Environ. Sci. Technol. 2001, 35, 3048-3054
Modeling of Copper Biosorption by Arthrobacter sp. in a UF/MF Membrane Reactor F. BEOLCHINI,† F. PAGNANELLI,‡ AND F. VEGLIO Ä * ,§ Dipartimento di Chimica, Ingegneria Chimica e Materiali, Universita` degli studi de L’Aquila, 67040 Monteluco di Roio, L’Aquila, Italy, Dipartimento di Chimica, Facolta` di S.M.F.N., Universita` degli Studi “La Sapienza”, P.le A. Moro, 5, 00185 Roma, Italy, and Dipartimento di Ingegneria Chimica e di Processo “G.B. Bonino”, Universita’ degli Studi di Genova, via Opera Pia, 15, 16145 Genova, Albaro, Italy
Copper biosorption by Arthrobacter sp. has been studied in this work. The process has been realized inside of a ultrafiltration/microfiltration (UF/MF) reactor in order to confine cells. A mathematical model has been developed that is able to predict experimental data under different operating conditions. The model takes into account different phenomena, which might occur during the process, such as a dependence of equilibrium parameters on pH, a partial cell disruption, and a change in the membrane retention properties at high biomass concentrations. Experimental tests have been performed under different operating conditions: a full factorial design has been implemented with pH (levels: 4, 5, and 6 units) and biomass concentration (levels: 1 and 5 g/L) as factors. A simple mathematical model based on metal mass balance taking into account the effect of pH on the Langmuir equilibrium adsorption parameters well fitted experimental data at low pH values and biomass concentrations. A more complex mathematical model, which considers a partial cell disruption during the biosorption trial, was proposed to understand and analyze the anomalous system behavior at pH ) 6 and biomass concentration equal to 5 g/L. The effect of mechanical stress on biomass performances was also examined by using a discontinuous system (test tube trials) simulating the membrane reactor apparatus. In this alternative system biosorption trials were carried out in test tubes in such a way to avoid or at least minimize the disruption due to mechanical stress. Experimental results obtained by using this system can be modeled up to pH ) 5 without considering cell disruption phenomenon, while at pH ) 6 possible chemical reactions of biomass constituents could happen.
1. Introduction Biosorption has been extensively studied as an alternative technology for toxic metal removal from wastewater: it is based on the metal binding capacities of various biological materials such as algae, bacteria, fungi, yeast, and plant biomaterial (1-7). * Corresponding author phone: +39-010-353 2583; fax: +39/010353 2586; e-mail:
[email protected] and
[email protected]. † Universita ` degli studi de L’Aquila. ‡ Universita ` degli Studi “La Sapienza”. § Universita’ degli Studi di Genova. 3048
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
In the present work the toxic metal removal from aqueous solutions by the bacterium Arthrobacter sp. (isolated from a natural environment) has been studied by using membrane reactors. Previous works have demonstrated the ability of Arthrobacter sp. to remove efficiently heavy metals. The equilibrium of single metal systems (copper, lead, and nickel) was studied in different experimental conditions, and experimental data have been well fitted by the Langmuir model (7, 8). Immobilization of the lyophilized biomass in a polymeric matrix has been also investigated in order to characterize the biomass support (9, 10) and to study the kinetics of the biosorption process (11). Furthermore, membrane processes have been also considered in some studies (12-14), as alternative to immobilization techniques which entrap biomass in polymeric matrices. Tangential flow membrane filtration (ultrafiltration or microfiltrationsUF and MF, respectively) is widely used in biotechnological separation processes (up and downstream processes)sin particular for cell harvesting and recoveries (15)sfor its particular advantages linked to the possibility of low-pressure conditions applied on the membrane coupled with large flux rates. In particular some preliminary biosorption tests were carried out in the membrane reactor (12), and simulation studies as a function of the operative conditions (residence time, reactors in series, biomass concentration, metal specific uptake) were elaborated (12, 13). Several authors have shown the potential application of this system on biosorption processes (5, 15), but major efforts have to be spent in the mathematical modeling to optimize this biosorption operative configuration. A simple model has been developed (13, 14) in order to predict time courses of copper concentration in the permeate. The Langmuir equation was used for equilibrium (13, 16, 17)
q)
qmaxCeq KS + Ceq
(1)
where q (mg/g) is the equilibrium metal specific uptake; Ceq (mg/L) is the equilibrium metal concentration in solution; KS (mg/L) is the model parameter related to the affinity between metal and active sites; and qmax (mg/g) is the model parameter corresponding to the maximum metal specific uptake. Coupling the metal material balance and eq 1, the following first-order differential equation was obtained for copper concentration variation during time
dC ) dt
F‚(Co - C) qmax‚KS‚X V‚ 1 + (C + KS)2
(
)
(2)
where C (mg/L) is the actual copper concentration in the reactor and in the permeate flux; Co (mg/L) is the copper concentration in the feeding stream; t (min) is the time; F (L/min) is the feed flow rate; P (L/min) is the permeate flow rate; V (L) is the reactor volume; and X (g/L) is the biomass concentration. The inlet flow rate, F (kept equal to the permeate flow rate, P, to have a constant reaction volume, V), was a function of time due to permeate flux decline (18); the following model can be used to represent this phenomenon:
F(t) ) a - d‚tc 10.1021/es000159b CCC: $20.00
(3)
2001 American Chemical Society Published on Web 06/13/2001
FIGURE 1. Typical configuration for biosorption processes in CSTR coupled with UF/MF membrane systems. Model (2) is not always able to describe the experimental data in a wide range of experimental conditions (14). In particular a disagreement between calculated and experimental results has been obtained for the largest cell concentration, and several hypotheses can be formulated to explain this phenomenon. The aim of the present work was the improvement of model (2), to take into account some phenomena, which might occur during the process, such as a partial cell disruption, a change in the membrane retention properties at high biomass concentrations, and a dependence of equilibrium parameters on pH. New experimental tests have been performed under different controlled operating conditions (three levels of pH, two levels of biomass concentration). The simple simulation of the system based on the metal mass balance (12) has been the base for the development of a new model which has been successfully fitted to these experimental data. An innovative and effective procedure to simulate discontinuously the membrane reactor system in test tubes was also proposed to analyze the effect of a minor cell damage.
2. Materials and Methods 2.1. Microorganisms. Arthrobacter sp. harvested from natural waters collected near L’Aquila (Italy) was supplied by the Dip. di Biologia di Base ed Applicata (L’Aquila University). Further details about cell cultivation, harvesting, and use can be found elsewhere (16). 2.2. Membrane Reactor Trials. Figure 1 shows schematically the experimental system here used. A biosorption trial was performed as follows: the lyophilized biomass (8) at different concentration levels was introduced in a temperature controlled glass reactor (volume 200 mL; temperature 30 °C). The dead biomass suspension was fed through a membrane module by a peristaltic pump (tangential velocity 0.3 m/s; transmembrane pressure 200 KPa). A polysulfone membrane was used with 100 000 Da Molecular Weight Cut Off (MWCO) and a total area of 36 cm2. The mixing in the reactor was ensured both by a magnetic stirrer placed into the reactor and by the recirculation flow of the retentate stream (Figure 1). Consequently the reactor can be assumed to be as a continuous stirred tank reactor (CSTR). The copper solution (containing about 10 mg/L of Cu2+ as CuSO4) was then fed to the bioreactor by a peristaltic pump, paying attention to control its flow rate: this was always equal to the permeate flux produced in the membrane module. This control was necessary because the permeate flux in general decreases during time due to the fouling characteristic of the suspension, as already reported for this cell harvesting system (18). The flux decreasing (F) was monitored during time and by this way the bioreactor worked at constant volume. The pH was controlled at a fixed value, according to the experimental design: during the biosorption process in the membrane reactor, the pH of the cell suspension becomes acidic because of the continuous addition of the metal bearing solution. In fact the acidic hydrolysis reaction of copper ion in solution releases continuously hydrogen ions in solutions.
The pH was monitored by a pH meter and kept constant by continuous additions of NaOH 0.1 and/or 0.01 N. In this particular case the pH variation cannot be used to understand biosorption mechanism operating in the system: in fact the original biosorbent was not protonated, and consequently the metal uptake can occur not only with the hydrogen but also with alkaline and alkaline-earth metal release (7). Work is in progress with protonated and calcium loaded biomass to evaluate both the mechanism and the stechiometry of the biosorption reaction. Different samples of permeate were collected during time and analyzed by an Atomic Absorption Spectrophotometer for copper concentration determination. Before the analysis, no procedure for cell separation was needed, since cells were completely retained by the membrane and did not pass in the permeate. The analysis with the Atomic Absorption Spectrophotometer was carried out at 327 nm with a flow rate of 7 mL/min; the calibration was performed by using a standard solution which is diluted by an automatic sampler obtaining five points for each calibration curve. The analysis of the metal concentration in solution is based on the mean values of three replicates, and after each measure the instrument executes a washing of 60 s by an aqueous solution of hydrochloric acid. A full factorial design was implemented with pH and biomass concentration as factors. In particular, pH levels were 4, 5, and 6 units, while biomass concentrations (X) were 1 and 5 g/L. All tests were replicated to evaluate the reproducibility and to estimate the experimental error variance. 2.3. Test Tube Trials. An original experimental procedure was introduced to simulate the membrane reactor system by a discontinuous trial avoiding or at least reducing the cell damage. These simulating tests were carried out in 50 mL test tubes: the biomass rests in the test tube (reactor), while a fresh metal solution was added discontinuously (feed) and successive samples (permeate) were collected after centrifugation (unit operation of separation analogous to the membrane module). The repeated exchange of a portion of the surnatant volume and the retaining of the biomass in the test tube simulate the flow through as in the membrane reactor. The use of centrifugation as solid/liquid separation device instead of the membrane apparatus permits one to obtain less mechanically stressing conditions and to validate the hypothesis of the effect of cell disruption on biosorption performances in the membrane reactor. The following experimental procedure was used in the test tube trials: (1) Lyophilized biomass was rehydrated under magnetic stirring for an hour in a volume V0 of distilled water. (2) A volume Vi of a copper solution at concentration Co was added to the cellular suspension. (3) The copper bearing cellular suspension was kept under agitation for 30 min (permitting the system to reach the equilibrium conditions), while the pH was maintained constant by NaOH additions. (4) The suspension was centrifuged (10 min at 7000 rpm), and a volume Vi of the surnatant was collected to measure the equilibrium metal concentration. (5) A volume Vi of fresh metal solution was added to the residual cell suspension in the test tube; this copper bearing cellular suspension was kept under agitation and at constant pH for 30 min and then it was centrifuged, and a second sample was taken for the determination of the metal residual concentration. A sequence of fresh metal bearing solution additions, centrifugation, and sampling was made to simulate discontinuously the biomass behavior in the membrane reactor. The same procedure was replicated without biomass to avoid confusion between biosorption and precipitation. The biomass concentration in the cellular suspension was 5 g/L; the effect of pH was studied at three levels (4, 5, and 6 units) VOL. 35, NO. 14, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3049
dC ) dt
[
V‚ 1 +
F‚(Co - C)
]
0.026‚exp(1.26‚pH)‚(-7.61‚pH2 + 69.7‚pH - 138.)‚X (C - 7.61‚pH2 + 69.7‚pH - 138.)2
(6)
where F is given by eq 3 and X is constant and equal to the biomass concentration at the beginning of the process. Equation 6 has been integrated by Runge-Kutta algorithm with the following initial conditions
t)0fC)0
FIGURE 2. qmax and Ks (equilibrium model parameters) vs pH profile. Points have been estimated from equilibrium tests described elsewhere (7), while the lines have been calculated by eqs 4 and 5, respectively, for qmax and Ks. and by using a variable concentration of the copper bearing solution as specified in the relative figure captions.
3. Results and Discussion 3.1. Modeling pH Effects. A first improvement of eq 2 was performed considering the dependence of equilibrium parameters, qmax and KS, on pH. The two main kinds of active sites that this biomass presents are the potentiometric titration of the Arthrobacter sp. and the relative modeling (7) outputs with pKa respectively around 6.5 and 10. These kinds of acidic sites can remove metallic ions from aqueous solutions through different mechanisms. For pH values greater than the pKa the sites are mainly in dissociated form and can exchange H+ with metal in solution. At pH lower than pKa values complexation phenomenon can also occur. The strong effect of pH on heavy metal biosorption can be explained by considering the deprotonation of the active sites as a pH function together with the competition among heavy metals and hydrogen ions in solution. Figure 2 shows the qmax and KS courses vs pH. To take into account the effect of pH in the membrane reactor biosorption modeling, the characteristic Langmuir parameters (estimated from equilibrium tests described elsewhere (7)) have been fitted by two empirical equations then introduced in the basic model (eq 2). In the investigated range of pH (4÷6 units), the following empirical equations (4 and 5) resulted to fit adequately the experimental data (regression coefficients, R2, (18), equal to 0.96 and 0.98, for eqs 4 and 5, respectively):
qmax(mg/g) ) 0.026‚exp(1.26 - pH)
(4)
KS (mg/L) ) -7.61‚pH2 + 69.7‚pH - 138
(5)
Empirical correlation is a simple tool to take into account the effect of pH: a mechanistic modeling approach to understand the effect of this experimental factor on biosorption performances was also considered (7) showing the phenomenon complexity and the necessity of a deeper experimental investigation at now in progress. Equations 4 and 5 have been introduced in eq 2, and the following eq 6 has been obtained for copper concentration in the retentate (equal to the permeate one), considering that copper retention coefficient was fixed at 0 (18) 3050
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
(7)
The calculated values have been compared to experimental data of the full investigated experimental factorial design with pH (three levels: 4, 5, and 6 units) and biomass concentration (two levels: 1 and 5 g/L) as factors. In Figures 3-6 some significant examples of the obtained experimental results have been shown. In each figure, the inlet flow rate profile (that is equal to the permeate one) is also shown: points represent experimental data, while the continuous line has been calculated by eq 3 after regression analysis for the estimation of parameters a, d, and c (13, 19). Furthermore, each figure reports the copper profile in the permeate in the case of no biomass inside the reactor. These simulations of the blank test have been performed for each test separately, because the inlet flow rate profile was different in each test, depending on operating conditions. The analysis of Figures 3-6 evidences the following aspects: (1) A very good agreement between experimental and simulated results has been achieved by using eq 6, for low biomass concentration and low pH, in the investigated range (Figures 3 and 5). (2) The simulation by eq 6 gives a complete disagreement for both levels of biomass concentration (1 and 5 g/L), at pH ) 6 (Figures 4 and 6). (3) Last but not least, a comparison between no biomass simulated data and the experimental values obtained in the presence of 1 g/L biomass and pH 4 (Figure 3) indicates that the copper sorption levels are quite low under these operating conditions. This aspect suggests that a model suitable for higher biomass concentration and pH is necessary for the simulation of an effective membrane biosorption process. In conclusion, it is evident that eq 6 does not succeed in describing the process as pH increases. Probably other phenomena occurring at pH ) 6 and not taken into account by eq 6, such as cell disruption and a certain copper retention by the membrane, could been also considered. 3.2. Mathematical Model for Cell Disruption. The model has been developed considering the previously developed one (13, 14), represented by eq 2, as a reference. Some modifications have been introduced in order to take into account other phenomena that might occur during the process. First of all, a dependence on pH of equilibrium parameters qmax and KS has been considered, as reported by eqs 4 and 5. Then, a partial cell disruption and a time dependent copper retention coefficient (σ) have been also considered. In fact, a probable cell disruption due to high stress associated with operating conditions might lead to a partial adsorption or pore plugging of cells debris on the membrane surface: as a consequence, the membrane is able to partially retain copper (as complex copper-cell debris), whose retention coefficient (8) becomes higher than zero
σ)1-
CP C
(8)
where CP is the concentration measured in the permeate and C is the concentration in the reactor.
FIGURE 3. Experimental (exp) and calculated (eq 6) data of permeate copper concentration (Cp) vs time at pH ) 4 with X ) 1 g/L. The blank curve has been calculated by eq 6 with X ) 0. Inlet flow rate profiles (Fi) as experimental (exp) and calculated (eq 3) data are also shown.
FIGURE 4. Experimental (exp) and calculated (eqs 6 and 13) data of permeate copper concentration (Cp) vs time at pH ) 6 with X ) 1 g/L. The blank curve has been calculated by eq 6 with X ) 0. Inlet flow rate profiles (Fi) as experimental (exp) and calculated (eq 3) data are also shown. An experimental confirmation of cell disruption was found by liquid chromatography analyses of permeate and retentate samples. Chromatograms (not shown here) evidenced the presence of many cell fragments both in the permeate and in the retentate with a σ factor of about 0.6. A first-order kinetics has been hypothesized for cell disruption
dX ) -kX dt
(9)
where X is the actual cell concentration (g/L) and k is the first-order rate constant (min-1). Integration of eq 9 with the following initial conditions
t ) 0 f X ) X0
(10)
X ) X0 exp(-kt)
(11)
FIGURE 5. Experimental (exp) and calculated (eq 6) data of permeate copper concentration (Cp) vs time at pH ) 4 with X ) 5 g/L. The blank curve has calculated by eq 6 with X ) 0. Inlet flow rate profiles (Fi) as experimental (exp) and calculated (eq 3) data are also shown.
FIGURE 6. Experimental (exp) and calculated (eq 6 and 13) data of permeate copper concentration (Cp) vs time at pH ) 6 with X ) 5 g/L. The blank curve has calculated by eq 6 with X ) 0. Inlet flow rate profiles (Fi) as experimental (exp) and calculated (eq 3) data are also shown. It is supposed that σ is strictly linked to disrupted cells’ concentration for two orders of reasons: (1) the cells’ fragments can bind free copper ions in solutions giving complexed copper species increasing its retention and (2) the cells’ fragments can plug membrane pores. Both events can occur even simultaneously probably causing an increase of the copper retention coefficient: in the first case for the larger dimensions of complexed copper with respect to free ion and in the latter case for the pore size decrease. By hypothesizing that the copper retention coefficient changes with a kinetic law of the same type of the cell degradation (9) it can be written that (12)
σ ) 1 - exp(-b‚t)
(12)
gives
where X0 represents cell’s concentration at the beginning of the process (g/L).
Even if the hypothesized kinetic laws are of the same type in both cases (9 and 12) the two rate constants (k and b) are generally different since different kinetics take place for cell disruption and for copper complexation and pore filling. Furthermore their difference takes also into account for a VOL. 35, NO. 14, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3051
FIGURE 7. Estimated values for k and b parameters (eqs 11 and 12) as a function of pH and initial cell concentration. stoichiometric consideration: a single cell disruption generates many cell fragments and each one of them might both complex copper ions and fill membrane pores. Combining eqs 2, 11, and 12 the following differential equation was obtained for copper concentration vs time profile in the retentate
dC ) dt
(
F‚(Co - C‚exp(-bt))
V‚ 1 +
)
qmaxKs‚X0 exp(-kt) (C + KS)2
(13)
where the inlet flow rate is given by eq 3 and equilibrium parameters are given by eqs 4 and 5. Equation 13 was integrated with the initial conditions (7). The copper profile in the permeate then can be calculated from the one in the retentate and copper retention coefficient, through eq 12. The two adjustable parameters, k and b, have been estimated through a nonlinear regression technique (19), minimizing the following objective function
φ)
∑(C
exp Pj
- CPjcalc)2
(14)
tion, from the other side also the cellular fragments can interact with the copper in solution (σ > 0). Comparing the experimental results with the predicted one (Figures 3-6), it seems that the negative effect due to X diminution prevails with respect to the positive interaction between cell debris and copper ions, except that at pH ) 6 and X ) 5 g/L. The mechanical stress in the system (caused by magnetic stirring, pumping, and recirculation) can be manipulated to evaluate the optimal operative conditions taking into account that the major is the cell debris, the major is the flux decline. 3.3. Test Tube Trials. Test tube trials consist of a discontinuous simulating system operating in less stressful conditions than the membrane reactor ones. These tests have the purpose of verifying the effect of cell disruption on biosorption by using a system working in a discontinuous mode and simulating the membrane reactor. The test tube system with respect to the membrane reactor presents milder operative conditions and consequently permits the evaluating of biosorption performances without or at least with a minor cell disruption. Experimental behavior was compared with simulation data obtained simply by combining the metal mass balance in the system and the Langmuir parameters at different pH values (7)
Vi‚Co + qi-1‚X‚V0 + V0‚Ci-1 ) Ci‚(V0 + Vi) + qi‚X‚V0 (15) where Vi (mL) is the added volume of fresh metal solution equal to the surnatant volume withdrawn after centrifugation; V0 (mL) is the volume of the initial cellular suspension; Co (mg/L) is the fresh metal bearing solution concentration; Ci (mg/L) is the equilibrium metal concentration of the ith sample; Ci-1 (mg/L) is the equilibrium metal concentration of the (i-1)th sample; X (g/L) is the biomass concentration; qi (mg/g) is the metal specific uptake with respect to the ith sample; and qi-1 (mg/g) is the metal specific uptake with respect to the (i-1)th sample. The expressions of qi and qi-1 are given by the Langmuir adsorption isotherms (eqs 16 and 17):
qi )
j
during the resolution of eq 13 by Runge-Kutta algorithms. Figures 4 and 6 show the good agreement between experimental and calculated profiles for copper concentration in the permeate obtained by using eq 13, at pH ) 6 for both biomass concentration levels (1 and 5 g/L). Figure 7 shows the estimated values for parameters k and b, as a function of pH and cell concentration. For X ) 1 g/L and pH ) 4 and pH ) 5, no regression analysis has been performed since the best results have been obtained by simulations with eq 6 (i.e. k ) b ) 0 in eq 13). As concerns the estimated values for parameters, it is evident in Figure 7 that both parameters are significantly dependent on pH. In fact they increase as the pH increases. In particular parameter k is quite high at pH ) 6 for both cell concentration levels: this implies a significant cell disruption during time, which leads to a consequent increasing copper retention coefficient. On the other hand, no significant cell disruption takes place at pH 5 with low biomass concentration (k ) 0). Furthermore, as expected, a constant copper rejection coefficient and equal to zero (b ) 0) was obtained just in the case of no cell disruption (k ) 0). Consequently, these data confirm the hypothesis that a cell disruption lead to precipitation of cells’ debris on the membrane surface, which partially plug membrane pores. From an operational point of view, the cell disruption could affect the metal removal in different ways. If a partial biomass damage diminishes the cell concentration in solu3052
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
qi-1 )
qmax‚Ci KS + C i
(16)
qmax‚Ci-1 KS + Ci-1
(17)
By replacing in eq 15 the expressions 16 and 17 it is possible to obtain a relation between the equilibrium concentration Ci and the number of samples (i) without considering any effect of the cell disruption. Also the blank trial is simulated simply by considering the system as a CSTR (18):
Vi‚Co + V0‚Ci-1 ) Ci‚(V0 + Vi)
(18)
The test tube system resulted in a good simulator for a membrane reactor permitting evaluation of the effect of mechanical stress on cells, which take place by using the membrane system at high biomass concentrations. In fact, chromatographic determinations on test tube samples (not shown here) have evidenced that no significant cell disruption took place. In Figures 8-11 experimental data are reported for biosorption trials in test tubes for Cu2+ removal at constant pH values (4, 5, and 6 units) with 5 g/L of biomass concentration. The experimental evidence shows that (1) at pH ) 4 and pH ) 5 simulated and experimental trends are very near even at high biomass concentration (5 g/L) without considering cells’ disruption effect in the model simulation as can be seen in Figures 8 and 9. It confirms that the
FIGURE 8. Test tube trial at pH ) 4 with inlet concentration of 16.7 mg/L of Cu2+ and X ) 5 g/L: biosorption and blank saturation curves experimentally determined (exp) and simulated (eqs 15 and 18, respectively).
FIGURE 9. Test tube trial at pH ) 5 with inlet concentration of 48.6 mg/L of Cu2+ and X ) 5 g/L: biosorption and blank saturation curves experimentally determined (exp) and simulated (eqs 15 and 18, respectively). deviations observed in the membrane reactor at high biomass concentrations are due to a partial cells’ disruption which instead has little impact in this milder operating system. (2) At pH ) 6 there are relevant differences between simulated and experimental trends: (i) at high inlet metal concentrations (66.6 mg/L of Cu2+) a nonmonotonic but reproducible trend was observed (Figure 10); (ii) at low concentrations of the inlet metal bearing solution (20.7 mg/L) the trend is again monotonic but very different from the predicted one (Figure 11). The first kind of anomaly (at 66.6 mg/L of Cu2+) was already observed in the membrane reactor (compare Figures 4 and 6 with 10) and confirms the goodness of test tube trials’ simulation. The nonmonotonic trend is probably due to a certain kind of chemical degradation of the bacterial cell wall which caused the availability of new acidic active sites. The main constituents of the cell wall of Gram-positive bacteria are peptidoglycan and lipopolysaccharides which form a tridimensional network covering the cell by different layers (20-22). These constituents can undergo a chemical degradation related to pH, discovering new sites of the inner layers. In fact the biomass at a certain point (reproducible) of the test tube trials starts to increase its accumulation capacity when the saturation curve presents a minimum.
FIGURE 10. Test tube trial at pH ) 6 with inlet concentration of 66.6 mg/L of Cu2+ and X ) 5 g/L: biosorption and blank saturation curves experimentally determined (exp) and simulated (eqs 15 and 18, respectively).
FIGURE 11. Test tube trial at pH ) 6 with inlet concentration of 20.7 mg/L of Cu2+ and X ) 5 g/L: biosorption and blank saturation curves experimentally determined (exp) and simulated (eqs 15 and 18, respectively). Using the 20.7 mg/L Cu inlet concentration, the minimum disappears even if the experimental and predicted trends are very unlike. The monotonic pattern observed in Figure 11 could be due to the fact that the possible new available sites generated by the chemical degradation exceed the metal in solution for 20.7 mg/L Cu inlet concentration. On the other side the difference between predicted and experimental data signifies that further phenomenon have to be considered. The test tube trials simulating the membrane reactor apparatus output that the partial cell disruption was due not only to the high stressing operative conditions in the membrane reactor but also to a probable chemical degradation of the cell membrane occurring at pH ) 6. In fact even using low NaOH concentrations (0.1 and 0.01 N) for the pH adjustment, when a NaOH drop is added, the local concentration could cause a partial cell membrane degradation of the biomass first coming in contact. In conclusion test tube trials can be considered a good and simple discontinuous simulation system of the membrane reactor. Test tube trials show that a model not considering cell disruption can represent quite well the experimental data with X ) 5 g/L up to pH ) 5. Moreover these tests, as a milder simulation of the membrane reactor, permit to isolate the effect of chemical degradation of the bacterial cell wall at pH ) 6. VOL. 35, NO. 14, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3053
Acknowledgments The authors are grateful to Mr. Marcello Centofanti and Mrs. Lia Mosca for their helpful collaboration in the experimental work.
Literature Cited (1) Veglio`, F.; Beolchini, F. Hydrometallurgy 1997, 44, 301. (2) Figueira, M. M.; Volesky, B.; Ciminelli, V. S. T.; Roddick F. A. Water Res. 2000, 34(1), 196-204. (3) Davis, T. A.; Volesky, B.; Vieira, R. H. S. F. Water Res. 2000, 34(17), 4270-4278. (4) Kratochvil, D.; Volesky, B. Water Res. 2000, 34(12), 3186-3196. (5) Chang, J. S.; Chen, C. C. Separation Sci. Technol. 1999, 34(8), 1607-1627. (6) Schiewer, S.; Wong, M. H. Chemosphere 2000, 41, 271-282. (7) Pagnanelli, F.; Petrangeli Papini, M.; Trifoni, M.; Toro, L.; Veglio `, F. Environ. Sci Technol. 2000, 34, 2773-2778. (8) Veglio`, F.; Beolchini, F.; Gasbarro, A.; Toro, L. Chem. Biochem. Eng. Q. 1999, 13(1), 9-14. (9) Veglio`, F.; Beolchini, F.; Gasbarro, A.; Lora, S.; Corain, B.; Toro, L. Hydrometallurgy 1997, 44, 317. (10) Veglio`, F.; Beolchini, F.; Boaro, M.; Lora, S.; Corain, B.; Toro, L. Proc. Biochem. 1999, 34, 367. (11) Veglio`, F.; Beolchini, F.; Toro, L. Ind., Eng. Chem. Res. 1998, 37, 3, 1107. (12) Veglio`, F.; Beolchini, F.; Quaresima, R.; Toro, L. In Biohydrometallurgy and the environment toward the mining of the 21st century; Amils, R., Ballester, A., Eds.; Elsevier: 1999.
3054
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 14, 2001
(13) Barba, D.; Beolchini, F.; Veglio`, F. Hydrometallurgy 2001, 59(1), 89-99. (14) Veglio`, F.; Beolchini, F.; Barba, D. Ind. Eng. Chem. Res. 2000, 39(7), 2480-2484. (15) Duncan, J. R.; Brady, D.; Stoll, A.; Wilhelmi, B. In Biohydrometallurgical Processing; Jerez, C. A., Vargas, T., Toledo, H., Wiertz, J. V., Eds; University of Chile: Santiago, Chile, 1995. (16) Veglio`, F.; Beolchini, F.; Gasbarro, A. Proc. Biochem. 1997, 32(2), 99. (17) Holan, Z. R.; Volesky, B.; Prasetyo, I., Biotechnol., Bioeng. 1993, 41, 819. (18) Cheryan, M. Ultrafiltration and microfiltration handbook; Technomic Publishing Co., Inc.: Lancaster, Basel, Switzerland, 1998. (19) Himmelblau, D. M. Process Analysis by Statistical Methods; John Wiley & Sons: 1978. (20) Voet, D.; Voet, J. G. Biochemistry; John Wiley & Sons: 1990. (21) Plette, A. C. C.; Van Riemsdijk, W. H.; Van der Wal, A. J. Colloid Interface Sci. 1995, 173, 354-363. (22) Plette, A. C. C.; Benedetti, M. F.; Van Riemsdijk, W. H. Environ. Sci. Technol. 1996, 30, 1902-1910.
Received for review July 18, 2000. Revised manuscript received March 23, 2001. Accepted April 12, 2001. ES000159B
