Applied Thermal Engineering 137 (2018) 368–376

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Research Paper

Physical properties and adsorption kinetics of silica-gel/water for adsorption T

chillers
⁎
Ramy H. Mohammeda,b, , Osama Mesalhya,b, Mohamed L. Elsayeda,b, Sichao Houc, Ming Suc,
Louis C. Chowa
a
Department of Mechanical and Aerospace Engineering, University of Central Florida, Orlando, FL 32816-2450, United States
b
Department of Mechanical Power Engineering, Zagazig University, Zagazig 44159, Egypt
c
Department of Chemical Engineering, Northeastern University, Boston, MA 02115, United States

H I GH L IG H T S

• Adsorption isotherm models are examined and discrepancies are observed/corrected.

• Physical properties and adsorption kinetics of silica-gel RD and RD-2060 are measured.
• Capillary condensation is observed at relative pressure higher than 0.4 for the silica-gels studied.
• A non-isothermal model is proposed to obtain diffusion coefficient in nanoporous materials from measured data.

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

Keywords: The choice of a suitable adsorbate/adsorbent pair is critical for an adsorption cooling cycle. The surface char-
Adsorption kinetics acteristics and thermo-physical properties of the adsorbent, and the adsorption rate of adsorbate are key
Surface diffusivity parameters in making the choice. Through literature review, it is found that there are disagreements among the
Inconsistencies experimental measurements and various equations/models used to describe adsorption isotherms and surface
Silica-gel
diffusivity of water in silica-gel. In this work, an experimental set-up is built to measure the isotherms and
Experimental set-up
kinetics of vapor adsorption for any working pair. Using the newly measured data, those from the manufacturers
and from the literature, these inconsistencies are eliminated by utilizing the Dubinin-Astakhov (D-A) model to fit
the entire adsorption isotherm curve. The Brunauer-Emmett-Teller (BET) method is used to calculate the surface
area, pore volume and pore diameter of two different types of silica-gel. Based on the adsorption rate and the
adsorbent temperature measured simultaneously, a new approach is proposed to measure the surface diffusivity
in the temperature and pressure ranges typical of those during the operating conditions of adsorption cooling
systems. Analysis of the results indicates that the surface diffusivity follows the Arrhenius-form equation. The
calculated activation energy at different adsorption conditions varies from 40.0 to 41.2 kJ/mol and the pre-
exponential factor varies from 2.5 × 10−4 to 2.8 × 10−4 m2/s. These values are close to those previously re-
ported in the literature. Thus, the proposed approach can be used to measure the surface diffusivity in nano-
porous materials.

1. Introduction
Adsorption refrigeration systems have been successfully utilized as
an environmentally friendly alternative technique for air conditioning
and refrigeration applications [1–3]. The performance of adsorption
refrigeration systems depends critically on the capacity of the adsorbent
to adsorb the vapor of the working fluid as well as on the rate at which
the adsorption bed can adsorb/desorb the working fluid [4–6]. Ad-
sorption working pairs are usually selected based on the available heat
source temperature and the required cooling load. The thermo-physical
properties, diffusion kinetics and equilibrium isotherms are the main
characteristics of adsorption working pair that control the performance
of an adsorption cooling system [7,8]. Silica-gel/water is one of the
preferred working pairs because it is non-toxic, stable over a wide range
of operating temperatures and compatible with most materials used in
adsorption cooling systems. Silica-gel/water also require only a low to
modest temperature for regeneration/desorption [9,10].
The surface area, pore volume and pore size of porous adsorbent are

⁎
Corresponding author at: Department of Mechanical and Aerospace Engineering, University of Central Florida, Orlando, FL 32816-2450, United States.
E-mail addresses: rhamdy@zu.edu.eg, rhamdy@knights.ucf.edu (R.H. Mohammed).

https://doi.org/10.1016/j.applthermaleng.2018.03.088
Received 23 October 2017; Received in revised form 1 March 2018; Accepted 26 March 2018
Available online 01 April 2018
1359-4311/ © 2018 Elsevier Ltd. All rights reserved.
R.H. Mohammed et al. Applied Thermal Engineering 137 (2018) 368–376

Nomenclature Xm maximum adsorbent capacity (kg/kg of adsorbent)

Xo equilibrium uptake (kg/kg of adsorbent)
D affinity coefficient (–)
d diameter (m) Greek symbols
Ds surface mass diffusivity (m2/s)
Dso pre-exponential factor (m2/s) τ tortuosity
E characteristic energy of adsorption (J/mol), error
Ea activation energy (kJ/mol) Subscripts
H heat of adsorption (kJ/kg)
n uploading factor (–) LC load cell
Ps saturation vapor pressure at Ts (Pa) e effective
Pv operating pressure (Pa) p particle, pressure
R universal gas constant (8314 J/kmol K) s solid phase, saturated, surface
RP relative pressure (–) T temperature
t time, sec v vapor phase
T temperature (K)

usually measured by using the Brunauer-Emmett-Teller (BET) method
[11]. The pore diameter of common silica-gels varies from 0.7 to 3 nm
according to the silica-gel type, while the specific surface area ranges
from 100 to 1000 m2/g. The adsorbent capacity is traditionally mea-
sured by either volumetric or gravimetric method [12]. The volumetric
approach depends on monitoring the pressure change during the ad-
sorption, which indicates the adsorption rate. The tested adsorbent
should be in a small amount for the pressure change to be small, so the
process could be assumed to be isobaric. The main disadvantage of this
method is the testing conditions are far away from the real operating
conditions of the adsorption cooling unit. On the other hand, the
gravimetric method measures directly the adsorbent capacity by means
of mass balances at specific operating conditions. This makes this
method more suitable to measure the adsorption isotherms of working
pairs.
The Freundlich, Tòth and Dubinin-Astakhov (D-A) equations are
used to fit experimental measurements of adsorption isotherm. The
Freundlich and Tòth equations do not estimate the uptake correctly
when the relative pressure is less than 0.15. The D-A equation, which is
widely used, was originally developed based on the Polanyi adsorption
theory and has different forms [12,13]. The uptake could be a function
of the operating pressure and adsorbent temperature as in Eqs. (1a) and
(1b) which are the same, or a function of the operating temperature as
in Eq. (1c).
n
Xo RT P
= exp ⎧−⎛⎜− s ln ⎛ v ⎞ ⎞⎟ ⎫
⎜ ⎟
Xm ⎨ ⎝ E ⎝ Ps ⎠ ⎠ ⎬ (1a)
⎩ ⎭
n come the last limitation, the amount of tested material should be small
Xo P
= exp ⎧−D ⎛⎜ln ⎛ s ⎞ ⎞⎟ ⎫
⎜ ⎟
Xm ⎨ ⎝ ⎝ Pv ⎠⎠ ⎬ (1b)
⎩ ⎭
n material are mixed with copper wires to keep the sample more iso-
Xo T
= exp ⎧−k ⎛ s −1⎞ ⎫
⎜ ⎟
Xm ⎨
⎩ ⎝ Tv ⎠ ⎬ ⎭ (1c)
where Xo is the equilibrium adsorbate mass per unit mass of adsorbent
(kg/kg) at vapor pressure Pv and adsorbent temperature Ts , Ps is the
saturation vapor pressure at Ts,Tv is the vapor temperature, Xm is the
maximum adsorbent capacity, Pv is the relative pressure (RP), D is the
Ps
affinity coefficient, R is the universal gas constant, n is the uploading
factor, k is constant depends on adsorption isotherm and E is the
characteristic energy of adsorption (J/mol).
The diffusion of adsorbed molecules in solid porous materials has
long been studied. Surface diffusion participates significantly in the
mass transport in these materials, especially in the adsorption, se-
paration and many other industrial applications. Although the me-
chanism of surface diffusion is rather complicated, two well-known
models have been developed to describe the dependence of the surface
diffusivity on the surface operating temperature and the surface cov-
erage. These two models are the hydrodynamic model [14] and the
hopping model [15]. Gilliland et al. [14] derived an equation for ad-
sorbed-layer movement based on thermodynamic principles. It was
assumed that the shear stress between the adsorbed layer and the solid
is proportional to the average rate of movement of adsorbed molecules
past the surface. The resulting equation has just one empirical constant
and its application requires adsorption-isotherm data. Although the
equation correlates their experimental measurements well with the
surface flow rates for the range of the studied variables, it is in integral
form which can be solved for special isotherm cases only. On the other
hand, the hopping or jumping mechanism is conceptually the most
basic mechanism for diffusion of adsorbate molecules. It assumes that
the gas molecules move over the surface from one site to another by
hopping with a certain velocity [15,16].
Many experimental measurements have been carried out to in-
vestigate the diffusion process inside porous materials. The Isothermal
Differential Step (IDS) method was considered to be the most reliable
method for identification of the kinetic parameters such as the diffusion
constant and the adsorbate diffusivity [17,18]. In this method, the
pressure and temperature are controlled, and a small amount of pure
liquid adsorbate is set inside the evaporator. The dry sample is exposed
to the evaporator and the decrease in the liquid indicates the adsorption
uptake of this adsorbent. This approach has some drawbacks like the
testing conditions are far away from the real operating conditions of
adsorption cooling cycle as well as the difficulty to keep the adsorbent
at constant temperature during the adsorption process to be able to use
the Arrhenius equation for calculating the surface diffusivity. To over-
(7–10 mg) to reduce the thermal effects on the adsorption kinetics [19].
For further reduction of the thermal effects, the particles of the tested
thermal. Still, it is difficult to achieve isothermal condition.
Strauss et al. [20] proposed an non-isothermal approach for ad-
sorption kinetic characterization to overcome the isothermal limitation.
In this case, a sample is placed inside an evacuated closed chamber and
subjected to temperature change. The adsorption rate is monitored by
recording the pressure evolution inside the system. As this methodology
cannot be considered isothermal, a complicated model based on solving
partial differential equations of energy and mass inside the adsorbent
material is used to fit the experimental data. Furthermore, this ex-
perimental approach cannot be considered satisfactory, because the
adsorption uptake is measured based on the change in the pressure and
these conditions are far from the actual operating condition of the ad-
sorption cooling system, where both adsorption and desorption process
are isobaric.
In this work, inconsistencies among the experimental measurements

369
R.H. Mohammed et al.
and the various equations/models used to describe the adsorption iso-
therms and the surface diffusivity of silica-gel/water are addressed.
Using inaccurate adsorption isotherms could lead to significant errors in
estimating the cooling capacity of adsorption cooling units.
Accordingly, new sets of coefficients for D-A model are proposed to
eliminate these discrepancies. At the same time, an experimental test
rig is designed and built to measure both the adsorption kinetics and
isotherms of any working pair. The adsorption characteristics of Fuji
silica-gel RD and RD-2060 are measured. The D-A model is modified to
fit appropriately the entire adsorption isotherm. A new approach, based
on the Arrhenius equation, is proposed to measure the surface diffu-
sivity in nanoporous materials. The new methodology overcomes the
complexity and limitations of the other approaches. One of the great
advantages of this approach is the possibility of testing adsorption ki-
netics of any working pair under the operating conditions of adsorption
cooling cycle. Also, it is a non-isothermal approach that considers the
change of the adsorbent temperature during the adsorption process.
The measured surface diffusivities in silica-gels are compared with the
previous measured ones from the literature. In addition, the thermo-
physical properties including specific surface area, pore volume and
pore size are measured and discussed for these silica-gels.
2. Inconsistencies in adsorption models of silica-gel/water
Silica-gel (SiO2) is widely used in many industries due to its strong
hydrophilicity towards water. There are many types of silica-gel such as
A, B, A5BW and A++. Silica-gel RD is commonly employed in ad-
sorption cooling systems and dehumidification applications. The ad-
sorption isotherm of silica-gel/water is measured by either the gravi-
metric or volumetric methodology, while the isothermal or non-
isothermal approach was proposed to extract the surface diffusivity.
Although there are many publications in the adsorption characteristics,
there are inconsistencies among the different experimental measure-
ments of the surface diffusivity. Also, there are significant discrepancies
among the adsorption equilibrium equations used to describe the ad-
sorption isotherm of silica-gel RD/water.
Chihara et al. [17] used the gravimetric method to measure the
adsorption properties of Fuji silica-gel with water vapor. Xia et al. [21]
used Eq. (1b) to fit these experimental measurements and obtained
values for Xm = 0.348 kg/kg, n = 1.6 and D = 0.449. Also, Eq. (1c)
with values for Xm = 0.346 kg/kg, n = 1.6 and k = 5.6 was used to
correlate the same experimental data, and this equation has been
widely used in numerical simulations of adsorption cooling systems
[22–25]. Fig. 1 shows large discrepancies among the experimental
measurements of the adsorption isotherm of silica-gel/water and the
calculated ones using the D-A model. So, new sets of parameters for the
D-A model should be proposed to appropriately fit the measurements.
Choosing Xm = 0.346 kg/kg, D = 0.64, and n = 1.0 for Eq. (1b) and
Xm = 0.346 kg/kg, k = 11, and n = 1.0 for Eq. (1c) can minimize the
discrepancy. The resulting isotherms using these new sets of parameters
are also plotted in Fig. 1. It is clear that the predicted uptakes using the
new sets of coefficients have a good agreement with the experimental
measurements.
The surface diffusivity of water in silica-gel was measured using
isothermal and non-isothermal approaches. Based on hopping me-
chanism, Sladek et al. [15] proposed a simple Arrhenius-form equation
for surface diffusion (Ds ) of water vapor onto silica gel as follows [16]:
1.6 × 10−6 A·H ⎞
Ds = exp ⎛−
⎜ ⎟
τ ⎝ Ts ⎠ (2a)
where τ is the solid tortuosity factor which is equal 2.8 for regular
density silica gels, Ts is the adsorbent temperature, A is a constant
which is equal to 0.974 for silica gel, and H is the heat of adsorption
that is calculated for silica-gel/water from the following relationship:
370
Applied Thermal Engineering 137 (2018) 368–376
kJ
H= −1400X o + 2950 for X o ⩾ 0.05 ⎜⎛ ⎟⎞
⎝ kg ⎠ (2b)
Based on the Isothermal Differential Step (IDS) method, a CAHN
2000 thermo-balance was used to measure the equilibrium uptake of
water vapor onto RD silica-gel at a temperature range from 30 to 150 °C
and a vapor pressure range from 1 to 5 kPa [26].
The surface diffusivity was found to be an Arrhenius function of
temperature (Eq. (3)) with activation energy (Ea ) of 41.5 kJ/mol and
pre-exponential factor (Dso ) of 2.9 × 10−4 m2/s. These measured values
are close to those obtained in [17] for Fuji Davison silica-gel type A.
This is may be because of the similar nanostructure in these silica-gels.
E
Ds = Dsoexp ⎛− a ⎞ ⎜ ⎟
⎝ RTs ⎠ (3)
Fig. 2 compares the aforementioned measured surface diffusivity of
water onto silica-gel. It is found that the measured surface diffusivity by
Aristov et al. [26] is close to the measured one by Chihara and Suzuki
[17]. The measured surface diffusivity by Sladek et al. [15], which
depends on the water concentration, is close to the others at low up-
take, but it is about 2.6 times higher at uptake of 0.3 at 350 K as pre-
sented in Fig. 2. Due to these inconsistencies, a new approach for
measuring the surface diffusivity is proposed and validated.
3. Experimental work
3.1. Physical properties measurements
The surface characteristics such as the specific surface area, pore
volume and pore size distribution of the silica gel RD and RD-2060 are
analyzed using a surface area analyzer (Quantachrome, NOVAe)
[27–29]. Nitrogen is used as the adsorbate under the environment of
liquid nitrogen (77 K). Fig. 3 shows the adsorption isotherms of Fuji
silica-gel RD and RD-2060. Both types show a type IV adsorption iso-
therm [30], which indicates that the nitrogen molecules first form
monolayer followed by multilayer on the silica walls. The pore sizes
within the silica beads fall in the mesoporous range, which facilitates
capillary condensation of vapor inside the silica. The specific surface
area, pore volume and average pore diameter of these two types of
silica beads are listed in Table 1. It is shown that silica gel RD has larger
surface area and pore volume than the other one, while the average
pore diameter of the two types are nearly the same.
Fig. 4 shows the statistical result of particle size distribution of the
two types of porous silica beads. Silica beads are well dispersed in a
petri dish with a diameter of 3 cm for the optical images. A MATLAB
program is used to distinguish the contour of the beads from the
0.4
0.35
0.3
0.25
Xo (kg/kg)
0.2
0.15 Exp. Data [17]
D-A model [21]
0.1
D-A model [22-25]
0.05 Xm=0.346, D=0.64, n=1.0
Xm=0.346, k=11, n=1.0
0
0 0.2 0.4 RP 0.6 0.8 1
Fig. 1. Comparison between the measured isotherm of silica-gel/water and the
predicated one using D-A model.
R.H. Mohammed et al. Applied Thermal Engineering 137 (2018) 368–376

7E-10 filled by the liquid refrigerant (adsorbate) and its temperature is con-
Sladek et al. [15]
trolled by thermal bath 1 (TB1). The other chamber has the tested
6E-10 Aristov et al. [26]
sample rested on 120 × 40 mm heat exchanger. The thickness of the
Chihara and Suzuki [17]
5E-10 sample on the heat exchanger should be small to be able to assume
Xo=0.2 uniform temperature within the sample during the adsorption process.
Ds (m2/s)

4E-10 The heat exchanger is connected to the thermal bath 2 and 3 using
flexible tubing to control the sample temperature during the adsorption
3E-10 Xo=0.05 and desorption process. The heat exchanger is set on a single point load
Xo=0.3 cell to monitor the mass change (i.e. adsorption rate) during the test.
2E-10 The time response of the load cell is less than 0.1 s. The load cell with
1E-10 the heat exchanger and flexible tubing is calibrated using a very precise
balance and its accuracy is estimated to be ± 0.02 g. The total error in
0 load cell reading due to wires connections and Data Acquisition (DAQ)
270 290 310 330 350 370 system is given in Appendix A. The load cell output is also recorded
Ts (K) without the adsorbent at different cooling water temperatures and mass
flow rates. It was determined that the flexible tubing and cooling water
Fig. 2. Surface mass diffusivity of water vapor onto silica gel obtained from do not lead to significant uncertainty in the load cell measurements. In
previous studies.
addition, tubing position are adjusted before the test to eliminate the
vibrations. During the test, the pressure inside the tank is always lower
350 than that of the outside. This will allow the acrylic top cover to push
onto the silicon gasket, causing a deformation, which provides a perfect
300 sealing and prevents the ambient air from going inside. The results from
the leakage test indicated that the pressure of measuring unit chamber
250 and evaporator/condenser chamber increased only 20 Pa/day. The
Volume, (cc/g)

sample temperatures at different locations during the test are measured

200 using T-type thermocouples. Further, these thermocouples measure the
temperature distribution within the sample to check whether uniform
150 temperature is achieved. Two MKS pressure transducers with accuracy
of ± 0.25% are used to measure the pressure of each chamber. All
100
measurements are connected to a data acquisition system.
Type-RD
50
RD-2060 3.2.2. Measurements methodology
The present testing methodology has three stages: preparation,
0
desorption, and adsorption. During the preparation stage, a few layers
0 0.2 0.4 0.6 0.8 1
RP of silica-gel is placed on the heat exchanger and its temperature is
controlled by the thermal bath 2 or 3. Subsequently, the load cell re-
Fig. 3. N2 uptake by two types of porous silica-gel beads.
sponse is recorded for 30 min to guarantee that the vibrations due to the
flexible tubes do not negatively affect the measurements and the load
Table 1 cell accuracy. The temperature of the evaporator/condenser is con-
The physical properties of type-RD and RD-2060 silica-gel beads. trolled using the thermal bath 1 (TB1). In the desorption stage, the
Silica-gel RD RD-2060 sample is heated up and evacuated by the vacuum pump until the re-
lative pressure is lower than 10−4. The sample is assumed to be com-
Surface area (m2/g) 827.5 686.3 pletely dry when the load cell reading is constant for one hour, then the
Pore volume (cc/g) 0.462 0.335
degassed silica-gel (i.e. dry silica gel) sample is weighed. For the ad-
Pore diameter (nm) 3.24 3.19
sorption step, valves are switched to allow cooling water to cool down
the sample to the desired adsorption temperature. Vacuum pump is
background based on the contrast difference as shown in Fig. 4A and B. turned off and then the two chambers are connected to start the ad-
The diameter (or equivalent diameter in type RD-2060) of each bead is sorption process. The sample temperatures and the load cell response
calculated by the program and the statistical results are shown in are monitored during the adsorption process. The equilibrium condi-
Fig. 4C and D. For type RD silica beads, the most probable diameter falls tions are achieved when the load cell response and thermocouples
in the range of 0.80–0.89 mm, while that of RD-2060 silica beads is measurements are constant for 0.5 h. Various silica-gel samples are
0.50–0.59 mm. Compared to RD-2060, RD has a wider distribution tested (see Table 2) in order to investigate the effect of the particles
(from 0.7 to 2.0 mm), which leads to a higher level of heterogeneity. sizes distribution on the adsorption characteristics of silica-gel/water.
The average diameters are 0.835 and 0.540 mm for RD and RD-2060
silica beads, respectively. 4. Results and discussion

4.1. Adsorption isotherms

3.2. Equilibrium uptake and adsorption kinetics measurements
3.2.1. Experimental test rig
The experimental facility has been described in detail elsewhere
[31–33]. The experimental setup is designed and built to measure the
adsorption kinetics and equilibrium uptake of the adsorbent under ty-
pical operating conditions. It has two insulated stainless chambers as
shown in Fig. 5. The small chamber works as an evaporator during the
adsorption process and a condenser during the desorption process. It is
The equilibrium uptakes of water vapor onto silica-gel RD-2060 and
RD at different pressure and temperature are measured and plotted in
Figs. 6 and 7, respectively. The experimental measurements show that
the maximum uptakes of silica-gel RD and RD-2060 are about 0.48 kg/
kg and 0.38 kg/kg, respectively. It is found that the adsorption isotherm
curve has an inflection point at relative pressure ≈ 0.4 for both silica-
gels. This is attributed to the fact that the region of the water adsorption
isotherm where the relative pressure is more than 0.4 (RP > 0.4) is

371
R.H. Mohammed et al. Applied Thermal Engineering 137 (2018) 368–376

Fig. 4. Optical images of (A) type-RD and (B) RD-2060 porous silica beads with computer-aided statistical analysis; particle size distribution of (C) RD and (D) RD-
2060 porous silica beads.

typically dominated by capillary condensation. As a result, the ad- temperature. The average effective diameter is calculated based on the
sorption isotherm is divided into two relative pressure regions to be exact solution of the diffusion equation (Fick’s law). The exact solution
accurately fit the experimental measurements by using the D-A equa- of vapor diffusion onto porous particle of diameter dp is [30]:
tion. It is found that the coefficients illustrated in Table 3 for D-A model
m v (t ) m (t ) 144Ds
appropriately describe the experimental data for uptakes for the entire = A t , for v < 0.3 where A2 =
mo mo πdp2 (5)
relative pressure range. Using one value of uploading factor (n) fits the
experimental measurements at relative pressure higher than 0.2 and 0.1 where m v is the mass adsorbed at certain time t , mo is the maximum
for silica-gel RD and RD-2060, respectively. These deviations at low mass adsorbed at t → ∞.
relative pressures make the D-A model with only one value of n not The uptake of a sample with different particles sizes after specific
sufficiently accurate to predict the performance of adsorption cooling time could be calculated as:
systems because the relative pressure at the beginning of adsorption
process in adsorption cooling systems is about 0.05. Using the D-A m v1 + m v2 + m v3 + …=mo1 A1 t + mo2 A2 t + mo3 A3 t + …=mo A t
model with only one value of n, the specific cooling power (SCP) could (6a)
be 14% higher than that predicted by the D-A model with the correct Then
coefficients.
mo1 A1 m A m A
+ o2 2 + o3 3 + …=A where mo1 = Xo1 ms1
mo mo mo (6b)
4.2. Adsorption kinetics
where mo A t is the total mass adsorbed, mo1 is the maximum mass
The adsorption kinetics characterizes the degree of saturation. adsorbed by particles of diameter dp1 whose mass is ms1.
Linear driving force (LDF) (Eq. (4)) for vapor adsorption kinetics is As the sample has the same adsorbent, Xo1 = Xo2 = Xo3 = Xo and Eq.
successfully used to fit the experimental adsorption rate with standard (6b) can be rewritten as:
deviation of 0.012 as shown in Fig. 8 [32]. ms1 A1 m A m A y y y 1
+ s2 2 + s3 3 + …=A → 21 + 22 + 23 + …= 2
ms ms ms d p1 d p2 d p3 de
dX 60Ds X (t )
= (Xo −X ) = km (Xo −X ) → = 1−exp(−km t ) n
dt de2 Xo (4) yi
= ∑
i=1 dpi2 (7)
where de is the effective diameter of the tested sample and Ds is the
surface diffusivity that can be calculated from Eq. (3). where yi is the mass fraction of the particles of diameter dpi .
During the adsorption period, the silica gel sample temperature is Accordingly, the effective particle size can be calculated based on
nearly isothermal. The difference between the maximum and minimum the particle size distribution of the tested sample.
temperature of the silica gel at all times is found to be less than 2 °C for Fig. 9 presents the procedure of calculating the activation energy
all tested cases. This small difference in temperature is due to using a (Ea ) and pre-exponential factor (Dso ) using the measured instantaneous
small thickness (<7 mm) of the adsorbent. Accordingly, uniform tem- adsorption uptake and the average adsorbent temperature. First, initial
perature within the sample during the adsorption test is assumed and values for the activation energy and pre-exponential factor are as-
the average sample temperature is used to represent the sample sumed. Based on the instantaneous average temperature of the sample

372
R.H. Mohammed et al. Applied Thermal Engineering 137 (2018) 368–376

Vacuum pump
(A)
Acrylic cover V2 V1 Silicon
gaskets

Insulation pad Vapor

Adsorber

T
P

Load cell Evaporator

TB3
TB2 TB1

(B) Pressure
Power supplies DAQ system
transducers
Vacuum Valves
pump
Readout

Measuring
chamber Evaporator/
condenser

Flow Pump Flow

meter meter

TB2
TB3
TB1

Fig. 5. (A) Experimental set-up (B) A pictorial view of the experimental facility.

Table 2 0.5
Tested silica-gels samples.
Silica gel type Fuji RD-2060 Fuji type-RD
0.4
Particles sizes (mm) 0.3–0.49 0.7–2.0
0.5–0.79
0.3
X (kg/kg)

0.3–0.79

during the adsorption, the surface diffusivity is calculated using Eq. (3), 0.2
then the instantaneous uptake is estimated from Eq. (4). According to
the difference between the estimated uptake and the measured one, the
values of the activation energy and pre-exponential factor are modified 0.1 Exp. measurements
to keep this difference is less than 0.001 kg/kg. Adsorption rates of the D-A model with two different n
two types of silica-gel are measured at different operating pressures D-A model with n=1.25, E=3465
0
(1.0–20.0 kPa) and adsorbent temperatures (10–70 °C). Multiple sam- 0 0.2 0.4 0.6 0.8 1
ples of silica-gel with different particles sizes are tested (see Table 2). It
RP
is found that the calculated activation energy at different adsorption
conditions varies from 40 to 41.2 kJ/mol and the pre-exponential factor Fig. 6. Data fitting of the experimental water sorption isotherm of silica-gel
from 2.5 × 10−4 to 2.8 × 10−4 m2/s. These values deviate from the Type-RD.
published data by less than 4.5% [32,34,35]. It is worth to mention that
using either the maximum or minimum temperature of the sample in- approach. It is evident from these figures that this proposed approach
stead of the average sample temperature to calculate the surface dif- shows a good agreement between the measured and predicted ad-
fusivity changes the calculated activation energy by less 0.3%. sorption rate despite of the particle size distribution. This good agree-
Figs. 10 and 11 depict the measured adsorption rate and the pre- ment is established for all tested samples shown in Table 2. Although
dicted one based on the calculated surface diffusivity using the new Eq. (5) is valid only for uploading ratio (X/Xo) less than 0.3, the

373
R.H. Mohammed et al. Applied Thermal Engineering 137 (2018) 368–376

0.5 Start

0.4 Enter
experimental data
Ts(t), X(t), Xo, de
0.3
X (kg/kg)

Assume initial value for

Dso and Ea
0.2

Error= 0, i = 0
0.1 Exp. measurements imax=2000
D-A model with two different n
D-A model with n=1.45, E=4375
0 Calculate Ds from Eq. 3
0 0.2 0.4 0.6 0.8 1

Change Dso or Ea
RP
Xp(i)/Xo=1-exp(-60Ds/
Fig. 7. Data fitting of the experimental water sorption isotherm of silica-gel RD- i=i+1
de*de)
2060.

Table 3 Error=Error+abs(X(i)-Xp(i)
D-A coefficients for the tested silica-gels.
Type RP D-A model Yes
i<imax
Xm n E
No
RD-2060 0.04–0.4 0.38 1.1 4375
0.4–0.95 0.38 1.6 4375 No
Error<0.001
Type-RD 0.04–0.4 0.48 1.1 3465
0.4–0.95 0.48 1.6 3465
Yes
Print Dso and Ea
50 1.2
Average sample temperature (oC)

1
40 End

0.8 Fig. 9. Procedure of calculating the activation energy and pre-exponential

30 factor.
X/Xo

0.6
20 1.2
0.4
Sample temperature
10 1
Load cell response 0.2
LDF fitting
0.8
0 0
0 500 1000 1500 2000
X/Xo

Time (s) 0.6

Fig. 8. Load cell response and silica gel RD temperature during adsorption
process. 0.4

calculated effective particle size of the sample still provides reasonable 0.2 Exp. data
results of the surface diffusivity. It is worth to mention that this ap- predicted data
proach is valid as long as the time response of the load cell and ther- 0
mocouple is comparable because both the instantaneous temperature 0 500 1000 1500
measurements and load cell response are used to calculate the surface Time (s)
diffusivity. Also, the thickness of the sample should be less than 5 mm, Fig. 10. Experimental and predicted adsorption rates for 5 g silica-gel RD-2060
which is comparable to the thickness of adsorbent layer in real ad- with particles sizes of 0.3–0.79 mm.
sorption bed, to be able to achieve uniform temperature distribution
within the tested sample. As a result, this proposed approach is ap-
5. Conclusion
plicable in measuring the surface diffusivity in nanoporous adsorbents
under operating conditions like those in the real adsorption cooling
Numerous inconsistencies among the limited available experimental
systems.
measurements of the adsorption characteristics of silica-gel/water and
the models used to describe these measurements are observed and
corrected to prevent their propagation in the future scientific literature.
Due to the drawbacks of the existing experimental facilities used to

374
R.H. Mohammed et al. Applied Thermal Engineering 137 (2018) 368–376

1.2 experimental set up is built to measure the adsorption isotherm and

kinetics of any working pair. Physical properties and adsorption char-
1 acteristics of Fuji silica-gel RD and RD-2060 commonly selected in
adsorption cooling and desalination cycles are experimentally in-
0.8 vestigated. The parameters in D-A model are adjusted so that the model
can appropriately fit the entire adsorption isotherm. A new approach
for calculating the surface diffusivity which allows the sample tem-
X/Xo

0.6
perature to be time dependent is proposed to overcome the limitations
and complexities of the other methodologies. The key advantage of this
0.4 approach, beside its simplicity, is the possibility of measuring the sur-
face diffusivity under the real operating conditions of adsorption
0.2 Exp. uptake cooling cycles. This proposed methodology is validated by comparing
Predicted uptake the calculated surface diffusivity of water vapor onto silica-gels with
0 the published data, and good agreement is established.
0 500 1000 1500 2000
Time (s)
Acknowledgement
Fig. 11. Experimental and predicted adsorption rates for 20 g silica-gel RD with
particles sizes of 0.7–2.0 mm. This work is supported by National Science Foundation (NSF)
through Grant No. 1603215 and the Egyptian Ministry of Research and
measure the adsorption characteristics of nanoporous materials, an Higher Education.

A: Error analyses

The factors that lead to errors in the load cell, temperature and pressure measurements are the accuracy of the load cell (E0), thermocouple (E1),
pressure transmitter (E2), data logger (E3) and errors due to connections (E4). The total errors in the load cell reading (ELC) temperature (ET) and
mean pressure (EP) measurements can be determined as follow [36–38]:

ELC = E02 + E32 + E42 (A.1)

ET = E12 + E32 + E42 (A.2)

EP = E22 + E32 + E42 (A.3)

As the surface diffusivity is a function of temperature, the error associated with it can be written as:
dDs DE E Ds E
E Ds = ET = s 2 ET ⇒ = ET
dT RT Ds RT 2 (A.4)
Based on the above equations, the maximum error of load cell, temperature and pressure measurements is found to be 1.5%, 2% and 1.5%,
respectively. The error in calculating the surface diffusivity is about 2.5%, and the error in the surface diffusivity due to the temperature variation of
the sample during the adsorption (maximum difference should be less than 2 °C) is found to be 3.5%.

References
[1] R.P. Sah, B. Choudhury, R.K. Das, A review on adsorption cooling systems with
silica gel and carbon as adsorbents, Renew. Sustain. Energy Rev. 45 (2015)
123–134.
[2] F.N. Al-Mousawi, R. Al-Dadah, S. Mahmoud, Different bed configurations and time
ratios: performance analysis of low-grade heat driven adsorption system for cooling
and electricity, Energy Convers. Manage. 148 (2017) 1028–1040.
[3] P.J. Vodianitskaia, et al., Experimental chiller with silica gel: adsorption kinetics
analysis and performance evaluation, Energy Convers. Manage. 132 (2017)
172–179.
[4] D.B. Boman, et al., Screening of working pairs for adsorption heat pumps based on
thermodynamic and transport characteristics, Appl. Therm. Eng. 123 (2017)
422–434.
[5] M. Louajari, A. Mimet, A. Ouammi, Study of the effect of finned tube adsorber on
the performance of solar driven adsorption cooling machine using activated car-
bon–ammonia pair, Appl. Energy 88 (3) (2011) 690–698.
[6] X. Li, et al., A review on development of adsorption cooling—novel beds and ad-
vanced cycles, Energy Convers. Manage. 94 (2015) 221–232.
[7] M.M. Younes, et al., A review on adsorbent-adsorbate pairs for cooling applications,
Appl. Therm. Eng. 114 (2017) 394–414.
[8] G.G. Ilis, Influence of new adsorbents with isotherm Type V on performance of an
adsorption heat pump, Energy 119 (2017) 86–93.
[9] D. Wang, et al., Progress in silica gel–water adsorption refrigeration technology,
Renew. Sustain. Energy Rev. 30 (2014) 85–104.
[10] D. Wang, et al., Investigation of adsorption performance deterioration in silica
gel–water adsorption refrigeration, Energy Convers. Manage. 58 (2012) 157–162.
[11] J.C. Maxwell, A Treatise on Electricity and Magnetism, Clarendon press, 1881.
[12] R. Wang, L. Wang, J. Wu, Adsorption Refrigeration Technology: Theory and
Application, John Wiley & Sons, 2014.
[13] H.N. Tran, et al., Mistakes and inconsistencies regarding adsorption of con-
taminants from aqueous solutions: a critical review, Water Res. (2017).
[14] E. Gilliland, R. Baddour, J. Russell, Rates of flow through microporous solids, AIChE
J. 4 (1) (1958) 90–96.
[15] K.J. Sladek, E.R. Gilliland, R.F. Baddour, Diffusion on surfaces. II. Correlation of
diffusivities of physically and chemically adsorbed species, Indust. Eng. Chem.
Fundam. 13 (2) (1974) 100–105.
[16] R.H. Mohammed, et al., Novel compact bed design for adsorption cooling systems:
parametric numerical study, Int. J. Refrig. 80 (2017) 238–251.
[17] K. Chihara, M. Suzuki, Air drying by pressure swing adsorption, J. Chem. Eng.
Japan 16 (4) (1983) 293–299.
[18] A. Sakoda, M. Suzuki, Fundamental study on solar powered adsorption cooling
system, J. Chem. Eng. Japan 17 (1) (1984) 52–57.
[19] A. Sapienza, et al., Dynamic study of adsorbers by a new gravimetric version of the
Large Temperature Jump method, Appl. Energy 113 (2014) 1244–1251.
[20] R. Strauss, K. Schallenberg, K. Knocke, Measurement of the kinetics of water vapor
adsorption into solid zeolite layers, Sci. Tech. Froid 1 (1992) 246–250.
[21] Z.Z. Xia, et al., Adsorption equilibrium of water on silica gel, J. Chem. Eng. Data 53
(10) (2008) 2462–2465.
[22] İ. Solmuş, et al., Transient behavior of a cylindrical adsorbent bed during the ad-
sorption process, Appl. Energy 142 (2015) 115–124.
[23] İ. Solmuş, et al., Numerical investigation of coupled heat and mass transfer inside
the adsorbent bed of an adsorption cooling unit, Int. J. Refrig. 35 (2012) 652.
[24] İ. Solmuş, et al., A two-energy equation model for dynamic heat and mass transfer
in an adsorbent bed using silica gel/water pair, Int. J. Heat Mass Transf. 55 (19)
(2012) 5275–5288.
[25] J. Di, et al., Theoretical and experimental study on characteristics of a novel silica
gel–water chiller under the conditions of variable heat source temperature, Int. J.
Refrig. 30 (3) (2007) 515–526.
[26] Y.I. Aristov, et al., Kinetics of water adsorption on silica Fuji Davison RD, Micropor.

375
R.H. Mohammed et al.
Mesopor. Mater. 96 (1) (2006) 65–71.
[27] S. Dervin, et al., Graphene oxide reinforced high surface area silica aerogels, J. Non-
Cryst. Solids 465 (2017) 31–38.
[28] T. Sato, K. Nakatani, Analysis of distribution and intraparticle diffusion of a
fluorescent dye in mesoporous silica gel by confocal fluorescence microspectro-
scopy, Anal. Sci. 33 (2) (2017) 179–183.
[29] M. von der Lehr, et al., Mesopore etching under supercritical conditions–a shortcut
to hierarchically porous silica monoliths, Micropor. Mesopor. Mater. 243 (2017)
247–253.
[30] D.M. Ruthven, Principles of Adsorption and Adsorption Processes, John Wiley &
Sons, 1984.
[31] R.H. Mohammed, et al., Revisiting the adsorption equilibrium equations of silica
gel/water for adsorption cooling applications, Int. J. Refrig. 86 (2018) 40–47.
[32] R.H. Mohammed, et al., Performance evaluation of a new modular packed bed for
376
Applied Thermal Engineering 137 (2018) 368–376
adsorption cooling systems, Appl. Therm. Eng. (2018).
[33] R.H. Mohammed, et al. Experimental and numerical investigation of a new silica-
gel/water packed bed for adsorption cooling applications, in: 3rd Thermal and
Fluids Engineering Conference (TFEC), ASTFE: Fort Lauderdale, FL, USA, 2018.
[34] E.S. Ali, et al., Weather effect on a solar powered hybrid adsorption desalination-
cooling system: a case study of Egypt’s climate, Appl. Therm. Eng. (2017).
[35] A.S. Alsaman, et al., Performance evaluation of a solar-driven adsorption desali-
nation-cooling system, Energy 128 (2017) 196–207.
[36] R.H. Mohammed, A simplified method for modeling of round and square ceiling
diffusers, Energy Build. 64 (2013) 473–482.
[37] M.A. Aziz, et al., Experimental and numerical study of influence of air ceiling dif-
fusers on room air flow characteristics, Energy Build. 55 (2012) 738–746.
[38] R.H. Mohammed, Numerical investigation of displacement ventilation effective-
ness, Int. J. Mech. Aero. Ind. Mech. Eng 8 (2014).