Selective Removal of Cobalt Species Using Nanofiltration Membranes

Selective removal of cobalt species from simulated nuclear liquid waste was investigated with different nanofiltration (NF) membranes at various solut...
15 downloads 0 Views 97KB Size
Download PDF
Environ. Sci. Technol. 2002, 36, 1330-1336

Selective Removal of Cobalt Species Using Nanofiltration Membranes KWANG-HO CHOO,† DAE-JOONG KWON,‡ KWANG-WON LEE,‡ AND S A N G - J U N E C H O I * ,‡ Department of Architectural, Civil, and Environmental Engineering, Daegu University, Gyeongsan, Gyeongbuk, 712-714 Korea, and Department of Environmental Engineering, Kyungpook National University, Daegu, 702-701 Korea

Selective removal of cobalt species from simulated nuclear liquid waste was investigated with different nanofiltration (NF) membranes at various solution pH levels, initial cobalt concentrations, and background ion concentrations. This study provides insight into the understanding of the relationships between rejections of a target compound (cobalt) and chemical equilibria of various species in the feed solution during NF. Particularly, the ratio of electrostatic rejection to steric rejection for different membranes used was quantitatively evaluated to find out the relative significance in NF. Substantial cobalt rejection by NF was achieved along with partial separation of monovalent ionic species, although it depended on the level of liquid pH and the presence of background species. Greater cobalt rejection at increased pH was attributed to the precipitation of CoCO3(s) associated with natural carbonates originating from atmospheric CO2 gas rather than that of Co(OH)2(s). A loose NF membrane (e.g., NTR7410) gave as high a rejection as other tighter ones due to the stronger influence of electrostatic rejection, particularly at low pH where no cobalt precipitation was occurring. The decrease of cobalt rejection with the addition of boric acid was found to occur due to the formation of complexes between cobalt and boric acid, which was verified by the analyses of solution turbidity and near-infrared spectroscopy.

Introduction Nuclear-fueled power-generating plants, which require large quantities of water to dissipate the heat released by nuclear reactions, are now in operation in many countries throughout the world (1). The operation of nuclear reactors thus creates a large amount of radioactive waste produced through fission and neutron capture, which is of great concern because of the induced radioactivity of the waste and the existence of some fission products (2-6). In particular, low-level radioactive wastes (LLRW), which arise from the operation of nuclear facilities as well as residue from medical and industrial use of radionuclides, account for more than 85% of the volume of all radioactive wastes, even though their level of radioactivity is relatively low as compared to high-level and * Corresponding author phone: +82-53-956-6582; fax: +82-53950-6579; e-mail: [email protected]. † Daegu University. ‡ Kyungpook National University. 1330

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

transuranic radioactive wastes (1, 7). Thus, volume reduction is essential for any generator of LLRW, along with radionuclides removal, since it eliminates far-reaching “cradle-tograve” liabilities and preserves space for the disposal of materials that do require long-term isolation. Recently, a number of different treatment methods have been applied to reduce the volume of liquid radioactive waste particularly related to LLRW as well as to remove radionuclides from the waste, such as thermal treatment, extraction, sorption, ion exchange, electrolytic reduction, and membrane separation (7-19). Thermal processes including evaporation, incineration, and vitrification helped increase the solids content of the waste considerably, and thermal treatment in combination with vitrification or wet oxidation has also been carried out for volume reduction and hazards control depending on the characteristics of the radioactive waste (8-10, 20, 21). The thermal processes, however, concentrate all species that are not evaporable including nonradioactive components with active ones and, furthermore, need lots of energy for operation. Extraction methods using solvent or supercritical fluid were employed to recover and concentrate highly radioactive contaminants such as uranium, plutonium, or other actinides, allowing the disposal of the process raffinate as a LLRW (11, 12, 22-24). But the solvent extraction process still suffers from many difficulties such as solvent degradation and inadequate decontamination efficiency (11, 12, 22). The supercritical fluid extraction method is still in the early stage of evaluation and requires an extreme operating condition in pressure and temperature. Several different sorbents such as zeolite, activated carbon, and microorganisms have been investigated in terms of decontamination of the liquid wastes and concentration of radioactive species (13-15, 25-29). The adsorption methods are confronting some issues yet to be solved, such as selectivity, regeneration, and sludge production. Ionexchange processes have been demonstrated to remove radioactive ions including cobalt and cesium from the liquid waste, but there is a need for the development of novel ionexchange resins with greater selectivity for the removal of contaminants alone (16, 30-32). The electrochemical treatment of mercury, nitrate, or nitrite in low or alkaline nuclear waste was studied and showed some promise for treating large volumes of waste (17, 33, 34), but further economic analyses are necessary to point out its advantages. Tight reverse osmosis (RO) membranes have been tested for the treatment of laundry or simulated liquid wastes with low-level radioactivity in order to minimize its volume before disposal (35, 36). The concentration of the wastes by RO was possible, but some limitations such as the fouling of membranes and the need for pretreatment should be further investigated. In addition, RO is not selective in the separation of radionuclides and nonactive species such as monovalent ions and moreover needs high operating pressures. To overcome these issues, the use of low-pressure membrane processes using loose membranes alone or in combination with other physicochemical techniques has been investigated (18, 19, 37-42). For instance, the membrane distillation through porous membranes (18), the use of microfilters with ion-exchange capabilities (19, 38), the combination of complexation and ultra/nanofiltration (39, 40), and the formation of liquid membranes (41, 42) have been studied, but they also have many drawbacks with respect to radionuclide removal efficiency, extent of volume reduction, and process stability. However, few studies have been conducted on nanofiltration (NF), which can make it possible not only to selectively separate radionuclides from nonactive com10.1021/es010724q CCC: $22.00

 2002 American Chemical Society Published on Web 02/07/2002

TABLE 1. Properties of the NF Membranes Used

membrane

manufacturer

material

pore radius (nm)

NTR7410 NTR7250 NTR729HF NF45

Nitto Denko Nitto Denko Nitto Denko FilmTech

sulfonated polysulfone poly(vinyl alcohol)/ polyamide poly(vinyl alcohol)/ polyamide polypiperazine amide

4.21 0.45 0.35 0.48

a Rejection of neutral solutes. b Rejection of NaCl. c na, not available. membranes, the predecessor of the NF45 membrane.

pounds but also to operate the system at relatively low pressures as compared to RO (40, 43). The separation of charged contaminants in NF is governed by two interesting factors: one is size exclusion (steric hindrance) by the membrane pores and the other is the electrostatic interaction between solutes and the membrane surfaces. Several researchers have focused on the transport phenomena of charged solutes across a charged NF membrane using different models described by the irreversible thermodynamics, the Donnan equilibrium, and the modified Nernst-Planck equation (44, 45). However, the size exclusion (steric rejection) of charged solutes by NF has been explained based on the analysis of neutral solute permeability. Thus, the exact physical properties of actual charged solutes and specific chemical reactions were not included in their approaches. In this work, therefore, NF methods were applied for the treatment of simulated nuclear wastewater containing cobalt, boric acid (BA), and salts. The NF performance was evaluated in terms of cobalt removal and selectivity between target and background species such as boron and sodium using a lab-scale cross-flow NF system. The effects of initial cobalt concentration, solution pH, and complexation phenomena on cobalt removal for different NF membranes were explored along with theoretical approaches to chemical equilibria of cobalt species and solution chemistries. Also, an attempt to discriminate between the steric and the electrostatic effects on overall rejection was made to investigate the significance of each effect at different pH levels for the membrane tested.

Experimental Section Feedwater. For the preparation of synthetic nuclear wastewater, composed of three representative components of cobalt, borate, and inert salts, analytical grade chemicals of Co(NO3)2, H3BO3, and NaCl were purchased from Aldrich Chemicals (Sigma-Aldrich, Korea) and dissolved in pure water. The exact concentration of each component in the feed solution was determined by the analytical methods described later. The target compound to be removed (cobalt) was actually nonradioactive because of ease of handling and operation in lab experiments, whose concentration was in the range of 1-30 mg/L as Co, corresponding to 0.085-0.51 mM. The concentration of BA, which is used to control nuclear reaction rates in actual plants, ranged from 0 to 100 mg/L as B (corresponding to 0-9.2 mM), whereas that of NaCl, which represented inert electrolytes in feed, was in the range of 0-200 mg/L as Na (corresponding to 0-8.7 mM). The composition of artificial feed wastewater for each experimental run, which is based on approximate concentrations of real low-level nuclear wastewater, varied depending on experimental conditions. The feed solution pH was adjusted by adding sodium hydroxide (0.1 N) or hydrogen chloride (0.1 N), ranging from 3 to 9. NF Membranes. The NF membranes used were the Nitto Denko NTR7410, NTR7250, and NTR729HF membranes (Nitto Denko Corp., Japan) and the FilmTech NF45 membrane (FilmTech Corp., Australia), whose detailed informa-

d

ζ-potential at neutral pH (mV)

isoelectric point

specific water permeability (L m-2 h-1 bar-1)

RNSa (%)

RNaClb (%)

-12.5 -5 na -2e

nac 4.0 na 4.6e-6.5

50.0 16.4 4.10 4.68

5 (sd) 94 (gd) 97 (gd) 93 (gd)

15 60 92 58

Model neutral solutes used: s, sucrose; g, glucose. e Value for NF40

tion that was obtained from the manufacturer and literature is given in Table 1 (44, 46-49). NF System Operation. A lab-scale plate-and-frame NF system with an effective membrane surface area of 25 cm2 was operated in a conventional cross-flow mode, with a typical tangential velocity of 1.0 m/s (1911 in Reynolds number) through the membrane channel, corresponding to a circulation flow rate of 1.5 L/min. During all NF tests, the operating temperature and pressure were always constant at 20 ( 2 °C and 4.9 bar (5.0 kgf/cm2), respectively. The applied pressure was at the reasonable, low end of those typically used for NF operations. Filtration runs were performed in closed loop operation with full recirculation of retentate and permeate to the feed tank except for taking 10 mL for each sampling. Initially, the 2-L feed tank was filled up with synthetic nuclear wastewater and then operated for 1 h at a predetermined pH value while taking 10 mL of retentate and permeate every 10 min for the Nitto Denko membranes. Subsequently, the feed pH was raised by a value of approximately 1 with NaOH addition repeating the same operating procedure as mentioned above until the feed pH reached more than 8-9. Sampling for 1-h runs at each pH level was done five times to check the consistency and reproducibility of the rejection data. During the whole operation, permeate flux was determined every 10 min by measuring the mass of permeate with a three-digit electronic balance at a given time interval. For the FilmTech NF45 membrane, the operation time and sampling interval at each pH were extended 2-fold due to a relatively low membrane flux, but the other experimental conditions are the same as those of the Nitto Denko membranes. During all experimental runs, continuous monitoring and control of operating parameters such as transmembrane pressure, flow rate, temperature, and solution pH was done to minimize the experimental errors. Preparation of Non-Carbonate Systems. To prepare a system with no carbonate species in the feed solution, all carbonate species should be removed from the solution and then the CO2 gas should be kept from dissolving from the air to the feed tank during NF. To do that, the feed pH first dropped to 4 with the addition of acid, then the solution was purged with N2 gas, and finally the upper dead space of the feed tank was continuously purged using N2 through the whole operation with the tank covered. All experiments except for the NF run described above were performed under a system open to the atmosphere. Determination of the Point of Zero Charges. For the determination of the point of zero charges (PZC) of the NF membranes used (corresponding to the isoelectric point of amphoteric membranes), the rejection of sodium by the NF membranes was measured using 0.2 M NaCl solution at the pH range of 3-7. The pH where a minimal rejection of NaCl (corresponding to a turning point in NaCl rejection) was obtained was considered the PZC of the membranes used (50). When the membrane was not amphoteric (e.g., NTR7410), no turning point in NaCl rejection was expected in the entire pH range tested, and so the low end of the pH VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1331

FIGURE 1. Average permeate flux vs feed pH for different NF membranes: feed composition, 0.0848 mM Co(NO3)2; operating pressure, 4.9 bar; tangential velocity, 1.0 m/s. examined (close to the low pH limit of NF system operation) was assumed to be the PZC. The PZC values obtained in this method were compared with those from the literature using streaming potential measurement techniques (47, 51, 52) Analytical Methods. The concentrations of cobalt and sodium in feed, retentate, and permeate were measured using an atomic adsorption (AA) spectrophotometer (Varian Spectra AA-250 Plus, Australia), while the measurement of boron concentration was done using an inductively coupled plasma spectrophotometer (Jobin-Yvon 38 plus, UK). The nearinfrared (NIR) spectroscopy (Varian CARY5G, Australia) was adopted to examine the complexation interaction between cobalt and BA. The wavelength NIR rate and interval were 120 nm/min and 1.0 nm, respectively, over the scan wavelength range of 1200-1350 nm. The pH of solutions was measured using a pH meter after calibrating the instrument with three buffer solutions every time. During all analyses, multiple analyses (at least three times) were done on a single sample for analytical precision. Equilibrium Calculations. Molar distribution diagrams of cobalt species involving different combinations of components dissolved in feed solution were computed as a master variable of pH using the program of MINEQL+ version 4.0 for Windows (Environmental Research Software, USA). The option of systems either open to atmosphere CO2 or closed was selected when carbonate was included in the component list. Under normal atmospheric conditions, CO2 was assumed to have a partial pressure of 10-3.5 atm.

Results and Discussion Membrane Permeability. The flux profiles for the NF membranes used with pH are shown in Figure 1. As anticipated from pure water permeability (Table 1), the flux for the NTR7410 membrane was always kept at its highest levels, while the NTR7250 membrane was in the middle, and the NRT729HF and NF45 membranes remained at the lowest levels. Similar trends at higher operating pressures or tangential velocities for the membranes used were expected with improved levels in permeate flux. This is because higher operating pressures offer a larger driving force to permeation in a pressure-driven NF and higher tangential velocities help decrease the concentration polarization due to turbulent flows at the membrane surface. For the NTR7410 membrane, the flux decreased slightly when pH increased, which might be caused by the changes of solution chemistry during membrane operations. This phenomenon will be discussed in detail in later sections. In addition, the time dependence of flux was nearly negligible during the short-term NF runs (data not shown), showing that minimal membrane fouling 1332

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

FIGURE 2. Effect of feed cobalt concentration and pH on cobalt rejection when no boric acid (BA) and NaCl exist in the feed: membrane, NTR7250; operating pressure, 4.9 bar; tangential velocity, 1.0 m/s. was occurring. Other than that, the flux decline could not be a significant issue yet in this study, so that during the rest of the study more attention was paid to the membrane performance associated with rejection efficiency. Effect of Feed pH and Cobalt Concentrations on Cobalt Rejection. Since the cobalt concentration and pH of actual nuclear wastewater is normally fluctuating within a certain range, cobalt rejection efficiencies were examined using feed wastewater with different pH levels and cobalt concentrations (Figure 2). Cobalt rejection started increasing around pH 6.5-7 and reached nearly complete rejection at the pH levels higher than 8.5. On the other hand, with increasing the feed cobalt concentration from 1.0 mg/L (0.017 mM) to 30 mg/L (0.51 mM), cobalt rejection tended to somewhat decrease particularly at low pH, but the extent was not so significant as the pH effect. The higher cobalt rejection at increased pH was hypothesized to stem from the change of chemical forms of cobalt species, such as formation of Co(OH)2, while the lower rejection efficiencies with increasing feed concentrations were possibly due to a consequence of a higher diffusive driving force at the membrane surface. To find out the precipitation potential of cobalt hydroxide with pH, a theoretical approach was made based on its chemical equilibria, and thus the molar distribution of cobalt species at different pH levels was computed (data not shown). The formation of Co(OH)2 solids began from a pH of approximately 8, so the difference between experimental and theoretical data seemed too large. This was contradictory to the assumption that the precipitation of cobalt hydroxides was responsible for the higher cobalt rejection when pH was increased. As a result, another point should be considered, and ultimately we suspected that the CO2 in the air might play a role in cobalt rejection. Thus, carbonate species were included in the calculation of molar distributions of cobalt species, assuming two different conditions: one is that the carbonate system is closed and the other is open to an atmosphere containing CO2. First, with the assumption of a closed system, CoCO3 solids started forming at pH 7, but the amount of precipitate was not significantly large enough yet to explain the experimental results as shown in Figure 2. When the second assumption of a heterogeneous, open system in which the carbonate species in feed are in equilibrium with the CO2 gas in the atmosphere above the solution was applied, CoCO3 precipitation was occurring from pH 6.5 and completed at pH 7.5. The starting point of precipitation nearly coincided with that of the experimental data, strongly supporting the involvement of CO2 in cobalt rejection.

FIGURE 3. Calculations of the percentage of precipitates of initial cobalt at various pH levels studied when different carbonate systems are involved in solution equilibria: (O) a system with no carbonate species; (0) an open system with an equilibrium dissolved CO2 concentration of 10-5 M; (b) a closed system with a total carbonate concentration of 10-5 M. The total cobalt concentration is 0.0848 mM.

FIGURE 5. Effect of membrane types on cobalt rejection: feed composition, 0.0848 mM Co(NO3)2; operating pressure, 4.9 bar; tangential velocity, 1.0 m/s.

TABLE 2. Measurements of Point of Zero Charges (PZC) and Cobalt Rejection Efficiencies at the PZC membrane

PZC

cobalt rejection (%)

NTR7410a

3.3 4.4 5.1 5.7

4.64 24.6 65.6 35.5

NTR7250 NTR729HF NF45

a The membrane is not amphoteric, so cobalt rejection at the PZC was based on the low end of the pH levels tested.

FIGURE 4. Cobalt rejection vs pH with and without carbonate species in feedwater: feed composition, 0.0848 mM Co(NO3)2; membrane, NTR7410; operating pressure, 4.9 bar; tangential velocity, 1.0 m/s. The system without carbonate species was prepared with the drop of initial solution pH to 4 and nitrogen purging during NF. The precipitated percent of initial cobalt with pH for the above three cases are summarized in Figure 3. Although the presence of carbonate species in the system helped explain the higher cobalt rejection due to precipitation, further explanation on the extent of precipitation and experimental verification are apparently needed. The actual NF operations must have been between closed and open carbonate systems because the feed tank was actually open to the CO2 gas in the air but the carbonate concentration in feed solution may depend on the rate of dissolution of CO2. To further confirm the effect of carbonate on cobalt rejection, another experimental run was performed in which all carbonate species were first removed by the drop of feed pH to 4 and N2 purging, and then throughout NF the feed tank was covered and the upper dead space was continuously purged with N2. It was found that cobalt rejection substantially decreased when feed solution was prevented from contacting the air (Figure 4). Consequently, there was no doubt that cobalt carbonates played a major role in the greater cobalt rejection with the increase of pH during NF instead of cobalt hydroxides. In addition, the deposition of such precipitates at the membrane surface might cause flux decline, particularly during longtime NF. In fact, as shown in Figure 1, the slight flux decline of the most porous NTR7410 membrane at high feed

pH levels could be associated with the deposition of cobalt precipitates. Effect of Membrane Types on Cobalt Rejection. The effect of the membrane types used on cobalt rejection at different pH levels is shown in Figure 5. Of the membranes used, the NTR729HF membrane that is densest according to the rejection data of Table 1 always gave pronounced cobalt rejection efficiencies of more than 75% over the wide range of pH tested, but no significant difference in cobalt rejection was observed for the other three membranes used. Interestingly, however, it should be noticed that the NTR7410 membrane with very low sucrose and NaCl rejection efficiencies of only 5% and 15%, respectively (Table 1), had substantial cobalt removal comparable or similar to the NTR7250 and NF45 membranes with a high NaCl rejection of 58-60% starting from low and neutral pH levels, respectively. This could be attributed to an effect of the electrostatic interaction between NTR7410 membrane surfaces and cobalt (which can be termed electrostatic rejection), since the NTR7410 membrane was most porous and thereby cobalt rejection by size exclusion (which can be termed steric rejection) was expected to be relatively small as compared to the other two NF membranes. Further cobalt rejection by the NTR7410 membrane was also achieved with increasing solution pH, indicating that open NF membranes such as NTR7410 would be as attractive for cobalt removal as other moderately dense NF membranes along with pH adjustments, being operated even at lower transmembrane pressures. An attempt was made to separate and compare the significance and contribution of electrostatic rejection by electrical repulsion and steric rejection by size exclusion on overall cobalt removal. To do that, the PZC of the membranes used and corresponding cobalt rejection at that point was needed. Experimentally, the PZC values were determined as the turning point of NaCl rejection with changing pH, i.e., the pH where the NaCl rejection efficiency was minimal (50). The PZC values and cobalt rejection efficiencies measured at the points are given in Table 2. The PZC values VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1333

FIGURE 6. Ratio of electrostatic rejection to steric rejection for different membranes used: feed composition, 0.0848 mM Co(NO3)2; operating pressure, 4.9 bar; tangential velocity, 1.0 m/s.

FIGURE 7. Effect of background components on cobalt rejection: basic feed composition, 0.0848 mM Co(NO3)2; membrane, NTR7250; operating pressure, 4.9 bar; tangential velocity, 1.0 m/s.

obtained in this study were very similar to those of Table 1 that were obtained from the literature by streaming potential measurements (47, 48). For the estimation of steric and electrostatic rejections during NF, it was assumed that cobalt rejection at PZC was occurring only by steric rejection and its contribution at different pH was always at the same rate. In addition to the exclusion of dissolved species, steric rejection also included the influence of chemical precipitation with the changes of pH, and thus electrostatic rejection was regarded as the remainder of overall cobalt rejection. The

steric (RS) and electrostatic (RE) rejection efficiencies can be estimated using the following equations:

RS ) 100fs + (1 - fs)RPZC

(1)

RE ) Ro - RS

(2)

where fs is the precipitated fraction of total amounts of cobalt in the feed; RPZC is the rejection efficiency at PZC; and Ro is the overall cobalt rejection efficiency. The quantitative

FIGURE 8. Effect of boric acid (BA) addition on cobalt rejection using the feed containing 0.0848 mM Co(NO3)2 alone or its mixture with 9.25 mM BA. The experiments were carried out at a transmembrane pressure of 4.9 bar and a tangential velocity of 1.0 m/s with (a) NTR7410, (b) NTR7250, (c) NTR729HF, and (d) NF45 membranes. 1334

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

FIGURE 9. Variation of feed turbidity with pH at different BA concentrations. The number in the parentheses of legend indicates the molar ratios of BA to cobalt. evaluation of steric and electrostatic rejections was possible using the above two equations, and thereby the relative importance of each of them could be assessed. The ratios of electrostatic rejection to steric rejection with pH for the membranes used are compared in Figure 6. The NTR7410 membrane offered a very large contribution of electrostatic effect to cobalt rejection, particularly at low pH levels. This supports the fact that great cobalt rejection by the loose NTR7410 membrane was caused by electrostatic interactions rather than steric hindrance when feed pH was acidic and thereby no chemical precipitation took place in the system. The importance of membrane surface charge in NF has been also pointed out by some investigators (50-52). Effect of Background Components on Cobalt Rejection. BA and sodium chloride were added as key background compounds to the feed solution containing target cobalt species (actually Co(NO3)2). Cobalt rejection efficiencies for NF of feed solution with different composition of background compounds are shown in Figure 7. The cobalt rejection efficiencies were not affected by the addition of sodium chloride, whereas when a relatively large amount of BA was added, cobalt rejection was greatly reduced and not so much improved by the pH increase in contrast to the results without the addition of BA. This suggested that BA might be interfering with the formation of cobalt precipitates at higher pH. To further verify the effect of the role of BA, cobalt rejection using feed solution with different BA concentrations was evaluated. When more BA was added to the feed solution containing 0.0848 mM Co at pH 7.9, that is, the molar ratio of BA to cobalt concentrations increased from 0 to 102, cobalt rejection by NTR7250 almost linearly declined from 76% to 42%. The effects of BA addition on cobalt rejection for the different membranes used were further compared and are shown in Figure 8. First, the NTR7410 membrane, which is most loose, showed a substantial reduction in cobalt rejection particularly at high pH levels when 9.25 mM BA was added to the feed solution containing 0.0808 mM cobalt nitrate. The denser NTR7250 membrane had nearly the same trend, even though the decline in rejection efficiency was relatively small as compared to the NTR7410 membrane. For the NTR729HF membrane, however, there was a very negligible difference in cobalt rejection between the absence and the presence of BA in the feed, indicating that the membrane was able to reject enough solute despite the addition of BA and consequent changes of water chemistries. The presence of BA also had nearly no effect on cobalt rejection for the NF45 membrane that had similar characteristics in glucose and NaCl rejection to the NTR7250 membrane. This might be caused by the difference of membrane material and structure (44, 45). Further research will be needed on the influence of physicochemical properties of the membranes.

FIGURE 10. Examination of complexation reactions between cobalt and BA by near-infrared (NIR) spectroscopy.

FIGURE 11. Separation of BA at different pH levels by NF. NF testings were done using the feed containing 0.0848 mM Co(NO3)2, 9.25 mM BA, and 4.35 mM NaCl at an operating pressure of 4.9 bar and a tangential velocity of 1.0 m/s. Although cobalt rejection somewhat depended on the type of membrane, the decrease of NF performance in the presence of BA was apparent. One possible explanation for this phenomenon is that the complexation of cobalt with BA is occurring and interfering with cobalt precipitation, so cobalt species are soluble and thereby pass through the NF membrane more easily. To find out that BA inhibited the formation of cobalt precipitates, the variations of turbidity in feed solution with different BA concentrations ranging from 0 to 3.42 mM were observed at different pH levels and are displayed in Figure 9. The turbidities decreased substantially even at high pH values when the BA concentration added was as low as 0.65 mM, which implies that BA is forming a dissolved complex with cobalt. To further verify the bonding between cobalt and BA, spectrophotometric investigation of the solutions with different combinations of cobalt and BA by NIR was done (Figure 10). The NIR band at a wavelength of approximately 1330 nm that was not monitored for the single solute systems tested exhibited the existence of a Co-O-B bond, which is indicative of complex formation reactions between cobalt and BA. Permeation of Nonhazardous Compounds. Another issue concerned with the application of NF for treatment of cobaltbearing nuclear wastewater was to separate nonhazardous compounds from cobalt species. Rejection efficiencies of BA at different pH levels are demonstrated in Figure 11. BA rejection was almost negligible for the NTR7410 and NTR7250 membranes used, while the rejection efficiency by the VOL. 36, NO. 6, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

1335

NTR729HF and NF45 membranes slightly increased but were still kept less than 30% over the entire pH range tested. Na rejection by the NTR7250 membrane was also always less than 30%, that is, more than 70% of Na passed through the NF membrane (data not shown). The partial permeation of sodium by NF (more than 70%) was still advantageous because RO membranes generally reject most of monovalent ions (>90%) that can cause additional serious issues in the final disposal process of consolidated radioactive waste, such as leaching problems. The higher permeation of Na occurred possibly due to a Donnan effect resulting from the retention of divalent cobalt ions. The results showed that the separation efficiency of nonhazardous compounds was reasonable and that the NF process was promising for selective removal of cobalt and nonhazardous compounds.

Acknowledgments This work was supported by the Korean Ministry of Science and Technology as a part of the 1999 Nuclear Research and Development Program. The authors thank Prof. Yamamoto of the Environmental Science Center at the University of Tokyo, Japan, for his providing us with Nitto Denko membranes.

Literature Cited (1) Bodansky, D. Nuclear Energy-Principles, Practices, and Prospects; American Institute of Physics: New York, 1996. (2) Biedscheid, J. A. Water Environ. Res. 1999, 71, 916-930. (3) Devarakonda, M. S. Water Environ. Res. 1993, 65, 452-459. (4) Matishov, G. G.; Matishov, D. G.; Namjatov, A. A.; Carroll, J.; Dahle, S. J. Environ. Radioact. 2000, 48, 5-21. (5) Hawickhorst, W. Nucl. Eng. Des. 1997, 176, 171-176. (6) Decamps, F.; Dujacquier, L. Nucl. Energy Des. 1997, 176, 1-7. (7) Gershey, E. L.; Klein, R. C.; Party, E.; Wilkerson, A. Low-Level Radioactive Waste-From Cradle To Grave; Van Nostrand Reinhold: New York, 1990. (8) Sheng, J.; Choi, K.; Yang, K. H.; Lee, M. C.; Song, M. J. Nucl. Technol. 2000, 129, 246-256. (9) Chou, S. F.; Tsern, I. P. J. Chin. Inst. Eng. 2000, 23, 161-170. (10) Park, J. K.; Song, M. J. Waste Manage. 1998, 18, 157-167. (11) Chamberlain, D. B.; Conner, C.; Hutter, J. C.; Leonard, R. A.; Wygmans, D. G.; Vandegrift, G. F. Sep. Sci. Technol. 1997, 32, 303-326. (12) Cuillerdier, C.; Musikas, C.; Nigond, L. Sep. Sci. Technol. 1993, 28, 155-175. (13) Park, S. M.; Park, J. K.; Kim, J. B.; Song, M. J. J. Environ. Sci. Health 1999, A34, 767-793. (14) Reddy, R. G. JOM 2001, 53, 21-24. (15) Matsuo, T.; Nishi, T. Carbon 2000, 38, 709-714. (16) Gokhale, A. S.; Venkateswaran, G.; Moorthy, P. N. Waste Manage. 1994, 14, 703-704. (17) Bockris, J.; Kim, J. J. Appl. Electrochem. 1997, 27, 623-634. (18) Zakrzewska-Trznadel, G.; Harasimowicz, M.; Chmielewski, A. G. J. Membr. Sci. 1999, 163, 257-264. (19) Park, J. K.; Lee, K. J. Waste Manage. 1995, 15, 283-291. (20) Tzeng, C. C.; Kuo, Y. Y.; Huang, T. F.; Lin, D. L.; Yu, Y. J. J. Hazard. Mater. 1998, 58, 207-220. (21) Fabiano, B.; Pastorino, R.; Ferrando, M. J. Hazard. Mater. 1998, 57, 105-125. (22) Nigond, L.; Condamines, N.; Cordier, P. Y.; Livet, J.; Madic, C.; Cuillerdier, C.; Musikas, C.; Hudson, M. J. Sep. Sci. Technol. 1995, 30, 2075-2099.

1336

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 6, 2002

(23) Lin, Y.; Smart, N. G.; Wai, C. M. Environ. Sci. Technol. 1995, 29, 2706-2708. (24) Smart, N.; Wai, C.; Phelps, C. Chem. Br. 1998, 34, 34-36. (25) Hasany, S. M.; Khurshid, S. J. Appl. Radiat. Isot. 1997, 48, 143146. (26) Lowry, J. D.; Lowry, S. B. J. Am. Water Works Assoc. 1988, 80, 50-64. (27) McLean, R. J. C.; Fortin, D.; Brown, D. A. Can. J. Microbiol./Rev. Can. Microbiol. 1996, 42, 392-400. (28) Macaskie, L. E.; Lloyd, J. R.; Thomas, R. A. P.; Tolley, M. R. Nucl. Energy 1996, 35, 257-271. (29) Brim, H.; McFarlan, S. C.; Fredrickson, J. K.; Minton, K. W.; Zhai, M.; Wackett, L. P.; Daly, M. J. Nat. Biotechnol. 2000, 18, 85-90. (30) Behrens, E. A.; Sylvester, P.; Clearfield, A. Environ. Sci. Technol. 1998, 32, 101-107. (31) Draye, M.; Lemaire, M.; Chevillotte, R.; Chomel, R.; Doutreluingne, P.; Foos, J.; Guy, A. Sep. Sci. Technol. 1995, 30, 1245-1257. (32) Harjula, R.; Lehto, J.; Tusas, E. H.; Paavola, A. Nucl. Technol. 1994, 107, 272-278. (33) Genders, J. D.; Hartsough, D.; Hobbs, D. T. J. Appl. Electrochem. 1996, 26, 1-9. (34) King, R. B.; Bhattacharyya, N. K.; Smith, H. D.; Wiemers, K. D. Environ. Sci. Technol. 1998, 32, 3178-3184. (35) Jenq, F. T.; Shih, C. J. In Proceedings of the 39th Industrial Waste Conference; Butterworth Publishers: Boston, 1984; pp 281290. (36) Lee, K. W.; Cho, S. H.; Park, H. H.; Kim, J. H. J. Kor. Soc. Environ. Eng. 1994, 16, 405-413. (37) Mann, N. R.; Todd, T. A. Chem. Eng. J. 2000, 80, 237-244. (38) Inoue, H.; Kagoshima, M. Appl. Radiat. Isot. 2000, 52, 14071412. (39) Mynin, V. N.; Terpugov, G. V. Desalination 1998, 119, 361-362. (40) Gaubert, E.; Barnier, H.; Maurel, A.; Foos, J.; Guy, A.; Bardot, C.; Lemaire, M. Sep. Sci. Technol. 1997, 32, 585-597. (41) Chiarizia, R.; Horwitz, E. P.; Ricket, P. G.; Hodgson, K. M. Sep. Sci. Technol. 1990, 25, 1571-1586. (42) Chiarizia, R. J. Membr. Sci. 1991, 55, 39-64. (43) Gaubert, E.; Barnier, H.; Nicod, L.; Favre-Reguillon, A.; Foos, J.; Guy, A.; Bardot, C.; Lemaire, M. Sep. Sci. Technol. 1997, 32, 2309-2320. (44) Bowen, W. R.; Mohammad, A. W. Chem. Eng. Res. Des. 1998, 76, 885-893. (45) Wang, X.; Tsuru, T.; Nakao, S.; Kimura, S. J. Membr. Sci. 1997, 135, 19-32. (46) Ikeda, K. Nitto Tech. Rep. 1996, 34, 65-74. (47) Nystro¨m, M.; Kaipia, L.; Luque, S. J. Membr. Sci. 1995, 98, 249262. (48) Ma¨ntta¨ri, M.; Nuortila-Jokinen, J.; Nystro¨m, M. Filtr. Sep. 1997, 24, 275-280. (49) Kiso, Y.; Kon, T.; Kitao, T.; Nishimura, K. J. Membr. Sci. 2001, 182, 205-214. (50) Xu, Y.; Lebrun, R. E. J. Membr. Sci. 1999, 158, 93-104. (51) Childress, A. E.; Elimelech, M. J. Membr. Sci. 1996, 119, 253268. (52) Childress, A. E.; Elimelech, M. Environ. Sci. Technol. 2000, 34, 3710-3716.

Received for review March 9, 2001. Revised manuscript received November 26, 2001. Accepted November 28, 2001. ES010724Q