Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk

Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk

LWT - Food Science and Technology xxx (2014) 1e8 Contents lists available at ScienceDirect LWT - Food Science and Technology journal homepage: www.e...

496KB Sizes 7 Downloads 109 Views

LWT - Food Science and Technology xxx (2014) 1e8

Contents lists available at ScienceDirect

LWT - Food Science and Technology journal homepage: www.elsevier.com/locate/lwt

Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk , Agnes de Paula Scheer, Marcos R. Mafra, Anderson Marcos Dias Canteli*, Danielle Carpine Luciana Igarashi-Mafra , Chemical Engineering Department, Graduation Program of Food Engineering, Centro Polit Federal University of Parana ecnico, Jardim das Am ericas, , 81531-990, Brazil Curitiba, Parana

a r t i c l e i n f o

a b s t r a c t

Article history: Received 21 November 2013 Received in revised form 30 May 2014 Accepted 5 June 2014 Available online xxx

This paper evaluated the performance of fixed-bed columns with activated carbon as the adsorbent for the removal of benzaldehyde present in an aqueous solution. The effects of the following parameters were evaluated: inlet concentration (91.9 mg L1 e 440.0 mg L1), feed flow rate (3.9 mL min1 e 11.8 mL min1), bed depth (30 mme90 mm) and column inner diameter (6.2 mme12.2 mm). All of the experiments were performed at 293.15 K. The results showed that the bed capacity, total bed capacity and saturation time decreased as the feed flow rate was increased. The opposite effect was observed with an increase in bed depth. Increasing the inlet concentration resulted in higher aroma adsorption. An increase in the inner diameter without changing the feed flow rate resulted in better aroma recovery. The results showed that the column performed well at all of the analyzed parameter values. Data obtained from the analysis of the effects of feed flow rate and bed depth were used for a scale-up study using the bed depth service time model, which showed good results with errors of less than 12%. The experimental data obtained in this study will be useful for further applications involving coffee aroma recovery. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Adsorption column Breakthrough studies Soluble coffee Benzaldehyde

1. Introduction Aroma compounds are very important in the beverage industry because these compounds contribute to the sensory characteristics of beverages. However, during the processing of some beverages, including coffee, some compounds responsible for the aroma and flavor may be lost, and these compounds should be recovered and reincorporated to maintain the sensory characteristics of the final product close to those of the pre-processed product. Several techniques, such as supercritical fluid extraction (Gracia, Rodríguez, García, Alvarez, & García, 2007), vacuum membrane distillation (Bagger-Jørgensen, Meyer, Varming, & Jonsson, 2004; Diban, Voinea, Urtiaga, & Ortiz, 2009) and pervaporation (Aroujalian & Raisi, 2007; Diban, Urtiaga, & Ortiz, 2008; del Olmo, danos, & Herna ndez, 2014), have been studied Blanco, Palacio, Pra

* Corresponding author. Tel.: þ55 18 3642 6157. E-mail addresses: [email protected], [email protected] ), [email protected] (A.M.D. Canteli), [email protected] (D. Carpine (A.P. Scheer), [email protected] (M.R. Mafra), [email protected] (L. Igarashi-Mafra).

for the recovery of flavors. In fact, the latter technique has been the subject of recent studies because it allows the use of a low temperature, which avoids flavor degradation, and requires low energy consumption compared with traditional techniques, such as steam distillation, liquid solvent extraction and vacuum distillation. In this context, the adsorption process becomes of interest to researchers. Adsorption is very accessible due to the simplicity of the process and the range of naturally available adsorbents (Singh, Srivastava, & Mall, 2009). This technique is widely used in wastewater treatment (Chen et al., 2012; Goel, Kadirvelu, Rajagopal, & Kumar Garg, 2005; Goshadrou & Moheb, 2011; Han et al., 2007; Ko, Porter, & McKay, 2000; Lin & Wang, 2002; Salman, Njoku, & Hameed, 2011; Tan, Ahmad, & Hameed, 2008). However, adsorption, which is typically performed using activated carbon as the main adsorbent, was recently used to recover aromas in aqueous solutions, such as pear aroma (Diban, Ruiz, Urtiaga, & Ortiz, 2007), essential oil distillation , Dagostin, da Silva, (Edris, Girgis, & Fadel, 2003), and coffee (Carpine Igarashi-Mafra, & Mafra, 2013; Lucas, Cocero, Zetzl, & Brunner, 2004; Zuim et al., 2011). The use of fixed bed columns in adsorption processes offers several advantages, such as simplified operation, construction,

http://dx.doi.org/10.1016/j.lwt.2014.06.015 0023-6438/© 2014 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

2

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8

scale up and process automation and allowing the recovery of a large amount of adsorbate with the use of a fixed bed (Aksu & € nen, 2004). The study of the dynamic equilibrium in columns Go provides important information, such as the system size, contact time and adsorbent usage rate, and this information can be obtained from breakthrough curves (Moreno-Castilla, 2004). The bed depth, feed flow rate, inlet concentration, inner diameter and pH of the solution have been observed to significantly affect the ongoing adsorption process (Chen et al., 2012; Han et al., 2007; Singh et al., 2009; Srivastava, Prasad, Mishra, Mall, & Swamy, 2008). The aim of this study was to evaluate the recovery of benzaldehyde from a synthetic aqueous solution through continuous adsorption in a fixed bed column as a potential industrial application. The influence of several operational conditions (inlet concentration, feed flow rate, bed depth and inner diameter) was analyzed. Furthermore, a scale-up study was performed using the experimental data and the bed depth service time model. 1.1. Column performance The fixed bed column performance is described using breakthrough curves, which graphs time versus Ct/Co (the ratio of the concentration of the solute in the column outlet at a given time t to the initial concentration of the solute at the column inlet). Certain parameters obtained from the breakthrough curves can be used to evaluate the fixed bed performance and efficiency. The total capacity of the column (qtotal, mg) provides the maximum amount of flavor that can be adsorbed by the fixed bed and can be estimated by the area under the breakthrough curve (Calero, inz, Bla zquez, Tenorio, & Martín-Lara, 2009; Salman et al., Herna 2011). If the bed is completely saturated and the inlet concentration is constant over time, the total capacity of the column is calculated from Equation (1):

qtotal

QC0 ¼ 1000

t¼t Z sat 

t¼0

 C 1 dt C0

(1)

where Q is the column feed flow rate (mg min1), C is the outlet concentration (mg L1), C0 is the inlet concentration (mg L1) and tsat is the time required for the bed to become saturated (min). The bed capacity (qbed, mg g1) is a parameter that determines the amount of flavor recovered by the fixed bed per gram of adsorbent present in the bed and is calculated from Equation (2), where m is the mass of activated carbon present in the bed (g) (Calero et al., 2009; Chen et al., 2012; Salman et al., 2011):

qbed ¼

qtotal m

(2)

The total amount of benzaldehyde fed into the column until full bed saturation (W, g) can be calculated using Equation (3) (Aksu & € nen, 2004; Calero et al., 2009): Go



QC0 tsat 106

(3)

The bed performance (P) relates the amount of aroma retained in the bed (qtotal) with the amount of aroma fed in the same run (W) €nen, 2004; Calero et al., 2009; Chen et al., 2012). A high (Aksu & Go performance indicates a good operational set up, and the performance can be calculated using Equation (4):

q Pð%Þ ¼ total  100 W

(4)

The adsorbent utilization (h) relates the total capacity obtained in the fixed bed (qtotal) with the total capacity obtained in a batch

experiment (qbatch) and therefore represents the amount of active sites that are not utilized in the fixed bed. This parameter is calculated from Equation (5):



qbed qbatch

(5)

The residence time, which affects the shape of the breakthrough curve and the breakthrough time (Singh et al., 2009), is the time required for the fluid to fill the empty column (Ko et al., 2000). This parameter is the most important in the design of a fixed bed (McKay & Bino, 1990), and effects in the residence time may be easily observed as a result of changes in the bed depth and feed flow rate. The true residence time (TRV) can be calculated from Equation (6):

TRV ¼

ε  VL Q

(6)

Where ε is the bed porosity and VL is the volume occupied by the adsorbent inside the bed (mL). The bed porosity can be estimated by the fraction of empty spaces (volume of distilled water present in the fixed bed after packing (mL) divided by the fixed bed volume (mL)). 1.2. Scale-up study The bed depth service time (BDST) model predicts the relationship between the bed depth, Z (cm), and the operation time, t (min). This model assumes that the adsorption rate is controlled by the surface reaction between the adsorbate and the unused capacity of the adsorbent (Srivastava et al., 2008; Zou, Zhao, & Zhu, 2013). Equation (7) expresses a linear relationship between the bed depth and the service time:



  N0 1 C Z ln 0  1 KBDST C0 C0 v Ct

(7)

where N0 is the adsorption capacity (mg mL1), v is the fluid velocity (cm min1), Ct is the outlet concentration at time t (mg mL1) and KBDST is the mass transfer coefficient (mL (mg min1)). KBDST and N0 can be calculated from the linear and angular coefficient, respectively, from the graph of t as a function of Z at a given Ct/C0 ratio (iso-concentration line). At 50% breakthrough (Ct/C0 ¼ 0.50 and t ¼ t0.50), the linear term of Equation (7) becomes indeterminate (ln(1)), and the equation is thus reduced to Equation (8):

t0;50 ¼

N0 Z C0 v

(8)

Thus, the graph of t0.50 at 50% breakthrough as a function of Z forms a line passing through the origin, and N0 can be calculated by the angular coefficient (Srivastava et al., 2008). A simplified form of the BDST model is expressed by Equation (9) (Goel et al., 2005):

t ¼ aZ  b a¼



(9)

N0 C0 v 1 KBDST C0

(10)  ln

 C0 1 Ct

(11)

where a is the angular coefficient (min cm1) and b is the linear coefficient (min) of the straight line obtained with Equation (7).

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8

3

The slope of Equation (9) at velocity v can be used to obtain the angular coefficient for a different velocity v0 using Equation (12):

a0 ¼ a

v v0

(12)

where a’ is the angular coefficient for the new velocity, v0 (cm min1). The model assumes that the linear coefficient does not change as the velocity is varied. 2. Materials and methods 2.1. Adsorbent Granular activated carbon derived from coconut husks (CARBOMAFRA® 119) was supplied by Carbomafra Chemical Industries S.A (Curitiba, Brazil) and was used as received. This adsorbent possessed a minimum iodine number of 850 mg of I2 per g of carbon, a low ash concentration (less than 10%) and an apparent density of 500 ± 50 kg m3. The activated carbon surface was irregular, heterogeneous and unordered with a high BET surface area (772 m2 g1), a micropore predominance (95.2%) and an average pore diameter of 22.1 Å (Zuim et al., 2011). 2.2. Adsorbate Benzaldehyde (CAS 100-52-7; chemical formula: C7H6O; molecular weight: 106.12; physical state at room temperature: liquid oil; relative density: 1.0415 g cm1 at 293.15 K; solubility in water: 6.55 g L1 at 298.15 K) was supplied by SigmaeAldrich (St. Louis, MO, USA) with a purity of 99.5%. This aroma possesses a powerful sweet odor similar to that of freshly crushed bitter almonds and has a burning taste that becomes sweet when appropriately diluted. The experimental solutions were prepared in distilled and deionized water. 2.3. Analytical measurement The benzaldehyde concentration was determined by measuring the absorbance of the aqueous solution (lmax ¼ 249 nm) using a UV spectrophotometer (Shimadzu 1800) and quartz cuvettes. The calibration curve of the absorbance as a function of the benzaldehyde concentration was linear between 0.2 and 6.0 mg L1. 2.4. Column adsorption studies The experiments were conducted in 250-mm-long glass columns with different internal diameters (6.2, 9.2 and 12.2 mm). Glass spheres (3 mm in diameter) were added to the base of the column to promote uniform flow through the bed. Glass wool was then fixed (5 mm in height) in the column to provide support to the adsorbent. The column was filled with type II water, and activated carbon was added to the column (1, 2 or 3 g). At the top of the column, glass wool was added (5 mm in height) to avoid fluidization of the activated carbon during the experiment. The column was fed an aqueous solution of benzaldehyde using a peristaltic pump (Masterflex® L/S Digital Drive) employing an upward flow. Samples were collected at regular intervals. The experiments were conducted until C/C0 ¼ 0.99; however, the experiments investigating the effect of the initial concentration were conducted until C/ C0 ¼ 0.90. A thermostatic bath was used to maintain the temperature of the system at 293 ± 0.5 K. Fig. 1 shows the diagram of the experimental set up. The following parameters were evaluated: inlet concentration (C0), feed flow rate (Q), bed depth (Z) and inner diameter (D). The

Fig. 1. Schematic diagram of the experimental set up. (1) Column support; (2) glass spheres; (3) glass wool; and (4) activated carbon.

inner diameter effect were evaluated for four different conditions: DQM (the feed flow rate and bed depth were maintained constant), DQZ (the feed flow rate and bed depth were maintained constant), DVM (feed velocity and activated carbon mass were maintained constant) and DVZ (feed velocity and bed depth were maintained constant). 2.5. Bed porosity After packing the bed, the volume of type II water present in the adsorbent (VV) was determined based on the relationship between mass and density, and the porosity was then estimated using Equation (13), where VL is the fixed bed volume.

ε¼

VV VL

(13)

Table 1 Experimental set up for benzaldehyde sorption onto activated carbon.

Run Run Run Run Run Run Run Run Run Run Run Run

1 2 3 4 5 6 7 8 9 10 11 12

Q (mL min1)

C0 (mg L1)

D (mm)

m (g)

Z (mm)

v (cm min1)

11.8 11.8 11.8 3.9 7.8 11.8 11.8 11.8 5.3 11.8 11.8 20.7

91.9 272.6 440.0 440.0 440.0 440.0 440.0 440.0 440.0 440.0 440.0 440.0

9.2 9.2 9.2 9.2 9.2 9.2 9.2 9.2 6.2 6.2 12.2 12.2

2 2 2 1 1 1 2 3 2 2 2 2

60 60 60 30 30 30 60 90 120 120 30 30

17.7 17.7 17.7 5.9 11.8 17.7 17.7 17.7 17.7 39.0 10.1 17.7

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

4

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8

Fig. 2. Comparison of experimental breakthrough curves for benzaldehyde adsorption onto an activated carbon packed column at T ¼ 293 K and D ¼ 9.2 mm. (a) Curves obtained with varied inlet concentrations (C0): △ ¼ 91.9 mg L1, B ¼ 272.6 mg L1 and , ¼ 440.0 mg L1. In these runs, Z ¼ 60 mm, Q ¼ 11.8 mL min1, M ¼ 2 g and v ¼ 17.7 cm min1. (b) Curves obtained with varied feed flow rates (Q): ◄ ¼ 3.9 mL min1, ✩ ¼ 7.8 mL min1 and þ ¼ 11.8 mL min1. In these runs, Z ¼ 30 mm, M ¼ 1 g, and v ¼ 5.9e17.7 cm min1. (c) Curves obtained with varied bed depths (Z): ; ¼ 30 mm, A ¼ 60 mm and  ¼ 90 mm. In these runs, Q ¼ 11.8 mL min1, M ¼ 1e3 g and v ¼ 17.7 cm min1.

2.6. Scale up The BDST model was used for the scale-up study based on data of the feed flow rate and bed depth (runs 4, 5, 6, 7 and 8 in Table 1). The OriginPro 8.5 software was used to perform the linear regression fits of the data. The goodness of the BDST fit was evaluated using the correlation coefficient (R2) and the sum of the squares of the errors (ERRSQ). The error (E (%)) between the experimental time (texp) and the predicted time (tcal) was obtained using Equation (14):

Eð%Þ ¼

 texp  t

cal

texp

 

 100

3.1.1. Effect of the inlet concentration Fig. 2a shows the effect of changing the inlet concentration in the range of 91.9e440.0 mg L1 while keeping the other parameters constant. These results are presented in Table 2 (runs 1, 2 and 3). The change in the concentration gradient significantly affects the carbon saturation rate. At lower values of the aroma inlet concentration, the shape of the curve is less pronounced, indicating that

Table 2 Comparison of activated carbon fixed-bed performance for benzaldehyde adsorption under different experimental conditions.

(14)

3. Results and discussion 3.1. Column study In a previous study, the total capacity of the adsorbent obtained from batch studies (qbatch) was 251.2 ± 2.3 mg g1, and this value was used to obtain the values of h. The average estimated porosity was 0.56 ± 0.013, and this value was used to obtain the TRV for each run.

Run Run Run Run Run Run Run Run Run Run Run Run

1 2 3 4 5 6 7 8 9 10 11 12

TRV (min) qbed (mg g1) qtotal (mg) h

W (g) P (%) tsat (min)

0.18 0.18 0.18 0.26 0.13 0.09 0.18 0.27 0.35 0.16 0.15 0.09

0.552 0.658 0.777 1.05 1.25 0.86 1.400 3.654 1.817 1.572 1.562 1.884

142.0 165.6 165.3 217.5 206.6 184.5 187.5 239.4 214.7 181.7 192.8 170.7

284.1 331.3 330.6 217.5 206.6 184.5 375.1 718.3 429.4 363.4 385.7 341.4

0.56 0.66 0.65 0.86 0.82 0.73 0.74 0.95 0.85 0.72 0.77 0.68

51.4 50.3 53.1 20.6 16.4 21.3 26.8 19.6 23.6 23.1 24.7 18.1

510 205 120 626 330 165 270 740 780 300 300 210

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8

adsorption occurs slowly, whereas an increase in the inlet concentration increases the steepness of the slope, causing the adsorbent to saturate in less time. In addition, lower concentration gradients cause the transport within the pores to occur slowly due to reductions in the coefficients of diffusion and/or mass transfer €nen, 2004), which results in an increase in the satura(Aksu & Go tion time. P (%) does not change significantly as C0 increases. The values of qbed, qtotal and h were very similar at higher C0 (272.6 and 440.0 mg L1). At the lowest C0 (91.9 mg L1), all of these parameters were lower. These results indicate that changes in the concentration gradient between 91.9 and 272.6 or 440.0 mg L1 affect the adsorbent saturation rate (as C0 increases, the driving force also increases, which implies a reduction in the size of the mass transfer zone (Goel et al., 2005)). Similar trends have been described in the literature (Chen et al., 2012; Salman et al., 2011; Tan et al., 2008). However, the similar qbed values obtained at high C0 were not an expected result because increasing the inlet concentration from 272.6 to 440.0 should increase qbed, a trend that was not observed. Therefore, there is a limit to the concentration difference be the driving force for adsorption. 3.1.2. Effect of the flow rate The effect of the feed flow rate on the adsorption of benzaldehyde onto activated carbon was investigated by varying Q from 3.9 to 11.8 mL min1 while holding the other parameters constant. Fig. 2b shows the breakthrough curves for different values of Q, and Table 2 presents these results (runs 4, 5 and 6). The data shown in Table 2 illustrate that tsat, qbed, qtotal and h decrease as Q increases. This result occurs because the time required for equilibrium between benzaldehyde and activated carbon is much higher than the retention time (6 h, Zuim et al., 2011). Therefore, as Q increases, TRV is reduced, which causes a negative effect on the mass transfer efficiency and thus results in a € nen, 2004; Ko et al., 2000; Singh et al., decrease in tsat (Aksu & Go 2009). Similar trends have been noted by other investigators in € nen, 2004; Carpine  et al., 2013; Ko et al., the literature (Aksu & Go 2000; Singh et al., 2009). According to Ko, Porter, and McKay (2001), an increase in the feed flow changes the diffusion of the film without changing the intra-particle diffusion. The movement of the adsorbate from the solution to the region surrounding the adsorbent causes a concentration gradient to develop at this interface, which allows the adsorbate to cross over the film and be adsorbed. An increase in the solution velocity reduces the adhesion of the adsorbate to the adsorbent and consequently reduces the efficiency of the process (Kundu & Gupta, 2007; Singh & Pant, 2006). The P (%) values obtained for flow rates equal to 3.9 and 11.8 mL min1 were similar; however, the P (%) value for the intermediate flow rate was only 16.4%. This result may be explained by the larger amount of adsorbate fed into the bed (W) to reach bed saturation. 3.1.3. Effect of the bed depth Fig. 2c shows the effect of changing the bed depth Z between 30 and 90 mm, and the data are shown in Table 2 as sets 6, 7 and 8; the other parameters remained constant. As Z increases, tsat and qtotal also increase as a result of the higher number of active sites that are available due to the increase in the total surface area (increasing the adsorbent mass) as the bed depth increases (Singh et al., 2009). At lowers depths (30 and 60 mm), the obtained values for qbed and h were similar; however, at a depth of 90 mm, these values were higher. The shape of the mass transfer zone resembles the characteristic S of the breakthrough curves, as illustrated in Fig. 2c, and offers an explanation for these increased values. Another possible explanation is that the residence time increases as Z

5

increases and thus results in improvements in the column parameters. Similar trends have been previously described in the literature (Han et al., 2007; Salman et al., 2011; Tan et al., 2008). As Z increases, P decreases; this result may have occurred as a result of the tail formed at higher bed depths. The increase in the tail implies a higher benzaldehyde loss, which reduces the column performance. An inverse relationship between h and P was observed. As the adsorbent becomes saturated, more adsorbate is consumed. According to Cooney (1991), the gradual approach of C/C0 to 1 (called the tail) in the breakthrough curves occurs when intraparticle diffusion is the rate-limiting step. 3.1.4. Effect of the inner diameter i) DQM Fig. 3a shows the effect of changing the inner diameter from 6.2 to 12.2 mm, and the data are shown in Table 2 as runs 10, 7 and 11. The shape of the mass transfer zone does not change significantly when the column inner diameter is altered. As D increases, qbed, qtotal and h increase due to the longer contact time between the adsorbate particle and the adsorbent particle caused by the reduction in fluid velocity, which favors aroma adsorption. ii) DQZ The effect of changing the inner diameter from 9.2 to 12.2 mm is shown in Fig. 3b, and the data are shown in Table 2 as runs 6 and 11. The values of qbed and h increase as D increases. This effect may be due to the lower velocity required to maintain a constant Q, which would result in a longer contact time for benzaldehyde with the adsorbent and thereby promote adsorption. The values of qtotal and tsat also increase as D increases because the larger diameter accommodates more adsorbent and increases the number of active sites present in the bed. P also increases as D increases, and this effect may be due to the more complete mass transfer zone of the larger-diameter column. This result agrees with similar trends that were previously reported in the literature (Singh et al., 2009; Srivastava et al., 2008). iii) DVM Fig. 3c shows the effect of changing the inner diameter from 6.2 to 12.2 mm, and the results are shown in Table 2 as runs 9, 7 and 12. The other parameters were maintained constant. The data in Table 2 illustrate that an increase in D results in decreases in qbed, h and tsat. The reduction of these parameters is due to the decrease in TRV as D increases and the bed depth decreases. Thus, at lower diameters, benzaldehyde has a longer contact time with the activated carbon, which promotes higher adsorption. This effect, in addition to the increase in turbulence, weakens the solideliquid interaction, which disfavors adsorption (Goshadrou & Moheb, 2011). iv) DVZ The effect of changing the internal diameter from 9.2 to 12.2 mm is shown in Fig. 3d, and the results are shown in Table 2 as runs 6 and 12. Increasing the inner diameter changes the tendency of the breakthrough curve. The breakthrough curve for higher D saturates faster until C/C0 is approximately 0.8. After this value, the saturation rate decreases, and the curve for the lower D saturates before the higher D. Table 2 indicates that the values of qbed, P and h decrease as D increases. This result may be due to the difficulties associated

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

6

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8

Fig. 3. Comparison of experimental breakthrough curves with different inner diameters at T ¼ 293 K. (a) Curves obtained with varied inner diameters (DQM): △ ¼ 12.2 mm, B ¼ 9.2 mm and , ¼ 6.2 mm. In these runs, Z ¼ 30e120 mm, Q ¼ 11.8 mL min1, M ¼ 2 g and v ¼ 10.1e39.0 cm min1. (b) Curves obtained with varied inner diameters (DQZ): þ ¼ 12.2 mm and < ¼ 9.2 mm. In these runs, Z ¼ 30 mm, Q ¼ 11.8 mL min1, M ¼ 1e2 g and v ¼ 10.1e17.7 cm min1. (c) Curves obtained with varied inner diameters (DVZ): ✩ ¼ 12.2 mm, A ¼ 9.2 mm and ; ¼ 6.2 mm. In these runs, Z ¼ 30 mm, Q ¼ 11.8e20.7 mL min1, M ¼ 1e2 g and v ¼ 17.7 cm min1. (d) Curves obtained with varied inner diameters (DVZ):  ¼ 12.2 mm, and ◄ ¼ 9.2 mm. In these runs, Z ¼ 30 mm, Q ¼ 11.8e20.7 mL min1, M ¼ 1e2 g and v ¼ 17.7 cm min1.

with packing a larger diameter bed. As expected, tsat and qtotal increase as D increases due to the increased number of active sites present in the fixed bed. 3.2. Scale up Fig. 4 shows the iso-concentration lines for benzaldehyde recovery in a fixed bed at 20, 40, 50 and 60% breakthrough. The angular coefficient values do not vary significantly, as shown in Table 3. The line obtained for 50% saturation does not pass through the origin (b s 0). Sharma and Forster (1995) and Zulfadhly, Mashitah, and Bhatia (2001) obtained the same behavior and developed a complex adsorption mechanism, which involves more than one rate-limiting step. The equation obtained using a flow rate of 11.8 mL min1 was used to predict the performance for the other

Table 3 Predicted parameters for the BDST model at 11.8 mL min1.

Fig. 4. Iso-recovery lines for breakthroughs of , ¼ 0.20, B ¼ 0.40, ✩ ¼ 0.50 and △ ¼ 0.60 for different bed depths. T ¼ 293 K; C0 ¼ 440.0 mg L1; v ¼ 17.7 cm min1; D ¼ 9.2 mm; and Z ¼ 30e90 mm.

Ct/C0

a (min cm1)

0.20 0.40 0.50 0.60

10.7 10.8 11.0 11.2

± ± ± ±

0.7 0.8 0.9 1.4

b (min) 24.0 9.0 2.6 6.3

± ± ± ±

4.7 5.6 6.2 9.1

KBDST (mL (mg min)1)

N0 (mg mL1)

R2

ERRSQ

0.130 0.101 e 0.144

82.23 82.87 84.1 86.06

0.9910 0.9873 0.9848 0.9697

9.375 13.500 16.666 35.042

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8 Table 4 Predicted breakthrough times based on the BDST parameters for new flow rates of 7.8 and 3.9 mL min1. Ct/C0

a' (min cm1)

Q ¼ 7.8 mL min1 0.20 16.1 0.40 16.3 0.50 16.5 0.60 16.8 Q ¼ 3.9 mL min1 0.20 32.2 0.40 32.4 0.50 32.9 0.60 33.7

b' (min)

tcal (min)

texp (min)

7

Nível Superior and the support provided by the Graduation Pro, Curitiba, gram of Food Engineering (Federal University of Parana Brazil).

E (%)

24.0 9.0 2.6 6.3

24.3 39.7 46.8 56.9

22 36 43 51

10.8 10.4 8.9 11.7

24.0 9.0 2.6 6.3

72.7 88.4 92.3 107.5

66 91 102 111

8.5 2.8 5.6 3.1

feeding flow rates (7.8 and 3.9 mL min1), and these results are presented in Table 4. The data shown in Table 4 indicate that the errors (E) were low and that good predictions were obtained with flow rates of 7.8 and 3.9 mL min1. Therefore, the constants obtained using the BDST model could be used to predict the times at which the ratio Ct/C0 equals 0.20, 0.40, 0.50 and 0.60 for other flow rates. 4. Conclusions The adsorption of benzaldehyde from an aqueous solution onto activated carbon was investigated in a fixed bed column at 293 K. The effects of the inlet concentration, feed flow rate, bed depth and inner diameter on adsorption were investigated. In general, the evaluated parameters improved as a result of increases in Z and decreases in Q. These performance metrics also improved as C0 was increased; however, this improvement in performance exhibited a limit at higher concentrations. An increase in the inner diameter without changing Q resulted in an improvement in the column parameters. However, when V was held constant, increasing D reduced the aroma adsorption. The effect of the inner diameter was studied in four different ways, and the results showed that a decrease in the inner diameter while maintaining the contact time between the adsorbate and adsorbent constant resulted in improvements in most of the response variables. However, the low surface area when V and Z were held constant (DVZ) resulted in a lower total adsorption at small inner diameters. The shape of the breakthrough curve was different in almost all of the runs with the exception of the run in which D was changed and the feed flow rate and the amount of adsorbent in the fixed bed were held constant. Additionally, there is likely more than one ratelimiting step in the adsorption of benzaldehyde onto activated carbon because the formation of a tail was observed in most of the runs. The scale-up study also indicated that there is more than one limiting rate. The BDST model was applied to the experimental data and used for the scale-up study, demonstrating that this model is suitable for scaling up the system. Zuim et al. (2011) proposed the use of the batch adsorption process for the recovery of benzaldehyde and achieved good results, although the process is usually not applicable to industry. The results obtained in this work are very promising, as demonstrated by the high aroma recovery rate achieved with the continuous system, and should be useful for further applications involving coffee aroma adsorption. Acknowledgments The authors are grateful for the scholarship and the research ~o de Aperfeiçoamento de Pessoal de grant provided by Coordenaça

References € nen, F. (2004). Biosorption of phenol by immobilized activated sludge Aksu, Z., & Go in a continuous packed bed: prediction of breakthrough curves. Process Biochemistry, 39, 599e613. Aroujalian, A., & Raisi, A. (2007). Recovery of volatile aroma components from orange juice by pervaporation. Journal of Membrane Science, 303, 154e161. Bagger-Jørgensen, R., Meyer, A. S., Varming, C., & Jonsson, G. (2004). Recovery of volatile aroma compounds from black currant juice by vacuum membrane distillation. Journal of Food Engineering, 64, 23e31. inz, F., Bl Calero, M., Herna azquez, G., Tenorio, G., & Martín-Lara, M. A. (2009). Study of Cr (III) biosorption in a fixed-bed column. Journal of Hazardous Materials, 171, 886e893. , D., Dagostin, J. L. A., da Silva, V. R., Igarashi-Mafra, L., & Mafra, M. R. (2013). Carpine Adsorption of volatile aroma compound 2-phenylethanol from synthetic solution onto granular activated carbon in batch and continuous modes. Journal of Food Engineering, 117, 370e377. Chen, S., Yue, Q., Gao, B., Li, Q., Xu, X., & Fu, K. (2012). Adsorption of hexavalent chromium from aqueous solution by modified corn stalk: a fixed-bed column study. Bioresource Technology, 113, 114e120. Cooney, D. O. (1991). The importance of axial dispersion in liquid-phase fixed-bed adsorption operations. Chemical Engineering Communications, 110, 217e231. Diban, N., Ruiz, G., Urtiaga, A., & Ortiz, I. (2007). Granular activated carbon for the recovery of the main pear aroma compound: viability and kinetic modelling of ethyl-2,4-decadienoate adsorption. Journal of Food Engineering, 78, 1259e1266. Diban, N., Urtiaga, A., & Ortiz, I. (2008). Recovery of key components of bilberry aroma using a commercial pervaporation membrane. Desalination, 224, 34e39. Diban, N., Voinea, O. C., Urtiaga, A., & Ortiz, I. (2009). Vacuum membrane distillation of the main pear aroma compound: experimental study and mass transfer modeling. Journal of Membrane Science, 326, 64e75. Edris, A. E., Girgis, B. S., & Fadel, H. H. M. (2003). Recovery of volatile aroma components from aqueous waste streams using an activated carbon column. Food Chemistry, 82, 195e202. Goel, J., Kadirvelu, K., Rajagopal, C., & Kumar Garg, V. (2005). Removal of lead(II) by adsorption using treated granular activated carbon: batch and column studies. Journal of Hazardous Materials, 125, 211e220. Goshadrou, A., & Moheb, A. (2011). Continuous fixed bed adsorption of C.I. Acid Blue 92 by exfoliated graphite: an experimental and modeling study. Desalination, 269, 170e176. Gracia, I., Rodríguez, J. F., García, M. T., Alvarez, A., & García, A. (2007). Isolation of aroma compounds from sugar cane spirits by supercritical CO2. The Journal of Supercritical Fluids, 43, 37e42. Han, R., Wang, Y., Yu, W., Zou, W., Shi, J., & Liu, H. (2007). Biosorption of methylene blue from aqueous solution by rice husk in a fixed-bed column. Journal of Hazardous Materials, 141, 713e718. Ko, D. C. K., Porter, J. F., & McKay, G. (2000). Optimised correlations for the fixed-bed adsorption of metal ions on bone char. Chemical Engineering Science, 55, 5819e5829. Ko, D. C. K., Porter, J. F., & McKay, G. (2001). Film-pore diffusion model for the fixedbed sorption of copper and cadmium ions onto bone char. Water Research, 35, 3876e3886. Kundu, S., & Gupta, A. K. (2007). As(III) removal from aqueous medium in fixed bed using iron oxide-coated cement (IOCC): experimental and modeling studies. Chemical Engineering Journal, 129, 123e131. Lin, S. H., & Wang, C. S. (2002). Treatment of high-strength phenolic wastewater by a new two-step method. Journal of Hazardous Materials, 90, 205e216. Lucas, S., Cocero, M. J., Zetzl, C., & Brunner, G. (2004). Adsorption isotherms for ethylacetate and furfural on activated carbon from supercritical carbon dioxide. Fluid Phase Equilibria, 219, 171e179. McKay, G., & Bino, M. J. (1990). Fixed bed adsorption for the removal of pollutants from water. Environmental Pollution, 66, 33e53. Moreno-Castilla, C. (2004). Adsorption of organic molecules from aqueous solutions on carbon materials. Carbon, 42, 83e94.  Blanco, C. A., Palacio, L., Pra danos, P., & Herna ndez, A. (2014). Pervadel Olmo, A., poration methodology for improving alcohol-free beer quality through aroma recovery. Journal of Food Engineering, 133, 1e8. Salman, J. M., Njoku, V. O., & Hameed, B. H. (2011). Batch and fixed-bed adsorption of 2,4-dichlorophenoxyacetic acid onto oil palm frond activated carbon. Chemical Engineering Journal, 174, 33e40. Sharma, D. C., & Forster, C. F. (1995). Column studies into the adsorption of chromium (VI) using sphagnum moss peat. Bioresource Technology, 52, 261e267. Singh, S., Srivastava, V. C., & Mall, I. D. (2009). Fixed-bed study for adsorptive removal of furfural by activated carbon. Colloids and Surfaces a: Physicochemical and Engineering Aspects, 332, 50e56. Singh, T. S., & Pant, K. K. (2006). Experimental and modelling studies on fixed bed adsorption of As(III) ions from aqueous solution. Separation and Purification Technology, 48, 288e296.

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015

8

A.M.D. Canteli et al. / LWT - Food Science and Technology xxx (2014) 1e8

Srivastava, V. C., Prasad, B., Mishra, I. M., Mall, I. D., & Swamy, M. M. (2008). Prediction of breakthrough curves for Sorptive removal of phenol by Bagasse Fly ash packed bed. Industrial & Engineering Chemistry Research, 47, 1603e1613. Tan, I. A. W., Ahmad, A. L., & Hameed, B. H. (2008). Adsorption of basic dye using activated carbon prepared from oil palm shell: batch and fixed bed studies. Desalination, 225, 13e28. Zou, W., Zhao, L., & Zhu, L. (2013). Adsorption of uranium(VI) by grapefruit peel in a fixed-bed column: experiments and prediction of breakthrough curves. Journal of Radioanalytical and Nuclear Chemistry, 295, 717e727.

, D., Distler, G. A. R., de Paula Scheer, A., Igarashi-Mafra, L., & Zuim, D. R., Carpine Mafra, M. R. (2011). Adsorption of two coffee aromas from synthetic aqueous solution onto granular activated carbon derived from coconut husks. Journal of Food Engineering, 104, 284e292. Zulfadhly, Z., Mashitah, M. D., & Bhatia, S. (2001). Heavy metals removal in fixedbed column by the macro fungus Pycnoporus sanguineus. Environmental Pollution, 112, 463e470.

Please cite this article in press as: Canteli, A. M. D., et al., Fixed-bed column adsorption of the coffee aroma compound benzaldehyde from aqueous solution onto granular activated carbon from coconut husk, LWT - Food Science and Technology (2014), http://dx.doi.org/10.1016/ j.lwt.2014.06.015