavior was obtained by breakthrough experiments. The water content in these experiments was monitored online by a transmission photometer. Based on the experimental data a physical model was developed in order to predict the dynamic drying behavior in

a fixed bed adsorber. Most of the investigated systems could be well described by this model.

Furthermore pore diffusivities of water could be obtained by fitting the model to the experimental data. The results of this study reveal a clear dependency of the dynamic drying behavior of primary alcohols from the solvent to be dried and the type of zeolite. Small alcohol molecules show significantly faster dry- ing rates than long-chain alcohols and the drying kinetics on a 3A zeolite is much faster than on a 4A zeo- lite. The experimental results were correlated to the molecular dimensions of adsorptives and the size of the pore openings of the zeolites as well as to the interactions between adsorptive, solvent and adsorbent. 2015 Elsevier B.V. All rights reserved.

1. Introduction

Due to technological innovations and more challenging environmental restrictions there are increasing requirements on raw and operating materials in several industries. In this context the water content of organic solvents is a critical factor for many processes. In new generations of high-compressing engines already a few ppm of water lead to an inhomogeneous combustion and an inefficient operation of the engine. Another example of the dis- turbing influence of trace water is the poisoning of Grignard reagents or Ziegler–Natta catalysts in multiple chemical reactions [1] . An efficient way to remove water from liquids and gases to less than 1 ppm is a treatment with oxidic adsorbents like zeolites, sili- cas and aluminas. Especially drying on synthetic type A zeolites is often used in industrial applications. In contrast to its industrial spreading there are only few publications on the adsorptive drying of organic solvents and no systematic overview on dominant parameters exists [2] . All published work can mainly be separated in two categories. On the one hand there are a lot of investigations

Corresponding author. Tel.: +49 203 379 1731; fax: +49 203 379 3119. E-mail addresses: bernd.burrichter@uni-due.de (B. Burrichter), christoph.pasel@ uni-due.de (C. Pasel), michael.luckas@uni-due.de (M. Luckas), dieter.bathen@ uni-due.de (D. Bathen).

1383-5866/ 2015 Elsevier B.V. All rights reserved.

on the adsorptive drying of non-polar solvents, like toluene or ben- zene, in the lower ppm-region [3–7] . These solvents are easy to dry due to their non-polar character and their poor solubility for water. On the other hand many publications deal with the drying of polar solvents, like alcohols, in the percent-region [8–17]. But up to now there is almost no work done on the adsorptive drying of polar organic solvents in the trace humidity region, although superdry- ing is one of the key strengths of adsorptive drying on zeolites. Only Goto et al. published adsorption isotherms of ethanol on a 4A zeolite with water concentrations from 0 to 3000 ppm w [18] . This lack of investigations might be a consequence of the high experimental effort, which is necessary to do reliable measure- ments with highly polar solvents in the trace humidity region. For a more detailed literature review it is referred to an earlier pub- lication of the authors [19] . Therefore systematic investigations on the drying of polar organic solvents on a 3A and 4A zeolite were conducted. In a first step adsorption equilibria for the drying of several primary and secondary alcohols as well as acetates were determined in shaker bottle experiments [20] . Afterwards a laboratory plant for continu- ous breakthrough experiments was build up in order to study the kinetic behavior of the drying process in a fixed bed. First experi- ments to determine optimal operation parameters and to identify dominant parameters of influence were carried out with the model

40

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

Nomenclature b R film c i R pore c i; 0 parameter of the Langmuir
Nomenclature
b
R film
c i
R pore
c i; 0
parameter of the Langmuir equation
concentration (mg H2O /kg solution )
inlet concentration (mg H2O /kg solution )
equilibrium concentration (mg H2O /kg solution )
molecular diffusion coefficient (m 2 /s)
axial dispersion coefficient (m 2 /s)
bead size of the particles (m)
diffusion coefficient in the pores (m 2 /s)
mass transfer resistance in the film (s)
mass transfer resistance in the particle (s)
t time (s)
u
superficial velocity (m/s)
c i; eq
LR
_
D 12
V solu
D AX
x H 2 O
d p
volume flow of solution (m 3 /s)
mass fraction of water in fluid phase
mass fraction of water in gas phase
y H 2 O
z axial dimension (m)
D Pore
k
k f
m
ADS
m
overall mass transfer coefficient (s 1 )
film mass transfer coefficient (m/s)
mass of adsorbent (kg)
mass of solution (kg)
Greek symbols
c
1
H 2 O
solu
L
p
pressure (bar)
saturation vapor pressure of water (bar)
activity coefficient of water in solvent
void fraction of the bed
void fraction of the particle
P
q
p s ; H 2 O
b bulk density of the adsorbent (kg/m 3 )
Ads;
q
loading of the adsorbent (g H2O /g adsorbent )
equilibrium loading of the adsorbent (g H2O /g adsorbent )
monomolecular loading (g H2O /g adsorbent )
w skeletal density of the adsorbent (kg/m 3 )
q Ads;
q
eq
density of the solvent (kg/m 3 )
q fluid
q
mon

substance isopropanol [19] . In this paper the results of a complete experimental series on the adsorptive drying of primary alcohols from methanol to 1-hexanol on 3A and 4A zeolites in a fixed bed are presented.

2. Experimental

2.1. Materials and methods

In this study 3A and 4A zeolites supplied by Zeochem AG ( Uetikon , Switzerland ) were used. The material was provided as spherical pellets with a bead size of 1.0–2.0 mm. More physical properties are presented in Table 1 . All alcohols used were of reagent grade and the water, which was given to the storage container to modify the water concentra- tion of the solvent, was ultra-purified. In Table 2 basic physical properties of alcohols and the adsorptive water are presented. The relative permittivity is used as a measure for the polarity of

Table 1 Basic physical properties of the adsorbents.

 

3A

4A

Commercial name Pore opening (Å) Cation Ionic radius d p (mm) q Ads ; b (kg/m 3 ) a

L (–) * P (–)

Zeochem Z3-01

Zeochem Z4-01

3.2

3.8

K +

Na +

1.38

1.02

1.0–2.0

1.0–2.0

0.70

0.69

0.4

0.4

0.55

0.55

a Data refers to an adsorber column with an inner diameter of 16 mm.

Table 2 Basic physical properties of water and alcohols.

the alcohol and therefore for the interactions of the alcohol mole- cules with the polar type A zeolites. Activity coefficients of water in the alcohols were determined using the UNIFAC activity coefficient model. Molecular dimensions of the alcohols were estimated by applying an approach of Gounaris and Helmich [21–23]. The most likely coordinates of all atomic nuclei of a molecule were deter- mined by an energy optimization of the molecule configuration. This optimization was done by quantum mechanical calculations using the MM2 method [24] . The coordinates of the molecule were rotated around its three axes, until the smallest projection is found. The diameter of this projection is the molecular diameter which is relevant for the crossing of the pore openings of a zeolite. The results of these calculations are in good accordance with other published data [25–29]. The water concentration of the alcohols was measured by a coulometric Karl-Fischer-Titrator ( Schott Instruments ) in shaker bottle experiments and by a transmission photometer ( Pier Electronic) in dynamic experiments. Additionally the titration was used as reference method for calibration of the transmission pho- tometer. The accuracy of the Karl-Fischer-Titrator was regularly checked with certified water-standards supplied by Hydranal .

2.2. Equilibrium experiments

Equilibrium data for the adsorptive drying of organic solvents were gained in static shaker bottle experiments. A defined amount of solvent with a known quantity of water and a defined mass of adsorbent were given to a flask and shaken under constant condi- tions (25 C and 150 rph) for up to 7 days until adsorption equilib- rium was reached. Afterwards the water content of the solvent was measured again and the loading of the adsorbent was calculated by following mass balance:

Name

Molecular formula

Relative permittivity 20 C (–) [30]

Activity coefficient (–)

Molecular diameter (Å)

Water

H 2 O CH 4 O

 

80.1

1

3.0

Methanol

33.0

1.761

4.3

Ethanol

C 2 H 6 O C 3 H 8 O

25.3

2.660

4.7

1-Propanol

20.8

3.141

5.2

1-Butanol

C 4 H C 5 H C 6 H

10

O

17.8

3.564

5.2

1-Pentanol

12

O

15.1 (25 C)

3.934

5.2

1-Hexanol

14

O

13.0

4.256

5.2

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

41

q

¼ m solu ð c i; 0 c i; eq Þ 10 6

m ADS

ð

1Þ

During these experiments all work was carried out in a glove- box purged with dry nitrogen to prevent contamination of the samples by ambient humidity. For an adsorption isotherm approx. 15–20 equilibrium points were determined for each adsorbent/sol- vent combination. The Langmuir equation was fitted to experimen- tal data by a non-linear regression [31] .

q ¼ q mon

b c i c i 1 þ b

ð 2Þ

The gas phase concentrations of water were calculated from the liquid phase concentrations by extended Raoult’s law (see Eq. (3)). By this means a comparison of the drying isotherms of alcohols with the corresponding water vapor isotherm is enabled. The water vapor isotherms represent the maximum adsorption capac- ity of an adsorbent for the adsorptive water under optimal condi- tions with no competitive adsorptive molecules. So by the comparison the extent of unused drying capacity can be quantified and the efficiency of drying can be assessed.

rel :Humidity ¼ y H

2

O p

p s;H 2 O

¼ x H 2 O c

1

H

2 O

ð 3Þ

A comprehensive overview of the experimental procedures for the uptake of equilibrium data is given in earlier papers [19,20,32].

2.3. Kinetic experiments

Information regarding the kinetic behavior of the adsorptive drying of alcohols has been obtained from dynamic experiments in a fixed bed adsorber. Solvent with a known water concentration is pumped through a stainless steel adsorber (inner diameter:

0.016 m/maximum height: 0.25 m) filled with zeolite with a vol- ume flow of 0.16 l/h. After leaving the adsorber the fluid enters the measuring cell of the transmission photometer, where the water concentration of the effluent is continuously monitored. Samples for the Karl-Fischer-Titration can be drawn by a syringe. All experiments were carried out at a temperature of 25 C and ambient pressure. A process flow chart and a photo of the experi- mental plant are shown in Fig. 1 . Beside kinetic information equilibrium data can be gained from dynamic experiments as well. The mass of adsorbed water during a

breakthrough experiment can be calculated by integrating the dif-

ference between the inlet and outlet concentration of the adsorp- tive water as a function of time (see Eq. (4) ).

q ¼

V solu q fluid 10 6

_

m

ADS

t

Z

t

0

eq

ðc i; 0 c i Þdt

ð 4Þ

A comparison of equilibrium points obtained from shaker bottle and dynamic experiments shows very good accordance between both methods [19] . However, the shaker bottle method enables the determination of equilibrium data with considerably less experimental effort. Therefore equilibrium data were mainly deter- mined in shaker bottle experiments.

2.4. Experimental errors

The experimental error of the equilibrium data from shaker bot- tle experiments (Eq. (1) ) was calculated by Gaussian error prop- agation. The relative standard deviation for the loading of the adsorbent is between 3% and 6%. It mainly depends on the differ- ence between initial and equilibrium concentration of the adsorp- tive water.

The error of the concentration measurements by the Karl- Fischer-Titrator results from the concentration tolerance of the water calibration standard which is given as 3%. The accuracy of the experimental breakthrough curves is defined by the accuracy of the photometer. Since the transmission photometer is calibrated against the titrator, the relative error of the photometer is about 3% as well.

3. Physical model

In addition to the experimental work a physical model was developed to simulate the dynamic behavior of the adsorptive dry- ing in the fixed bed adsorber. A Dispersed Plug Flow (DPF) model based on the mass balances in the solid and fluid phase was used. The corresponding energy balances were neglected due to the insignificance of the heat of adsorption when adsorbing traces of water from a liquid. The mass transfer through the surrounding film of the adsorbent particle and in the pores of the zeolite was simplified by using a Linear Driving Force (LDF) approach. Additional assumptions were a homogeneous liquid and solid phase, no radial gradients inside the column and the absence of

no radial gradients inside the column and the absence of Fig. 1. Flow chart (left) and

Fig. 1. Flow chart (left) and photograph (right) of the laboratory plant.

42

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

competitive adsorptives. Considering all these simplifications the mass balances in the liquid and solid phase can be expressed as fol- lows [33,34]:

Liquid phase : @ c i ¼ D AX @ 2 c i

@

t

u LR

@ z 2 ð L þ ð 1 L Þ P Þ

@ c i

@ z

ð q eq qÞ

k ð 1 L Þ ð 1 P Þ q Ads ; w 10 6

ð L þ ð 1 L Þ P Þ q fluid

Solid phase : @ q ¼ k ð q eq qÞ

@

t

ð 5Þ

ð

6Þ

The overall mass transfer coefficient k of the LDF approach can be written in the form of a series connection of the resistances in the surrounding film and in the pores of the zeolite, the latter fol- lowing the approach of Glueckauf [35] .

1 k ¼ d p 6k f

þ

2

p

d

60 p

D Pore

ð 7Þ

The physical parameters used in this model, the axial dispersion coefficient D AX [36] , the film mass transfer coefficient k f [37] and the molecular diffusion coefficient of water in the alcohol D 12 [38] , which is necessary to calculate k f , were determined mathe- matically by well-known correlations. In case of the diffusion coefficient of the water in the pores of the zeolite D Pore there is no reliable correlation. Therefore the diffusivity of water in the pores is used to fit the simulations to the experimental results. For more detailed information regarding the simplifying assumptions, the calculation methods for the parameters and the initial and boundary conditions it is referred to [19] .

4. Results and discussion

4.1. Equilibrium data

Adsorption equilibria of the drying of primary alcohols from methanol to 1-hexanol on a 3A zeolite are displayed in Fig. 2 . Additionally the water vapor isotherm on a 3A zeolite is shown. Experimental data are presented as filled symbols while the fitted Langmuir isotherms are shown as lines.

All isotherms are measured in the concentration range from 0 to 2200 ppm w in the liquid phase. Due to the conversion of the water concentration to relative humidity (Eq. (3) ) with different activity coefficients (see Table 2 ) the concentration range covered on the

abscissa is different for different alcohols. For this reason in the case of methanol drying all equilibrium points are between a rela- tive humidity of 0–0.7%, while the experimental data for the drying of 1-hexanol go up to a relative humidity of 4%. However a conver- sion of the water concentration only stretches the isotherms and has no impact on the relative position of the isotherms and the gap to the water vapor isotherm. Equilibrium data for alcohol dry- ing on 4A zeolites are displayed in the same way in Fig. 3 . Langmuir parameters and correlation coefficients of the fitting are summar- ized in Table 5 . All adsorption isotherms of water from primary alcohols show almost rectangular shapes. The plateau of the isotherm is already reached at relative humidities around 0.5%. But with respect to the maximum drying capacity there are significant differences between the six primary alcohols. Whilst the three long-chain alcohols 1-butanol, 1-pentanol and 1-hexanol show very high dry- ing capacities of approx. 15–16 wt.%, the three short-chain alcohols exhibit a clearly reduced drying capacity on the 3A zeolite. The comparison to the water vapor isotherm illustrates that 1-hexanol and 1-pentanol show no co-adsorption on a 3A zeolite. The drying capacities of these two alcohols almost reach the level of the water vapor isotherm. The uptake capacity for water from 1-butanol is slightly reduced. In contrast to the long-chain alcohols the three short-chain alcohols show a significant gap to the water vapor iso- therm. This clearly proves that primary alcohols up to a chain length of at least three carbon atoms are able to enter the microp- ores of a 3A zeolite and to co-adsorb. This finding contradicts the published and calculated sizes of alcohol molecules and pore open- ings of the zeolites [25–29] . Those data seem to indicate that alco- hol molecules are substantially bigger than the pore opening of a 3A zeolite and therefore should be excluded from the micropores of the zeolite. From the experiments it is evident that data for molecule sizes and pore openings cannot be used to precisely pre- dict co-adsorption in type A zeolites. In case of a 4A zeolite the drying isotherms of the four primary alcohols from 1-propanol to 1-hexanol show very similar shapes. The slope at low water concentrations is slightly smaller than on the 3A zeolite and maximum capacities of approx. 17 wt.% are reached. In contrast, ethanol and methanol drying exhibit flatter

20 3A 16 12 8 4 0 loading of the adsorbent [g/100g]
20
3A
16
12
8
4
0
loading of the adsorbent [g/100g]

water vapor

1-pentanol

1-hexanol

1-butanol

1-propanol

ethanol

methanol

01234

relative humidity [%]

Fig. 2. Adsorption isotherms of water from different alcohols on a 3A zeolite.

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

43

20 4A 16 12 8 4 0 loading of the adsorbent [g/100g]
20
4A
16
12
8
4
0
loading of the adsorbent [g/100g]

water vapor

1-butanol

1-pentanol

1-hexanol

1-propanol

ethanol

methanol

01234

relative humidity [%]

Fig. 3. Adsorption isotherms of water from different alcohols on a 4A zeolite.

isotherms with a significantly lower uptake capacity for the adsorptive water. Especially methanol drying on a 4A zeolite yields very poor results. Even with large amounts of adsorbent water con- tents below 350 ppm w cannot be achieved. Due to this behavior a description of the drying equilibrium of methanol by a Langmuir- type isotherm is not possible, wherefore no fit can be seen in Fig. 3 . Comparing the drying isotherms of 1-propanol on 3A and 4A zeolites reveals the influence of molecular size and polarity of the solvent molecules on the adsorption equilibrium. The isotherm of water from 1-propanol on a 3A zeolite clearly proves the accessibility of the micropores of type A zeolites for 1-propanol molecules. However, since no significant co-adsorption of 1-propa- nol on the 4A zeolite can be found, the polarity of the solvent must have a dominant influence on the adsorption equilibrium. The absence of co-adsorption in case of the 4A zeolite is due to stronger interactions of the polar water molecule with the sites of the more polar 4A zeolite compared to the 3A zeolite. Sodium ions in the cages of the 4A zeolite have a higher charge density than potas- sium ions in the 3A zeolite due to a smaller ionic radius (see Table 1 ). This change seems to favor interactions between water and zeolite to a higher degree than interactions between alcohol and zeolite and prevent co-adsorption of the alcohol 1-propanol. Another interpretation could be that there are stronger interac- tions between 1-propanol and 3A zeolite than between 1-propanol and 4A zeolite. It cannot be clearly decided at this moment which argument is more relevant. In any case we conclude that molecular size determines whether co-adsorption is possible or not, while interactions between adsorbent and adsorptive/solvent specify the extent of co-adsorption. With regard to the initial slope water adsorption isotherms from alcohols on 3A and 4A zeolite exhibit a complex behavior. At low concentrations the 3A isotherm is steeper while at higher concentration the 4A isotherm shows a higher plateau. There are several theories for this behavior which cannot be discussed in every detail in this place. The most convincing argument for the steep initial slope on 3A zeolite may be the smaller window aper- ture decelerating the rate determining step of alcohol adsorption [20] . Another valid argument to explain the steeper initial slope may be multi cation–guest interactions between the larger potas- sium ion and water. At higher water concentration water displaces alcohol on the adsorbent which is favored on 4A zeolite compared

to 3A zeolite due to a strong interaction between sodium and water.

4.2. Adsorptive drying of primary alcohols in a fixed bed adsorber

4.2.1. Drying on a 3A zeolite Optimal operation conditions were determined in a parameter study on the adsorptive drying of isopropanol [19] . According to the outcomes of this study a superficial velocity of 0.8 m/h and a bed height of 12.4 cm were chosen for the 3A zeolite. These set- tings result in retention times of the solvent in the fixed bed of approx. 3.5 min. The operating conditions of all dynamic runs are listed in Table 6 .

(a) Concentration dependency

All primary alcohols were dried with varying inlet concentra- tion of water. Exemplary the drying of the short-chain alcohol ethanol and the long-chain alcohol 1-hexanol are shown in the fol- lowing figures. Fig. 4 displays the drying of ethanol and Fig. 5 depicts the drying of 1-hexanol. Experimentally determined

2000 experiment 2000ppmw 3A simulation 2000ppmw experiment 1500ppmw simulation 1500ppmw 1500 experiment 1000ppmw
2000
experiment 2000ppmw
3A
simulation 2000ppmw
experiment 1500ppmw
simulation 1500ppmw
1500
experiment 1000ppmw
simulation 1000ppmw
experiment 500ppmw
simulation 500ppmw
1000
500
0
0
500
1000
1500
2000
time [min]
water concentration [ppm w ]

Fig. 4. BTCs of water from ethanol on a 3A zeolite with varying inlet concentration.

44

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

2000 experiment 2000ppmw 3A simulation 2000ppmw experiment 1500ppmw simulation 1500ppmw 1500 experiment 1000ppmw
2000
experiment 2000ppmw
3A
simulation 2000ppmw
experiment 1500ppmw
simulation 1500ppmw
1500
experiment 1000ppmw
simulation 1000ppmw
experiment 500ppmw
simulation 500ppmw
1000
500
0
0
500
1000
1500
2000
2500
3000
3500
4000
4500
time [min]
water concentration [ppm w ]

Fig. 5. BTCs of water from 1-hexanol on a 3A zeolite with varying inlet concentration.

breakthrough curves (BTC) are shown by symbols, while the lines represent the results of the simulation. Drying of both alcohols reveal longer mass transfer zones (MTZ) and therefore steeper BTCs for higher water concentrations. Similar results can be found for the drying of the other four primary alco- hols. This proves that the dynamics of the adsorptive drying of alcohols in the trace humidity region depends on the adsorptive concentration. This finding can be quantified by the results of the simulation. Diffusivities of water in the pores of the zeolite D Pore increase with growing inlet concentration of the adsorptive (see Table 6 ). In addition pore diffusivities are in the area of 10 12 m 2 /s. These small values indicate surface diffusion as the dominant mass transfer mechanism [19] . Similar results have already been published by Yamamoto et al. and Goto et al. [17,18]. There are different reasons for this behavior. Firstly the adsorp- tion equilibrium in case of a concave isotherm and an increasing concentration gradient has a steepening effect on the BTC. Furthermore the adsorption enthalpy decreases at high surface loadings [33,39] which has a positive impact on the mobility of the adsorptive molecules. At higher water concentrations high sur- face loadings are reached earlier. Therefore the enhanced mobility of water molecules leads to increasing diffusivities at higher inlet concentrations of the adsorptive. Although this behavior has not yet been seen for the adsorptive drying of organic solvents on zeo- lites, Kümmel and Worch have already made similar observations for the adsorption on mesoporous systems [39] . In addition a quantitative analysis of the transport resistances

d

in the surrounding film of the particle R film ¼ k f and in the inner

p

6

D Pore has been executed in order to demon-

of the zeolite R pore ¼

strate the dominating influence of the mass transfer resistance in the pores of the zeolite (see Table 3 ). Assuming an average particle diameter of 0.015 m and a super- ficial velocity of 0.8 m/h, which result in Reynolds numbers between 0.2 and 2, the mass transfer resistance in the particle is at least 100 times higher than in the surrounding film. This clearly

2

d

p

60 p

Table 3 Comparison of mass transfer resistances in and around the particle.

Name

R film (s)

R pore (s)

R pore = R film (–)

Methanol

25

2679

108

Ethanol

39

5952

153

1-Propanol

57

12,755

224

1-Butanol

67

13,393

200

1-Pentanol

84

17,857

213

1-Hexanol

97

20,604

212

reveals that film diffusion has no relevant impact on the overall mass transfer coefficient in our liquid adsorption system. Similar results have already been obtained in previous works [6,17,40] .

(b) Solvent dependency

The BTCs of water from all six primary alcohols with inlet con- centrations of 2000 ppm w water are displayed in the following dia- grams. In Fig. 6 the experimental data of the dynamic drying of methanol, ethanol and 1-propanol as well as the results of the theoretical modeling are presented. The drying of the three long- chain alcohols in a fixed bed adsorber is shown in Fig. 7 . Drying of short-chain primary alcohols on a 3A zeolite reveals very steep BTCs with low uptake capacities for the adsorptive water. The 12.4 cm long fixed bed enables complete drying of the alcohols just for a few minutes before the water content of the effluent rises. In addition the slope of the BTCs shows a clear dependency from the solvent to be dried. The length of the MTZ constantly increases with growing chain length of the primary alcohol. This proves that the drying kinetics decelerates with grow- ing molecular dimension of the primary alcohol to be dried. This finding can be underpinned by pore diffusivities of water in the pores of the zeolite as well. While water in methanol shows a pore diffusivity of 2.0 10 11 m 2 /s diffusivities in ethanol (9.0 10 12 m 2 /s) and 1-propanol (4.2 10 12 m 2 /s) decrease con- stantly (see Table 1 ).

2000 1500 3A 1000 experiment methanol simulation methanol experiment ethanol 500 simulation ethanol experiment
2000
1500
3A
1000
experiment methanol
simulation methanol
experiment ethanol
500
simulation ethanol
experiment 1-propanol
simulation 1-propanol
0
0
500
1000
1500
2000
time [min]
water concentration [ppm w ]

Fig. 6. Drying of short-chain primary alcohols on a 3A zeolite.

2000 1500 3A 1000 experiment 1-butanol simulation 1-butanol 500 experiment 1-pentanol simulation 1-pentanol
2000
1500
3A
1000
experiment 1-butanol
simulation 1-butanol
500
experiment 1-pentanol
simulation 1-pentanol
experiment 1-hexanol
simulation 1-hexanol
0
0
500
1000
1500
2000
time [min]
water concentration [ppm w ]

Fig. 7. Drying of long-chain primary alcohols on a 3A zeolite.

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

45

Simulations of the BTCs of water from methanol and ethanol are in good accordance with the experimental results. In contrast there are huge differences between experiment and simulation in case of 1-propanol drying. These deviations are a consequence of a signifi- cant overshoot of the water concentration at the end of the adsor- ber. After approx. 1000 min the water concentration of the effluent exceeds the inlet concentration by about 120 ppm w due to a subse- quent displacement of adsorbed water through alcohol molecules. Since the theoretical model assumes a single component adsorp- tion with no competing adsorptives a sufficient modeling of this effect is not possible. The subsequent displacement of water will be discussed more detailed later on. The BTCs of water from the three long-chain primary alcohols on a 3A zeolite exhibit a similar pattern. The drying capacity is nearly equal in all three cases and much higher than for short- chain primary alcohols. This finding is in line with the results of the shaker bottle experiments. In the first 4 h the effluent concen- tration remains almost constant at approx. 50 ppm w . Humidities below 50 ppm w cannot be achieved due to marginal short circuit flows in the small fixed bed with a length of 12.4 cm. The incom- plete drying at the beginning of the experiment is also the reason for small deviations from the simulation, which predicts a com- plete drying for the first liquid passing the adsorber. Regarding the kinetic behavior there are only small differences between the three long-chain alcohols. However, the trend toward faster kinetics in case of smaller solvents molecules can be con- firmed by the BTCs. The MTZ of 1-butanol drying is slightly shorter than in case of 1-pentanol, which is again steeper than the MTZ of water from 1-hexanol. Pore diffusivities obtained from the sim- ulation confirm this finding (see Table 6 ), although the dependency is much weaker than in case of the three short-chain primary alcohols. Although there are deviations between simulations and experi- mental results in some cases the simulations still reflect the gen- eral shape of the BTCs satisfactory. Therefore the comparison of different simulations and pore diffusivities is considered to be jus- tified. In Fig. 8 the modeled BTCs of all six primary alcohols on a 3A zeolite are shown in one diagram. By this figure the dependency of the dynamic drying behavior from the solvent becomes obvious. Besides the uptake capacity for water the length of the MTZ, which is characteristic for the kinetics of the drying, is clearly depending on the solvent. With every additional methyl group of the primary alcohol the BTC of water from the particular alcohol becomes shal- lower, what indicates a slower kinetics of the drying process.

Since no other parameter has been changed in these experi- ments, the reason for the decreased kinetics of alcohol drying must be a hindrance of water diffusion by solvent molecules blocking the narrow pores of the 3A zeolite. From equilibrium data it can be seen that small alcohols can pass the pore openings of the 3A zeolite. With increasing molecular size the probability increases that a solvent molecule blocks a pore opening. This reduces the velocity of water diffusion considerably. Both methanol and etha- nol are small enough to move almost freely through the narrow pores of a 3A zeolite. Therefore they do not hinder water diffusion markedly. This is the reason for significantly higher pore diffusivi- ties of water in case of methanol and ethanol drying. Long-chain primary alcohols, like 1-pentanol or 1-hexanol, show almost no co-adsorption on a 3A zeolite. This leads to the assumption, that water removal from these solvents should be much faster due to the absence of alcohol molecules in the microp- ores of the zeolite. However, water diffusion in case of drying these non-adsorbing alcohols is further decelerated. This finding sug- gests that already a hindrance of water diffusion through compet- ing alcohol molecules at the entrance of the microporous system of the type A zeolites is responsible for the reduced kinetics of the process. In order to learn more about the behavior of solvent molecules in the inner area of type A zeolites gravimetric uptake measure- ments of pure ethanol and 1-hexanol vapors on 3A and 4A zeolites were executed. Ethanol and 1-hexanol were chosen as representa- tive of short-chain and long-chain alcohols, respectively. Results of these experiments are shown in Fig. 9 . Sorption isotherms of ethanol exhibit a sharp increase at low partial pressures and a relative flat plateau at 9 resp. 11 wt.% for 4A and 3A zeolites. High uptake capacities at low partial pressures clearly prove a filling of the micropores of the type A zeolites. In contrast to ethanol 1-hexanol shows very poor adsorption rates at low relative saturations. At relative saturations between 0% and 55% only 1–2 g alcohol adsorb per 100 g of zeolite. The uptake of 1-hexanol rises rapidly at a relative saturation of 60%, what indi- cates capillary condensation of 1-hexanol in the meso- and macropores of the zeolites. This is clear evidence that at higher alcohol concentrations at least the supply pores of the zeolite and the entrances to the microporous part of the zeolite are filled with solvent molecules, even in case of big, non-adsorbing sol- vents. The interpretation of the small amount of adsorbed alcohol at low concentrations is difficult. On the one hand the small alcohol uptake could show few molecules which were able to enter the

2000 3A 1500 methanol ethanol 1000 1-propanol 1-butanol 500 1-pentanol 1-hexanol 0 0 500 1000
2000
3A
1500
methanol
ethanol
1000
1-propanol
1-butanol
500
1-pentanol
1-hexanol
0
0
500
1000
1500
2000
time [min]
water concentration [ppm w ]

Fig. 8. Comparison of the adsorptive drying of all six primary alcohols on a 3A zeolite.

16 12 8 ethanol 3A 4 ethanol 4A 1-hexanol 3A 1-hexanol 4A 0 0 10
16
12
8
ethanol 3A
4
ethanol 4A
1-hexanol 3A
1-hexanol 4A
0
0
10
20
30
40
50
60
70
80
90
100
loading of the adsorbent [g/100g]

relative saturation [%]

Fig. 9. Adsorption isotherms of pure ethanol and 1-hexanol vapors on 3A and 4A zeolites.

46

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

entrances of the micropores of the zeolite. Another explanation could be a monolayer in the mesopores of the zeolite. For ethanol as well as for 1-hexanol uptake capacities on the 3A zeolite are approx. 20% higher than on the 4A zeolite. This finding is surprising, because both zeolites consist of the same zeolite structure and just the monovalent cations are changed from sodium (4A) to potassium (3A). Therefore the 4A zeolite should exhibit a slightly higher uptake capacity due to the higher charge density of the sodium ions (see the water vapor isotherms). This issue could not be properly answered within this work. However, experimental artifacts can be excluded due to reproduction mea- surements. Knowing the pore size distribution would be helpful to better understand these findings. However volumetric measure- ments of nitrogen sorption in a BET apparatus to calculate a pore size distribution of 3A and 4A zeolites are a challenging problem because of very slow kinetics in the microporous system at 77 K. Nevertheless the results of the alcohol sorption experiments underline the explanation for the slower drying kinetics of primary alcohols with increasing chain length. Alcohol molecules blocking the narrow pores of the zeolite may hinder water diffusion, leading to smaller pore diffusivities of the water and slower drying rates. In addition the drying of long-chain, non-adsorbing alcohols indicate that already a hindrance of water diffusion through alcohol mole- cules at the entrances to the micropores of the zeolite may be deci- sive for the velocity of the whole drying process.

(c) Displacement of adsorbed water through alcohol molecules

Subsequent displacement of adsorbed water molecules through alcohol molecules in case of 1-propanol drying is at first unex- pected due to the much higher polarity of the water compared to 1-propanol (see relative permittivity in Table 2 ). However, since there are only traces of water in the alcohol, 1-propanol is able to occupy approx. half of the uptake capacity of the zeolite in state of equilibrium (see Fig. 2 ). Furthermore it has to be considered that in case of contact times of 3.5 min in a 12.4 cm long fixed bed adsorber water breakthrough curves at the end of the adsorber dis- play just a temporary equilibrium. Equilibrium state in the fixed bed is not attained until the inlet and outlet concentrations of the adsorptive water are equal for a longer period of time. If contact times in the fixed bed are very short, diffusion veloci- ties of the participating substances will be crucial for the tempor- ary equilibrium. In contrast to small and mobile water molecules long-chain alcohol molecules are more bulky. Hence they show significantly slower diffusion rates in the narrow pores of type A zeolites. As a consequence 1-propanol molecules need much more time to reach the adsorption sites than fast diffusing water mole- cules and all adsorption sites are captured by water molecules at first. In this temporary equilibrium much more water is adsorbed

than in the final state of equilibrium. Due to the absence of the slowly diffusing 1-propanol molecules the temporary equilibrium is very close to the water vapor isotherm (see Fig. 10 ). After a cer- tain time 1-propanol molecules reach the adsorption sites as well and replace water molecules according to the final state of equi- librium. This leads to the characteristic overshot of the water con- centration in the effluent in case of 1-propanol drying. Bigger alcohol molecules, e.g. 1-butanol, diffuse even more slowly in the microporous part of the zeolite than 1-propanol. But in contrast to 1-propanol, these alcohols are not able to co-ad- sorb on the zeolite in a noticeable amount. They are not polar enough to displace water molecules from their adsorption sites even though their concentration is much higher than the water concentration. Therefore the temporary equilibrium in the fixed bed is next to the final state of equilibrium and no subsequent dis- placement occurs. Smaller alcohol molecules, like ethanol, are polar enough to occupy a significant amount of the adsorption capacity. However, these molecules are sufficiently small and mobile to diffuse almost as fast as water and reach the adsorption sites at the same time as water molecules. For this reason even in fixed beds with very short contact times a temporary equilibrium is achieved which is similar to the final equilibrium. Subsequent displacement of adsorbed water molecules through 1-propanol molecules is enabled by a large excess of solvent mole- cules, a relatively high polarity of the solvent and a significantly slower diffusion of the solvent compared to the water in the nar- row pores of the zeolite.

4.2.2. Drying on a 4A zeolite Results for the dynamic drying of primary alcohols on a 4A zeo- lite are displayed in the following diagrams. Fig. 11 shows experi- mental results and simulations for the drying of the three short- chain alcohols. The drying of 1-butanol, 1-pentanol and 1-hexanol is presented in Fig. 12 . The dynamic runs were carried out in the same experimental plant as in case of 3A drying. Compared to 3A zeolite measuring of complete BTCs required a longer bed length of 23.7 cm. In contrast to the drying on a 3A zeolite a subsequent displace- ment of adsorbed water by ethanol can be seen. Displacement of water by 1-propanol on a 4A zeolite is not possible because 1-pro- panol shows no co-adsorption on the 4A zeolite and hence no tem- porary equilibrium occurs (see Fig. 3 ). For this reason a description of the BTC of water from ethanol by the physical model is not suc- cessful, while drying of methanol and 1-propanol is in good agree- ment with the model. Displacement of water by ethanol molecules seems to be a con- sequence of significantly decelerated ethanol diffusion in compar- ison to the 3A zeolite. In addition water diffusion must be less

to the 3A zeolite. In addition water diffusion must be less Fig. 10. Displacement of adsorbed

Fig. 10. Displacement of adsorbed water by alcohol molecules in case of a 3A zeolite.

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

47

2000 1500 4A 1000 experiment methanol simulation methanol 500 experiment ethanol simulation ethanol experiment
2000
1500
4A
1000
experiment methanol
simulation methanol
500
experiment ethanol
simulation ethanol
experiment 1-propanol
simulation 1-propanol
0
0
500
1000
1500
2000
2500
3000
3500
time [min]
water concentration [ppm w ]

Fig. 11. Drying of the short-chain primary alcohols on a 4A zeolite.

2000 1500 4A methanol ethanol 1000 1-propanol 1-butanol 500 1-pentanol 1-hexanol 0 0 500 1000
2000
1500
4A
methanol
ethanol
1000
1-propanol
1-butanol
500
1-pentanol
1-hexanol
0
0
500
1000
1500
2000
2500
3000
3500
time [min]
water concentration [ppm w ]

Fig. 13. Comparison of the adsorptive drying of all six primary alcohols on a 4A zeolite.

decelerated compared to the diffusion of ethanol. The variation of the pore opening between 3A and 4A zeolite cannot be the reason for the slower ethanol diffusion, because pore openings become larger. Therefore changed interactions between adsorptives and zeolite due to the modification of cations in the zeolite framework seem to be the cause for this behavior. Both adsorptives (water and alcohol) show slower diffusion rates on the 4A zeolite due to more intensive interactions with the sodium ion with its higher charge density (see also discussion in the next Section 4.2.3 ). The reason why ethanol diffusion is more decelerated than water diffusion is not yet known. Again dynamic drying of primary alcohols on a 4A zeolite is dependent on the chain length of the alcohol to be dried. As already observed in the shaker bottle experiments the alcohols from 1-bu- tanol to 1-hexanol reveal almost the same high drying capacity, whereas the BTC from 1-propanol is slightly shifted leftwards on the x -axis. Ethanol and methanol coadsorb significantly on the 4A zeolite as well. In contrast to the adsorption equilibrium the kinetics shows a reverse dependency. While methanol shows the steepest BTCs, the kinetics of drying is much slower in case of longer primary alcohols. In order to enable a detailed comparison of the dynamic drying of all six primary alcohols on a 4A zeolite the simulations of the drying of all alcohols with an initial water concentration of 2000 ppm w are displayed in Fig. 13 . With increasing chain length the MTZ of the dynamic water removal becomes subsequently

2000 1500 1000 4A experiment 1-butanol simulation 1-butanol 500 experiment 1-pentanol simulation 1-pentanol
2000
1500
1000
4A
experiment 1-butanol
simulation 1-butanol
500
experiment 1-pentanol
simulation 1-pentanol
experiment 1-hexanol
simulation 1-hexanol
0
0
500
1000
1500
2000
2500
3000
3500
time [min]
water concentration [ppm w ]

Fig. 12. Drying of the long-chain primary alcohols on a 4A zeolite.

longer. The calculated diffusivities of water in the pores of the 4A zeolite deliver similar results. Pore diffusivities become smaller with increasing chain length of the alcohol (see Table 6 ). The rea- son for this behavior is the same as in case of the 3A zeolite. Solvent molecules, which are blocking the micropores of the zeo- lite and their entrances, hinder water diffusion and lead to smaller pore diffusivities. Due to the higher probability that bigger mole- cules get stuck in a narrow pore opening the diffusivities of water are lower in case of drying long-chain alcohols than for small alco- hol molecules.

4.2.3. Comparison of 3A and 4A zeolite The comparison of the adsorptive drying of 1-pentanol on a 3A and a 4A zeolite is exemplary displayed in Fig. 14 . In order to guar- antee a comparability of the BTCs both experiments are carried out in fixed bed with a bed height of 23.7 cm. While the mass of adsorbed water from 1-pentanol is almost equal for both zeolites there are significant differences regarding the shape of the BTCs. On a 3A zeolite a complete drying of the alcohol is achieved for almost 300 min and afterwards a relatively sharp MTZ follows. In contrast on 4A zeolite 1-pentanol can be dried to water concentrations below 50 ppm w just for a few min- utes. The MTZ is much flatter and takes almost the whole length of the fixed bed of 23.7 cm. This proves that the kinetics of drying on the 4A zeolite is much slower than on the 3A zeolite. Similar results can be observed for the drying of all other alcohols

2000 1500 1000 experiment 1-pentanol 3A 500 simulation 1-pentanol 3A experiment 1-pentanol 4A simulation 1-pentanol
2000
1500
1000
experiment 1-pentanol 3A
500
simulation 1-pentanol 3A
experiment 1-pentanol 4A
simulation 1-pentanol 4A
0
0
500
1000
1500
2000
2500
3000
3500
time [min]
water concentration [ppm w ]

Fig. 14. Adsorptive drying of 1-pentanol on 3A and 4A zeolites.

48

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

Table 4 Comparison of pore diffusivities of water on 3A and 4A zeolite.

Solvent

D Pore – 3A (m 2 /s)

D Pore – 4A (m 2 /s)

Methanol

2.0 10 11 9.0 10 12 4.2 10 12 a 4.0 10 12 3.0 10 12 2.6 10 12

3.0 10 11 6.0 10 12 a 2.0 10 12 1.6 10 12 1.2 10 12 1.05 10 12

Ethanol

1-Propanol

1-Butanol

1-Pentanol

1-Hexanol

a Poor accordance between experiment and simulation due to subsequent dis- placement of adsorbed water.

Table 5 Fitting parameter of Langmuir isotherms.

Alcohol

Zeolite

q mon (g/100 g)

b (kg/mg)

R 2 ()

Methanol

3A

2.27

0.00542

0.942

Methanol

4A

No Langmuir fit possible

Ethanol

3A

6.97

0.00528

0.985

Ethanol

4A

6.98

0.00232

0.966

1-Propanol

3A

9.21

0.02849

0.981

1-Propanol

4A

16.31

0.00690

0.976

1-Butanol

3A

13.98

0.02853

0.990

1-Butanol

4A

17.61

0.00726

0.965

1-Pentanol

3A

16.70

0.01634

0.982

1-Pentanol

4A

17.82

0.00709

0.976

1-Hexanol

3A

15.74

0.02407

0.988

1-Hexanol

4A

17.78

0.00758

0.994

investigated in this study. Additionally the modeled diffusivities of water in the pores of the zeolites support this finding. With excep- tion of methanol as the shortest alcohol of the homologous series the pore diffusion of water is always faster in case of the 3A zeolite (see Table 4 ). At first these results seem to be surprising. Since 4A zeolites have slightly bigger pore openings than 3A zeolites, a faster diffu- sion of the adsorptives would be expected for the 4A zeolite. But in

Table 6 Overview of experimental parameters of dynamic experiments.

this case two other effects are more important. On the one hand the sodium ions with their higher charge density compared to the potassium ions in the 3A zeolite lead to more intensive interac- tions of the adsorbent with adsorptive molecules. This results in slower diffusion rates of polar molecules on the 4A zeolite in general. On the other hand more solvent molecules are able to enter the micropores of the 4A zeolite due to its bigger pore openings. For this reason the hindrance of water diffusion through blocking alco- hol molecules is slightly increased in comparison to the 3A zeolite. This results in lower pore diffusivities of the water and flatter BTCs. A combination of both effects will be responsible for slower dry- ing rates on the 4A zeolite. In order to define the dominant parameter of influence more detailed investigations on the diffu- sion of molecules in the inner of type A zeolites would be necessary.

5. Conclusions

This study reveals that adsorption equilibrium and kinetics of the drying of primary alcohols on 3A and 4A zeolites in the lower ppm w -region significantly depend on the solvent to be dried and the type of zeolite. The drying of short-chain primary alcohols on 3A and 4A zeo- lites shows considerably steeper BTCs than the drying of primary alcohols with a longer aliphatic chain. The reason for the slower kinetics and the smaller pore diffusivities of water are solvent molecules, which are hindering the water diffusion. Since bigger solvent molecules are blocking the narrow micropores of type A zeolites with a higher probability, water diffusion rates decrease subsequently with increasing chain length of the alcohol. However, since the kinetics even decelerates for large molecules like 1-hexanol, which are not able to enter the cages of type A zeo- lites, the main resistance must already be a blocking of the entrances to the microporous part of the zeolite by solvent mole- cules. This finding can be confirmed by sorption experiments of pure ethanol and 1-hexanol vapors on 3A and 4A zeolites. In comparison of both zeolites much faster kinetics can be observed for the drying on 3A zeolite. In this case the effect of

No.

Zeolite

c i ; 0 (ppm w )

V _ solu (l/h)

u LR (m/h)

m ADS (g)

H (cm)

D Pore (m 2 /s)

(1) Methanol

Me-3A

3A

1998

0.16

0.796

17.46

12.37

2.0 10 11 3.0 10 11

Me-4A

4A

1974

0.16

0.796

32.99

23.74

(2) Ethanol

Eth-3A-1

3A

478

0.16

0.796

17.47

12.37

5.0 10 12 7.0 10 12 9.0 10 12 9.0 10 12 6.0 10 12

Eth-3A-2

3A

980

0.16

0.796

17.46

12.37

Eth-3A-3

3A

1533

0.16

0.796

17.46

12.37

Eth-3A-4

3A

1969

0.16

0.796

17.47

12.37

Eth-4A

4A

2017

0.16

0.796

32.98

23.74

(3) 1-Propanol

Prop-3A

3A

1979

0.16

0.796

17.47

12.37

4.2 10 12 2.0 10 12

Prop-4A

4A

2005

0.16

0.796

32.99

23.74

(4) 1-Butanol

But-3A

3A

2018

0.16

0.796

17.46

12.37

4.0 10 12 1.6 10 12

But-4A

4A

2008

0.16

0.796

32.99

23.74

(5) 1-Pentanol

Pent-3A

3A

1983

0.16

0.796

17.46

12.37

3.0 10 12 1.2 10 12

Pent-4A

4A

2012

0.16

0.796

32.99

23.74

(6) 1-Hexanol

Hex-3A-1

3A

508

0.16

0.796

17.47

12.37

6.5 10 13 1.4 10 12 1.8 10 12 2.6 10 12 1.05 10 12

Hex-3A-2

3A

991

0.16

0.796

17.47

12.37

Hex-3A-3

3A

1465

0.16

0.796

17.46

12.37

Hex-3A-4

3A

1952

0.16

0.796

17.46

12.37

Hex-4A

4A

2014

0.16

0.796

32.99

23.74

B. Burrichter et al. / Separation and Purification Technology 145 (2015) 39–49

49

slightly increased pore opening from 3A to 4A zeolite is out- weighed by changed interactions on the surface of the zeolite. The change of cations from sodium (4A) to potassium (3A) leads to weaker interactions between adsorbent and adsorptive mole- cules, which results in an increased mobility of the adsorptive molecules on the 3A zeolite and higher diffusion rates. In addition the wider pore openings of the 4A zeolite enable more solvent molecules to enter the micropores of the zeolite further enhancing the hindrance of water diffusion. To sum up the sizes of solvent molecules and pore openings as well as the interactions between adsorbent and adsorptives are the main factors of influence to the dynamic drying behavior of pri- mary alcohols in a fixed bed adsorber. Furthermore for two adsorbent–alcohol combinations a subse- quent displacement of adsorbed water through alcohol molecules can be observed. This displacement is a consequence of an inter- play of the diffusion velocities of both adsorptives, which are determined by the molecular sizes and the polarity of the adsorp- tives, and the affinity of the adsorbent for the adsorptive, which is determined by the polarities.

Acknowledgements

The authors gratefully acknowledge financial support by Deutsche Forschungsgemeinschaft DFG (GZ: BA 2012/5-1). In addi- tion the authors thank Zeochem AG for the supply of adsorbents and the INC in Leipzig for assistance with the uptake of alcohol vapor isotherms.

References

[1] G. Natta, ‘‘Polymerisate’’ in Ullmanns Encyklopädie Der Technischen Chemie, 14. Band, Urban & Schwarzenberg, München, 1963. [2] D. Basmadjian, The adsorption drying of gases and liquids, Adv. Dry. (1984) 3307–3357 . [3] D.R. Burfield, R.H. Smithers, Desiccant efficiency in solvent drying. 3. Dipolar aprotic solvents, J. Org. Chem. 43 (20) (1978) 3966–3968 . [4] D.R. Burfield et al., Molecular sieves – desiccants of choice, J. Appl. Chem. Biotechnol. 28 (1) (1978) 23–30 . [5] S. Joshi, J.R. Fair, Adsorptive drying of toluene, Ind. Eng. Chem. Res. 27 (11) (1988) 2078–2085 . [6] S. Joshi, J.R. Fair, Adsorptive drying of hydrocarbon liquids, Ind. Eng. Chem. Res. 30 (1) (1991) 177–185 . [7] M.J. Rivero et al., Scale-up of adsorptive styrene drying, Polym. Int. 51 (9) (2002) 792–799 . [8] D.R. Burfield, R.H. Smithers, Desiccant efficiency in solvent and reagent drying. 7. Alcohols, J. Org. Chem. 48 (14) (1983) 2420–2422 . [9] W.K. Teo, D.M. Ruthven, Adsorption of water from aqueous ethanol using 3-A molecular sieves, Ind. Eng. Chem. Process Des. Dev. 25 (1) (1986) 17–21 . [10] S.M. Ben-Shebil, Effect of heat of adsorption on the adsorptive drying of solvents at equilibrium in a packed bed of zeolite, Chem. Eng. J. (Lausanne) 74 (3) (1999) 197–204 . [11] A. Carton et al., Separation of ethanol–water mixtures using 3A molecular sieve, J. Chem. Technol. Biotechnol. 39 (2) (1987) 125–132 . [12] A.K. Jain, A.K. Gupta, Adsorptive drying of isopropyl alcohol on 4A molecular sieves: equilibrium and kinetic studies, Sep. Sci. Technol. 29 (11) (1994) 1461–