Вы находитесь на странице: 1из 9

Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

Contents lists available at ScienceDirect

Journal of Environmental Chemical Engineering


journal homepage: www.elsevier.com/locate/jece

Research paper

Adsorptive recovery of neodymium and dysprosium in phosphorous MARK


functionalized nanoporous carbon

Dipendu Sahaa, , Sel Didem Akkoyunlua, Ryan Thorpeb, Dale K. Hensleyc, Jihua Chenc
a
Department of Chemical Engineering, Widener University, One University Place, Chester, PA 19013, USA
b
Department of Physics and Astronomy, Rutgers University, Piscataway, NJ 08854, USA
c
Center for Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Phosphorus functionalized nanoporous carbon was synthesized from lignin by periodical reactions with po-
Porous carbon tassium hydroxide and triphenylphosphine. It has the BET surface area of 837 m2/g and total pore volume
Adsorption 0.41 cm3/g along with 0.9 atom% phosphorus. The adsorption of neodymium (Nd (III)) and dysprosium (Dy(III))
Rare earth element (REE) were in the range of 335–344 mg/g upto their initial concentration of 500 ppm in water and such adsorption
Neodymium
amount is higher than that of majority of the adsorbents reported in literature. The adsorption capacity showed
Dysprosium
negligible dependency on the solution pH. The distribution coefficients for Nd(III) and Dy(III) were within 1000
to 10,000 mL/g. The adsorption capacity of iron (Fe(III)) under the similar conditions was one order of mag-
nitude lower and it suggests a possible separation of these rare earth elements from iron. The maximum se-
lectivity of separation of Nd(III) and Dy(III) from iron was ca. 32 to 61. Kinetic study revealed that the ad-
sorption of Dy was faster than that of Nd(III). The results of kinetic study were better fit with pseudosecond order
kinetics. XPS studies on both Nd(III) and Dy(III)-adsorbed carbons revealed a small shift in the P 2p3/2 energy
level of phosphorus towards higher energy level hereby suggesting a possible formation of metallic phosphates.
Additionally, Similar studies on Nd(III) and Dy(III)-adsorbed carbon indicated small amounts carbonates and
oxides of rare earth elements that might have also been formed. The overall results suggest that this carbon can
be used as a potential sorbent for enrichment and separation of Nd(III) and Dy(III).

1. Introduction elements in the periodic table, which play very critical role in the en-
ergy sectors of United States, and they are not presently mined, refined
Rare-earth elements (REEs) or rare-earth metals (REMs) are seven- or traded in large quantities. These elements, mostly constituting of rare
teen elements including fifteen lanthanides, scandium (Sc) and yttrium earths are termed as critical elements or ‘energy-critical elements’
(Y). In a broad spectrum, the rare earth elements can be classified into (ECEs). It was also suggested that the unavailability of these ECEs
two general categories of light rare-earths and heavy rare-earths with would limit the competitiveness of U.S. based industries and scientific
varying degree of demands and utilization [1]. Light rare-earths include sectors and eventually may lower the quality of life [3].
La, Ce, Pr, Nd, Sm, whereas heavy rare-earths consist of Eu, Gd, Tb, Dy, The rare earth elements possess very similar chemical properties
Ho, Er, Tm, Yb, Lu and Y. Heavy rare-earths have slightly more valuable with each other and with other neighboring elements of the periodic
and less available. Majority of these elements have a specific yet table and therefore, it is an arduous task to separate them. In order to
widespread utilizations in electronics, power sources and military ap- extract these metals from ores, different processes, like solid extraction,
plications [2]. Nd, Pr, Sm and Dy are utilized to produce strong per- co-precipitation and ion-exchange techniques were developed. Solvent
manent magnets that are used in electric motors, computers and gui- extraction with specific type of solvents is regarded as the most
dance systems. Y, Ce, La, Gd, Lu and Nd are used in sensors, optical common technique to enrich REEs from their ores [4–9]. In a global
devices, camera lenses, fiber optics and other electric materials, like perspective, China, USA and Australia are the leading REE producing
resistors and capacitors. Y, La, Nd, Eu, Tb and Dy are used in phosphors countries [10], while China controls 82.8% of worldwide REE pro-
for optical displays and lasers. Four rare-earth metals, Ce, La, Nd and Y duction followed by Australia (7.9%) and USA (3.2%) [11]. In order to
contribute to 85% of world’s rare earth element production [2]. Beside maintain a steady supply and recycle the existing rare earth elements
such usages, it has been suggested that there are few less familiar that are already used, a strategy has been developed to recover the rare-


Corresponding author.
E-mail address: dsaha@widener.edu (D. Saha).

http://dx.doi.org/10.1016/j.jece.2017.09.009
Received 19 June 2017; Received in revised form 5 September 2017; Accepted 6 September 2017
Available online 07 September 2017
2213-3437/ © 2017 Elsevier Ltd. All rights reserved.
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

earths from respective wastes in an aqueous solution, typically upon equilibrium and kinetic measurements were performed upto 500 ppm
acid digestion [12,13]. Recently, a strict export restriction of REE in of metal solution. To compare and calculate selectivity with iron (Fe
China created a shortage of REEs worldwide [14]. Despite acid ex- (III)), the equilibrium run with iron was also investigated. The binding
traction is commonly used in the purification of REEs from their ores, it of nature of REEs with phosphorous or oxygen functionalities was in-
is not effective in extracting the low concentration of REEs from the vestigated by XPS analysis.
wastes. Additionally, solvent extraction is considered as hazardous
owing to the involvement of large amount of toxic solvents [15]. It is 2. Experimental
also time-consuming, labor intensive and may leave undesired residues
[16–18]. 2.1. Synthesis of oxygen and phosphorous functionalized carbon
Neodymium (Nd) and dysprosium (Dy) are two of the important
rare earth elements. Dy is used in laser elements, metal halide lamps, Typically, 3 g de-alkaline lignin (TCI America) and 3 g potassium
magnetorestriction materials and to measure ionization radiation in hydroxide (KOH) (Sigma-Aldrich) were mixed in a coffee grinder and
dosimeters. Dy also has a very high thermal neutron absorption cross put in a porcelain boat. The boat was inserted within the combustion
sectional area and therefore it is used in neutron absorbing material in tube of a tube furnace (Lindberg-blue) and temperature was ramped
nuclear reactor. Because of its unusually high heat capacity, Nd is used from room temperature to 800 °C at a ramp rate of 10 °C/min. The final
is cryocoolers. The most important use of Nd is in magnets and it uti- temperature was maintained for 2 mins and then it was cooled in the
lized to make the strongest permanent magnets. Because of their per- room temperature before taking it out. All the heating and cooling
formance, the global demand of Nd and Dy may grow as high as 700% profiles are performed under nitrogen flow. The resultant carbons were
and 2600% respectively in the next two decades [19,12]. Recovery as washed with copious amounts of DI water several times and filtered and
well as recycle of few REEs, like neodymium (Nd) and dysprosium (Dy) dried. The resultant carbon is mixed with triphnylphosphine (P(C6H5)3)
from the waste NdFeB magnets is one of the key REE separation needs in 1:5 ratio of carbon to triphnylphosphine [38] and heated in a por-
of today [14,20]. Recently, European Commission declared Dy as one of celain boat inside the same tube furnace. Temperature was raised to
key critical metals with highest supply risk [21]. Inappropriate disposal 800 °C in 10 °C/min in nitrogen flow and then cooled down to room
of waste NdFeB magnets may also cause environmental pollution that temperature. Thus obtained carbon is again washed several times with
provides an added need to separate REEs from such waste magnets. DI water, filtered and dried for further analysis.
Adsorption-based processes may provide an excellent alternative to
separate REEs from wastes compared to the liquid–liquid extraction 2.2. Characterization of carbons
processes. It is a sustainable, benign and inexpensive process compared
to almost all other processes that have been employed to recover REEs. The carbon was characterized for pore textural properties by N2 and
Over the years, varieties of natural, synthetic and bio-based sorbents CO2 adsorption, surface functionality by x-ray photoelectron spectro-
have been employed for this purpose [1]. For Nd and Dy adsorption, scopy (XPS) and shape and size scanning electron microscopy (SEM).
different types of composite sorbents, like calcium alginate-poly glu- The pore textural properties including BET surface area and pore vo-
tamic acid hybrid gels [22], silica-based urea formaldehyde composites lume were calculated from N2 adsorption isotherm at 77 K. The overall
impregnated with organophosphorous extractant [23], magnetic nano- pore size distribution was obtained by employing non-local density
hydroxyapatite [24], EDTA- and DTPA-functionalized chitosan biopo- function theory (NLDFT) on both N2 adsorption isotherm at 77 K and
lymers [25], imprinted mesoporous silica [14], flower-like [26], nano CO2 adsorption isotherm at 273 K. Pore size below 12 Å was obtained
Mg(OH)2, functionalized silica (KIT-6-N-DGA) and porous organic fra- from CO2 adsorption isotherm whereas pore size above 12 Å was ob-
mework (a type of MOF) [27] were successfully employed to enrich tained from N2 adsorption isotherm. The gas adsorption experiments
them from aqueous solutions. It has also been suggested that oxygen were carried out in Quantachrome’s Autosorb iQ-Any gas instrument
functionality including carboxylate group and phosphorous function- and all the necessary calculations including BET analysis and non-local
ality favor the affinity of REEs to bind to the adsorbent surface, prob- density function theory (NLDFT) calculations were performed in in-
ably by forming chelates and complexes. Binding with oxygen func- strument’s built-in software. Before all the gas adsorption measure-
tionality has clearly been demonstrated by attachment of Nd, Dy and Pr ments, the samples were outgassed in 300 °C for 3 h and below 1 Torr
onto EDTA- and DTPA-functionalized chitosan biopolymers [25]. pressure. XPS studies were performed using a Thermo-Fisher K-Alpha
Binding of REEs to phosphorous functionalized materials were suc- instrument. Photoelectrons were excited with a monochromatic Al-Kα
cessfully demonstrated in the separation of REEs in a packed column x-ray source with an energy of 1486 eV, and the total instrumental
loaded with DNA-immobilized filter paper [28], freeze-dried salmon resolution was 0.5 eV. The pass energy, step size and dwell time were
milt [29] and phosphoric acid immobilized on silica backbone [30]. LIII- 50 eV, 0.1 eV and 50 ms, respectively. The sample was mounted on a
edge extended x-ray absorption fine structure (EXAFS) confirmed the carbon tape and charge neutralization was performed by using 2 eV
binding of phosphorous atom with REEs [28,29]. Affinity of REEs to- Ar+ ions. Scanning Electron Microscopic (SEM) images were captured
wards oxygen functionalities was also the underlying reason of em- in Carl Zeiss Merlin SEM microscope operating at 1 kV.
ploying bio-based sorbents [2], like bacterial cell wall [31,32], to in-
teract and capture REEs. Different other types of functionalized 2.3. Adsorption of metals in carbon
adsorbents that were employed to adsorb Dy are Imprinted styr-
ene–divinylbenzene copolymer [33], oxidized multi-walled carbon na- All the REE adsorption experiments were performed in a round-
notubes [34], 11-Molybdo-vanadophosphoric acid supported on Zr bottom flask with 25 mL of aqueous salt solution in DI water and
modified mesoporous silica SBA-15 [35], Phosphonic acid-functiona- 0.025 g of carbon. The temperature of mixing was approximately room
lized porous microspheres [36], and DETA-functionalized chitosan temperature (298 K). Neodymium nitrate (Nd(NO3)3·6H2O), dyspro-
magnetic nano-based particles [37]. sium nitrate (Dy(NO3)3·5H2O) and ferric nitrate (Fe(NO3)3·9H2O) were
In the past, majority of the phosphorous-doped carbons were syn- used as sources of neodymium, dysprosium and iron, respectively. For
thesized by activating with phosphoric acid. In this work, we have re- equilibrium studies, REE salts were dissolved in DI water to generate 50
ported the synthesis of phosphorous-functionalized microporous carbon to 500 ppm solution with respect to the metal and each of them was
obtained from bio-based (lignin) precursor along with phosphorous stirred with carbon for 4 h. Kinetics studies were performed with Nd
functionalization by triphenylphosphine. This adsorbent was char- (III) and Dy(III) for 500 ppm solution only and samples were withdrawn
acterized with pore textural properties, electron microscopy (SEM) and in the time interval of 30 s, 2 min, 15 min, 40 min, 1 h, 2 h, 3 h and 4 h.
x-ray photoelectron spectroscopy (XPS). Nd and Dy adsorption In order to investigate pH dependency of adsorption, the 36% HCl

4685
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

solution was added to the solution of REE salts (for 500 ppm only). The
pH values of 6.1 (for Nd(III)) or 6.6 (for Dy(III)) were achieved by direct
mixing of the metal salts without any effort of additional adjustments.
pH adjustment with NaOH was not performed as it causes precipitation
of metal hydroxides. All the quantitative analysis of metals was per-
formed via UV spectroscopic method by using arsenezo-III method. The
linear range of UV calibration was obtained for 0.1 to 0.4 ppm con-
centration of metal and hence all the solutions are diluted as needed to
meet the concentration requirements. Typically, 2.3 mL of diluted
metal solution is added with 600 μL of acetate buffer and then mixed
with 120 μL 0.1% (w/v) arsenezo-III in DI water. The mixture was
soaked for 10 min and then analyzed in Thermo-Scientific Genesis
UV–vis spectroscope against a blank. The peak obtained at 616 and
622 nm wavelength for Nd(III) and Dy(III), respectively, (Supporting
information, Fig. S3) were used for the study. Arsenezo-III solution was
prepared fresh everyday.

3. Results and discussion

3.1. Materials characteristics

All the pore textural properties of the oxygen and phosphorus


functionalized carbon were calculated by analyzing N2 adsorption-
desorption plot at 77 K and CO2 adsorption at 273 K (Fig. 1a and b).
According to IUPAC nomenclature, the isotherm was of type-I sug-
gesting predominantly microporous structure within the material. The
Brunauer-Emmett-Teller (BET) surface area was of the carbon is
837 m2/g. The overall pore size distribution is shown in Fig. 1c. Fig. 1c
suggests that the carbon has narrow micropore widths in the regions of
3.5, 4.8, 5.2 and 8.2 Å and larger micropore widths in 14.7 and 18.5 Å.
Significant mesopore contribution was not detected in the pore size
distribution analysis. The total pore volume of the carbon is 0.41 cm3/
g. In order to understand the change in porosity upon phosphorous
functionalization, the pore textural properties of the pristine (KOH
activated) carbon before phosphorous functionalization was also mea-
sured. The BET surface area of non-functionalized carbon is 2250 m2/g
with the total pore volume of that carbon is 0.9 cm3/g. (Detailed por-
osity information of unfunctionalized carbon is shown in Fig. S1 of
Supporting information) So, it is evident that phosphorous functiona-
lization reduced the surface area and pore volume of the carbon.
The scanning electron microscopic images are shown in Fig. 2a–d in
different magnification levels. The overall particle size was in the range
of few microns to few hundred microns. These images confirmed that
the carbon has different sizes of macroporosity, including 1–10 μm and
500–699 nm. Fig. 2c shows the presence of an array of larger porous
entities of about 1 μm width. These larger pores may be beneficial in
improving the kinetics of adsorption as discussed later. For comparison,
the SEM images of unfunctionalized carbon are shown in Fig. S2 of
Supporting information.
Surface functionality of this carbon was analyzed in details by X-ray Fig. 1. N2 adsorption-desorption plot at 77 K(a), CO2 adsorption at 273 K (b) and pore
photoelectron spectroscopy (XPS) and the overall spectra are given in size distribution obtained by NLDFT technique (C).
Supporting information (Fig. S4). Peak fitting results for C-1s, O–1 s and
P-2p are shown in Fig. 3a–c. The elemental compositions of carbon, been reported. In the overall survey (Fig. S4 of Supporting informa-
oxygen and phosphorous are 70.5, 27.6 and 0.9 atom%, respectively. tion), the remaining two major contributions originated from carbon
The two deconvoluted peaks for phosphorous appear at binding en- and oxygen. Different carbon functionalities that were present in the
ergies of 132.2 eV and 133.8 eV and are characteristic of the 2p3/2 and system are C- sp2, C-sp3, C-O/C-N and C=O/COOH/OeC=O. Their
2p1/2 states of a single P chemical state. These values are slightly lower quantitative contributions were in the range of 50.0, 14.8, 2.0 and 3.7
than the 132.5 eV and 134.2 eV binding energies reported in the lit- atom%, respectively. Primary oxygen functionalities were C-O/OH/
erature [39]. According to literature, these peaks were ascribed to both N=O and C-O-H and their quantitative contributions were 19.6 and 8.0
higher oxidation states [39–41], as well as bound phosphorous on atom%, respectively. The oxygen functionality originated on the carbon
carbon surface [39] and hence the only phosphorous functionality that surface from both original carbon precursor (lignin) and KOH during
was detected by XPS is C-P-O. The slight decrease in binding energy for the time of activation. The rest of the elements were Na, K and traces of
phosphorous may be attributed to the interaction of phosphorous with Mg. Potassium (K) was originated from KOH, whereas the rest of ele-
enlarged aromatic carbon ring system [42]. In the literature, both lower ments were probably originated from porcelain boat during the time of
(0.26 to 1.84 at.% as measured by XPS [43,44,39,41]) and higher carbonization or activation. XPS analysis was also performed on
phosphorous content (3–15%) [45,46], compared to our results, have

4686
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

Fig. 2. SEM images of phosphorous-doped carbon in different


magnifications; scale bar 100 μm (a), 10 μm (b), 1 μm (c) and
200 nm(d).

neodymium (Nd) and dysprosium (Dy) adsorbed carbons and those different kinetic behavior. Adsorption of Dy(III) was rapid, over 67% of
discussed in details later in the manuscript total adsorption was completed in 2 mins only, and rest of the ad-
sorption took place in 4 h. On the other hand, adsorption of Nd(III) was
quite slow, it took about one hour to complete 67–68% of total ad-
3.2. Adsorptive recovery and separation of neodymium and dysprosium sorption and rest of the adsorption took place in remaining 3 h. Inter-
estingly, the total adsorption amount of both the metals remained same,
Neodymium (Nd(III)), dysprosium (Dy(III)) and iron (Fe(III)) equi- which also supports the equilibrium adsorption study given in Fig. 4.
librium adsorption in aqueous phase was performed with their corre- Both slow adsorption of Nd [25,24,27], and rapid adsorption of Dy(III)
sponding nitrate salts with the maximum concentration of 500 ppm. [14] can also be seen in different past studies already discussed in this
The equilibrium adsorption isotherms of the metals are shown in Fig. 4 article. Owing to the limited information on the detailed behavior or
and the iron adsorption isotherm is shown in magnified in the inset of characteristics of Nd(III) or Dy(III) ion, it is quite challenging to identify
the same figure. All the adsorption isotherms are at pure or single the key reason that make the adsorption of Dy(III) faster than Nd (III).
component basis. It is observed that Nd(III) and Dy(III) demonstrated Dy(III) ion is only slightly larger than that of Nd(III) and therefore, the
very similar adsorption behavior including their overall adsorption size may not be the key contributing factor. The differences in ionic
capacity. The overall adsorption capacity of Nd(III) and Dy(III) was dissociation and competition with anions could be the contributing
335.5 ± 33.5 mg/g and 344.6 ± 23.23 mg/g, respectively (corre- factor, but it is not possible to make a realistic hypothesis without a
sponding to 2.35 and 2.12 mmol/g for Nd(III) and Dy(III), respec- series of different kinds of relevant experiments, which are beyond the
tively.) On the contrary, Fe(III) adsorption was very low. The adsorp- scope of this work.
tion isotherm demonstrated a linear nature upto 300 ppm and then The mechanism of adsorption of any species onto porous adsorbents
relatively sharper increase to about 47.6 mg/g. The overall uptake of Fe (i.e., carbon) can comprise of four stages, (a) migration of the species
(III) remained one order of magnitude lower compared to Nd(III) and from bulk of the solution to the boundary layer surrounding the carbon
Dy(III). (bulk diffusion), (b) diffusion of the species from boundary layer to the
In order to investigate the difference in adsorption capacity between external surface of carbon, (c) transport of the species from the external
phosphorous-functionalized and non-functionalized carbon, we also surface of carbon to the pores of the carbon (intraparticle diffusion) and
measured the Nd(III) and Dy(III) adsorption capacity in un- (d) adsorption or chemical complexation of the species on the pores or
functionalized (without phosphorous) carbon before its exposure to at the active sites of the carbon. Out of these four steps, the intraparticle
triphenylphosphine. It is to be noted that the BET surface area of non- diffusion can be modeled as [47].
functionalized carbon 2250 m2/g, which is about over 2.5 times higher
than that of phosphorous-doped carbon (837 m2/g). It was observed qt = Kidt1/2 + C (1)
that the Nd(III) and Dy(III) adsorption capacities in unfunctionalized
where qt is the amount adsorbed at time t and Kid is the intraparticle
carbon were 254 ± 9 and 286 ± 4 mg/g, respectively, for 500 ppm
diffusion rate constant (mg g−1 min−1/2). The intraparticle diffusion is
initial concentration. Lower adsorption capacity for both the REEs in a
the sole limiting step only if the linear regression of qt versus t1/2 passes
unfunctionalized carbon with very high surface area may suggest the
through the origin [48]. We found that, for only Nd(III), such linear
importance of phosphorous functionalization.
regression passes through the origin giving rise to the Kid value of
To the best of our knowledge, Nd(III) and Dy(III) equilibrium ad-
24 mg g−1min−1/2 (R2 value of 0.93). Therefore, it can be stated that
sorption in our adsorbent is higher than that of majority of other ad-
the slow intraparticle diffusion could be the underlying reason for
sorbents reported in literature, in some cases, these values are orders of
sluggish diffusion of Nd(III).
magnitude lower. The detailed comparison of the adsorption values
The pseudofirst order rate equation is given by
with other published reports were shown in Table 1.
The kinetics of REE adsorption was investigated by studying the k
adsorption amounts of Nd(III) and Dy(III) for 500 ppm concentration log(qe − qt ) = logqe − ⎛ 1 ⎞ t
⎝ 2.303 ⎠ (2)
only and in the interval of 30 s, 2 min, 15 min, 40 min, 1 h, 2 h, 3 h and
4 h (Fig. 5). It was observed that Nd(III) and Dy(III) demonstrated where, k1 is pseudofirst order rate constant and calculated by linear

4687
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

Furthermore, the micropore diffusion in an adsorbent material can


be modeled by the following equation,

mt 6 −π 2Dc t ⎞
1− = 2 exp ⎛⎜ 2 ⎟
m∞ π ⎝ rc ⎠ (4)

where mt and m∞ are mass adsorbed in time t and final adsorption


amount, respectively, D c is intracrystalline diffusivity in micropores and
Dc
rc is intracrystalline radius. 2 is referred to as diffusive time constant
rc
m
and can be calculated by the linear regression of 1 − m t versus t. The

diffusive time constants of Nd(III) and Dy(III) are 0.70 × 10−5 s−1and
−5 −1
1.01 × 10 s , respectively. Faster diffusive time constant of Dy(III)
supports its rapid kinetics. The peak fitting results of Eq. (4) is shown in
Fig. S6(a) and (b) of the Supporting information.
The dependency of solution pH on the adsorption of Nd(III) and Dy
(III) was investigated by using 3 pH values of 2, 4 and 6.1 (Nd(III)) or
6.6 (Dy(III)) and the results are shown in Fig. 6. The adsorption at
higher pH could not be investigated as any addition of base resulted in
precipitation of metal hydroxides. The results suggested that the de-
pendency of adsorption of both Nd(III) and Dy(III) on solution pH is
insignificant. Such a trend provides an excellent advantage of our ad-
sorbents over other sorbents reported in literature as most them de-
monstrated lower adsorption at lower pH [14,32,25,26].
For aqueous phase adsorption, it is a common practice to calculate
distribution coefficient (Kd) defined as,

Ci − Cf V
Kd = ⎛ ⎞
Cf ⎝m⎠ (5)

where, Ci is initial concentration, Cf final concentration, V volume of


solution (mL) and m mass of the adsorbent (g) used. The distribution
coefficient values as a function of initial concentration is shown in
Fig. 7. These values lie between ca. 6000 to 679 mL/g and 1223 to
215 mL/g for Dy(III) and Nd(III), respectively. Owing to low adsorp-
tion, the distribution coefficients for Fe(III) are within 187 to 126 mL/g
within the same initial concentration. One order of magnitude differ-
ence in distribution coefficients between Fe(III) and Nd(III) or Dy(III)
signifies the possible separation between them. To the best of our
knowledge, our distribution coefficient values for Nd(III) and Dy(III)
are higher than majority of reports available in literature, especially at
lower pH. The Kd values for Nd(III) and Dy(III) was ca. 175 and 25 mL/
g in imprinted mesoporous silica [14]. These values were about 500 to
900 for Nd(III) and Dy(III) in microbial strain-based adsorbents [32]. In
DTPA chitosan-based sorbents, the maximum Kd values for Nd(III) and
Dy(III) reached about 10,000 mL/g, but quickly fell below 1000 mL/g
when it pH dropped below [25], 2. The consistency of adsorption ir-
respective to solution pH for our adsorbents suggests a much higher
distribution coefficient in lower pH. In carboxylic acid functionalized
porous aromatic framework [27], the Kd value for Nd at the higher pH
was reported to be higher than ca. 30,000 mL/g, it also dropped below
Fig. 3. Detailed XPS peak fitting of for C–1 s (a), O–1 s (b) and P-2p (c) of the phos- 20 at a pH lower than 2. In a surface modified porous silica KIT-6, the
phorous-doped carbon. The curve at the top of each plot shows the residuals of peak
distribution coefficients for Nd(III) and Dy(III) in the mixture of REEs
fitting results.
were reported to be 4000 to 4500 mL/g.
The enrichment factor can simply be defined as the ratio of moles of
regression of log(qe − qt) versus t plot. On the other hand, pseudose- desired component in the adsorbed phase over moles of the same
cond order rate constant is given by component in the remaining solution after adsorption when they are in
equilibrium. Quite obviously, the enrichment factor is a function of
t 1 1
= + ⎜⎛ ⎞⎟ t initial concentration of the metals (REEs). From Fig. 8a, it is clear that
qt k2 qe2 ⎝ qe ⎠ (3) the highest enrichment factors are obtained at the lowest concentration
where, k2 is the pseudosecond order rate constant and can be calculated of 50 ppm and these values are 6300 to 11800 for Nd(III) and Dy(III),
by linear regression of t versus t plot. It is observed that pseudofirst respectively. At the higher concentration, the enrichment factors
qt
dropped and remained almost steady value of 670 for Nd(III) and Dy
order rate constant fits with moderate accuracy with R2 value
(III).
∼0.90–0.91. However, pseudosecond order rate equation fits very well
The selectivity of component 1 (preferred component) over com-
both Nd(III) and Dy(III) with R2 ∼0.93–0.99. Both the rate constants
ponent 2 (non-preferred component) is defined as [49,50],
for Nd and Dy are shown in Table 2.

4688
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

Fig. 4. Equilibrium adsorption plots for Nd(III), Dy(III) and


Fe(III). The inset shows the details and full adsorption iso-
therm of Fe(III). The dose was 0.025 g carbon with 25 mL
solution of REEs. The temperature of mixture was approxi-
mately 298 K. Error bar shows the standard deviation.

(x1/ y1)
S1/2 =
(x2 / y2 ) (6)

where, component 1 and 2 signify the preferred and non-preferred


adsorbate, respectively, and x and y are adsorbed and bulk phase
concentration in moles when are they are in equilibrium. The common
technique for calculation of selectivity is Ideal Adsorbed Solution
Theory (IAST) and it has been applied to find the selectivity for REEs
[29]. However, there is no well-investigated study on how IAST can be
used for aqueous phase adsorption as, according to the original pub-
lished paper [51], it could be applied for gas phase adsorption only.
Therefore, we did not employ this technique. The x values can be ob-
tained by simply converting the adsorption amount in (mg/g) to (mol/
g). The y values are the residual strengths of the corresponding solution
after adsorption and converted from ppm to mol/l. Obviously, se-
lectivity values are the functions of initial concentrations and the re-
sults for SNd(III)/Fe(III) and SDy(III)/Fe(III) (in 1:1 mass ratio for Nd(III)/Fe
(III) or Dy(III)/Fe(III)) are shown in Fig. 8b. It was observed that both Fig. 5. Kinetics of Nd(III) and Dy(III) adsorption at the initial concentration of 500 ppm.
the selectivity values were higher at the lower concentration, but de- The dose was 0.025 g carbon with 25 mL solution of REEs. The temperature of mixture
creased at the elevated concentration. In this context, it is worth was approximately 298 K. Error bar shows the standard deviation.
mentioning that selectivity of Nd(III) over Dy(III) or vice versa is about
unity owing to their similar adsorption capacity. Therefore, they cannot

Table 1
Comparison of Nd and Dy adsorption in different types of adsorbents.

Adsorbent type Element separated Maximum adsorption Equilibrium Concentration Reference


capacity

Imprinted mesoporous silica in dialysis bag Dy(III) 22.63 mg/g 80 mg/g [14]
Silica-based urea-formaldehyde composite with impregnated organophosphorous Nd(III) 4.96 mg/g – [23]
Carboxylic Acid-Functionalized Porous Aromatic Framework Nd(III) 2.2 mmol/g 288.4 mg/g 1–4 mM [27]
Freeze-dried microbial strains Dy(III), Nd(III) 3 μmol/g (∼10−5 mg/g) – [32]
EDTA and DTPA-functionalized chitosan biopolymers Dy(III), Nd(III) 80 mg/g 1800 mg/L [25]
Magnetic nano-hydroxyapatite Nd(III) 300 mg/g 200 mg/L [24]
Imprinted styrene–divinylbenzene copolymer Dy(III) 40.15 – [33]
Oxidized multi-walled carbon nanotubes Dy(III) 22 mg/g 50 mg/L [34]
11-Molybdo-vanadophosphoric acid supported on Zr modified mesoporous silica Dy(III) 50 mg/g – [35]
SBA-15
Mesoporous silicas functionalized with phosphonic acid groups Dy(III) 0.3 mmol/g 2.5 mmol/L [36]
Diethylenetriamine-functionalized chitosan magnetic nanoparticle Dy(III), Nd(III) 50 mg/g 300 mg/L [37]
Phosphorous functionalized nanoporous carbon Nd(III), Dy(III) 335.5 mg/g (Nd (III) 175 ppm (Nd (III) This work
344.6 mg/g (Dy (III) 162 ppm(Nd (III)

4689
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

Table 2
Rate constants Nd and Dy adsorption.

Rate constants Nd(III) R2 values Dy(III) R2 values

Pseudofirst order (k1), 9.21 × 10−3 0.9 8.06 × 10−3 0.91


(min−1)
Pseudosecond order (k2), 1.14 × 10−4 0.93 3.39 × 10−4 0.99
(g mg−1 min−1)

Fig. 6. Dependence of Nd(III) and Dy(III) adsorption on pH of solution. All initial con-
centration was 500 ppm with respect to Nd and Dy. Error bar shows the standard de-
viation.

Fig. 8. Enrichment factor (a) and selectivity (b) for Nd(III) and Dy(III) recovery.

phosphorous compounds may be present in the system and possible


complexation of REE metals with phosphorus functionalities on carbon
surface in a small extent. Besides that, the XPS spectra also demon-
strated a small amount of carbonate (CO3−) peak at about 289.9 eV in
Nd(III) adsorbed carbon, which may indicate formation of Nd(CO3)3 or
possible complexation of neodymium with oxygen functionalities on
carbon surface, like with carboxylic acid groups. Additionally, O–1 s
spectrum of Nd(III) adsorbed carbon showed evidence of a small peak at
529 eV, which is consistent with metal oxides. It also corroborates the
possible formation a small amount of Nd2O3 or similar complexation
with oxygen functionalities.

Fig. 7. Dependence of distribution coefficient (Kd) on initial concentration of Nd(III) and 4. Conclusions
Dy(III) in the equilibrium adsorption study. Error bar shows the standard deviation.
In this work, phosphorous and oxygen functionalized nanoporous
be separated by this carbon. carbon was synthesized from lignin as carbon precursor along with si-
In order to further understand the interaction of Nd(III) and Dy(III) multaneous activation with potassium hydroxide and triphnylpho-
with the carbon surface, we have performed the XPS analysis of these sphine. The resultant carbon demonstrated a predominantly micro-
metal adsorbed carbons and the results are shown in Fig. 9a–d. The porous structure with BET surface area of 837 m2/g along with
doublet neodymium peaks appear at about 1005.4 eV and 983.4 eV for phosphorous content of 0.9 atom%. Nd(III) and Dy(III) demonstrated
Nd-3d3/2 and Nd-3d5/2 respectively, whereas same peaks for dyspro- equilibrium adsorption of ca. 335–344 mg/g, which is higher than
sium appear at about 1335.1 eV and 1296.4 eV. These peaks in the majority of the adsorbents reported in literature. This adsorption
carbons samples are unchanged from their reference pure Nd amount of Nd(III) and Dy(III) was one order of magnitude higher that of
(NO3)3.6H2O and Dy(NO3)3.5H2O samples, which means that these iron (Fe(III)) thereby suggesting their possible separation from iron. pH
metals retained their +3 oxidation states, and their chemical change is results suggested that the equilibrium adsorption is almost independent
insignificant. However, from P-2p3/2 analysis (Supporting information, of solution pH. The kinetics of adsorption of Dy(III) was much faster
Fig. S5), it is observed that the binding energy of P-2p3/2 is changed than that of Nd(III). The kinetic data was successfully modeled with
from 132.2 eV to 132.8 eV. This change is close to expected energy pseudofirst order, pseudosecond order and micropore diffusion models.
level metal phosphates (∼133 eV). It could indicate multiple The maximum distribution coefficients for Nd(III) and Dy(III) were
1000–10,000 mL/g. The maximum calculated selectivity of separation

4690
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

Fig. 9. Peak fitting of XPS analysis in Nd(III) and Dy(III) −adsorbed carbon, Dy-3d (a), Nd-3d (b), C–1 s (c) and O–1 s (d). The curve at the top of each plot shows the residuals of peak
fitting results.

of Nd(III)/Fe(III) and Dy(III)/Fe(III) were about 60. The XPS analysis of Talanta 45 (1997) 437–444.
[6] C.H. Xiong, X.Z. Liu, C.P. Yao, Effect of pH on sorption for RE(III) and sorption
Nd(III) and Dy(III)-adsorbed carbon suggested the primary chemical behaviors of Sm(III) by D152 resin, J. Rare Earths 26 (2008) 851–856.
components of Nd(III) and Dy(III) were not changed. However a de- [7] C.H. Xiong, Sorption behavior of D155 resin for Ce(III), Ind. J. Chem 47A (2008)
tailed analysis suggested a small but possible complexation with 1377–1380.
[8] Y. Zhu, Y. Zheng, A. Wang, A simple approach to fabricate granular adsorbent for
phosphorous, carbon or oxygen in the form of phosphates, oxides or adsorption of rare elements, Int. J. Biol. Macromol. 72 (2015) 410–420.
carbonates. The future direction of the work includes designing of a [9] F. Xie, T.A. Zhang, D. Dreisinger, F. Doyle, A critical review on solvent extraction of
packed bed composed of this sorbent and investigate the physical se- rare earths from aqueous solutions, Minerals Eng. 56 (2014) 10–28.
[10] Rare Earths, Mineral Commodity Summaries, U.S.G. Survey, 2014, pp. 128–129.
paration of the rare earth elements in a continuous fashion.
[11] https://www.ausimmbulletin.com/feature/an-outlook-on-the-rare-earth-elements-
mining-industry/ (Accessed August 2017).
Acknowledgement [12] K. Binnemansa, P.T. Jones, B. Blanpain, T.V. Gerven, Y. Yangd, A. Waltone,
M. Buchert, Recycling of rare earths: a critical review, J. Cleaner Prod. 51 (2013)
1–22.
This work was partly supported by faculty development award and [13] K. Nansai, K. Nakajima, S. Kagawa, Y. Kondo, Y. Shigetomi, S. Suh, Global mining
provost grant from Widener University. TEM (J.C. and H.C.H.) and SEM risk footprint of critical metals necessary for low-carbon technologies: the case of
(D.K.H) experiments were partially conducted under the user proposal neodymium, cobalt, and platinum in Japan, Environ. Sci. Technol. 49 (4) (2015)
2022–2031.
(CNMS2016-302) at the Center for Nanophase Materials Sciences, [14] X. Zheng, E. Liu, F. Zhang, Y. Yan, J. Pan, Efficient adsorption and separation of
ORNL, which is a DOE Office of Science User Facility. dysprosium from NdFeB magnets in an acidic system by ion imprinted mesoporous
silica sealed in a dialysis bag, Green Chem. 18 (2016) 5031–5040.
[15] C.P. Yao, Adsorption and desorption properties of D151 resin for Ce(III), J. Rare
Appendix A. Supplementary data Earths 28 (2010) 183–188.
[16] G.M. Ritcey, A.W. Ashbrook, Solvent Extraction: Principle and Applications to
Supplementary data associated with this article can be found, in the Process Metallurgy, Part I, Amsterdam Elsevier Press, 1984, p. 603.
[17] O. Samuelson, Ion Exchangers in Analytical Chemistry, John Wiley, 1972, p. 415.
online version, at http://dx.doi.org/10.1016/j.jece.2017.09.009. [18] C.H. Xiong, Z.W. Zheng, Evaluation of D113 cation exchange resin for the removal
of Eu(III) from aqueous solution, J. Rare Earths. 28 (2010) 862–867.
References [19] E. Alonso, A.M. Sherman, T.J. Wallington, M.P. Everson, F.R. Field, R. Roth,
R.E. Kirchain, Evaluating rare earth element availability: a case with revolutionary
demand from clean technologies, Environ. Sci. Technol. 46 (2012) 3406–3414.
[1] I. Anastopoulosa, A. Bhatnagar, E.C. Limac, Adsorption of rare earth metals: a re- [20] B. Sprecher, R. Kleijn, G.J. Kramer, Recycling potential of neodymium: the case of
view of recent literature, J. Mol. Liq. 221 (2016) 954–962. computer hard disk drives, Environ. Sci. Technol. 48 (2014) 9506–9513.
[2] N. Das, D. Das, Recovery of rare earth metals through biosorption: an overview, J. [21] E. Commission, European Commission, DG Enterprise & Industry, Brussels, 2014.
Rare Earths 31 (2013) 933–943. [22] F. Wang, J. Zhao, X. Wei, F. Huo, W. Li, Q. Hu, H. Liu, Adsorption of rare earths (III)
[3] https://www.aps.org/policy/reports/popa-reports/upload/elementsreport.pdf by calcium alginate–poly glutamic acid hybrid gels, J. Chem. Technol. Biotechnol.
(Accessed January 2017). 89 (2014) 969–977.
[4] J.G. Sengupta, Determination of scandium, yttrium and lanthanides in silicate rocks [23] A. Naser, G.S. El-deen, A.A. Bhran, S. Metwally, A. El-Kamash, Elaboration of im-
and four new Canadian iron formation reference materials by flame atomic-ab- pregnated composite for sorption of europium and neodymium ions from aqueous
sorption spectrometry with micro sample injection, Talanta 31 (1984) 1045–1051. solutions, J. Ind. Eng. Chem. 32 (2015) 8–272.
[5] J.S. Kim, C.H. Lee, S.H. Han, M.Y. Suh, Studies on complexation and solvent ex- [24] C. Gok, Neodymium and samarium recovery by magnetic nano-hydroxyapatite, J.
traction of lanthanides in the presence of diaza-18-crown-6-di-isopropionic acid, Radioanal. Nucl. Chem. 301 (2014) 641–651.

4691
D. Saha et al. Journal of Environmental Chemical Engineering 5 (2017) 4684–4692

[25] J. Roosen, K. Binnemans, Adsorption and chromatographic separation of rare earths (III)], Cellulose 22 (2015) 2589–2605.
with EDTA- and DTPA-functionalized chitosan biopolymers, J. Mater. Chem. A 2 (5) [38] C. Zhang, N. Mahmood, H. Yin, F. Liu, Y. Hou, Synthesis of phosphorus-doped
(2014) 1530–1540. graphene and its multifunctional applications for oxygen reduction reaction and
[26] C. Li, Z. Zhuang, F. Huang, Z. Wu, Y. Hong, Z. Lin, Recycling rare earth elements lithium ion batteries, Adv. Mater. 25 (2013) 4932–4937.
from industrial wastewater with flowerlike Nano-Mg(OH)2, ACS Appl. Mater. [39] J. Wu, Z. Yang, X. Li, Q. Sun, J. Jin, P. Strasserd, R. Yang, Phosphorus-doped porous
Interfaces 5 (2013) 9719–9725. carbons as efficient electrocatalysts for oxygen reduction, J. Mater. Chem. A 1
[27] S. Demir, N. Brune, J.F.V. Humbeck, A.J. Mason, T.V. Plakhova, S. Wang, G. Tian, (2013) 9889–9896.
S.G. Minasian, T. Tyliszczak, T. Yaita, T. Kobayashi, S.N. Kalmykov, H. Shiwaku, [40] L.S. Dake, D.R. Baer, D.M. Friedrich, Auger parameter measurements of phosphorus
D.K. Shuh, J.R. Long, Extraction of lanthanide and actinide ions from aqueous compounds for characterization of phosphazenes, J. Vac. Sci. Technol. A 7 (1989)
mixtures using a carboxylic acid-functionalized porous aromatic framework, ACS 1634–1638.
Cent. Sci. 2 (4) (2016) 253–265. [41] C. Zang, N. Mahmood, H. Yin, F. Liu, Y. Hou, Synthesis of phosphorus-doped gra-
[28] Y. Takahashi, K. Kondo, A. Miyaji, M. Umeo, T. Honma, S. Asaoka, Recovery and phene and its multifunctional applications for oxygen reduction reaction and li-
separation of rare earth elements using columns loaded with DNA-filter hybrid, thium ion batteries, Adv. Mater 25 (2013) 4932–4937.
Anal. Sci. 28 (2012) 985–992. [42] A.M. Puziy, O.I. Poddubnaya, R.P. Socha, J. Gurgul, M. Wisniewski, XPS and NMR
[29] Y. Takahashi, K. Kondo, A. Miyaji, Y. Watanabe, Q. Fan, T. Honma, K. Tanaka, studies of phosphoric acid activated carbons, Carbon 46 (2008) 2113–2123.
Recovery and Separation of Rare Earth Elements Using Salmon Milt. 9 (12) (2014) [43] Z.W. Liu, F. Peng, H.J. Wang, H. Yu, W. Zheng, J. Yang, Phosphorus-doped graphite
e114858. layers with high electrocatalytic activity for the O2 reduction in an alkaline
[30] H.J. Park, L.L. Tavlarides, Adsorption of Neodymium(III) from aqueous solutions medium, Angew Chemie Int. Ed. 50 (14) (2011) 3257–3261.
using a phosphorus functionalized adsorbent, Ind. Eng. Chem. Res. 49 (24) (2010) [44] D. Yu, Y. Xue, L. Dai, Vertically aligned carbon nanotube arrays Co-doped with
12567–12575. phosphorus and nitrogen as efficient metal-free electrocatalysts for oxygen reduc-
[31] Y. Takahashi, X. Châtellier, K.H. Hattori, K. Kato, D. Fortin, Adsorption of rare earth tion, J. Phys. Chem. Lett. 3 (2012) 2863–2870.
elements onto bacterial cell walls and its implication for REE sorption onto natural [45] M. Myglovets, O.I. Poddubnaya, O. Sevastyanova, M.E. Lindström, B. Gawdzik,
microbial mats, Chem. Geol. 219 (2005) 53–67. M. Sobiesiak, M.M. Tsyba, V.I. Sapsay, D.O. Klymchuk, A.M. Puziy, Preparation of
[32] H. Moriwaki, R. Masuda, Y. Yamazaki, K. Horiuchi, M. Miyashita, J. Kasahara, carbon adsorbents from lignosulfonate by phosphoric acid activation for the ad-
T. Tanaka, H. Yamamoto, Application of freeze-dried powders of genetically en- sorption of metal ions, Carbon 80 (2014) 771–783.
gineered microbial strains as adsorbents for rare earth metal ions, ACS Appl. Mater. [46] Y. Guo, D.A. Rockstraw, Physical and chemical properties of carbons synthesized
Interfaces 8 (2016) 26524–26531. from xylan, cellulose, and Kraft lignin by H3PO4 activation, Carbon 44 (2006)
[33] V. Biju, J.M. Gladis, T.P. Rao, Ion imprinted polymer particles: synthesis, char- 1464–1475.
acterization and dysprosium ion uptake properties suitable for analytical applica- [47] D. Saha, S. Barakat, S. Van Bramer, K.A. Nelson, D.K. Hensley, J. Chen, Non-com-
tions, Anal. Chim. Acta 478 (2003) 43–51. petitive and competitive adsorption of heavy metals in sulfur-functionalized or-
[34] S.M.A. Koochaki-Mohammadpour, M. Torab-Mostaedi, A. Talebizadeh-Rafsanjani, dered mesoporous carbon, Appl. Mater. Interfaces 8 (49) (2016) 34132–34142.
F. Naderi-Behdani, Adsorption isotherm, kinetic, thermodynamic, and desorption [48] S. Pan, H. Shen, Q. Xu, J. Luo, M. Hu, Surface mercapto engineered magnetic Fe3O4
studies of lanthanum and dysprosium on oxidized multiwalled carbon nanotubes, J. nanoadsorbent for the removal of mercury from aqueous solutions, J. Colloid
Dispersion Sci. Technol. 35 (2) (2014) 244–254. Interface Sci. 365 (2012) 204–212.
[35] H. Aghayan, A. Mahjoub, A. Khanchi, Samarium and dysprosium removal using 11- [49] D. Saha, G. Orkoulas, J. Chen, D.K. Hensley, Adsorptive separation of CO2 in sulfur-
molybdo-vanadophosphoric acid supported on Zr modified mesoporous silica SBA- doped nanoporous carbons: selectivity and breakthrough simulation, Microporous
15, Chem. Eng. J. 225 (2013) 509–519. Mesoporous Mat. 241 (2017) 226–237.
[36] I.V. Melnyk, V.P. Goncharyk, N.V. Stolyarchuk, L.I. Kozhara, A.S. Lunochkina, [50] D. Saha, H.A. Grappe, A. Chakraborty, G. Orkoulas, Postextraction separation, on-
B. Alonso, Y.L. Zub, Dy (III) sorption from water solutions by mesoporous silicas board storage, and catalytic conversion of methane in natural gas: a review, Chem.
functionalized with phosphonic acid groups, J. Porous. Mater. 19 (2012) 579–585. Rev. 116 (19) (2016) 11436–11499.
[37] A.A. Galhoum, M.G. Mahfouz, S.T. Abdel-Rehem, N.A. Gomaa, A.A. Atia, [51] A.L. Myers, J.M. Prausnitz, Thermodynamics of mixed gas adsorption, AIChE J. 11
T. Vincent, E. Guibal, Diethylenetriamine-functionalized chitosan magnetic nano- (1965) 121–127.
based particles for the sorption of rare earth metal ions [Nd(III), Dy(III) and Yb

4692

Вам также может понравиться