You are on page 1of 14

Adsorption

https://doi.org/10.1007/s10450-019-00111-8

Adsorption behavior of oxalic acid at water–feldspar interface:


experiments and molecular simulation
Xiaopeng Xue1,2 · Wei Wang1,2 · Hao Fan1,2 · Zhonghao Xu1,2 · Israel Pedruzzi1,2 · Ping Li1,2 · Jianguo Yu1,2

Received: 10 October 2018 / Revised: 7 May 2019 / Accepted: 9 May 2019


© Springer Science+Business Media, LLC, part of Springer Nature 2019

Abstract
Feldspar belongs to aluminosilicate minerals with a huge reserve, accounting for at least 98% in soil minerals. Oxalic acid is
the common organic matter in the natural world, having three configurations (­ H2C2O4, ­HC2O4− and ­C2O42−) dependent on pH
value in water solution. In this work, the adsorption behavior of oxalic acid at water–feldspar interface were investigated at
the molecular level through molecular dynamic simulation and advanced characterization technologies in order to provide the
useful information for bioleaching and flotation industry. Classical molecular dynamic (MD) simulation, density functional
theory (DFT) calculation and frequency calculation were performed to analyze adsorption behavior of oxalic acid at water–
feldspar interface. Adsorption of ­H2C2O4 on feldspar surface belonged to physical outer-sphere adsorption with hydrogen
bond while adsorption of ­HC2O4− and ­C2O42− belonged to inner-sphere adsorption with Al–O bond based on the molecular
dynamic simulation results. Frequency calculation demonstrated both H ­ C2O4− and ­C2O42− complexed with Al active site
rather than Si site on feldspar surface, and the detected ATR-FTIR spectra were also in agreement with the simulated results.
Dissolution experiments of feldspar with oxalic acid and TGA–DSC analysis of feldspar after oxalic acid adsorption were
carried out to evaluate the adsorption behavior. Adsorption mechanism of oxalic acid on feldspar surface was proposed.

Keywords  Adsorption · Oxalic acid · Feldspar · Molecular simulation · Mineral dissolution

1 Introduction Ullman 1996; Strobel 2001; Cama and Ganor 2006; Oelkers
et al. 2011; Ramos et al. 2015; Zhang and Jun 2015). Feld-
Oxalic acid is the common organic matter in the natural spar (K-feldspar, Na-feldspar, Ca-feldspar) reserve is huge,
world, which is generated frequently from biomass degrada- accounting for at least 98% in soil minerals. Feldspar is com-
tion, microbial metabolites and industrial wastes et al. There posed of Al–O and Si–O tetrahedron structures to form the
are three kinds of configurations (­H2C2O4, ­HC2O4− and stable framework, where Al–O tetrahedron connects with
­C2O42−) dependent on pH value in aqueous solution. It is Si–O tetrahedron resulting in extra negative charges and K ­ +
found that oxalic acid, as the simplest carboxylic acid, can locating in the tetrahedron structures space balances nega-
adsorb on many minerals, which would promote minerals tive charges to obtain the stable structure (Yang et al. 2014).
dissolution obviously, such as feldspar, kaolinite, albite and Dissolution of feldspar could increase significantly the nutri-
montmorillonite (Stillings et al. 1996, 1998; Welch and ent element K concentration in soil for plant growth (Ramos
et al. 2017). Moreover, feldspar belongs to aluminosilicate
mineral (Crundwell 2014), and aluminosilicate ore is also
* Ping Li the main impurity in the hydrometallurgy process of cop-
liping_2007@ecust.edu.cn
per, gold, zinc, phosphorus et al. So it’s meaningful to study
* Jianguo Yu interactions between feldspar and organic ligands.
jgyu@ecust.edu.cn
For adsorption behavior of organic ligands on minerals,
1
State Key Laboratory of Chemical Engineering, School two types of adsorption mechanisms were reported, outer-
of Chemical Engineering, East China University of Science sphere adsorption and inner-sphere adsorption (Strawn and
and Technology, Shanghai 200237, China Sparks 1999; Johnson et al. 2004a, b, 2005). By outer-sphere
2
National Engineering Research Center for Integrated adsorption, minerals surface was hydrated by water to pro-
Utilization of Salt Lake, East China University of Science mote the formation of surface hydroxyls. Organic ligands
and Technology, Shanghai 200237, China

13
Vol.:(0123456789)
Adsorption

were adsorbed through physical hydrogen bonds with sur- than 120° (Yeh et al. 2015) while chemical bonds length
face hydroxyls, which was relatively weaker and easier to be was often less than 2.5 Å based on molecular simulation
desorbed. Normally, outer-sphere adsorption couldn’t pro- results. Moreover, orbitals calculation and frequency cal-
mote dissolution rate of minerals as a protective layer (Duck- culation based on DFT method could figure out the inner-
worth and Martin 2001). By inner-sphere adsorption, metal sphere adsorption behavior, such as electrons transfer and
ions on the minerals surface with positive charges could con- complexation types. Here, complexation types included
nect with organic ligands directly through chemical bonds, mainly mononuclear monodentate and mononuclear biden-
which was relatively stronger. Inner-sphere adsorption could tate which could result in different frequency in ATR-FTIR
lower active energy to promote dissolution rate of minerals spectra (Kubicki et al. 1997, 1999; Zhang and Jun 2015).
significantly (Johnson et al. 2004b). However, it should be To the best of our knowledges, there were no reports
noticed that the inner-sphere adsorption could inhibit disso- about molecular simulation for oxalic adsorption on Feld-
lution of minerals sometimes, since inner sphere adsorption spar. So, our objective was to investigate the adsorption
of organic ligands with relatively bigger volume could result behavior of oxalic acid at water–feldspar interface from the
in huge space resistance (Biber et al. 1994). molecular level through advanced characterization technolo-
To probe adsorption behavior of organic acid on min- gies and molecular dynamic simulation. Firstly, molecular
erals surface, some advanced analytical instruments have dynamic simulation was adopted to seek and predict adsorp-
been developed to help figure out adsorption mechanism. tion sites of oxalic acid on feldspar surface. And then ATR-
For example, Atomic Force Microscope (AFM) could detect FTIR and TGA–DSC analysis were used to validate the
organic acids on minerals surface at nano scale (Filby et al. molecular prediction. In the end, the adsorption mechanism
2012; Xian et al. 2017; Xiong et al. 2018). Normally, sur- of oxalic acid on feldspar was proposed. The research infor-
face of minerals was smooth before adsorption, while surface mation would provide theoretical data for the utilization of
became rough after adsorption. AFM could detect adsorption aluminate ores and the purification of valuable minerals with
thickness and coverage fraction on mineral surface. Electrons aluminate impurities.
of organic ligands with negative charges could transfer to
metal ions on minerals surface during chemical inner-sphere
adsorption, X-ray photoelectron spectroscopy (XPS) (War- 2 Experimental and theoretical sections
man et al. 2016; Feng et al. 2017) could detect the electrons
transfer at different electrons orbitals. Normally, outer-sphere 2.1 Experimental section
adsorption wouldn’t change Attenuated Total Reflection
Flourier Transformed Infrared Spectroscopy (ATR-FTIR 2.1.1 Feldspar mineral
spectra), while inner-sphere adsorption could change ATR-
FTIR spectra significantly. Chemical adsorption of organic Feldspar was obtained from Hebei province, China. Minerals
ligands on minerals surface could produce new frequency were crushed and finely milled. The fine ground minerals
peaks in ATR-FTIR spectra (Kubicki et al. 1999; Duckworth were sieved through 200 mesh sieve and washed by dilute
and Martin 2001; Ha et al. 2008). In addition, TGA–DSC HCl (0.01 M) one time, by dilute NaOH (0.01 M) one time,
analysis could distinguish between outer-sphere adsorption by ultra pure water three times. After this procedure, miner-
and inner-sphere adsorption. With temperature increasing, als powder were totally dried at 80 °C. And then, minerals
outer sphere adsorption of organic acid on minerals sur- were characterized by X ray diffraction(XRD) and Scan-
face started to be desorbed firstly at low temperature about ning electron microscope (SEM), as shown in Fig. 1. Main
100–200 °C, while inner-sphere adsorption of organic acids component of minerals was microcline, one kind of feldspar
was desorbed at least 400 °C due to formation of complexa- minerals. The strongest peak intensity corresponded to facet
tions between organic ligands and metal ions. (0 1 0) of feldspar, the biggest surface area of feldspar and
In the other side, with development of computational easier to be cleaved (Xu et al. 2017). As a result, facet (0
chemistry (Han et al. 2012), molecular simulation could 1 0) was mainly considered in following experiments and
investigate theoretically the adsorption behavior of oxalic theoretical calculations (calculated by Material Studio 6.1).
ligands on mineral surface at an atomic level (Rai et al.
2011; Xu et al. 2013; Xiong et al. 2018). Molecular dynamic 2.1.2 Dissolution experiment
simulation could seek the optimum adsorption site on min-
eral surface for organic ligands to distinguish between inner- 100 mL 0.05 mol/L oxalic acid ­(C2H2O4·2H2O, AR, > 99.5%)
sphere adsorption and outer-sphere adsorption. The adsorp- with different pH (pH 0.00, pH 2.76, pH 6.00) were mixed
tion energy of outer-sphere adsorption was obviously lower with 5 g feldspar minerals into flasks. Because both ­Cl− and
than that of inner sphere adsorption, and hydrogen bonds ­Na+ ions have the negligible complexation with Al and
length was between 2.5 and 3.5 Å and angle was greater Si sites on feldspar surface, the pH values in the aqueous

13
Adsorption

Fig. 1  XRD spectra and SEM


image for feldspar. a Compari-
son between experimental and
computational XRD spectra
of feldspar; b SEM image of
feldspar mineral showing main
facet (0 1 0)

solutions with 0.05 mol/L oxalic acid were adjusted through provided more accurate charges assignments than that by
external HCl and NaOH titration in order to keep oxalic acid classical MD calculations. Unit cell of feldspar (volume of
existing in the single form of ­H2C2O4 at pH 0.00, the form of 802.20 Å3) imported from software library were optimized
­HC2O4− at pH 2.76 and ­C2O42− at pH 6.00. And then, flasks in module ­Dmol3 on geometry and three forms of oxalic
were put into shaker at 25 °C and 180 rpm for 2 days. 1 mL ­ 2C2O4, ­HC2O4− and C
acids including H ­ 2O42− based on pro-
samples taken from flasks every 12 h were centrifuged at tonation constant (Fig. 2) were built based on research by
12000 rpm and 4 °C for 10 min to obtain supernatant that Weber group (Weber et al. 2014). In this work, 0.05 mol/L
would be analyzed by Inductively Coupled Plasma Optical oxalic acid was selected for experimental verification, and
Emission Spectrometer (ICP-OES) to obtain the dissolution then based on ­pKa1 and ­pKa2, the oxalic acid forms at dif-
amounts of Al and Si elements from feldspar mineral. ferent pH value were calculated. It was found that 99.98%
of oxalic acid existed the form of ­H 2C 2O 4 at pH 0.00,
2.1.3 ATR‑FTIR detection and TGA–DSC analysis 99.98% of oxalic acid exists in the form of ­HC2O4− at pH
2.76, 99.98% of oxalic acid was C ­ 2O42− at pH 6.00. For
After dissolution experiment, solid fraction from centrifuga- the molecular simulation, the oxalic acid solution with
tion was dried at 25 °C naturally. The dried solids were ana- pH 0.00, 2.76 and 6.00 were selected in order to investi-
lyzed by ATR-FTIR (Nicolet 6700, USA). All spectra were gate the interaction between feldspar and each oxalic acid
obtained using KBr pellets at 25 °C with a resolution 256
scans and 4 cm−1. At the same time, the dried solids were
analyzed by TGA–DSC (SDT/Q600, TA company, USA).
Temperature increased from room temperature to 500 °C
by 20 °C/min rate under ­N2 environment when weight loss
curves were measured.

2.2 Theoretical section

Material Studio 6.0 developed by Accelrys incorporation


was adopted to carry out classical molecular dynamic (MD)
simulation in Forcite module and density functional theory
(DFT) calculations in D­ mol3 module for simulating inter-
actions between oxalic acid and feldspar. Furthermore,
frequency calculations were performed in Gaussian09 W
developed by Gaussian incorporation.

2.2.1 Molecular models of feldspar and oxalic acid

Charges assignments of atoms were extremely important Fig. 2  Three forms of oxalic acid in the aqueous solution with differ-
and critical for classical MD simulation. DFT calculations ent pH value

13
Adsorption

2.2.2 Molecular model and method of classical molecular


dynamic simulation

Classical molecular dynamic (MD) simulation was applied


to investigate oxalic acid adsorption on feldspar surface.
Oxalic acid had three forms at different pH value, as H ­ 2C2O4,
­HC2O4− and C ­ 2O42−, respectively. Crystal facet (0 1 0) of
feldspar was most stable and had the biggest surface area. So
the adsorption of three forms of oxalic acid was considered
mainly on feldspar crystal facet (0 1 0). MD simulation of
interactions models between feldspar and oxalic acid were
built in Forcite module as following procedures. Firstly, facet
Fig. 3  Molecular models of water, feldspar and oxalic acid with three (0 1 0) was cleaved and expand to 4*2*2 super cell. Then two
configurations. Black number: charges; blue numbers: bond length; water layers were added on the super cell of feldspar. Water
green number: angles (Color figure online) layer near surface of feldspar was relaxed, while another water
layer was fixed. At the same time, first layer of feldspar surface
form, respectively. Water molecule was obtained as same was relaxed while the remained parts of feldspar were fixed.
describing method as oxalic acid molecule. Finally, three forms of oxalic acid were laid accidently close
In these geometry optimization calculations, gener- to the relaxed feldspar surface as shown in Fig. 4.
alized gradient approximation (GGA) with the Perdew- Three adsorption models (A, B and C) were optimized
Burke-Ernzerhof (PBE) parametrization of the exchange on geometry to obtain the stable initial adsorption models.
correlation functional and basis set DNP 3.5 (comparable In this step, only oxalic acid molecules, one water layer near
to Gaussian 6–31 g** method) were adopted. Convergence super cell and surface of super cell were unrestrained. At last,
criteria included energy change 1 × 10−5Ha, SCF tolerance molecular dynamics simulation of each adsorption model was
1 × 10 −6, max force 0.002 Ha/Å and max displacement performed using UFF force field. UFF force field was univer-
0.005 Å. In addition, a fermi smearing 0.005 Ha was used sal force field to describe interfacial phenomenon of organic
to improve the convergence speed (Solíscalero et al. 2013). matter-mineral. The potential parameters for UFF were:
Geometry optimization results were shown in Fig. 3 and
E = ER + E𝜃 + E𝜑 + E𝜔 + Evdw + Eel
Table 1.
1 ( )2
ER = Kij r − rij
2

Table 1  Crystal parameters of Lattice length Lattice angle


feldspar
a (Å) b (Å) c (Å) 𝛼(◦ ) 𝛽(◦ ) 𝛾(◦ )

Calculation 8.573 12.962 7.219 90.567 115.917 87.750


Experiment (Ref. 1) 8.560 12.984 7.209 90.28 116.03 89.03

Ref. 1: Negro et al. (1978)

Fig. 4  Designed adsorption
models of oxalic acid with three
forms on feldspar surface. a
­H2C2O4 on feldspar; b ­HC2O4−
on feldspar; c ­C2O42− on
feldspar

13
Adsorption

was total energy of adsorption energy of adsorption models


1
without surface of feldspar, ­E4 was total energy of two water
[ ]
E𝜃 = Kijk 1 + cos (p𝜃 + 𝜑)
2 layers, respectively. N was number of unit cells where oxalic
acids contacted. Here, N was 8.
m

E𝜑 = Kijkl Cn cos n𝜑ijkl
2.2.3 Molecular model and method of DFT calculation
n=0

Based on previous reports, organic ligands were easily


absorbed on mineral surface by chemical adsorption. And
( )
E𝜔 = Kijkl C0 + C1 cos 𝜔ijkl + C2 2 cos 𝜔ijkl
our MD simulation also demonstrated that oxalic ligands
( [ ]6
x [ x ]12 ) would react with Al site on feldspar surface. In order to
Evdw = DLJ −2 LJ + −2 LJ figure out how electrons transfer in chemical adsorption,
x x
adsorption of ligand H­ C2O4− and ­C2O42− on feldspar was
designed by DFT calculations in ­Dmol3 module as shown
Qi Qj in Fig. 5. Generalized gradient approximation (GGA) with
( )
Eel = 332.0637
∈ Rij the Perdew–Burke − Ernzerhof (PBE) parametrization of
the exchange − correlation functional and basis set DNP
where E was total energy, ER was energy of bond stretch, 3.5 was used. Convergence criteria included energy change
E𝜃 was energy of angle bend, E𝜑 was energy of torsion, E𝜔 1 × 10−5Ha, SCF tolerance 1 × 10−6, max force 0.002Ha/Å
was energy of inversion, Eel was energy of van der waals, and max displacement 0.005 Å. In addition, a Fermi smear-
respectively. ing 0.005Ha was used to improve the convergence speed.
Because the energy-minimized structure had been And then, the geometry optimization was carried out for
obtained by geometry optimization and partial of adsorption two models in Fig. 5. Here, water molecules weren’t con-
structures were fixed, the ensemble NPT were not applied. sidered, since the classical MD simulation results showed
Here, NVT was adopted and molecular dynamic simulation that organic ligands had much stronger adsorption ability
time was 1 ns with time step 1 fs to reach equilibrium state. than water.
In addition, the initial velocities were random, and energy
deviation was 50000.0 kcal/mol. Electrostatic was described 2.2.4 Molecular model and method of frequency
by Ewald method while van der waals was described by calculation
atom based method.
Parameters ­Ead was defined to evaluate the adsorption Chemical adsorption of organic ligand on mineral surface
ability of oxalic acids on surface of feldspar in equation (Shi would produce new frequency peaks in ATR-FTIR spec-
et al. 2013): tra. So theoretical calculations of frequency based on DFT
methods were carried out as shown in Fig. 6. Four methods
E1 − E2 − E3 + E4
Ead = including HF/3–21 g*, HF/6–31 g*, B3lyp/3–21 g* and
N B3lyp/6–31 g* were adopted to calculate the adsorption
where E­ ad was adsorption energy of organic acids on sur- frequency of oxalic acid on feldspar surface (Al cluster and
face of feldspar, ­E1 was total energy of adsorption models Si cluster). Every model was optimized on geometry. And
after molecular dynamics simulation, ­E2 was total energy then, the suitable methods were picked up to predict the
of adsorption models without organic acids molecules, ­E3 adsorption frequency of oxalic acid on feldspar. Blue arrow

Fig. 5  Adsorption models of
­HC2O4− (a) and C­ 2O42− (b) on
feldspar surface facet (0 1 0) for
DFT calculations

13
Adsorption

Fig. 6  Designed models for


bond frequency calculation

represented symmetric vibration of bond C–O (marked as on feldspar surface for oxalic acid with three configura-
C–Os) while green arrow represented anti-symmetric vibra- tions are shown in Fig. 7, and the relative adsorption ener-
tion of bond C–O (marked as C–Oas), which were two main gies are calculated through classical molecular dynamics
adsorption frequencies of oxalic ligands, while Al–Os and simulation, listed in Table 2. It is found that the adsorp-
Si–Os were the typical frequencies of feldspar. As shown in tion energy of H ­ 2C2O4 is − 0.10 eV, which is extremely
Fig. 6, four frequencies were mainly considered. lower. Moreover, adsorption equilibrium state of ­H2C2O4
on feldspar surface shows that the bond length between
H atom ­(H2C2O4) and O atom (feldspar) is 2.919 Å with
3 Results and discussions angle 129.4°, which belongs to the hydrogen bond. There-
fore, it’s deduced that ­H2C2O4 is adsorbed physically. The
3.1 Adsorption behavior based on classical adsorption energies of ­C2O42− and ­HC2O4− are − 1.17 eV
molecular dynamics simulation and − 0.68  eV, respectively, which are much higher.
Based on the adsorption equilibrium states in Fig. 7, the
In water solution, oxalic acid has three configurations, bond lengths between Al atom (feldspar) and O atom
­ 2C 2O 4 at pH 0.00, ­H C 2O 4− at pH 2.76 and ­C 2O 42− at
H ­( C 2O 42− and H
­ C 2O 4−) are 3.206 Å and 3.392 Å, which
pH 6.0, respectively. The adsorption equilibrium states belong to the strong electrostatic force attraction. During

Fig. 7  Adsorption equilibrium
states of oxalic acid with three
configurations on feldspar
surface

13
Adsorption

Table 2  Adsorption energies System Energy


of oxalic acid with three
configurations on feldspar E1 eV E2 eV E3 eV E4 eV Ea eV
surface
H2C2O4-feldspar A − 4128.60 − 4129.06 − 23.98 − 25.23 − 0.10 Out-sphere adsorption
HC2O4−-feldspar B − 4117.07 − 4116.79 − 18.54 − 23.69 − 0.68 Inner-sphere adsorption
C2O42−-feldspar C − 4114.09 − 4111.56 − 12.20 − 19.03 − 1.17 Inner-sphere adsorption

Table 3  Adsorption energies of organic acids on minerals in litera-


tures

Mineral Adsorption energy (eV)


Formic acid Acetic acid

Calcite − 1.06 (Ref. 2) − 0.98 (Ref. 4)


− 0.97 (Ref. 4) − 1.23 (Ref. 2)
quartz − 0.63 (Ref. 2) − 0.67 (Ref. 2)
kaolinite − 0.76 (Ref. 2) − 0.84 (Ref. 2)
TiO2 − 1.13 (Ref. 3) − 1.18 (Ref. 5)

Ref. 2: Budi et al. (2018); Ref. 3: Nunzi et al. (2010); Ref. 4: Ataman
et al. (2016); Ref. 5: Spreafico et al. (2014)

deprotonation of H ­ 2C 2O 4, charges of O atom increase


obviously (Fig. 2), as a result, adsorption transforms from
weak hydrogen bond adsorption (system A) to strong elec- Fig. 8  RMSD curves for oxalic acid with three configurations
trostatic force adsorption (System B and C).
Table 3 lists some reported data for the adsorption ener-
gies of organic acids on calcite, quartz, kaolinite and ­TiO2, Adsorption of ­C2O42− and H ­ C2O4− with much lower RMSD
respectively. These adsorption energies from literatures values on feldspar are stronger than that of ­H2C2O4 with
are at the ranges from − 0.63 eV to − 1.23 eV, where it much higher RMSD value.
is found that formic and acetic acid are adsorbed through There are two kinds of active sites on feldspar surface,
the coordination bond, such as O–Ca, O–Si, O–Al and Al site and Si site. Which site would be coordinated with
O–Ti. In this work, the calculated adsorption energies O atom (carrying most negative charges) of oxalic acid?
of ­C2O42− and H ­ C2O4− (adsorbed through Al–O coordi- Figure 9 shows the radial density functional spectra between
nation) are at the ranges from − 0.68 eV and − 1.17 eV, O and Al(Si) calculated by classical molecular dynamics
which is in agreement with the reference values. simulation method. It can be seen clearly that there are
The characterization parameter, root mean square dis- no obvious peaks between O and Si, and there are obvi-
placement (RMSD) can be used to determine the adsorp- ous peaks between Al and O in three adsorption systems
tion dynamics of oxalic acid on feldspar. Higher RMSD at 3.07 Å, 2.97 Å and 3.21 Å. RDF spectra indicate that O
value means that oxalic acid is in water phase rather than atom with most negative charges prefer to coordinate with
adsorbed on feldspar surface. While lower RMSD value Al site rather than Si site on feldspar surface. This result cor-
means that oxalic acid prefers to be adsorbed on feldspar responds highly with the previous experimental and theoreti-
surface. Here, RMSD is defined as the following equation cal analysis (Welch and Ullman 1996; Criscenti et al. 2005;
(Teklebrhan et al. 2014): Alstadt et al. 2016).
Competitive adsorption behavior caused by water mole-
N
cule should be considered when adsorption of organic matter

1∑
RMSD = [ri(t) − ri(0)]2
N i happens at mineral–water interface. Which molecule would
prefer to coordinate with Al active site, H ­ 2O or organic
where N is total number of oxalic acid molecules, ri(t) and ligand? RDF spectra between O atom (water) and Al atom
ri(0) represent positions of carboxylic acid at initial and t (feldspar) is calculated as shown in Fig. 10. Shortest distance
time. Figure 8 shows RSMD calculation results, which are in between O and Al is 2.93 Å in all three systems. Coordina-
agreement with the calculated results of adsorption energy. tion complexation between Al and O mainly relies on elec-
trostatic force, which could be used to evaluate coordination

13
Adsorption

Fig. 9  RDF spectra between O atom of oxalic acid and Al, Si atoms


of feldspar Fig. 11  Electrostatic force between O atom from water, oxalic acid
and Al atom of feldspar

ability. Electrostatic force is defined as F = KQ1Q2/R2, 3.2 Adsorption behavior based on DFT calculation
where K is electrostatic constant, as 9*109N·m2/C2, ­Q1
and ­Q2 are charges of two atoms, R is distance between Based on the classical molecular dynamics results,
two atoms. Here, only O atom and Al atom are consid- ­HC2O4− and ­C2O42− would incline to go on the mononuclear
ered. It can be seen that electrostatic force between H­ 2O is monodentate chemical adsorption on feldspar surface. Here,
lower than ­C2O42− and H­ C2O4− but higher than H ­ 2C2O4 as DFT calculations in ­Dmol3 module are carried out to inves-
shown in Fig. 11. It can be deduced that water would block tigate how ­HC2O4− and ­C2O42− react with feldspar. After
­H2C2O4 to coordinate with Al active site, while C ­ 2O42− and

­HC2O4 would be prior to coordinate with Al active site
than water. Previous study (Alstadt et al. 2016) found water
would displace acetic molecule and prior to be absorbed on
kaolinite surface. Because mineral surface properties would
change obviously in the presence and absence of water.

Fig. 10  RDF spectra between O atom of water and Al atom of feld- HC2O4−(a) and
Fig. 12  Optimized structures for adsorption of ­
spar ­C2O42−(b) on feldspar surface

13
Adsorption

Table 4  Al–O bond length between carboxylic ligand and aluminum In order to investigate how electrons change during
This paper Ref. 6 Ref. 7 Ref. 8
reaction, partial density of state for Al, Si (feldspar) and O
(oxalic ligand) before and after reactions is studied, as shown
Al–O Bond length (Å) 1.840 1.851 1.880 1.800 in Fig. 13. It can be seen clearly that energy partial electrons
1.842 1.857 1.819 1.807 for Al, Si and O are very close to Fermi level, which indi-
1.835 1.903 1.874 cate these electrons are very active and easier to participate
Ref. 6: Jin et al. (2011); Ref. 7: Ramos et al. (2015); Ref. 8: Kubicki
in chemical reactions. Partial density of state (Pdos) for Si
et al. (1996) doesn’t change obviously after reaction, which explain that
electrons of Si doesn’t participate in reactions. While Pdos
shape for Al and O change obviously after reactions, not
the geometry optimization for the adsorption of C ­ 2O42− and only many peaks move to lower energy, but also intensities

­HC2O4 on feldspar surface, the reasonable adsorption con- of peaks decrease or increase obviously. Especially, s orbit-
figuration is obtained as shown in Fig. 12, where Al–O bond als of O from organic ligands and p, d orbitals of Al atom
length is 1.840 Å and 1.842 Å for system A, 1.835 Å for sys- overlap significantly, which demonstrate that O atom reacts
tem B. Compared to 1.800 Å before geometry optimization with Al atom strongly. This is another proof that organic
in Fig. 5, this result is very similar to previous research as ligands prefer to coordinate with Al atom rather than Si
listed in Table 4. As described on molecular dynamic simu- atom.
lation, adsorption ability of ­HC2O4− and ­C2O42− are stronger
than that of water. So, water is not considered in these DFT 3.3 Comparison of bond frequencies
calculations. According to the literatures, Al–O bond lengths between ATR‑FTIR detection values
between organic ligands and Al atom distribute from 1.800 and calculated values
to 1.903 Å. Calculation bond length in this work are very
close to reference values listed in Table 4, which indicate ATR-FTIR spectra is used to investigate interaction between
oxalic ligands complex with Al site on feldspar surface. oxalic acid and feldspar as shown in Fig. 14. It can be found

Fig. 13  Partial density of state for Al, Si (feldspar) and O (oxalic ligand) before and after reactions

13
Adsorption

precisely with good correlation coefficient and error, which


will be applied to predict frequency Al–oxa model and
Si-oxamodel.
As shown in Fig. 16, slope of fit line of Si–Oxa model
is 1.51 which indicate that Si–Oxa model can’t explain
experimental frequency value though correlation coefficient
is 0.99. The frequency of Al–Oxa model is very close to
experimental frequency value with slope 1.12 and accept-
able error. This calculation results prove again that oxalic
acid would prefer to complex with Al rather than Si on the
surface of feldspar. Frequency calculation results highly cor-
respond with classical molecular dynamic simulation.

3.4 Evaluation of adsorption behavior


through dissolution experiment and TGA–DSC
analysis
Fig. 14  ATR-FTIR spectra for original feldspar and feldspar after
reaction with three forms of oxalic acid The dissolution experiment of feldspar mineral with oxalic
acid is carried out to evaluate the adsorption behavior. Feld-
spar minerals are mixed with 0.05 mol/L oxalic acid with
there are two new peaks (1317 cm−1 and 1620 cm−1) appear- different pH (pH 0.00, pH 2.76, pH 6.00) in plastic bottles
ing at higher pH 2.76 and 6.00 values. However, there is for 2 days reactions. Al and Si leaching results are shown in
not any new peak appearing at pH 0.00 value, indicating Fig. 17. It is found that Al and Si are preferentially released
no chemical interaction between oxalic acid and feldspar. from feldspar at higher and lower pH, respectively. Usually,
As listed in Table 5, based on previous reports, 1317 cm−1 the dissolution rate of mineral by organic acid is defined
and 1620 cm−1 could be ascribed as vibrations of C–Os and as R = K1*aH+ n
 + K2*anligand (Furrer and Stumm 1986), where
C–Oas bonds, respectively. the first term represents the dissolution by ­H+ attack and
In order to explain why there are new peaks appear- the second term represents the dissolution by ligand compl-
ing, four methods including HF/3–21  g*, HF/6–31  g*, exation. For feldspar dissolution by oxalic acid, at pH 0.00,
B3lyp/3–21 g* and B3lyp/6–31 g* are adopted to calcu- ­H+ with extremely high reactivity attack on Si site causes
late the adsorption frequency, and the predicted results are faster dissolution, so both Al and Si release amounts are
compared with ATR-FTIR spectra. Firstly, the four methods much higher. With H ­ + concentration decreasing, feldspar
are used to calculate models as follows to determine which dissolution is mainly controlled by H ­ C2O4− complexing
method is suitable for this system. C–Os and C–Oas are typi- with Al site at pH 2.76, so the Al and Si release amounts
cal vibration frequencies of ­H2C2O4, ­HC2O4− and ­C2O42−. become slower than that at pH 0.00. When pH 6.00, the
While Al–O and Si–O are typical vibration frequencies of Al and Si release amounts are more than that at pH 2.76
feldspar. The four typical vibrations are used to evaluate due to stronger complexation ability of ­C2O42− than that of
accuracy of four calculation methods as shown in Fig. 6. ­HC2O4−, but Al existing in the form of Al(OH)3 results in a
As shown in Fig. 15, correlation coefficient of HF/6–31 g* lower Al concentration detection in the solution.
method is highest, but the calculated frequency value is And then, TGA–DSC characterization is used to analyze
far higher than the observed value. Frequency calculated different feldspar samples after interaction with oxalic acid
by B3lyp/3–21 g* method is very close to the observed as shown in Fig. 18. According to MD simulation and DFT
value, but correlation coefficient is lower. Both HF/3–21 g* calculation results, E ­ ad value equals − 0.10 eV at pH 0.00,
and B3lyp/6–31 g* methods can predict frequency more which means outer-sphere adsorption (physical adsorption),

Table 5  ATR-FTIR frequencies of carboxylic ligand adsorption on Al-bearing minerals


Bond frequency This work exp Cal (Ref. 9) Exp (Ref. 10) Exp (Ref. 11) Exp (Ref. 12) Exp (Ref. 13)

V(C–Os)/cm−1 1317 1360 1318 1381 1429 1300


V(C–Oas)/cm−1 1619 1671 1619 1534 1617 1695

Ref. 9: Kubicki et al. (1999); Ref. 10: Alstadt et al. (2016); Ref. 11: Belber and Stumm (1994); Ref. 12: Ramos et al. (2014); Ref. 13: Pettibone
et al. (2008)

13
Adsorption

Fig. 15  Comparisons between the calculated frequencies by different methods and experimental frequencies

and − 1.17 eV at pH 6.00, there are obvious decomposi-


tion peaks of chemical adsorption of oxalic acid at 450 °C,
which means inner-sphere adsorption (chemical adsorp-
tion) due to ­H C 2O 4− complexing with Al site to form
­[HC2O4Al(OH)3]− cluster at pH 2.76 and ­C2O42− complex-
ing with Al site to form [­ C2O4Al(OH)3]2− cluster at pH 6.00.

3.5 Adsorption mechanism of oxalic acid at water–


feldspar interface

According to molecular simulation, ATR-FTIR, TGA–DSC


and dissolution experiments, the adsorption mechanism of
oxalic acid at water–feldspar interface can be predicted as
shown in Fig. 19, where the dissolution of feldspar by oxalic
acid includes two parts, one is from the contribution of H ­ +
in aqueous solution attacking Si site on the feldspar surface,
Fig. 16  Comparison of calculated frequency between Al–oxa and Si– and the other part is from the contribution of the complex
oxa models
reaction between Al site on the feldspar surface and ­H2O,
­HC2O4− or ­C2O42− in the aqueous solution. The dissolution
so TG-DSC characterization result does not show obvi- rate of feldspar is represented as R = K1*anH+ + K2*anligand,
ous decomposition peak of chemical adsorption of oxalic where the first term is the controlling step if the oxalic acid
acid at 450 °C. E
­ ad values equaling − 0.68 eV at pH 2.76 solution with a lower pH value, such as pH 0.00, and the

13
Adsorption

Fig. 17  Al and Si dissolution amount from feldspar at different pH by oxalic acid

Fig. 18  TGA-DSC analysis of original feldspar and feldspar after oxalic acid adsorption at different pH value

second term is the controlling step for the oxalic acid solu- 4 Conclusion
tion with a higher pH value, such as pH 6.00. There are the
co-existing ­H+, ­H2O, ­HC2O4− ­C2O42− and ­H2C2O4 in some The adsorption behavior of oxalic acid at water–feld-
scenarios, it should be mentioned that the rank of the com- spar interface can be explored from the molecular level
plex reaction is ­H2O–Al site < HC2O4−–Al site < C2O42−–Al through molecular dynamic simulation. Although feldspar
site and without complex reaction between H ­ 2C2O4–Al site. is composed of Al–O and Si–O tetrahedron structures to

13
Adsorption

Budi, A., Stipp, S.L.S., Andersson, M.P.: Calculation of entropy of


adsorption for small molecules on mineral surfaces. J. Phys.
Chem. C 122, 8236–8243 (2018)
Cama, J., Ganor, J.: The effects of organic acids on the dissolution of
silicate minerals: a case study of oxalate catalysis of kaolinite
dissolution. Geochim. Cosmochim. Acta 70, 2191–2209 (2006)
Criscenti, L.J., Brantley, S.L., Mueller, K.T., Tsomaia, N., Kubicki,
J.D.: Theoretical and 27 Al CPMASS NMR investigation of alu-
minum coordination changes during aluminosilicate dissolution.
Geochim. Cosmochim. Acta 69, 2205–2220 (2005)
Crundwell, F.K.: The mechanism of dissolution of minerals in acidic
and alkaline solutions: partII application of a new theory to sili-
cates, aluminosilicates and quartz. Hydrometallurgy 149, 265–275
(2014)
Duckworth, O.W., Martin, S.T.: Surface complexation and dissolution
of hematite by C ­ 1-C6, dicarboxylic acids at pH 5.0. Geochim.
Cosmochim. Acta 65, 4289–4301 (2001)
Feng, Q., Wen, S., Deng, J., Zhao, W.: Combined DFT and XPS inves-
tigation of enhanced adsorption of sulfide species onto cerussite
by surface modification with chloride. Appl. Surf. Sci. 425, 8–15
Fig. 19  Interaction mechanism between oxalic acid and feldspar at (2017)
different pH Filby, A., Plaschke, M., Geckeis, H.: AFM force spectroscopy study
of carboxylated latex colloids interacting with mineral surfaces.
Colloid. Surf A 414, 400–414 (2012)
form the stable framework, oxalic acid with three con- Furrer, G., Stumm, W.: The coordination chemistry of weathering: I.
figurations, ­H 2 C 2 O 4 , ­H C 2 O 4 − and C
­ 2 O 42 , can adsorb Dissolution kinetics of δ-Al2O3 and BeO. Geochim. Cosmochim.
on feldspar surface to promote the mineral dissolution. Acta. 50, 1847–1860 (1986)
Ha, J., Yoon, T.H., Wang, Y., Musgrave Jr., C.B., Brown, G.E.: Adsorp-
Adsorption of H ­ 2C 2O 4 on feldspar surface belongs to tion of organic matter at mineral/water interfaces: 7. ATR-FTIR
physical outer-sphere adsorption with hydrogen bond for- and quantum chemical study of lactate interactions with hematite
mation while adsorption of ­HC2O4− and ­C2O42− belong nanoparticles. Langmuir 24, 6683–6692 (2008)
to inner-sphere adsorption and prefer to coordinate with Han, B., Viswanathan, V., Pitsch, H.: First-principles based analysis of
the electrocatalytic activity of the unreconstructed Pt(100) surface
Al active site rather than Si active site on feldspar sur- for oxygen reduction reaction. J. Phys. Chem. C 116, 6174–6183
face. The rank of adsorption matter on feldspar surface is (2012)
­C2O42−>HC2O4−>H2O > H2C2O4. At lower pH 0.00, water Jin, X., Yan, Y., Shi, W., Bi, S.: Density functional theory studies on the
instead of H
­ 2C2O4 will attack Al active site to promote structures and water-exchange reactions of aqueous Al(III)–oxa-
late complexes. Environ. Sci. Technol. 45, 10082–10090 (2011)
Al dissolution. At higher pH 2.76 and 6.00, ­HC2O4− and Johnson, S.B., Yoon, T.H., Kocar Jr., B.D., Brown, G.: Adsorption
­C2O42− instead of water will attack Al active site to pro- of Organic matter at mineral/water interfaces: 2. Outer-sphere
mote Al dissolution. At acidic conditions, ­H+ would attack adsorption of maleate and implications for dissolution processes.
Si active site to promote Si dissolution. Langmuir 20, 4996–5006 (2004a)
Johnson, S.B., Yoon, T.H., Slowey Jr., A.J., Brown, G.: Adsorption
of organic matter at mineral/water interfaces: 3. Implications
Acknowledgements  The authors wish to acknowledge National Natu- of surface dissolution for adsorption of oxalate. Langmuir 20,
ral Science Foundation of China (No. 21776089, No. U1610102, No. 11480–11492 (2004b)
21506063) and the International S&T Cooperation Program of China Johnson, S.B., Brown, G., Healy, T.W., Scales, P.J.: Adsorption of
(No. 2016YFE0132500). organic matter at mineral/water interfaces: 6. effect of inner-
sphere versus outer-sphere adsorption on colloidal stability. Lang-
muir 21, 6356–6365 (2005)
References Kubicki, J.D., Blake, G.A., Apitz, S.E.: Molecular orbital models of
aqueous aluminum-acetate complexes. Geochim. Cosmochim.
Acta 60, 4897–4911 (1996)
Alstadt, V.J., Kubicki, J.D., Freedman, M.A.: Competitive adsorp- Kubicki, J.D., Itoh, M.J., And, L.M.S., Apitz, S.E.: Bonding mecha-
tion of acetic acid and water on kaolinite. J. Phys. Chem. A 120, nisms of salicylic acid adsorbed onto illite clay: an ATR-FTIR and
8339–8346 (2016) molecular orbital study. Environ. Sci. Technol. 31, 1151–1156
Ataman, E., Andersson, M.P., Ceccato, M., Bovet, N., Stipp, S.L.S.: (1997)
Functional group adsorption on calcite: I. Oxygen containing and Kubicki, J.D., Schroeter, L.M., Itoh, M.J., Nguyen, B.N., Apitz, S.E.:
nonpolar organic molecules. J. Phys. Chem. C 120, 16586–16596 Attenuated total reflectance fourier-transform infrared spectros-
(2016) copy of carboxylic acids adsorbed onto mineral surfaces. Geo-
Biber, M.V., Afonso, M.D.S., Stumm, W.: The coordination chemistry chim. Cosmochim. Acta 63, 2709–2725 (1999)
of weathering: IV. Inhibition of the dissolution of oxide minerals. Negro, A.D., Pieri, R.D., Quareni, S., Taylor, W.H.: The crystal struc-
Geochim. Cosmochim. Acta 58, 1999–2010 (1994) tures of nine K feldspars from the Adamello Massif (northern
Biber, M.V., Stumm, W.: An in situ ATR-FTIR study: the surface Italy) addendum. Acta Cryst. 34, 3843 (1978)
coordination of salicylic acid on aluminum and iron(III) oxides.
Environ. Sci. Technol. 28, 763–768 (1994)

13
Adsorption

Nunzi, F., Angelis, F.D.: DFT investigations of formic acid adsorption Teklebrhan, R.B., Ge, L., Bhattacharjee, S., Xu, Z., Sjöblom, J.: Initial
on single-wall T ­ iO2 nanotubes: effect of the surface curvature. J. partition and aggregation of uncharged polyaromatic molecules at
Phys. Chem. C 115, 2179–2186 (2010) the oil–water interface: a molecular dynamics simulation study. J.
Oelkers, E.H., Golubev, S.V., Pokrovsky, O.S., Bénézeth, P.: Do Phys. Chem. B 118, 1040–1051 (2014)
organic ligands affect calcite dissolution rates? Geochim. Cos- Waiman, C.V., Arroyave, J.M., Chen, H., Tan, W., Avena, M.J., Zanini,
mochim. Acta 75, 1799–1813 (2011) G.P.: The simultaneous presence of glyphosate and phosphate at
Pettibone, J.M., Cwiertny, D.M., Scherer, M., Grassian, V.H.: Adsorp- the goethite surface as seen by XPS, ATR-FTIR and competitive
tion of organic acids on tio2 nanoparticles: effects of pH, nanopar- adsorption isotherms. Colloid. Surf. A 498, 121–127 (2016)
ticle size, and nanoparticle aggregation. Langmuir 24, 6659–6667 Weber, K.H., Liu, Q., Tao, F.M.: Theoretical study on stable small
(2008) clusters of oxalic acid with ammonia and water. J. Phys. Chem. A
Rai, B., Sathish, P., Tanwar, J., Moon, K.S., Fuerstenau, D.W.: A 118, 1451–1468 (2014)
molecular dynamics study of the interaction of oleate and Welch, S.A., Ullman, W.J.: Feldspar dissolution in acidic and organic
dodecylammonium chloride surfactants with complex alumino- solutions: compositional and pH dependence of dissolution rate.
silicate minerals. J. Colloid. Interf. Sci. 362, 510–516 (2011) Geochim. Cosmochim. Acta 96, 2939–2948 (1996)
Ramos, M.E., Emiroglu, C., García, D., Sainzdiaz, C.I., Huertas, F.J.: Xian, Z., Hao, Y., Zhao, Y., Song, S.: Quantitative determination of
Modeling the adsorption of oxalate onto montmorillonite. Lang- isomorphous substitutions on clay mineral surfaces through AFM
muir 31, 11825–11834 (2015) imaging: a case of mica. Colloid. Surf. A 533, 55–60 (2017)
Ramos, M.E., Garcia-Palma, S., Rozalen, M., Johnston, C.T., Huer- Xiong, Y., Li, Z., Cao, T.T., Xu, S.M., Yuan, S.L., Sjoblom, J., Xu,
tas, F.J.: Kinetics of montmorillonite dissolution: an experimental Z.H.: Synergistic adsorption of polyaromatic compounds on sil-
study of the effect of oxalate. Chem. Geol. 363, 283–292 (2014) ica surfaces studied by molecular dynamics simulation. J. Phys.
Ramos, C.G., Querol, X., Dalmora, A.C., Pires, K.C.D.J., Schneider, Chem. C 122, 4290–4299 (2018)
I.A.H., Oliveira, L.F.S.: Evaluation of the potential of volcanic Xu, Y., Liu, Y.L., He, D.D., Liu, G.S.: Adsorption of cationic collec-
rock waste from southern brazil as a natural soil fertilizer. J. tors and water on muscovite (001) surface: a molecular dynamics
Clean. Prod. 142, 2700–2706 (2017) simulation study. Miner. Eng. 53, 101–107 (2013)
Shi, W., Xia, M., Wu, L., Wang, F.: Molecular dynamics study of Xu, L., Tian, J., Wu, H., Deng, W., Yang, Y., Sun, W.: New insights
polyether polyamino methylene phosphonates as an inhibitor of into the oleate flotation response of feldspar particles of different
anhydrite crystal. Desalination 322, 137–143 (2013) sizes: anisotropic adsorption model. J. Colloid. Interf. Sci. 505,
Solíscalero, C., Ortegacastro, J., Hernándezlaguna, A., Muñoz, F.: A 500–508 (2017)
DFT study of the amadori rearrangement above a phosphatidyle- Yang, Y., Min, Y., Lococo, J., Jun, Y.S.: Effects of Al/Si ordering on
thanolamine surface: comparison to reactions in aqueous environ- feldspar dissolution: part I. Crystallographic control on the stoi-
ment. J. Phys. Chem. C 117, 8299–8309 (2013) chiometry of dissolution reaction. Geochim. Cosmochim. Acta
Spreafico, C., Schiffmann, F., Vandevondele, J.: Structure and mobility 126, 574–594 (2014)
of acetic acid at the anatase (101)/acetonitrile interface. J. Phys. Yeh, I.C., Lenhart, J.L., Rinderspacher, B.C.: Molecular dynamics sim-
Chem. C 118, 6251–6260 (2014) ulations of adsorption of catechol and related phenolic compounds
Stillings, L.L., Drever, J.I., Brantley, S.L., Sun, Y., Oxburgh, R.: Rates to alumina surfaces. J. Phys. Chem. C 119, 7721–7731 (2015)
of feldspar dissolution at pH 3–7 with 0–8 Mm oxalic acid. Chem. Zhang, L., Jun, Y.S.: Distinctive reactivities at biotite edge and
Geol. 132, 79–89 (1996) basal planes in the presence of organic ligands: implications for
Stillings, L.L., And, J.I.D., Poulson, S.R.: Oxalate adsorption at a pla- organic-rich geologic ­Co2 sequestration. Environ. Sci. Technol.
gioclase (An47) surface and models for ligand-promoted dissolu- 49, 10217–10225 (2015)
tion. Environ. Sci. Technol. 32, 2856–2864 (1998)
Strawn, D.G., Sparks, D.L.: The use of XAFS to distinguish between Publisher’s Note Springer Nature remains neutral with regard to
inner- and outer-sphere lead adsorption complexes on montmoril- jurisdictional claims in published maps and institutional affiliations.
lonite. J. Colloid. Interf. Sci. 216, 257–269 (1999)
Strobel, B.W.: Influence of vegetation on low-molecular-weight car-
boxylic acids in soil solution—a review. Geoderma 99, 169–198
(2001)

13

You might also like