Batch and fixed-bed-column studies for Methylene Blue

removal and recovery by untreated coffee residues

Dye removal by coffee residues

ABSTRACT

This paper conduces to the Industrial Ecology concept through the use of an everyday

urban solid waste, i.e., coffee residues, to treat industrial wastewaters colored with

basic dyes, e.g., Methylene Blue. The result of various batch system parameters

(stirring speed, pH, initial dye concentration and contact time) was investigated.

Further, in continuous fixed-bed column systems, the effects of bed-depth, flow rate

and initial dye concentration were investigated. The data obtained from the

experimental batch system were replicated with the use of (a) Freundlich, Langmuir

and Sips isotherm models and (b) 1st order, 2nd order, and intra-particle diffusion

kinetic models. For the data obtained from the continuous fixed-bed column system,

two common models were implemented, i.e. (a) Bohart and Adams and (b) Clark. The

Bohart and Adams capacity was up to N = 46166 mg L-1 or q0 = 104.5 mg g-1 for bed-

depth 15 cm, initial dye concentration 800 mg L-1 and flow rate 20 mL min-1. The

outcome showed that the Methylene Blue is adequately adsorbed on coffee residues.

As a result the process in question can be adopted as a low cost technique for cleaning

basic dyes from the aquatic environment.

1
Keywords: wastewaters, methylene blue, coffee residues, adsorption, desorption,

column.

Introduction

Many industrial plants, such as plastics, paper, food, cosmetics, printing, textile and

leather use dyes to color their products (Robinson and al., 2001). In textile industries

about 10–15% of the dye gets lost in the dyeing process effluent which are harmful

products and may stimulate cancer epidemics (Sparado and al., 1992- Mastrangelo

and al., 2002). Dyes usually are of synthetic origin and complex aromatic molecular

structures that make them more stable and difficult to biodegrade (Sparado and al.,

1992- Robinson and al., 2001- Mastrangelo and al., 2002- Song and al., 2011).

Normally the industrial runoffs are disposed of into rivers and lakes, thus altering the

biological balance of surrounding ecosystems (Deng et al., 2011). As a result, over

the past decades, particular attention has been given to the removal of dyestuffs from

wastewater.

In treating wastewater, various methods are applied to remove dyes that include

photocatalytic degradation (Sohrabi and Ghavami, 2008), membrane separation

(Ciardelli and al., 2001), chemical oxidations (Tang and al., 2012) and

electrochemical process. Among the techniques of dye removal mentioned above, the

adsorption process can be used to remove different types of coloring materials as it

gives the best results (Rafatullah and al., 2010).

2
Adsorption onto activated carbon (Kaouah and al., 2013 – Sathishkumar at al., 2012)

is the technology most widespread for removal of pollutants from water and

wastewaters. The drawback of activated carbon is its high cost (El-Sayed and al.,

2011). Therefore, it is of paramount importance thence of low cost substitute

absorbents to replace activated carbons. Various types of untreated biomass are

considered to have a use in dye removal: peanut husk (Song and al., 2011), palm

kernel fibre (El Sayed and al., 2011), Turbinaria turbinate alga (Altenor, and al.,

2012), Caulerpa lentillifera (Punjongharn et al., 2008), defatted jojoba (Al-Anber and

al., 2011), Shorea dasyphylla sawdust (Hanafiash et al., 2012) grapheme (Liu and al.,

2012), and sugar bet pulp (Malekbala and al., 2012). Table 1 shows the sorption

efficiency of several adsorbents.

Additionally, numerous pretreated lignocellulosic materials are employed to remove

dyes in water and wastewater. Pyrolyzed date pits (Theydan and Ahmed, 2012), date

stones (Ahmed and Dhedanb, 2012) and Turbinaria turbinate alga (Altenor and al.,

2012) have been determined as effective adsorbents.

Tests have been made with coffee residues and other similar byproducts like coffee

husks and coffee grounds to remove various dyes and heavy metals from aqueous

solutions. Coffee husks have been used for the removal of Cu (II), Cr (VI), Cd (II) and

Zn (II) (Oliveira and al., 2008), coffee grounds for Nylosan Red N-2RBL (Reffas

and al., 2010) and degreased coffee bean for Malachite Green (Baek and al., 2010).

Kyzas et al. (2012) reported the removal of Remazol Brilliant Blue RN and Basic

Blue 3G using coffee residues, focusing on batch processes and not on continuous

systems (Kyzas and al., 2012).

3
The adsorption capabilities of coffee residues to remove Methylene Blue (MB) from

aqueous solutions were investigated in the current study. Both batch and continuous

fixed-bed-column systems were reviewed. With regard to the fixed-bed-column

systems filled with coffee residues, various bed-depths, flow rates and initial dye

concentrations were taken into account for the first time.

2. Materials and Methods

2.1 Adsorbents

Coffee residues, used as adsorbent, is a low-cost material acquired from different

cafeterias in the city of Piraeus (Greece). The subject adsorbent was washed with

distilled watern then dried at 110oC for 24 hrs.

2.2 Adsorbate

Methylene Blue dye (C16H18ClN3S. 3H2O, molecular weight = 373.90 10-3 kg mol-1)

was used herein in batch and column experiments, supplied by Sigma-Aldrich. We

prepared a stock solution by dissolving a specific amount of MB (humidity 22%) in

distilled water. Working solutions were 3 - 140 mg L-1. MB concentrations were

determined by measuring the absorbent values in each experiment with HACH

DR4000U UV-VIS spectrophotometer at λ=664nm.

2.3. Adsorption isotherm studies

4
Βatch experiments gave isotherms. Quantities of approximately 0.5g coffee residues

accurately weighted, were transferred to 0.8-L bottles, where 0.5 L adsorbate solution

was added. The temperature was 23 o C, the MB initial concentration ranged from C0

= 1.5 mg L-1 to 150 mg L-1. The bottles, sealed, were mechanically agitated for a

period of 7 days. A 7-day period was figured after optimization analysis. Agitation

periods varied from 4 hrs to 14 days, so that nearly equilibrium conditions were

reached. The final concentrations were determined. Concentrations (before and after

equilibrium) from each bottle represented one point on the adsorption isotherm plots.

The pH was near 7.5.

2.4. Isotherm equations

To fit the experimental data three isotherm models, including Freundlich (1906),

Langmuir (1916) and Sips (1948) equations, were used. Freundlich isotherm model

presumes that the surface of absorbent is heterogeneous and polymolecular layer

adsorption takes place. This model can be described by the equation:

1
q  K F  (Ce ) n (1)

q represents the amount adsorbed per unit mass of the adsorbent (mg g-1).

Ce is the equilibrium concentration of MB adsorbed (mg L-1).

KF [(mg g-1)(L mg-1)1/n] is the Freundlich isotherm constant related to adsorption

capacity and n is the Freundlich isotherm constants connected with adsorption

intensity. Where the isotherm experimental data approximates the Freundlich

equation, the parameters KF and n can be calculated either by plotting logq versus

logCe either by non-linear regression analysis (NLRA).

5
The linear form of the Freundlich (1906) isotherm model can be determined by the

following equation:

1
log q  log K F  log Ce (2)
n

The Langmuir (1916) isotherm model is given as:

K L qmCe
q (3)
1  K LCe

Moreover, this equation in linearized form is:

1  1   1   1 
    (4)
q  qm   K L  qm   Ce 

KL represents the Langmuir constant related to the adsorption energy (L mg-1).

qm is the amount of MB adsorbed (mg g-1) when saturation is achieved. Where the

isotherm experimental data approximates the Langmuir equation, the parameters KL

and qm can be estimated either by plotting 1/q versus 1/Ce either by NLRA.

Another important parameter, RL, namely the separation factor, is incorporated in

multiple studies. The value of RL indicates the type of the isotherm to be either

unfavorable (RL > 1), linear (RL = 1), favorable (0 < RL < 1) or irreversible (RL = 0) and

is expressed by the following equation:

6
 1
RL  (5)
1  K L  C0

C0 is the initial dye concentration (mg L-1) and KL is the Langmuir constant (L mg-1).

Sips (1948) isotherm model is a combination of the Langmuir (1916) and Freundlich

(1906) isotherm type models and is considered to better describe heterogeneous

surface. The Sips equation is presented by:

q m   K L  Ce  1 / n
q (6)
1   K L  Ce  1 / n

where KL (L mg-1), qm (mg g-1) is the Langmuir constants and n the Freundlich (1906)

constant. NLRA can estimate these parameters.

The Standard Error of Estimate (SEE) was calculated in each of the following cases:

n'
SEE  ( y
i 1
i  yi ,theor ) 2 /(n' p' ) (7)

yi represents the experimental value of the depended variable

yi,theor represents the theoretical or estimated value of the depended variable

n΄ is the number of the experimental measurements and

p΄ is the number of parameters (the difference n –p΄ being the number of the degrees

of freedom).

7
2.5. Kinetic studies

Batch experiments were carried out at stirring speeds from 65 to 664 rpm, pH differed

from 1.54 to 12.9, the adsorption temperature varied from 23 to 70 oC, initial dye

concentration from 3 to 140 mg L-1 for a maximum contact time up to 95 min.

Samples were taken at 5 minutes intervals and the MB concentration was measured.

The reactor, containing V = 1 L aqueous solution of dye was placed in a water bath to

maintain constant temperature at the desired level.

2.6. Kinetic equations

To make adsorption mechanism clearer, the adsorption data were fitted using the

pseudo-first order, the pseudo-second order and the intra-particle diffusion kinetic

model.

2.6.1. Pseudo-first order kinetic model

The Lagergren (1898) non-linear pseudo-first order equation is given here:

q  qt  q  e  k t (8)

q and qt (mg g-1) represent amounts of MB dye adsorbed per unit mass of the

adsorbent at equilibrium time ( t   ) and time of adsorption t(min), respectively,

while k (min-1) is the pseudo-first order rate constant for the adsorption process.

Additionally:

8
q  ( C0  Ce )V / m and qt  ( C0  C )V / m (9)

C, C0 , Ce (mg L-1) represent the MB concentrations in the bulk solution at time t, 0,

and  , respectively, while m (g) stands for the weight of the adsorbent used, and V
(mL) is the solution volume. Further modification of eq. (9) in logarithmic form gives:

ln( q  qt )  ln q  k  t (10)

The rate constant k of the Lagergren model follows the Arrhenius law:

k  p  exp(  E / RT ) (11)

E represents the activation energy (in kJ mol-1) for MB adsorption on untreated coffee

residues and p is the frequency factor (in min-1).

2.6.2 Pseudo-second order kinetic model

The equation of pseudo-second order (Ho and al., 2000) equilibrium adsorption model

is given as:

1
qt  q  q 1
 k 2t  1
or
qt  q 
1
 k 2t (12)
q

where k2 (min-1) is the rate constant of second order adsorption.

2.6.3 Intra-practice diffusion kinetic model

9
According to the theory proposed by Weber and Morris (1963), the intra-particle

diffusion model, is expressed as:

qt  c  k p  t (13)

where kp [mg (g min0.5)-1] represents the intra-particle diffusion rate constant and c (mg

g-1) constant is related to the thickness of boundary. A c value near zero signifies that

the only controlling step of the adsorption process is diffusion.

2.7. Continuous fixed-bed column studies

Experiments of continuous-flow were made on Stainless steel columns with

dimensions 15 X 2.5 and 25 X 2.5cm. The bed height was x = 15 cm and 25 cm,

respectively. The adsorbent weight was m = 32 g and 54 g, respectively. The pH was

8.0. The flow rates were fixed at approximately 10, 20, 40, 70 and 80 mL min-1 using

a preparative HPLC pump, LaPrep P110 - VWR - VWR International. The initial

concentrations of MB were 165, 800 and 1600 mg L-1. To determine the MB

concentration in the effluent, samples of outflow were taken at 100 mL intervals.

2.8. Continuous fixed-bed-column adsorption models

Bohart and Adams (1920) established a widely used continuous fixed-bed-column

model. They assumed that the adsorption rate is controlled by the surface binding (by

means of chemical reaction or physical interaction) between unused solid capacity and

adsorbate, i.e., adsorption rate = K.C.Cu, where K is the coefficient of adsorption rate,

10
C is the concentration of adsorbate at the solid phase at distance x, and Cu is the

surface adsorptive capacity unused at time t, expressed as mass per volume of bed.

The material balance for adsorbate is given by the partial differential equation

Cu
  K  C  Cu (14)
t

while the corresponding partial differential equation for the Cu decrease is

C K
   C  Cu (15)
x u

where u is the superficial liquid velocity. The equations above are obtained

disregarding diffusion and accumulation terms, hypotheses that are valid in chemical

engineering practice, on condition that strict scale up specifications are applied in the

design stage and successful operation conditions are applied in the industrial operation

stage.

The differential equations can be applied over the total length x of the bed to give:

C   K . N .x 
ln i  1  ln[exp   1]  K  Ci  t (16)
C   u 

where N (mg L-1); is the initial or total adsorption capacity coefficient, also quoted as

Cu,0 (Bohart and Adams,1920); C=effluent concentration (mg L-1); Ci=influent

concentration (mg L-1); K=adsorption rate coefficient (L mg-1 min-1); x=bed depth

11
(cm); u=linear velocity (cm.min-1); and t=time (min). Given that exp( K .N .x / u ) is

commonly much greater than unity, this equation can be reduced to:

C  KNx
ln i 1   K  Ci  t (17)
C  u

which the researchers usually adopt, for its convenience in estimating the values of

parameters K and N through linear regression either of ln[(C0 / Ci )– 1] vs t or t vs x

when the following suffling is adopted:

Nx 1  C 
t    ln i  1 (18)
C i  u K  Ci  C 

t, in this array, is the time to breakthrough, i.e., the time period needed for

concentration to reach a specific value. To use the last expression as a linear

regression model, wastewater passed through beds of varying depths, keeping

constant Ci and u, at values similar to those expected to prevail under actual

conditions at full scale. Differently, it can be performed with the aid of at least three

columns placed in series. In such a case, samples are taken at the bottom of each

column and measured for adsorbate concentration, increasing the frequency of

measurements when approaching the breakthrough concentration C. The time at

which the effluent reaches this concentration is used as the dependent variable while x

plans the role of the independent one. Apparently, using such a regression model

indicates the additional error to measure the independent variable with less precision

compared to the dependent. A common error in both models comes from the

estimation of concentration from measuring adsorbance although the reference

12
curve/relation has been drawn/structured in the inverse mode, i.e., for predetermined

concentrations the respective adsorbances have been measured.

In this work we used the model of eq. (17) for estimation of parameter values through

linear regression to get numerical results comparable with respecrtive data found for

other fixed bed adsorption studies in literature. The non-linear form of this model is:

Ci
C (19)
1  Ae  rt

where A  e K  N  x / u ; r  K  Ci .

On the other hand, Clark (1987) has advanced the Bohart and Adams (1920) model by

incorporating the parameter n of the Freundlich adsorption isotherm. We used the

model of eq. (19) for parameter values estimation through NLRA to get more reliable

numerical results, i.e., with a lower total standard error of estimate (SEE). In this last

case, the regression model is written under the form for parameter identification

through Ci value estimation (a) endogenously and (b) exogenously.

1
 Ci n 1  n 1
C  rt 
(20)
1  Ae 

n=inverse of the slope of the Freundlich (1906) isotherm.

The Bohart and Adams model (1920) can be reduced for n=2 from Clark model

(1987).

2.9. Desorption studies

13
At the end of each one of the above sorption experiments, we carried out desorption

tests using distilled water as influent. The flow rate was fixed at about 20 mL min-1.

To determine the concentration of MB in the effluent, outflow samples were taken at

intervals of 100 mL.

The kinetic equation used for desorption is the following:

C  C '0 e  k 't ' (21)

C’0 represents the initial MB concentration of desorption effluent, k’ is desorption rate

constant assuming first order desorption kinetics and t’ represents desorption time.

2.10. Physico-chemical characterization

The coffee residues study by scanning electron microscopy (SEM) was made at the

Institute of Materials Science, National Centre for Scientific Research ‘Demokritos’,

using an FEI INSPECT SEM.

The concentration of output solution was measured at λ=664 nm and using HACH

DR4000U UV-vis spectrophotometer.

Finally, pH measurements were made using a digital pH meter, MultiLab model 540.

14
3. Results and Discussion

3.1. Microstructure

Fig. 1 presents the SEM micrographs for coffee residues before and after MB

adsorption. The coffee residues particle in Fig. 1(a) with magnification 7500X shows

some pores. This particle surface texture is relatively rough as it can be seen in Fig

1(b) with magnification 30000X. The conditions for adsorption were C0=140 mg L-1,

t=95 min, m/V=1 g L-1, agitation speed=664 rpm. The coffee residues particles texture

after MB adsorption in Fig. 1(d) is rougher compared to the same material before MB

adsorption, shown in Fig. 1(b). This condition indicates the swelling effect on the

lignocellulosic particles after MB adsorption.

3.2. Isotherms results

Fig. 2 presents MB adsorption isotherms experimental data for untreated coffee

residues. An estimation is made for the theoretical curves of the three models

mentioned above. Table 1 shows the parameter values of these models. These

parameters were obtained by NLRA.

Fitting of the Sips model of adsorption to the present data was the most satisfactory

for MB adsorption, against the other two isotherm models, as shown by the

corresponding SEE-values given in Table 1. Fitting of the Langmuir’s adsorption

model to the present data was satisfactory also, yet to a lesser degree than the Sips

15
model. We found the RL values to be 0.236-0.935, i.e. 0 < RL < 1 for all MB

concentrations C0 in the range of 3-140 mg L-1.

3.3. Kinetic results

3.3.1. Effect of stirring speed

The stirring speed used at the study was between 65 and 664 rpm at constant dye

concentration 14 mg L-1, pH equals 7, stirring time adopted was 95min. Fig. 3 shows

the correlation between the first-order rate constant k for the MB absorption by coffee

residues. Generally, the rate constant increases with stirring speed up to 200 rpm were

a level-off is reached.

3.3.2 Effect of pH

The pH of the aqueous solution is one of the most important parameters of the

adsorption process, according to the literature. Fig. 4 illustrates the effect of the initial

pH on the MB adsorption onto coffee residues. The scalke of pH studied was from 2

to 12. As can be seen the rate constant increases linearly by pH increasing.

3.3.3 Effect of initial MB concentration

Fig. 5 shows the effect of initial MB concentration on adsorption on coffee residues.

The kinetic parameters of the above three kinetic models were estimated using NLRA

and are presented in Table 2. The SEE values are lower for the intra-particle diffusion

16
kinetic model indicating better fitting to the experimental data. In addition the pseudo-

second order kinetic model had better fitting than the pseudo-first one.

3.3.4 Effect of temperature

The variation in temperature from 20 to 70 oC on MB uptake by coffee residues

showed that increase in temperature of the process improved the adsorption of MB

from bulk solution. The conditions under which the experiments were conducted with

different temperature were pH = 8, the initial concentration was 14 mg L-1, the stirring

speed was 600 rpm and the dose of coffee residues was 1 g. The system entropy was

increased with temperature increase , which enhanced the chances of adsorption. Fig.

6 shows the change in adsorption rate constant k as a function of adsorption

temperatures.

Estimation of the activation energy E (kJ mol-1) of eq. (11) for the adsorption of MB

on untreated coffee residues can be made by linear regression of lnk on 1/T. This

activation energy was found to be 27.4 kJ mol-1 or 6.57 kcal mol-1. The frequency

factor of eq. (11) was p= 1236 min-1. Therefore, the physical process of intra-particle

diffusion, is probably the controlling step of the adsorption process.

3.4. Continuous fixed-bed-column results

Continuous fixed-bed-column systems have been investigated in this study. The

column arrangements were filled with coffee residues at various initial dye

concentrations, flow rates and bed-depths.

17
3.4.1. Effect of initial concentration

Investigation of the effects of initial MB concentration has been made at 165-1600 g

L−1, respectively. The bed height was 15 cm and the temperature was 23 oC. The flow

rate was fixed at 20 mL min−1 and the pH of MB solution was 8. Fig.9 presents the

breakthrough curves, plotted in according to the Bohart and Adams model. Fig. 9

illustrated that breakthrough time decreased with increasing MB concentration.

3.4.2. Effect of flow rate

One of the most important characteristics in evaluating sorbents for continuous

treatment of dyes effluents on an industrial scale is flow rate (Zhao and al., 1999). The

flow rate effect in the fixed bed column, filled with coffee residues, was investigated

varying the flow rate from 10–80 mL min−1 with bed depths held constant at 15 cm.

The pH was 8. Concentration of the influent MB in the feed was 165 mg L−1. The N

adsorption capacity was higher at lower flow rate values, as shown in Fig.10 and

Table 3.

The residence of the adsorbate was longer at lower flow rate and hence the adsorbent

got more time to bind with the dye efficiently (Malkoc and Nuboglu, 2006). This

means that if the residence time of the solution in the column is not large enough for

the adsorption equilibrium to be reached at the given flow rate, the dye solution leaves

the column before the equilibrium takes place. It was observed that the adsorbent got

saturated easily at higher flow rates. The MB absorption decreased with increase in

flow rate. Breakthrough time increased with decrease in flow rate. Higher flow rates

saturated he breakthrough curve earlier since the front of the adsorption zone quickly

18
reached the top of the column. In contrast, longer contact tim and lower flow rate

resulted in a shallow adsorption zone (Vinodhini and Nilanjana Das, 2010).

3.4.3 Effect of bed height

To determine the effect of bed height on the breakthrough curve, the adsorbate MB

solution (initial MB concentration 165 mg L−1, pH 8) was passed through the

adsorption column at a flow rate 70 mL min−1 by varying the bed height. Fig. 11

shows the breakthrough curves at bed heights of 15 and 25 cm.

The results revealed slight difference of the shape and gradient of the breakthrough

curve with the variation in the bed depth. Concentration of MB in the effluent

increased fast after the breakthrough point. As shown in Fig.11 the lower bed depth

(15 cm) gets saturated earlier than the higher bed depth (25 cm).

3.4.4. Continuous fixed-bed-column models parameters

As presented in Figs. 9-11, the values of parameters A and r according to (i) the

Bohart and Adams model (1920) and (ii) the Clark model (1987) were estimated by

NLRA from the column effluent data for all the cases, through equations (19) and

(20), respectively. For calculation of the parameter K and N values after having

performed NLRA the expressions K=r/Ci and N=u.lnA/(x.K)= Ci.u.lnA/(x.r) are used.

The effluent dye solution volume V (in L) is V=Q.t, where Q is the dye solution flow

rate. According to the Bohart and Adams model (1920) the theoretical estimations

19
sufficiently simulate the experiments data in the examples shown in Figs. 9-11. The N

and K values are presented in Table 3. The SEE-values are given in the same Table.

In accordance with them, the fitting of the Bohart and Adams model to experimental

data was found to be better than that of the Clark model. The parameter n

incorporated in the Clark model (1987) of the Freundlich adsorption isotherm, shown

in Table 3, was taken from Table 1 results.

3.5. Continuous fixed-bed-column desorption results

The experimental adsorption/desorption data given in Fig. 12 show that desorption

was approximately 17% of the maximum adsorbed MB quantity which was q0=83.0

mg g-1. Here, the bed depth was x=15, the flow rate was Q=20 mL min-1, the MB

initial concentration was Ci=1600 mg L−1 for adsorption and equal to zero for

desorption (i.e., we used pure water for desorption). The theoretical adsorption curve

in this figure is according to the Bohart and Adams model, while the theoretical curve

of desorption is according to the eq. (21).

3.6. Discussion.

Our research aimed at studying the adsorption capacity of the coffee residue and its

application to industrial scale. There is scope for further research to specify the coffee

residues quantity by region distribution and corresponding exploitation by industrial

scale unit.

The capacity of various lignocellulosic materials to absorb Methylene Blue found in

20
the literature was compared to the coffee residues adsorption capacity estimated in the

present work (see Table 4). The adsorption capacity of other waste biomass such as

peanut husk (Al-Anber and al., 2011), turbinaria turbinate alga (Altenor and al.,

2012) was inferior to that of the coffee residue and wheat straw (Batzias and al.,

2009). Table 5 shows the adsorption capacities of various coffee residues/byproducts

found in the literature. The coffee residues adsorption capacity found in the present

work was close to the average of the Freundlich and Langmuir capacities, i.e, KF and

qm respectively, reported in the literature.

Our Research Group, as the present study shows, is continually involved up in the

study of wastewater treatment using original and modified (pretreated) lignocellulosic

biomass as potential adsorbents. A previous study of our Group (Sidiras and al., 2011)

gave Langmuir capacity qm=38.7 mg g-1 for pine sawdust, significantly lower

comparing to qm =78.9 mg g-1 for the coffee residues that were the subject of this

work. Further, gave Bohart-Adams capacity N = 8046 mg L-1 for pine sawdust,

notably lower comparing to N = 46166 mg L-1 for the coffee residues. Therefore,

coffee residues are of great interest for industrial scale application, particularly in

countries producing coffee like Brazil (Oliveira and al., 2008-, Oliveira and al., 2008-

Oliveira and al., 2008). The adoption of coffee residues could provide a solution,

considering its ample availability as a common waste in urban and rural environment.

Finding, therefore, a solution to the problem of wastewater cleaning is the challenge

of our times.

Average coffee consumption in Greece amounts to 841,000 bags, out of which

359,000 bags of roasted coffee and 483,000 bags of soluble coffee. In the near future

21
consumption, with very small annual growth rates, is expected to present stability. The

greek market is growing at a very slow rate which makes it clear that we are going

through a period of maturity. The adoption by Greek consumer the taste of the

Europeans in recent years has significantly affected coffee consumption positively.

Currently coffee is available packaged or in bulk and made available directly or

indirectly (through wholesalers) in retailing (mainly supermarkets), in coffee shops

and foodservice (cafeterias, hotels, etc).

We have also surveyed the readiness of storeowners to get involved with municipal

schemes of gathering coffee residues, thus endorsing an environmental effort and

minimizing the volume of their waste. As for the drying of the residues, this can be

done by exposure to sunlight, notably in Mediterranean and tropical countries. Drying

in the sun is a low energy consuming method, however a wide area is required to

spread the biomass, which increases the overall cost. This drawback is balanced by

the high absorbing capacity of coffee residues. Consequently, the importance of coffee

residues as a potential industrial adsorbent is apparent.

The coffee residues are easy to collect and process. They constitute a large percentage

of the municipal waste that can be transported without any treatment or separation at

source, i.e. at cafeterias in the city. Another promising source to obtain coffee residues

is the coffee industry. Extra processes though for collecting the waste in industrial

scale would decrease the coffee production cost due to scale economies. However the

quantity of this waste represents only a segment of the overall coffee distributed

commercially. The coffee residues can be gathered by store owners in special bags to

22
be collected directly by the absorbent producing industry. The municipality of the city

can help in the process of gathering and transporting the coffee residue by

coordinating a special network. This would provide an opportunity for citizens to

show their readiness to support environmental protection through a process of urban

waste management. With participation of the people and municipal authorities a

reverse supply chain network can be formulated, in which, the taken material is

derived from the gathering and transfer of urban wastes. No coffee waste management

has been applied to date, which ends in municipal solid waste without any treatment.

The proposed use of coffee residues guarantees a double benefit. First, using

economic and technical criteria, the calculation of the radius of influence of the

material should be measured and secondly, transport costs can be treated with an

unprecedented spatial resolution, which exploits the material from local industrial

plants.

4. Conclusions

An innovative technique uses coffee residues within the framework of waste biomass

management from urban and rural areas for wastewater cleaning in the Industrial

Ecology sector. Further, that is an indicator of the awareness of citizens for the

prudent management of waste and their contribution to sustainable development. Our

work proves the viability of using coffee residues to remove basic dyes like MB from

wastewaters . The result of different system parameters on (a) batch and (b)

continuous fixed-bed column systems were studied. The data collected from

experimental systems were simulated with the aid of the most commonly used

isotherm and kinetic models. The results of continuous fixed-bed column proved that

23
the MB is actually adsorbed on coffee residues giving maximum Bohart and Adams

capacity N = 46166 mg L-1 or q0 = 104.5 mg g-1 for bed-depth 15 cm, initial dye

concentration 800 mg L-1 and flow rate 20 mL min-1. This outcome allow us to

recommend this low cost residue for removal of basic dyes in industrial scale

applications.

Acknowledgments

The authors would like to acknowledge Senior Researcher N. Boukos at the Institute

of Materials Science of the National Center for Scientific Research ‘Demokritos’ for

the SEM measurements.

References

Ahmad, M.A., Rahman, N.K., 2011. Equilibrium, kinetics and thermodynamic of

Remazol Brilliant Orange 3R dye adsorption on coffee husk-based activated

carbon. Chem. Eng. J. 170, 154–161.

Ahmed, M.J., Dhedanb, S.K., 2012. Equilibrium isotherms and kinetics modeling

of methylene blue adsorption on agricultural wastes-based activated carbons.

Fluid Phase Equilibr. 317, 9–14.

Al-Anber, Z.A., Al-Anber, M.A., Matouq, M., Al-Ayed, O., Omari, N.M., 2011.

Defatted Jojoba for the removal of methylene blue from aqueous solution:

Thermodynamic and kinetic studies. Desalination. 276, 169–174.

24
Altenor, S., Ncibi, M.C., Emmanuel, E., Gaspard, S., 2012. Textural

characteristics, physiochemical properties and adsorption efficiencies of

Caribbean alga Turbinaria turbinata and its derived carbonaceous materials for

water treatment application, Biochem. Eng. J. l 67, 35– 44.

Baek, M.H., Ijagbemi, C.O., Se-Jin, O., Kim, D.S., 2010. Removal of Malachite

Green from aqueous solution using degreased coffee bean. J. Hazard. Mater. 176,

820–828.

Batzias, F., Sidiras, D., Schroeder, E., Weber, C., 2009. Simulation of dye

adsorption on hydrolyzed wheat straw in batch and fixed-bed systems. Chem. Eng.

J. 148, 459–472.

Bohart G., Adams E.N., 1920. Some aspects of the behavior of charcoal with

respect to chlorine. J. Am. Chem. Soc. 42, 523–544.

Ciardelli, G., Corsi, L., Marcucci, M., 2001. Membrane seperation for watewater

reuse in the textile industry. Resour. Conserv. Recycl. 31, 189-197.

Clark, R.M., 1978. Modeling TOC removal by GAC: The general logistic

function. J. Am. Wat. Works Assoc. 79, 33-37.

Deng, H., Lu, J., Li, G., Zhang, G., Wang, X., 2011. Adsorption of methylene

blue on adsorbent materials produced from cotton stalk. Chem. Eng. J. 172, 326–

334.

25
Dural, M.U., Cavas, L., Papageorgiou, S.K., Katsaros, F.K., 2011. Methylene blue

adsorption on activated carbon prepared from Posidonia oceanica (L.) dead leaves:

Kinetics and equilibrium studies. Chem. Eng. J. 168, 77–85.

El-Sayed, A.M., Mitchell,V., Manning,L.A., Cole,L., Suckling, D.M., 2011. New

sex pheromone blend for the lightbrown apple moth, Epiphyas postvittana. J.

Chem. Ecol. 37, 640-646.

Freundlich, H.M.F., 1096. Über die adsorption in lösungen, Zeitschrift für

Physikalische. Chemie. 57, 385-471.

Hanafiash, M., Wan Ngah, W.S., Zolkafly S.H., Teong L.C., Majid Z., 2012. Acid

Blue 25 adsorption on base treated Shorea dasyphylla sawdust: Kinetic, isotherm,

thermodynamic and spectroscopic analysis. J.E.S. 24(2), 261-268

Ho, Y.S., Ng J.C.Y., McKay, G., 2000. Kinetics of pollutants sorption by

biosorbents: review. Sep. Purif. Methods. 29, 189-232.

Kaouah, F., Boumaza, S., Berrama, T., Trari, M., Bendjama, Z., 2013. Preparation

and characterization of activated carbon from wild olive cores (oleaster) by H3PO4

for the removal of Basic Red 46. J. Clean. Prod. 54, 296-306.

Kyzas, G.Z., Lazaridis, N.K., Mitropoulos, A.CH., 2012. Removal of dyes from

aqueous solutions with untreated coffee residues as potential low-cost adsorbents:

26
Equilibrium, reuse and thermodynamic approach. Chem. Eng. J.189– 190, 148–

159.

Lagergren, S., 1989. Zur theorie der sogenannten adsorption gelöster stoffe,

Kungliga Svenska Vetenskapsakademiens. Handlingar 24, 1-39.

Langmuir, I., 1916. The constitution and fundamental properties of solids and

liquids. J. Am. Chem.Soc. 38, 2221-2295.

Liu, T., Li, Y., Du, Q., Sun, J., Jiao, Y., Yang, G., Wang, Z., Xia, Y., Zhang, W.,

Wang, K., Zhu, H., Wu, D., 2012. Adsorption of methylene blue from aqueous

solution by grapheme. Colloids and Surf. B: Biointerfaces. 90, 197– 203.

Malekbala, M., Hosseini, S., Yazdi, S.K., Masoudi Soltani, S., 2012. The study of

the potential capability of sugar beet pulp on the removal efficiency of two

cationic dyes. Chem. Eng. Res. Des. 90, 704–712.

Malkoc, E., Nuhoglu, Y., Abali, Y., 2006. Cr (VI) adsorption by waste acorn of

Quercus ithaburensis in fixed beds: prediction of breakthrough curves. Chem.

Eng. J. 119, 61–68.

Mastrangelo, G., Fedeli, U., Fadda, E., Mila, G., Lange, J., 2002. Epidermiologic

evidences of cancer risk in textile industry workers: a review and update. Toxicol.

Industr. Health 18, 171-181.

27
Nashua, N., Hameed, B.H., 2011. Adsorption of methylene blue from aqueous

solution onto NaOH-modified rejected tea. Chem. Eng. J. 166, 783–786.

Oliveira, L.S., Franca, A.S., Alves, T.M., Rocha, S.D., 2008. Evaluation of

untreated coffee husks as potential biosorbents for treatment of dye contaminated

waters. J. Hazard. Mater. 155(3), 507-512.

Oliveira, L.S., Franca, A.S., Camões, J.G.S., Oliveira, W.E., 2008. Low-cost

adsorbents from chemically modified coffee husks for the treatment of wastewater

containing heavy metals. J. Biotechnol. 136, S655.

Oliveira, W.E., Franca, A.S., Oliveira, L.S., Rocha, S.D., 2008. Untreated coffee

husks as biosorbents for the removal of heavy metals from aqueous solutions. J.

Hazard. Mater.152, 1073–1081.

Punjongharn P., Meevasana K., Pavasant P., 2008. Influence of particle size and

salinity on adsorption of basic dyes by agricultural waste: dried Seagrape

(Caulerpa lentillifera). J.E.S. 20, 760-768.

Rafatullah, M., Sulaiman, O., Hashim, R., Ahmad, A., 2010. Adsorption of

methylene blue on low-cost adsorbents: A review. J. Hazard. Mater. 177, 70–80.

Reffas, A., Bernardet,V., David,B., Reinert, L., Bencheikh Lehocine,M., Dubois,

M., Batisse, N., Duclaux, L., 2010. Carbons prepared from coffee grounds by

H3PO4 activation: Characterization and adsorption of methylene blue and Nylosan

Red N-2RBL. J. Hazard. Mater. 175, 779–788.

28
Robinson, T., Mcmullan, G., Marchant, R., Nigam, P., 2001. Remediation of dyes

in textile effluent: a critical review on current treatment technologies with a

proposed alternative. Bioresour. Technol. 77, 247–255.

Sathishkumar, P., Arulkumar, M., Palvannan, T., 2012. Utilization of agro-

industrial waste Jatropha curcas pods as an activated carbon for the adsorption of

reactive dye Remazol Brilliant Blue R (RBBR). J. Clean. Prod. 22, 67-75.

Sidiras, D., Batzias, F., Schroeder,E., Ranjan,R., Tsapatsis, M., 2011. Dye

adsorption on autohydrolyzed pine sawdust in batch and fixed-bed systems Chem.

Eng. J. 171, 883–896.

Sips, R., 1948. Structure of a catalyst surface. J. Chem. Phys.16, 490-495.

Sohrabi, M.R., Ghavami, M., 2008. Photocatalytic degradation of Direct Red 23

dye using UV/TiO2: Effect of operational parameters. J. Hazard. Mater. 153,1235.

Song, J., Zou, W., Bian, Y., Su, F., Han, R., 2011. Adsorption characteristics of

methylene blue by peanut husk in batch and column modes. Desalination 265,

119–125.

Sparado, J.T., Gold, M.H., Renganathan, V., 1992. Degradation of azo dyes by

lignin degrading fungus Penicillium chrysosporium. Appl. Environ. Microbiol.

58, 2397–2401.

29
Tang, H., Yin, L., Lu, H., 2012. Synthesis, Conformations and Cell-Penetrating

Properties. Biomacromolecules. 13, 2609.

Theydan, S.K., Ahmed, M.J., 2012. Adsorption of methylene blue onto biomass-

based activated carbon by FeCl3 activation: Equilibrium, kinetics, and

thermodynamic studies. J. Anal. Appl. Pyrol. 97, 116–122.

Vinodhini, V., Das, N., 2010. Packed bed column studies on Cr (VI) removal

from tannery wastewater by neem sawdust. Desalination 264, 9–14.

Vučurović, V.M., Razmovski, R.N., Tekic, M.N., 2012. Methylene blue (cationic

dye) adsorption onto sugar beet pulp: Equilibrium isotherm and kinetic studies. J.

Taiwan Inst. Chem. E. 43, 108–111.

Wang, L., Huang, Z., Zhang, M., Chai, B., 2012. Adsorption of methylene blue

from aqueous solution on modified ACFs by chemical vapor deposition. Chem.

Eng. J. 189– 190, 168– 174.

Weber, W.J., Morris, J.C., 1963. Kinetics of adsorption on carbon from solution. J.

Sanit. Eng. Div. Am. Soc. Civ. Eng. 89, 31–60.

Zhang, W., Yang, H., Dong, L., Yan, H., Li, H., Jiang, Z., Kan, X., Li, A., Cheng,

R., 2012. Efficient removal of both cationic and anionic dyes from aqueous

solutions using a novel amphoteric straw-based adsorbent. Carbohyd. Polym. 90,

887– 893.

30
 Zhao, M., Duncan, J.R., Van Hille, R.P., 1999. Removal and recovery of zinc

from solution and electroplating effluent using Azolla filiculoides. Water Res. 33,

516–1522.

FIGURE LEGENDS

Fig. 1: SEM micrographs for coffee residues before (a, b) and after (c, d) MB

adsorption (C0=140 mg/L, t=95 min, m/V=1 g/L, 660 rpm, pH=8, 23°C).

Fig. 2: Isotherms of MB adsorption on coffee residues (C0=1.6-140 mg/L, t=7 d,

m/V=1 g/L, 660 rpm, pH=8, 23°C).

Fig. 3: k values for MB adsorption on coffee residues vs. agitation speed (C0= 12 mg

L-1, m/V=1g L-1, 65-660 rpm, pH=8, 23oC).

31
Fig. 4: k values for MB adsorption on coffee residues vs. pH (C0= 12 mg L-1, m/V=1g

L-1, 660 rpm, pH=1.5-13, 23oC).

Fig. 5: Lagergren curves of MB adsorption coffee residues as affected by MB initial

concentration (C0= 3-140 mg L-1, m/V=1g L-1, 660 rpm, pH=8, 23oC).

Fig. 6: The plots of lnk (rate constant k in min-1) for the removal of MB by adsorption

on coffee residues vs.1/T (T in K), according to the Arrhenius law (C0= 12 mg

L-1, m/V=1g L-1, 660 rpm, pH=8, 23-70oC).

Fig. 7: Second order curves of MB adsorption coffee residues as affected by MB

initial concentration (C0= 3-140 mg L-1, m/V=1g L-1, 660 rpm, pH=8, 23oC).

Fig. 8: MB amount adsorbent on coffee residues vs. t 0.5 according to the intra-

particle diffusion model (C0= 3-140 mg L-1, m/V=1g L-1, 660 rpm, pH=8,

23oC).

Fig. 9: Column experimental data and theoretical curves of MB adsorption on coffee

residues; the effluent concentration is presented vs. the effluent volume for

Ci=165-1600 mg L-1, Q=20 mL min-1, x=15 cm (the theoretical curves are

according to the Bohart and Adams model).

Fig. 10: Column experimental data and theoretical curves of MB adsorption on coffee

residues; the effluent concentration is presented vs. the effluent volume for

Q=20-80 mL min-1, Ci=165 mg L−1, x=15 cm (the theoretical curves are

according to the Bohart and Adams model).

Fig. 11: Column experimental data and theoretical curves of MB adsorption on coffee

residues; the effluent concentration is presented vs. the effluent volume for

x=15 and 25 cm, Q=70 mL min-1, Ci=165 mg L−1, (the theoretical curves are

according to the Bohart and Adams model).

32
Fig. 12: Column experimental data and theoretical curves of MB adsorption and

desorption on coffee residues; the effluent concentration is presented vs. the

effluent volume for x=15, Q=20 mL min-1, Ci=1600 mg L−1, (the adsorption

theoretical curve is according to the Bohart and Adams model).

33
Nomenclature

C concentrations of MB in the bulk solution at time t in the case of batch

adsorption process; also, the effluent concentration (mg L−1) in the case of the

column adsorption process

c intercept of the intra-particle diffusion equation (mg g−1)

C0 initial dye concentration in the case of batch adsorption process (mg L−1)

Ce equilibrium concentration of the adsorbate (mg L−1) for t→∞

Ci influent concentration (mg L−1) in the case of column adsorption process

E adsorption activation energy (kJ mol−1)

K adsorption rate coefficient (Lmg−1min−1) in the case of column adsorption

process

k first order rate constant for the batch adsorption process (min−1)

k2 second order rate constant for the batch adsorption process (min−1)

KF Freundlich constant related to adsorption capacity [(mg g−1) (L mg−1)]1/n

KL Langmuir constant related to the energy of adsorption (L mg−1)

kp intra-particle diffusion rate constant (mg g−1min−0.5)

m weight of the adsorbent used (g)

MB methylene blue

N adsorption capacity coefficient (mg L−1) in the case of column adsorption

process

n inverse of the slope of the Freundlich isotherm, it is related to adsorption

intensity

NLRA non-linear regression analysis

p frequency factor at the Arrhenius law (min−1)

34
q amount adsorbed per unit mass of the adsorbent fort→∞ (mg g−1)

qm Langmuir constant related the amount of dye adsorbed (mg g−1) when the

saturation is attained

qt amounts of dye adsorbed per unit mass of the adsorbent (in mg g−1) at time t

R2 coefficient of determination

RL dimensionless constant called ‘equilibrium parameter’ or ‘separation factor’

expressing the essential characteristics of the Langmuir isotherm

SEE standard error of estimate

SEM scanning electron microscope

t adsorption time (in min)

T ads

35