A Rational Approach in the Design of Selective ... - ACS Publications

Homeira Ebrahimzadeh , Nafiseh Shekari , Najmeh Tavassoli , Mostafa M. Amini , Maryam Adineh , Omid Sadeghi. Microchimica Acta 2010 170 (1-2), 171-178...
0 downloads 0 Views 244KB Size
Download PDF
9632

Langmuir 2006, 22, 9632-9641

A Rational Approach in the Design of Selective Mesoporous Adsorbents Koon Fung Lam,†,‡ King Lun Yeung,*,‡ and Gordon McKay‡ EnVironmental Engineering Program and the Department of Chemical Engineering, Hong Kong UniVersity of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, P.R. China ReceiVed May 18, 2006. In Final Form: July 17, 2006 Two MCM-41 derived adsorbents have been tailor-made for the separation of silver and copper ions using the hard-soft, acid-base (HSAB) principle as the design guideline. NH2-MCM-41 containing “hard” Lewis base adsorption sites (i.e., RNH2) was prepared for the adsorption of the “hard” Lewis acid, Cu2+, and SH-MCM-41 with a grafted “soft” thiolpropyl base was prepared for the selective removal of Ag+, a “soft” Lewis acid. Single- and binarycomponent adsorption studies were conducted at different metal concentrations, solution compositions, and pH values. The experimental results showed that SH-MCM-41 has excellent affinity and capacity for silver adsorption and adsorbed only the silver ions with copper remaining in the solution. The selectivity was not affected by the metal concentration and composition, anion, and pH. Under similar experimental conditions, NH2-MCM-41 selectively adsorbed copper from the binary solution. The selectivity of NH2-MCM-41 remained for the copper at different pH values, although the adsorption capacity diminished at lower pH values. The type of anions used affected copper adsorption on NH2-MCM-41 with an increased copper uptake in the presence of the sulfate ions. A simple Freundlich adsorption model was sufficient to describe metal adsorption on SH-MCM-41 and NH2-MCM-41, and the LeVan and Vermeulen model was successfully used to predict the adsorption capacity and selectivity for binary-component adsorptions.

1. Introduction There is a growing interest in the application of mesoporous materials for the selective adsorption of cations and anions,1,2 organic compounds (e.g., dyes, aromatic hydrocarbons),3,4 and biomolecules5 from aqueous solutions. Following the report of selective mercury adsorption using thiolated MCM-41 by Feng et al.,6 several thiolated mesoporous silicas including SBA-15 and MCM-48 were also shown to be selective for mercury adsorption.7,8 Brown and co-workers9 used the thiolated mesoporous silica to separate mercury from a metal solution containing Hg2+, Cd2+, Pb2+, Zn2+, Co2+, Fe3+, Cu2+, and Ni2+. Liu and co-workers10 compared the adsorptions of thiol- and aminocontaining SBA-15. They reported that SBA-15 with thiolpropyl groups is capable of separating mercury and copper from a fivecomponent metal solution, whereas SBA-15 with aminopropyl groups indiscriminately adsorbs all metals from the solution. However, a recent report11 shows that MCM-41 containing RNH2, R2NH, and R3N groups displays gold-only adsorption and can selectively adsorb the precious gold from binary solutions containing copper and nickel. Indeed, SBA-15 grafted with nitrogen-containing imidazole groups is very selective for Pt2+ * To whom correspondence should be addressed. Tel: 852-2358-7123. Fax: 852-2358-0054. E-mail: [email protected]. † Environmental Engineering Program. ‡ Department of Chemical Engineering. (1) Yoshitake, H.; Yokoi, T.; Tatsumi, T. Chem. Mater. 2003, 15, 1713. (2) Lam, K. F.; Ho, K. Y.; Yeung, K. L.; McKay, G. Stud. Surf. Sci. Catal. 2004, 154, 2981. (3) Ho, K. Y.; McKay, G.; Yeung, K. L. Langmuir 2003, 19, 3019. (4) Choudhary, V. R.; Mantri, K. Langmuir 2000, 16, 7031. (5) Yiu, H. H. P.; Botting, C. H.; Botting, N. P.; Wright, P. A. Phys. Chem. Chem. Phys. 2001, 3, 2983. (6) Feng, X.; Fryxell, G. E.; Wang, L.-Q.; Kim, A. Y.; Liu, J.; Kemner, K. M. Science 1997, 276, 923. (7) Aguado, J.; Arsuaga, J. M.; Arencibia, A. Ind. Eng. Chem. Res. 2005, 44, 3665. (8) Pe´rez-Quintanilla, D.; del Hierro, I.; Fajardo, M.; Sierra, I. Microporous Mesoporous Mater. 2006, 89, 1-3, 58. (9) Brown, J.; Mercier, L.; Pinnavaia, T. J. Chem. Commun. 1999, 69. (10) Liu, A. M.; Hidajat, S.; Zhao, D. Y. Chem. Commun. 2000, 1145. (11) Lam, K. F.; Yeung, K. L.; McKay, G. J. Phys. Chem. B 2006, 110, 2187.

and Pd2+ ions and can separate these precious metals from solutions that also contain Ni2+, Cu2+, and Cd2+.12 This work attempts to rationalize the design of selective mesoporous adsorbents using the hard-soft, acid-base (HSAB) principle as a guide for selecting surface functional moieties to achieve the desired selectivity for metal ion adsorption. Pearson’s HSAB principle13 states that hard (Lewis) acids prefer to bind to hard (Lewis) bases and soft (Lewis) acids prefer to bind to soft (Lewis) bases and is related to the nature and stability of the chemical interactions between the acid-base species. Klopman’s frontier molecular orbital (FMO) analysis showed that hard acid-hard base interactions form stable chargecontrolled, ionic complexes, while soft acid-soft base interactions give strong FMO-controlled covalent complexes.14 Although HSAB-Klopman FMO analysis failed to address the interactions between a hard acid and a soft base or that of a soft acid and a hard base, it does not exclude the possibility of a stable hardsoft interaction, and indeed many form stable albeit strained complexes. Also, HSAB is ambivalent regarding borderline acids and bases. Despite these shortcomings, HSAB provides a good prediction of Lewis acid-base interactions without resorting to rigorous computation. The selective adsorption of Cu(II) and Ag(I) cations was investigated in this study. Table 1 lists the absolute hardness and absolute electronegativity for copper and silver metals and cations. The concept of absolute hardness and absolute electronegativity was introduced by Pearson17 to better quantify the HSAB principle. The Cu2+ cation is considered to be a hard Lewis acid, (12) Kang, T.; Park, Y.; Choi, K.; Lee, J. S.; Yi, L. J. Mater. Chem. 2004, 14, 1043. (13) Peasron, R. G. J. Am. Chem. Soc. 1963, 85, 22, 3533. (14) Klopman, G. J. Am. Chem. Soc. 1968, 90, 2, 223. (15) Mattigod, S. V.; Parker, K.; Fryxell, G. E. Inorg. Chem. Commun. 2006, 9, 96. (16) Martell, A. E.; Smith, R. M. Critical Stability Constants; Plenum Press: New York, 1974; Vol. 4. (17) Pearson, R. G. Absolute electronegativity and absolute hardness. In The Concept of the Chemical Bond; Maksic´, Z. B., Ed.; Springer-Verlag: Berlin, 1990; Chapter 2.

10.1021/la061410p CCC: $33.50 © 2006 American Chemical Society Published on Web 10/05/2006

Design of SelectiVe Mesoporous Adsorbents

Langmuir, Vol. 22, No. 23, 2006 9633

Table 1. Chemical Properties of Copper and Silver Metals and Cations absolute absolute hardnessb electronegativityc Misono metal eV eV softnessd Cu0 Cu+ Cu2+ Ag0 Ag+

3.25 6.28 8.27 3.14 6.96

4.48 14.01 28.56 4.44 14.53

stability constanta OH-

NH3

3.45 2.84

6.30

4.00

4.05

2.00

3.30

Reference 16. Absolute hardness is defined as (I - A)/2 where I and A are the ionization potential and electron affinity of the ions, respectively. It measures “the resistance to change of the electron cloud”. c Absolute electronegativity is defined as (I + A)/2. It measures “the tendency of an ion to gain or lose electrons”. d Reference 15. a

b

and the Ag+ cation is a soft Lewis acid according to Pearson’s HSAB principle. Grafting a hard Lewis base such as RNH2 on MCM-41 (NH2-MCM-41) should create an adsorbent that is selective for Cu2+ adsorption, while the SH-MCM-41 grafted with soft RSH base should display a greater affinity for adsorption of Ag+. Table 1 also lists the value of the Misono softness parameter that, according to Mattigod et al.,15 correlates well with metal adsorption on a thiol-terminated self-assembly monolayer on mesoporous supports (thiol-SAMMS). Also included in the table are the stability constants for the hydroxyl and ammonia complex of Cu(II) and Ag(I) cations. The SH-MCM-41 and NH2-MCM-41 adsorbents were prepared and characterized. Single- and binary-component adsorption studies were conducted for silver (I) and copper (II) solutions. The effects of anion, solution concentration, composition, and pH were investigated, and X-ray photoelectron spectroscopy (XPS) was also carried out to examine the chemical states of the adsorbed species and surface functional moieties. A possible scheme for the adsorption of silver and copper on SH-MCM-41 and NH2-MCM-41 was proposed based on the experimental data. The simple LeVan and Vermeulen model was successfully used to describe the binary-component adsorption on these adsorbents using single-component adsorption data. 2. Experimental Section 2.1. Adsorbent Preparation and Characterization. The preparation of mesoporous silica, MCM-41, was described in a previous publication.11 A micron-sized, MCM-41 powder with a uniform platelike morphology was prepared from the alkaline synthesis solution containing tetraethyl orthosilicate (TEOS, 98%, Aldrich), cetyltrimethylammonium bromide (CTABr, 99.3%, Aldrich), and ammonium hydroxide (NH4OH, 28-30 wt %, Fisher Scientific) at molar ratios of 6.58 TEOS:1 CTABr:292 NH4OH:2773 H2O. The synthesis was carried out at room temperature under vigorous mixing to obtain well-dispersed particles. The MCM-41 was filtered, washed, dried, and ground to obtain free-flowing powder. The powder was calcined at 823 K for 24 h to burn away the organic templates. The particle size distribution of MCM-41 was obtained by a light scattering method using a Coulter Beckman ζ-potential analyzer. A detailed description of the particle morphology and pore structure was obtained using a high-resolution, transmission electron microscope (TEM, JEOL JEM 2010), while the average surface area and pore size were calculated from nitrogen physisorption experiments (Coulter SA 3100) and X-ray diffraction (XRD) data (Philips 1830). MCM-41 adsorbents containing thiol and amino groups were prepared by grafting RSH and RNH2 (where R is a propyl group) on the calcined MCM-41 surface according to the scheme depicted in Figure 1. The thiolpropyl groups on SH-MCM-41 were grafted by reflux in a 250 mL dry toluene solution containing 0.1 mole of (3-mercaptopropyl)triethoxysilane (95%, Fluka) at 383 K for 18 h in nitrogen atmosphere. The adsorbent was filtered and washed with

Figure 1. Schematic diagram of the preparation of SH-MCM-41 and NH2-MCM-41. fresh toluene before drying in an oven at 333 K. The aminopropyl groups were added on NH2-MCM-41 by reflux in a dry toluene solution containing 3-aminopropyltrimethoxysilane (97%, Aldrich) at 383 K for 18 h. Different amounts of aminopropyl groups (i.e., 1.01 and 2.26 mmol/g) were grafted by changing the concentration of the precursor. The NH2-MCM-41 adsorbent was recovered by a series of filtration and washing steps. Thermogravimetric and differential thermal analyses (TGA/DTA, Setaram 31/1190) were carried out to identify and quantify the number of organic moieties grafted on the MCM-41. During the analysis, the samples were heated in flowing dry air (i.e., 25 sccm) from 298 to 873 K at 5 K/min, and the resulting weight loss and heat flow were recorded. The surface chemistry and composition of the SH-MCM-41 and NH2-MCM-41 adsorbents were analyzed by Fourier transform infrared spectroscopy (FTIR, Perkin-Elmer GX 2000) and XPS (Physical Electronics PHI 5000). The changes in the surface area and pore structure caused by the grafting of organic moieties were determined by nitrogen physisorption and XRD experiments. 2.2. Single- and Binary-Component Adsorptions. Singlecomponent adsorption isotherms of silver and copper were measured using 0.1 g of adsorbent for 100 mL of aqueous solutions containing 0.5-3 mM metal ions. Silver nitrate (99.8%, BDH), copper (II) nitrate (99%, Nacalai Tesque, Inc.) and copper (II) sulfate (99%, RDH) were used to prepare the metal salt solutions. The pH of the solutions was adjusted to 5.0 ( 0.1 by adding a small amount of dilute nitric acid or sodium hydroxide solutions. The batch adsorption experiments were conducted in a shaker bath kept at a constant temperature of 295 ( 2 K. The adsorption rate was determined by taking samples of the adsorption solution at fixed time intervals, while the equilibrium adsorption was measured at the end of a 5 day adsorption experiment. The initial and final concentrations of the metal in the solution were analyzed by inductively coupled plasma, atomic emission spectrometry (ICP-AES, Perkin-Elmer Optima 3000XL). Three measurements were made for each sample, and the results were averaged. Calibration was made before each set of measurements using ICP standard solutions: 1000 ppm Ag (99.999%) in 2% HNO3 and 1000 ppm Cu (99.999%) in 2% HNO3 purchased

9634 Langmuir, Vol. 22, No. 23, 2006

Lam et al.

from High-Purity Standards. The spent adsorbents were analyzed by XPS to determine the chemical states of the surface functional groups (i.e., RSH and RNH2) and adsorbed metals (i.e., Ag and Cu). Binary AgNO3/Cu(NO3)2 solutions of different metal ion concentrations and pH were prepared for selective adsorption studies. The effect of the anion was also studied using AgNO3/CuSO4 solutions. In the adsorption experiments, 0.1 g of adsorbent powder was added to the 100 mL solution and allowed to reach an equilibrium at a temperature of 295 ( 2 K. The solution was filtered to remove and recover the adsorbents. The remaining metal ions in the solution after adsorption were measured by ICP-AES, which permits the simultaneous analysis of up to 40 elements. Three measurements were averaged, and the equilibrium adsorption capacity was calculated.

3. Adsorption Models The grafting of thiolpropyl and aminopropyl groups onto MCM-41 introduces surface heterogeneity, and therefore the Freundlich adsorption model (i.e., qe ) KCen) was used to model metal adsorption on the SH-MCM-41 and NH2-MCM-41 adsorbents. The adsorption capacity was calculated from the experimental data using eq 1 and plotted with respect to Ce to obtain the equilibrium adsorption isotherm.

qe )

(Co - Ce)V m

(1)

where qe (mmol/g) is the adsorption capacity; Co (mM) and Ce (mM) are the initial and equilibrium concentrations of metal in the solution, respectively; V (L) is the volume of the solution; and m (g) is the mass of the adsorbent. The LeVan and Vermeulen model,18 derived from the ideal adsorption solution theory of Myer and Prausnitz,19 was used to predict binary adsorptions from the single-component adsorption data. Equations 2 and 3 are the simplified form of the LeVan and Vermeulen equations based on the single-component Freundlich isotherms of components 1 and 2.

()

nj qe,1 )

[( ) K1 n1

1/n1

()

[( ) K1 n1

1/n1

1/n1

Ce,1

() ]

K2 Ce,1 + n2 nj

qe,2 )

K1 n1

K2 n2

1-nj

1/n2

+ ∆F2

(2)

+ ∆F2

(3)

Ce,2

1/n2

Ce,1 +

Ce,2

() ] K2 n2

1-nj

1/n2

Ce,2

where qe,1 and qe,2 are respectively the amount of components 1 and 2 adsorbed at equilibrium; Ce,1 and Ce,2 are the equilibrium concentrations of components 1 and 2, respectively; K1, K2 are the values of the Freundlich constants from the single-component adsorption isotherms of components 1 and 2, respectively; and nj and ∆F2 are given by eqs 4 and 5, respectively.

() ()

n1 nj )

K1 n1 K1 n1

1/n1

1/n1

() ()

Ce,1 + n2

K2 n2

K2 Ce,1 + n2

1/n2

Ce,2 (4)

1/n2

Ce,2

(18) LeVan, M. D.; Vermeulen, T. J. Phys. Chem. 1981, 85, 3247. (19) Myer A. L.; Prausnitz, J. M. AIChE J. 1965, 11 (1), 121.

() [( ) K1 n1

() () ] [( ) ( ) ]

K2 1/n2 Ce,2 n2 ∆F2 ) (n1 - n2) 2-nj K1 1/n1 K2 1/n2 Ce,1 + Ce,2 n1 n2 K1 1/n1 K2 1/n2 ln Ce,1 + Ce,2 (5) n1 n2 1/n1

Ce,1

4. Results and Discussion 4.1. Mesoporous Silica Adsorbents. The dry, calcined MCM41 powder was free flowing and had little tendency to agglomerate. The scanning electron microscopy (SEM) picture of the calcined MCM-41 in Figure 2a shows that the MCM-41 has a flat, disklike shape with an average diameter of about 0.80 µm and a thickness of 0.10 µm. It is evident from the micrograph that the particles have a uniform size. Indeed, the light scattering measurement in Figure 2b shows that the powder has a narrow particle size distribution with a mean particle diameter of 0.75 ( 0.15 µm that is comparable to the SEM results. The well-ordered, hexagonal pore structure of MCM-41 is revealed by the highresolution transmission electron micrograph in Figure 2c. An average pore diameter of 3.09 nm was calculated according to the method described by Kruk and co-workers20 using the d spacing measured by XRD (Figure 3a) and the pore volume determined from the N2 physisorption experiments. The powder sample displays the characteristic XRD peaks for MCM-41. The prepared sample has good crystallinity as indicated by the presence of the (110) and (200) diffraction peaks.21 The fresh MCM-41 powder before calcination had a BET surface area of 8 m2/g, but, after burning away the CTA+ molecules from the pores, the specific surface area measured 1070 m2/g. The pore volume was measured to be 1.032 cm3/g. Elemental analysis of calcined MCM41 by XPS detected only silicon and oxygen with a trace amount of carbon from adsorbed organic contaminants originating from the air. This indicates a complete removal of organic template molecules by the air calcination. The FTIR detected only surface silanol and hydroxyl groups (Figure 4a), and there were no signals from organic compounds. Table 2 summarizes the BET surface area, the average pore diameter, and the predominant surface chemical moieties found on the MCM-41. SH-MCM-41 adsorbent was prepared by grafting thiolpropyl groups on the calcined MCM-41. Sulfur-containing compounds are known for their strong affinity for late transition metals (i.e., Pd, Pt, Ag, and Au).22 Introducing organic moieties on the pore wall increases disorder and leads to a decrease in the intensity of (110) and (200) peaks for the SH-MCM-41, as shown in Figure 3b. The surface elemental analysis of SH-MCM-41 by XPS showed a carbon-to-sulfur (C/S) ratio of 5.3. This suggests that one of the ethoxysilane group in the (3-mercaptopropyl)triethoxysilane remained unreacted and the thiolpropyl group was attached to the surface by two Si-O-Si bonds instead of three. Thermogravimetric analysis (TGA) is commonly used to measure the amount of organic groups grafted on mesoporous silica.23 The SH-MCM-41 contained 1.0 mmol of thiolpropyl groups per gram of adsorbent according to TGA and confirmed by the XPS data. Figure 4a,b displays the FTIR spectra of the MCM-41 and SH-MCM-41 samples. The MCM-41 spectrum (20) Kruk, M.; Jaroniec, M.; Sayari, A. Chem. Mater. 1999, 11, 492. (21) Kleitz, F.; Schmidt, W.; Schuth, F. Microporous Mesoporous Mater. 2003, 65, 1. (22) Grabar, K. C.; Griffith Freeman, R.; Hommer, M. B.; Natan, M. J. Anal. Chem. 1995, 67, 735. (23) Lindlar, B.; Lu¨chinger, M.; Ro¨thlisberger, A.; Haouas, M.; Pirngruber, G.; Kogelbauer, A.; Prins, R. J. Mater. Chem. 2002, 12, 528.

Design of SelectiVe Mesoporous Adsorbents

Langmuir, Vol. 22, No. 23, 2006 9635

Figure 4. FTIR spectra of (a) MCM-41, (b) SH-MCM-41 and (c) NH2-MCM-41.

Figure 2. (a) The particle size and morphology, (b) particle size distribution, and (c) pore structure of the MCM-41 powder were characterized by SEM, the light-scattering method, and TEM, respectively.

Figure 3. XRD patterns of (a) MCM-41, (b) SH-MCM-41 and (c) NH2-MCM-41.

shows the characteristic signals belonging to the surface silanol groups at 3743 cm-1, the Si-O-Si bridge between 1000 and 2000 cm-1, as well as the broad peak at 3428 cm-1 for the adsorbed water molecules. The SH-MCM-41 sample shows a characteristic peak at 2571 cm-1 for S-H stretching. The signals at 2925 and 2849 cm-1 are assigned to the asymmetric and symmetric stretching modes of the aliphatic chain (i.e., n-propyl). The figure shows a decrease in the infrared signals belonging to surface silanols after the thiolpropyl groups were grafted. A concomitant decrease in both the BET surface area and the average pore size is shown in Table 2. Similar loss of surface area had

been reported by other researchers and was attributed to the addition of the surface organic moieties.24 NH2-MCM-41 adsorbents were prepared from MCM-41 by grafting aminopropyl groups on the pore walls. Most metal ions readily form complexes with amine compounds, and many adsorbents that contain nitrogen-bearing compounds, including polymeric resins25 and chitosan,26 are good metal adsorbents. Figure 3c shows the concomitant decrease in XRD intensity after grafting aminopropyls on MCM-41. XPS analysis of the surface elemental composition of NH2-MCM-41 showed the carbon-to-nitrogen (C/N) ratio was 3.38, about 12% higher than the expected value of 3. This means that about 90% of the methoxy groups in the 3-aminopropyltrimethoxysilane had reacted to form a bond between the organic amine group and the silica surface. NH2-MCM-41 adsorbents with 1.01 and 2.26 mmol aminopropyl groups per gram of adsorbent were prepared and characterized. Figure 4c shows the disappearance of the characteristic peaks belonging to surface silanols of the original MCM-41 after grafting the aminopropyls. The infrared spectrum of NH2-MCM-41 displays the characteristic signals for the aminopropyl at 3267 and 3352 cm-1 belonging to amine stretching27 and peaks at 2925 and 2849 cm-1 belonging to the n-propyl. The BrunauerEmmett-Teller (BET) surface area and average pore size of NH2-MCM-41 shown in Table 2 are smaller than that of the original MCM-41, with values of 774 m2/g and 2.92 nm, respectively. The specific surface area of NH2-MCM-41 is also smaller than that of SH-MCM-41, which was similar to the observation made by Liu et al.10 for amino- and thiolfunctionalized SBA-15. (24) Yoshitake, H.; Koiso, E.; Horie, H.; Yoshimura, H. Microporous Mesoporous Mater. 2005, 85 (1-2), 183. (25) Donia, A. M.; Atia, A. A.; El-Boraey, H. A.; Mabrouk, D. H. Sep. Purif. Technol. 2006, 48 (3), 281. (26) Vold, I. M. N.; Vårum, K. M.; Guibal, E.; Smidsrød, O. Carbohydr. Polym. 2003, 54 (4), 471. (27) Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 76th ed; CRC Press: Boca Raton, FL, 1988; pp 9-80.

9636 Langmuir, Vol. 22, No. 23, 2006

Lam et al.

Table 2. Physical and Chemical Properties of the Mesoporous Silica Adsorbents adsorption capacity mmol/g (mg/g)

MCM-41 NH2-MCM-41a NH2-MCM-41b SH-MCM-41 a

surface area m2/g

pore sizea nm

primary moiety

1070 772 774 990

3.09 2.82 2.92 3.02

OH NH2 NH2 SH

functional groups mmol/g

Ag+

Cu2+

1.01 2.26 1.00

0 0.11 (11.8) 0.62 (66.3) 0.97 (103)

0 0.25 (15.8) 0.84 (53.2) 0.02 (1.27)

Pore size was calculated based on d spacing measured by XRD, BET surface area, and pore volume by nitrogen physisorption.

Figure 5. Single-component adsorption of Ag+ (]) and Cu2+ (0) on (a) SH-MCM-41 and (b) NH2-MCM-41 as a function of time. (1 mmol RX/g adsorbent, 0.1 g of adsorbent per 100 mL of solution, pH ) 5, and T ) 295 ( 2 K).

4.2. Single-Component Adsorption. The single-component adsorption study was conducted at a pH of 5. The experimental results showed that calcined MCM-41 does not adsorb silver or copper from the solutions. This suggests that the surface silanol groups of the MCM-41 have poor affinity for Ag+ and Cu2+ adsorptions (cf. Table 2). On the other hand, the modified MCM41 was able to adsorb the metal ions. Figure 5a,b plots the metal adsorptions as a function of time on SH-MCM-41 and NH2MCM-41, respectively. Adsorption on mesoporous adsorbents is expected to be fast because of the large, easily accessible pore channels.28 Silver and copper adsorptions on SH-MCM-41 are rapid, and equilibrium is reached in less than 15 min, as shown by Figure 5a. The SH-MCM-41 has a strong affinity for silver. It adsorbs a large quantity of silver, but only a trace amount of copper, as shown in the figure. Figure 5b shows that silver adsorption on NH2-MCM-41 was also rapid, but the adsorption equilibrium for copper takes a longer time of about 30 min. NH2-MCM-41 adsorbs more copper than silver, as shown in the figure. These results indicate that the chemical moieties grafted inside the MCM-41 pores are responsible for metal adsorptions and that they remained readily accessible despite the decrease in the pore size following the surface chemical modifications (cf. Table 2). (28) Walcarius, A.; Etienne, M. Lebeau, B. Chem. Mater. 2003, 15 (11), 2161.

Figure 6. Single-component adsorption isotherms of Ag+ (]) and Cu2+ (0) for (a) SH-MCM-41 and (b) NH2-MCM-41. (1 mmol RX/g adsorbent, 0.1 g of adsorbent per 100 mL of solution, pH ) 5, and T ) 295 ( 2 K). Please note that the symbols represent the experimental data, and the lines represent the model calculation.

Figure 6a plots the equilibrium adsorption isotherms for silver and copper on SH-MCM-41. The SH-MCM-41 adsorbs 0.97 mmol silver per gram of adsorbent (i.e., 103 mg/g) equivalent to one silver atom for each grafted thiolpropyl. On the other hand, only a trace amount of copper (i.e., 0.02 mmol/g or 1.27 mg/g) was adsorbed under the same experimental conditions. The NH2-MCM-41 adsorbs both silver and copper but has a larger capacity for the latter, as shown in Figure 6b. It adsorbed 0.11 mmol/g or 12 mg/g of silver and 0.25 mmol/g or 16 mg/g of copper. The presence of thiolpropyl and aminopropyl groups introduced surface heterogeneity on an otherwise uniform MCM41 pore surface, and the Freundlich adsorption equation (i.e., qe ) KCen) provides a more appropriate description of the metal adsorption on these functionalized adsorbents. Indeed, the plots show that there is a good fit between the model and experiment. The calculated values for the constant K and the exponent n are summarized in Table 3. Table 2 shows that NH2-MCM-41 with 2.2 mmol/g aminopropyl loading had adsorption capacities for silver and copper of 0.62 and 0.84 mmol/g, respectively. The values are higher than the data obtained using NH2-MCM-41 with 1 mmol/g loading. Calculations show that the amount of copper adsorbed per surface moiety increased from 0.25 to 0.35 copper per

Design of SelectiVe Mesoporous Adsorbents

Langmuir, Vol. 22, No. 23, 2006 9637

Table 3. Fitting Parameters of Single-Component Freundlich Adsorption Isotherms adsorbent SH-MCM-41 NH2-MCM-41

metal Cu Ag Cu Ag

K

n

0.0259 0.9694 0.2453 0.1077

0.0100 0.1202 0.0511 0.0581

aminopropyl at the higher aminopropyl loading. The amount of silver adsorbed also increased from 0.10 to 0.27 silver per aminopropyl. The poorer performance of 1 mmol/g NH2-MCM41 is due to the interaction between the grafted aminopropyls and the unreacted silanols. This phenomenon was also observed for Au-Cu and Au-Ni systems.11 A SH-MCM-41 with a higher thiolpropyl loading was also prepared, but the resulting adsorbent was very hydrophobic and could not be dispersed in aqueous solutions. XPS analysis of the spent adsorbents was carried out to determine the chemical states of the surface functional groups as well as that of the adsorbed metals. Figure 7a shows that the sulfur atom on fresh SH-MCM-41 has a S2p3/2 binding energy of 164.0 eV. This is close to the value of 163.8 eV found in alkylthiol (RSH) molecules.29 The binding energy for sulfur remained unchanged after copper adsorption, but shifted to a lower value (i.e., 163.13 eV) after silver was adsorbed. The binding energy of 163.13 eV is similar to that of silver sulfide (AgS). In the AgS, the silver perturbs the electron around S, weakening its binding energy. A similar effect is believed to happen during the adsorption of silver on the thiolpropyl groups, resulting in the observed decrease in the binding energy of the sulfur atom. The XPS data for N1s are shown in Figure 7b for the fresh and spent NH2-MCM-41. The N1s value for fresh NH2-MCM-41 is 400 eV, comparable to the reported value of 399.5 eV for alkylamines (RNH2).30 Additional peaks with binding energies of 402 and 407 eV appeared after copper adsorption.

Figure 7. X-ray spectroscopy of surface functional moieties (a) S2p3/2 and (b) N1s and adsorbed metals (c) Ag MVV and (d) Cu3d5/2.

The formation of a dative bond between copper and the surface amino group involves the free electron pair in the nitrogen atom, resulting in a positive polarization that could explain the higher binding energy. Indeed, reports showed that protonated amino (i.e., HNH3+, RNH3+) has a N1s binding energy of 402 eV. Nitrate (i.e., NO3-) has a binding energy of 407 eV and most likely originates from the copper nitrate salt used in the adsorption experiment. The XPS data in Figure 7b shows that the N1s binding energy of the spent NH2-MCM-41 used in silver adsorption has the same value as that of the fresh adsorbent. The binding energy for silver adsorbed on SH-MCM-41 and NH2-MCM-41 was obtained from detecting MVV signal by XPS and is shown in Figure 7c. The binding energy of 1130.8 eV for silver adsorbed on SH-MCM-41 is lower compared to standard values of 1132-1137 eV for Ag+. The strong interaction between silver and the thiolpropyl results in a lower positive charge on Ag and thus a lower binding energy of the AgMVV electron. Silver is adsorbed on NH2-MCM-41 as Ag+ and has a binding energy of 1133.4 eV. Figure 7d displays the XPS data for Cu3d5/2. The copper was adsorbed on NH2-MCM-41 as Cu2+, as indicated by the presence of the satellite peak (944 eV) in Figure 7d, and the Cu3d5/2 binding energy of 935.25 eV is close to the value of 935.5 eV reported for Cu(NO3)2. The trace amount of copper found on SH-MCM-41 is in a reduced state (i.e., Cu+ or Cu0), as shown by the absence of the satellite peak in the XPS data in Figure 7d. The binding energy of 933.25 eV is comparable to that of a metallic copper. Figure 8 describes a possible scheme for the adsorption of silver and copper on SH-MCM-41 and NH2-MCM-41. The thiolpropyl groups were attached to the surface of the MCM-41 pores via the condensation of two ethoxysilane groups in the (3-mercaptopropyl)triethoxysilane (1) with two surface silanol groups (2) forming two Si-O-Si bonds for each grafted thiolpropyl (3). The methoxysilane groups in the aminopropyltrimethoxysilane (8) are more reactive, and the surface aminopropyl groups (9) were anchored to the pore surface by three Si-O-Si bonds. The grafted thiolpropyls (4) and aminopropyls (10) are respectively categorized as a weak and a strong base according to Pearson.13 The grafted organic moieties could be dispersed singly or aggregated into islands on the surface of the pores and could also interact with each other and the surface hydroxyl groups through hydrogen bonding. This creates adsorption sites of different energetics that is best described by the Freundlich equation, as demonstrated in Figure 6. A recent work of Mattigod et al.15 reported a good correlation between the adsorption capacity of thiol-SAMMS and the Misono softness parameter. The Misono softness parameter arises from the attempt to quantify Pearson’s HSAB theory,13 and the Ag(I) and Cu(II) have values of 4.05 and 2.84, respectively. Mattigod et al.15 reported that the adsorption capacities of thiol-SAMMS were 1.23 silver and 0.24 copper per thiolpropyls. In comparison, the SH-MCM-41 prepared in this work had adsorption capacities of 0.97 silver and 0.02 copper per thiolpropyls. This difference in adsorption capacities could be a reflection of the dissimilarities in the adsorption sites of thiol-SAMMS and SH-MCM-41. The thiol-SAMMS, with its monolayer thiolpropyl coverage, has a uniform adsorption site, while SH-MCM-41 possesses heterogeneous sites (4a, 4b, 4c, 4d). Figure 8a illustrates the adsorptions of silver and copper cations on SH-MCM-41. The Ag+ cations (5), being a weak acid, are preferentially adsorbed by the (29) Castner, D. G. Langmuir 1996, 12, 5083. (30) Hooper, A. E.; Werho, D.; Hopson, T.; Palmer, O. Surf. Interface Anal. 2001, 31, 809.

9638 Langmuir, Vol. 22, No. 23, 2006

Lam et al.

Figure 8. Proposed scheme for the adsorption of silver and copper on SH-MCM-41 and NH2-MCM-41.

thiolpropyls, and the formation of a strong bond is apparent from the XPS data of S2p3/2 (Figure 7a) and the MVV signal of Ag (Figure 7c). The trace amount of coppers (6) found on SHMCM-41 was in a reduced state (i.e., Cu+, Cu0) and could proceed with the oxidation of two thiolpropyl groups into a disulfide (7), as shown in the figure. According to HSAB principle, the aminopropyl groups grafted on NH2-MCM-41 are a strong base. On the basis of the absolute hardness defined by Pearson,31 Cu2+ is classified to be a harder metal ion than Ag+, and therefore the aminopropyl groups should have a greater affinity for Cu2+ adsorption. Figure 8b illustrates metal adsorptions on NH2-MCM-41. Interactions between the aminopropyls (10a) and surface hydroxyls are expected (10b), diminishing their availability to adsorb silver (11). This could explain the lower adsorption capacity of NH2-MCM-41 with a low aminopropyl loading for the silver cations compared to the adsorbent with higher aminopropyl loading (i.e., 0.10 versus 0.27 Ag+ per aminopropyl). Cu2+ is a stronger acid and could displace the hydroxyl groups complexed to the aminopropyls (12) and thus adsorbed at a greater amount than Ag+. Also, Cu2+ adsorption led to a measurable change in the N1s binding energy that was not observed during Ag+ adsorption. This suggests that the aminopropyls form a stronger bond with Cu2+ than with Ag+. Nitrates were detected on the adsorbed Cu2+ and not on Ag+ (12). The strong affinity of thiolpropyl for Ag+ and aminopropyl for Cu2+ and the formation of strong chemical bonds between Ag+-thiolpropyl and Cu2+-aminopropyl are in accordance with the HSAB theory. 4.3. Binary-Component Adsorption. Hankzlı´k et al.32 used carbonaceous adsorbents from spruce wood, pine bark, cork, and peat for the adsorption of Ag(I), Cd(II) and Cu(II). They observed that adsorbents from spruce wood, pine bark, and peat have a high adsorption preference to Cu2+, while highly carbonified adsorbents display better adsorption of silver. However, the complex nature of the natural adsorbents prevents (31) Pearson, R. G. J. Am. Chem. Soc. 1988, 110, 7684. (32) Hanzlı´k, J.; Jehlicˇka, J.; Sˇebek, O.; Weishauptova´, Z.; Machovicˇ, V. Water Res. 2004, 38, 2178.

a conclusive explanation regarding the selectivity. Lee and coworkers33 performed a more controlled experiment using silica beads modified with Adogen 364 (a mixture of trialkyl tertiary amines) for the selective adsorption of Ag+ from a solution containing Cu2+ and Ni2+. The selectivity was explained mainly by the thermodynamics of metal ion solvation and neglected the role of the surface functional moieties. Figure 9 plots the equilibrium adsorption capacity of the adsorbents as a function of equilibrium metal concentration in the solution. SH-MCM-41 and NH2-MCM-41 have functional group loadings of 1 mmol/g. The binary-component adsorption experiments were conducted using solutions with equimolar metal ion concentrations at a fixed pH of 5 and a temperature of 295 ( 2 K. The SH-MCM-41 adsorbs only silver from the binary solutions (Figure 9a). This means that SH-MCM-41 has a 100% selectivity for silver under the experimental conditions. The maximum amount of silver adsorbed by SH-MCM-41 is 0.91 mmol/g (i.e., 97 mg/g) and is close to 0.97 mmol/g (i.e., 103 mg/g) obtained from the single-component adsorption experiment (Figure 6a). Unlike the single-component adsorption data (Figure 6b), which shows that NH2-MCM-41 adsorbs both silver and copper, Figure 9b shows that NH2-MCM-41 adsorbs only copper from the binary solutions containing silver and copper nitrate salts. The adsorption capacity for copper is 0.24 mmol/g (15 mg/g) comparable to the value obtained in Figure 6b. The binarycomponent adsorptions were modeled using the LeVan and Vermeulen equations (i.e., eqs 2-5) using the single-component adsorption data. There is good agreement between the model calculation and the binary-component adsorption data, as shown in Figure 9. The model correctly predicted the selectivity of the two adsorbents. This is particularly important when selectivity is not immediately apparent from the single-component adsorption data, as in the case of silver and copper adsorption on NH2MCM-41. A previous study showed that the concentration of the anion could affect the adsorption of gold on NH2-MCM-41.11 Jia and (33) Lee, W.; Kim, C.; Yi, J. J. Chem. Technol. Biotechnol. 2002, 77, 12551261.

Design of SelectiVe Mesoporous Adsorbents

Figure 9. Ag+ (]) and Cu2+ (0) adsorptions on (a) SH-MCM-41 and (b) NH2-MCM-41 from AgNO3/Cu(NO3)2 binary solutions (1 mmol RX/g adsorbent, 0.1 g of adsorbent per 100 mL of solution, [Ag+]/[Cu2+] ) 1, pH ) 5, and T ) 295 ( 2 K). Please note that the symbols represent the experimental data, and the lines represent the model calculation.

Figure 10. Effects of the anion on (a) silver adsorption on SHMCM-41 and (b) copper adsorption on NH2-MCM-41 from the AgNO3/Cu(NO3)2 (]) and AgNO3/CuSO4 (0) binary solutions (1 mmol RX/g adsorbent, 0.1 g of adsorbent per 100 mL of solution, [Ag+]/[Cu2+] ) 1, pH ) 5, and T ) 295 ( 2 K). Please note that the unadsorbed metals are not shown in the graph.

Demopoulos34 reported that the presence of sulfate ions decreases the adsorption capacity of activated carbon for Ag+. Figure 10 plots the results of the binary adsorption experiments obtained from solutions prepared from silver and copper nitrates and from (34) Jia, Y.; Demopoulos, G. P. Ind. Eng. Chem. Res. 2003, 42, 72.

Langmuir, Vol. 22, No. 23, 2006 9639

silver nitrate and copper sulfate. The anions do not affect the selectivity of the adsorbents. The SH-MCM-41 adsorbs only silver, while NH2-MCM-41 adsorbs only copper, as shown in Figure 10a,b, respectively. Silver adsorption on SH-MCM-41 is insensitive to the type of anion present in the solution, and the adsorption capacity remained unchanged, as shown in Figure 10a. On the other hand, Figure 10b shows that the adsorption capacity of NH2-MCM-41 for copper depends on whether Cu(NO3)2 or CuSO4 salt was used. The figure shows that the adsorption capacity of NH2-MCM-41 for copper nearly doubled when CuSO4 instead of Cu(NO3)2 was used in the binarycomponent adsorption. Separate adsorption experiments were carried out at a fixed total metal ion concentration of 3 mM, but with a varying silver content of 0-100 mol %. The solutions were prepared from silver and copper nitrate salts, and the adsorption was allowed to reach equilibrium at a constant temperature and pH. The results in Figure 11a show that SH-MCM-41 has excellent selectivity for silver adsorption. Copper was not adsorbed, regardless of the starting composition of the solution, and only silver was removed by the adsorbent. SH-MCM-41 is able to remove even trace amounts of silver from solutions with high concentrations of copper, as shown in Figure 11b. The Cu(NO3)2 concentration in the solutions was kept at 1 mM (i.e., 64 ppm), while the AgNO3 concentration was varied from 0 to 0.4 mM (i.e., 5-40 ppm). The results show that only silver was adsorbed, even when the silver concentration was as low as 0.05 mM or 5 ppm. A 100% silver removal was obtained over the entire range of silver concentrations tested (i.e., 0-3 mM). It is clear that the excellent selectivity of SH-MCM-41 for silver is not affected by the metal concentration, and complete silver removal by adsorption is possible, even in solutions containing very dilute amounts of silver. Figure 11c plots the equilibrium adsorption capacity of NH2MCM-41 for copper and silver as a function of copper content in the Ag-Cu solutions. The solutions have a fixed metal ion concentration of 3 mM and were prepared from AgNO3 and Cu(NO3)2 salts. The results show that NH2-MCM-41 preferentially adsorbs the copper and not the silver ions from the solutions (Figure 11c). A 100% selectivity for copper is obtained, and the adsorption capacity of NH2-MCM-41 for copper is comparable to the value obtained from the single-component adsorption (Table 3). Solutions containing dilute concentrations of Cu(NO3)2 (i.e., 0-0.25 mM) in concentrated AgNO3 solutions (i.e., 1 mM) were prepared. Figure 11d shows that NH2-MCM41 is very efficient in removing trace amounts of copper from the solution and complete copper removal was obtained under the experimental conditions. Although NH2-MCM-41 prefers to adsorb copper, it can also adsorb silver, as shown by the single-component adsorption data in Figure 6b. This means that, when the copper concentration is low and not all the adsorption sites on NH2-MCM-41 are occupied, silver can adsorb on the available vacant sites, as shown in Figure 11d. Figure 11 shows that the simple binary adsorption model of LeVan and Vermeulen is sufficient to describe the binary AgCu adsorption on the mesoporous SH-MCM-41 and NH2MCM-41 adsorbents. The model calculation suggests that SHMCM-41 should have greater capacity for silver adsorption than the experimental observation (cf. Figure 11a). The figure shows that the deviation between model and experiment is worse at higher copper concentrations, where the higher anion concentration of the binary solution could affect the dissociation of the metal salts, resulting in fewer cations and therefore less adsorption. Indeed, this was observed in the case of Au-Cu and Au-Ni

9640 Langmuir, Vol. 22, No. 23, 2006

Lam et al.

Figure 11. Plots of the effects of solution composition on Ag+ (]) and Cu2+ (0) adsorptions on (a,b) SH-MCM-41 and (c,d) NH2-MCM-41 from the AgNO3/Cu(NO3)2 binary solutions. (1 mmol RX/g adsorbent, 0.1 g of adsorbent per 100 mL of solution, [Ag+] + [Cu2+] ) 3 mM for panels a and b, [Cu2+] ) 1 mM for panel c, and [Ag+] ) 1 mM for panel d, pH ) 5, and T ) 295 ( 2 K). Please note that the symbols represent the experimental data, and the lines represent the model calculation.

Figure 12. Plots of the effects of pH on Ag+ (]) and Cu2+ (0) adsorptions on (a) SH-MCM-41 and (b) NH2-MCM-41. (1 mmol RX/g adsorbent, 0.1 g of adsorbent per 100 mL of solution, [Ag+]/ [Cu2+] ) 1, [Ag+] + [Cu2+] ) 6 mM, pH ) 5, and T ) 295 ( 2 K).

adsorptions.11 However, silver adsorption on SH-MCM-41 is insensitive to the nature and concentration of anions, as shown in Figure 10a. A more plausible explanation is based on the XPS data in Figure 7d, which shows that the trace amount of copper adsorbed by SH-MCM-41 from concentrated copper solutions is at a reduced +1 or metallic state instead of Cu2+. The reduction of Cu2+ is accompanied by the oxidation of two thiolpropyl groups into a disulfide, decreasing the available number of

thiolpropyls for silver adsorption according to the proposed scheme in Figure 8a. This results in the net decrease in the adsorption capacity of the adsorbent for silver. Figure 11c,d shows that the model provides an excellent prediction of adsorbent selectivity and copper adsorption on NH2-MCM-41. The effects of pH on metal adsorption from the binary solutions were investigated, and the results are shown in Figure 12a,b. The study was conducted between a pH of 2.5 and 5.5. The adsorption of silver and copper on SH-MCM-41 is insensitive to pH, as shown in Figure 12a. The selectivity for silver adsorption remains excellent, and the adsorbent removes only silver and not copper from the solutions. ζ potential measurements showed that SHMCM-41 had an isoelectric point at around a pH of 3.5, and experiments showed that more than 90% the adsorbed silver could be recovered at high purity (i.e., >99%) by a simple acid wash using 5 M nitric acid. It can be seen from Figure 12b that NH2-MCM-41 remained selective to copper adsorption, regardless of the solution pH, but the adsorption capacity for copper decreases at low pHs. Similar to SH-MCM-41, the NH2-MCM41 has an isoelectric point at a pH of 2.95, and the amino group is protonated at a pH lower than this value. The protonated xNH3-MCM-41 found at low pH values is less able to accommodate the adsorption of Cu2+ ions. The adsorbed metals (i.e., copper or silver) can be recovered from the NH2-MCM-41 by a simple acid wash. The regenerated adsorbents could be reused without loss of performance.

Concluding Remarks This work demonstrated that the HSAB principle could provide a useful guide in the design of selective adsorbents for metal separations. A “hard” Lewis acid, Cu2+, was selectively removed by NH2-MCM-41 containing “hard” Lewis base adsorption sites (i.e., RNH2), while Ag+, a “soft” Lewis acid, was selectively adsorbed by the “soft” thiolpropyl base on SH-MCM-41. The HSAB principle further suggests that the observed chemical

Design of SelectiVe Mesoporous Adsorbents

affinity is the result of the stable interactions between the acidbase species. Indeed, the XPS analysis carried out in this study indicated strong interactions between Cu2+ and RNH2 in NH2MCM-41 and between Ag+ and RSH in SH-MCM-41. However, the HSAB principle failed to properly address the important issue of “hard-soft” interactions. The adsorption experiments showed that the “hard” Lewis base, NH2-MCM-41, was able to adsorb both “hard” Cu2+ and “soft” Ag+ Lewis acids. It also becomes unwieldy for predicting the subtleties of interactions between different surface adsorption sites (e.g., RNH2, RSH, Si-OH), which was shown to affect the metal adsorption on the mesoporous adsorbents. This work also demonstrated that a simple Freundlich adsorption model is sufficient to describe metal adsorptions on the mesoporous SH-MCM-41 and NH2-MCM-41 adsorbents. The single-component adsorption data was successfully used in the LeVan and Vermeulen model to predict the adsorption capacity and selectivity for the binary-component adsorptions. A good agreement between the calculation and experiment suggests the

Langmuir, Vol. 22, No. 23, 2006 9641

possible use of the model in the adsorbent design. However, the model was unable to predict the effect of pH on the metal adsorption because it simultaneously affects both metal speciation and the surface functional moieties. Thus, it is difficult to predict a priori the optimum adsorption and regeneration pH values for a given adsorbate-adsorbent couple. Despite these shortcomings, this work clearly showed that a rational design of selective adsorbent is not only possible but also practical. Acknowledgment. The authors gratefully acknowledge the funding from the Hong Kong Research Grant Council (Grant RGC-HKUST 6037/00P) and the Environmental Conservation Fund (ECWW05/06.EG01). We thank the Material Characterization and Preparation Facility at the HKUST for the use of the XRD, SEM, TEM, XPS, and TGA/DTA equipment and the Advanced Engineering Material Facility for the use of the Coulter SA3100 surface area and pore size analyzer and the Beckman Coulter ζ-potential analyzer. LA061410P