ARTICLE pubs.acs.org/JPCA
Adsorption of Trimethyl Phosphate on Maghemite, Hematite, and Goethite Nanoparticles ,†,|| € Peter M€akie,†,‡ Gunnar Westin,§ Per Persson,‡ and Lars Osterlund* †
)
FOI, CBRN Defence and Security, SE-901 82, Umea, Sweden ‡ Department of Chemistry, Umea University, SE-901 87, Umea, Sweden § Department of Materials Chemistry, The Ångstrom Laboratory, Uppsala University, P.O. Box 538, SE-751 21, Uppsala, Sweden Department of Engineering Sciences, The Ångstrom Laboratory, Uppsala University, P.O. Box 534, SE-751 21, Uppsala, Sweden
ABSTRACT: Adsorption of trimethyl phosphate (TMP) on well-characterized hematite, maghemite and goethite nanoparticles was studied by in situ DRIFT spectroscopy as a model system for adsorption of organophosphorous (OP) compounds on iron minerals. The iron minerals were characterized by X-ray diffraction (XRD), Raman spectroscopy, transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), specific surface area, and pore size distribution. The minerals were found to consist of stoichimetrically and morphologically well-defined maghemite, hematite, and goethite nanoparticles. Analysis of in situ diffuse reflectance Fourier transform (DRIFT) spectroscopy shows that TMP bonds mainly to Lewis acid Fe sites through the O phosphoryl atom (PdO—Fe) on hematite and maghemite. On goethite most TMP molecules bond to Brønstedt acid surface OH groups and form hydrogen bonded surface complexes. The vibrational mode analysis and uptake kinetics suggest two main reasons for the observed trend of reactivity toward TMP (hematite > maghemite > goethite): (i) larger number of accessible Lewis acid adsorption sites on hematite; (ii) stronger interaction between the Lewis acid Fe sites and the phosphoryl O atom on TMP for hematite and maghemite compared to goethite with concomitant formation of surface coordinated TMP and dimethyl phosphate intermediates. As a result, on the oxides a surface oxidation pathway dominates during the initial adsorption, which results in the formation of surface methoxy and formate. In contrast, on goethite a slower hydrolysis pathway is identified, which eventually yields phosphoric acid. The observed trends of the reactivity and analysis of the corresponding surface structure and particle morphology suggest an intimate relation between the surface chemistry of exposed crystal facets on the iron minerals. These results are important to understand OP surface chemistry on iron minerals.
1. INTRODUCTION Iron oxides are known to be chemically reactive and are extensively used in many technical applications. Iron oxides are used as oxidationreduction or acidbase catalysts due to their amphoteric character. Examples of hematite (R-Fe2O3) catalyzed reactions include production of hydrocarbons (FischerTropsch synthesis), oxidation of alcohols to aldehydes and ketones, dehydrogenation of ethylbenzene to styrene,1 and decomposition of pesticides and chemical warfare agents.2,3 Goethite (R-FeOOH) catalyzes the hydrolysis of carboxylate and phosphorothioate esters,1 and maghemite (γ-Fe2O3) is efficient in catalyzing air oxidation of alcohols at atmospheric pressures.4 Hematite nanorods have recently attracted large interest as photocatalysts.5 Despite this, very few studies have been reported on the adsorption and surface chemistry of organophosphorous (OP) on welldefined iron oxide particles.3,68 Such studies are important to unravel elementary reactions steps on mineral surfaces9,10 and are of fundamental importance to understand and quantify processes that control the transformation of anthropogenic OP compounds in soils to asses their bioavailability and biodegradation. These r 2011 American Chemical Society
reactions are, however, complex and have hitherto received little attention.11 OP compounds are in some cases the dominant phosphate source in soils (up to 80% in pasture soils).12 Of the anthropogenic sources it is of utmost importance to determine the bioavailability and biodegradation originating from pesticides, industrial wastes, military deposits, etc., due to their environmental and health risks. A unique property of phosphates is the unusually large reactivity toward solid particles in the natural aquatic environment.13 Thus reactions at environmental interfaces largely govern the fate of phosphates in ecosystems. Due to the large abundance of iron oxides in the earth’s crust, the uptake and interaction of OP with iron oxides constitute an important part of the anthropogenic phosphor cycle. Of the possible transformations of OP sorption is thought to be of particular importance for the bioavailability and biodegradation.810 Iron (hydr)oxides may promote decomposition of OP, yielding a Received: February 1, 2011 Revised: June 27, 2011 Published: June 28, 2011 8948
dx.doi.org/10.1021/jp201065w | J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A variety of reaction products (some of which may be hazardous and persistent) and can also act as a sink for OP.6 OP compounds may affect iron (hydr)oxides particles by promoting mineral dissolution (OFe bond dissociation) and deactivate sorption sites by reducing the number of available active surface sites (site inhibition), impinging sorption of other compounds.14 In the present study we employ different polymorphs of wellcharacterized iron oxides (hematite, maghemite) and hydroxide (goethite) nanoparticles as model systems of iron minerals to investigate fundamental sorption processes.1 With this approach we are able to make comparative studies on structure-specific adsorption processes using trimethyl phosphate (TMP) as an OP model compound. Hematite (R-Fe2O3) is a red, ferromagnetic material that is widespread in soils and rocks. R-Fe2O3 is often the end product in the transformation of other iron oxides due to its thermodynamic stability.1 Maghemite (γ-Fe2O3) is a redbrownish ferrimagnetic material. It is a weathering product of magnetite and a product from heating of other iron oxides with organic matter.1 It is formed by aerial oxidation of magnetite (weathering) and thermal dehydroxylation of goethite in the presence of organic matter in anaerobic environments.1 Goethite (R-FeOOH) is yellow-brownish antiferromagnetic material and occurs throughout the entire global ecosystem and is (like RFe2O3) one of the major products in many mineral transformations due to its thermodynamic stability.1 Nanoparticles have different surface structure and electronic properties compared to their bulk counterpart and exhibit a larger fraction of lowcoordinated surface atoms, which are expected to influence OP sorption processes.15 They have different morphologies exposing crystal facets with varying nearest neighbor cation distances, FeOFe bond angles, basicity of surface O atoms, and hydroxylation state depending on reaction conditions.1619 Here we present results of well-characterized iron(III) oxide nanoparticles exhibiting different physical properties and show that these variations influence their different reactivities toward TMP. We employ in situ diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy to follow TMP adsorption and reaction to show the adsorption structure and reactivity. A mechanistic explanation is presented that accounts for the experimental observations.
2. EXPERIMENTAL SECTION 2.1. Materials. Maghemite nanopowder was obtained from a commercial source consisting of 99.5% purity iron oxide with an average primary particles size of d = 28 nm (Pi-Kem Ltd.). The powder consisted entirely of the γ-Fe2O3 modification as determined by powder X-ray diffraction (XRD). The commercial powder exhibited a rather heterogeneous particle size distribution, as shown by transmission electron microscopy (TEM). Therefore the maghemite powder was further refined to produce smaller nanoparticles with a narrower size-distribution. The agglomerated maghemite powder was mixed with deionized water, treated in an ultrasonic bath for 30 min (to facilitate dispersion and deagglomeration of the powder), and left for sedimentation for 48 h. The brown-reddish solution part was transferred to a vial and used for nanophase powder preparation by evaporating the water in air. The sedimented light-brown solid containing coarse particles with larger size distribution than those in solution were removed from solution after sedimentation following addition of water and vigorous shaking. The light brown powder obtained in this manner was shaken with water to
ARTICLE
form opaque slurries that were used to prepare powders and films. Maghemite films for absolute concentration calibration measurements were produced by depositing the nanoparticle solution on Si wafers with a syringe. To produce smooth films, the liquid had to be spread out over the surface using a steel spatula during the final part of the drying. The films were subsequently dried at 423 K. The procedure was repeated several times, typically 35 times, to build the desired thickness, judged by the strength of the color. The film thicknesses were typically ∼13 μm, which was estimated from white beam interferometry (Veeco Wkyo NT2000). Hematite was obtained by heating the refined small sized maghemite nanopowders at 893 K for 2 min, which is enough for a complete phase change from the starting γ phase. In spite of the short annealing time employed the particles grew and sintered to some degree during the phase change, as evident by the XRD and TEM analysis (see below). Goethite nanoparticles were prepared and characterized in a manner similar to that previously described by Boily et al.20 Briefly, a solution of 2.5 M NaOH (EKA Chemicals) was slowly pumped into a solution of 0.5 M Fe(NO3)3 3 9H2O (Merck) under stirring and bubbling with Ar gas until the pH reached 12. The product was aged at 60 C for 1 week prior to ca. 2 months dialyzing in Millipore 12-14000 D tubes and subsequently stored in polyethene bottles. The precipitate obtained in this way was identified as goethite by XRD. The goethite suspensions were thoroughly purged with nitrogen gas to remove any carbonate species in solution or at the goethite surface. The OP compound trimethyl phosphate (liquid TMP, 98% GC purity, Merck) was used without further purification. 2.2. Materials Characterization. Grazing incidence XRD was performed at room temperature on thin films of maghemite on Si wafers using a Siemens D5000 X-ray diffractometer equipped with a Cu KR1 radiation source.21 Powder XRD diffraction was measured on hematite and goethite nanopowders using a Bruker d8 Advance instrument in θθ mode with an optical configura tion consisting of a primary G€obel mirror and a Vantec-1 22 detector. Published reference data from the PDF-2 database (hematite, card no. 00-003-0664; maghemite, card no. 00-0391346; goethite, card no 01-081-0464)23 and literature24 were used to analyze the collected diffraction data. Transmission electron microscopy (TEM) analysis was performed on a Jeol JEM 1230 microscope equipped with a digital multiscan camera (Gatan MSC 600CW). Samples were suspended in water and 5 μL of the suspension was added to a Formvar coated 3 mm grid and dried prior to measurements. A rough dimensional analysis of the particle sizes from TEM images was done in the following way: For hematite 7 wellresolved particles were picked out. In general, the nanoparticles appeared asymmetric and major and minor axes were defined and the mean of the projected diameters was calculated and finally averaged over all 7 particles. In the case of maghemite two subsets of particles were selected to quantify the apparent bimodal size distribution apparent in TEM (cf. Figure 2): 20 small-to-medium sized and 20 large-to-medium sized particles, respectively, were separately averaged. The rod-shaped goethite particles exhibit three dimensions, i.e., length, width, and height (Figure 2c). Again two subsets of 30 particles in the small-to-medium and large-to-medium size range, respectively, were selected and the particle length was obtained by averaging all 60 particles. These values represent projections onto the imaging plane and thus underestimate the true length. The orientation distribution of the particles around the long-axis (i.e., around the Æ010æ direction in 8949
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A Figure 2b, inset) prevented, however, accurate separation of width and height with sufficient statistics due to their similar magnitude. Instead, an average width and height defined as (height width/2) was calculated. A detailed morphology analysis was performed using Wulff constructions to determine the crystal shapes observed in TEM (WinXMorph Software).25 The input parameters in the analysis were crystal structure, unit cell dimensions, and the experimentally determined dimensions and particle morphologies from XRD and TEM as well as literature data on morphology.19,26,27 Hematite is isostructural with corundum and belongs to the hexagonal crystallographic system (space group D2h16-R3c), with 30 atoms (six Fe2O3 units) per unit cell with lattice parameters a = b = 5.034 Å and c = 13.752 Å. Maghemite (γ-Fe2O3) has an inverse spinel structure and belongs to the cubic crystallographic system (space group Oh7-Fd3m), with 40 atoms (eight Fe2O3 units) per unit cell with lattice parameters a = b = c = 8.347 Å. Goethite is isostructural with diaspore and belongs to the orthorhombic system (space group D2h16-Pnma), with 16 atoms (four FeOOH units) per unit cell with lattice parameters a = 4.608 Å, b = 9.956 Å, and c = 3.022 Å. For goethite the Æ110æ, Æ002æ, and Æ020æ diffraction peaks were used and projected onto the orthogonal crystal axis to estimate the asymmetric particle dimensions. From the TEM images 40 particles were used in the morphology analysis for each oxide. The final particle morphology is shown as insets in Figure 2. Due to the fabrication procedure of hematite particles described above, a majority of these particles appeared coalesced in TEM, which prevented a similar detailed morphology analysis as for the other particles. Specific surface area and pore size distribution (PSD) were determined by the BrunauerEmmettTeller (BET) and BarrettJoynerHalenda (BJH) methods, respectively employing a Micromeritics ASAP 2020 analyzer for maghemite and a Micromeritics TriStar 3000 analyzer for hematite and goethite. X-ray photoelectron spectroscopy (XPS) characterization was performed with a Kratos Axis Ultra electron spectrometer using a monochromatic Al KR X-ray source operated at 150 W, a hybrid lens system with magnetic lens, and a charge neutralizer. Both wide spectrum (pass energy 160 eV) and narrow scans of all detected elements (pass energy 20 eV) were acquired. The binding energy (BE) scale was referenced to the C1s line of aliphatic carbon set at 285.0 eV. Processing of the spectra was accomplished with Kratos software and the CasaXPS program package.28 Raman spectroscopy of oxide films was carried out in backscattering geometry using a Horiba Jobin-Yvon Labram HR800 confocal Raman microscope equipped with a gas reaction cell (Linkam TS 1500). An Ar ion laser beam employing λ = 514.5 nm and with output power of 5.0 mW was focused on the sample by a 50 long working objective.29 Attenuation of the laser power was found to be of utmost importance when spectra were aquired on maghemite and goethite due to facile phase transformation to the hematite phase. By using notch filters that reduce laser power to 1.25 mW during analysis of maghemite and goethite, respectively, we observed no laser induced transformations to hematite. The spectral resolution was 3.5 cm1 employing 600 grooves mm1 grating. The spectra were frequency calibrated against the intense Raman active mode of the underlying silicon substrate at 520.7 cm1. Raman measurements were conducted on samples subjected to the same pretreatment conditions applied in the DRIFT experiments (see below) with the difference that the samples were annealed
ARTICLE
in synthetic air (20% O2 and 80% N2; AGA research grade) with a slightly lower gas flow of 60 mL min1 using a home-built gas generator connected to the Raman reaction cell. 2.3. In Situ DRIFT Spectroscopy. In situ diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy measurements were performed in a modified reaction cell (Harrick Scientific) using a Bruker IFS-66v/S spectrometer equipped with a LN2 cooled broad band MCT detector. Spectra were acquired with 4 cm1 resolution and each spectrum was averaged over 128 scans. Repeated in situ DRIFT spectra were acquired as a function of dosing time with background spectra acquired in synthetic air ∼6 min before TMP adsorption. All spectra presented here are raw spectra without any smoothing or baseline corrections. The DRIFT reaction cell was connected to a gas generator system that allows controlled dosing of reaction gas in the reaction cell as described elsewhere.30 Briefly the reaction gas was generated by evaporation of TMP into a carrier gas stream (synthetic air, 100 mL min1) from a diffusion tube setup connected to a temperature controlled reservoir held at T = 62 ( 0.2 C, yielding a gas injection rate of 6.08 μg min1 into the feed gas (equivalent to 175 ppmv). The samples were pretreated to remove organic residues as determined by DRIFT spectroscopy in the following manner: Maghemite samples were calcined at 300 C for 80 min and washed in 4 mM NaOH and Milli-Q-water to remove organic residues as described elsewhere.30 The hematite and goethite samples were subsequently annealed in situ in the reaction cell in synthetic air (100 ml min1) at 400 and 200 C, respectively, for 20 min and then cooled to 20 C. Phase purity (determined by Raman spectroscopy) and spectroscopic purity of the samples (determined by in situ DRIFT during pretreatment) were confirmed prior to measurements. Liquid reference spectra of TMP were acquired on a DuraSamplIRII ATR accessory equipped with a 9 internal reflection diamond element and collected at 2 cm1 resolution. Each spectrum was averaged over 300 scans and ATR corrected.31 To quantify the TMP coverage as a function of dosing time on powder samples in DRIFT, the following procedure was performed: (i) To relate the IR absorbance to absolute concentrations of TMP, a multivariate calibration measurement was first performed using transmittance FTIR spectroscopy using a liquid sample cell (Omni cell, Specac) with TMP (15 vol % of TMP) diluted in CS2 (>99.1% purity, Merck) covering the absorbance region of interest. The strong phosphoryl ν(PdO) and ν(C—O —(P)) TMP bands located at ∼1280 and ∼1040 cm1, respectively, were utilized. It was found that one absorbance unit (1 AU) in transmission FTIR associated with the ν(PdO) and ν(C—O—P) modes corresponded to 4.34 1022 and 1.49 1022 TMP molecules m2, respectively. (ii) Analogous in situ transmittance FTIR spectroscopy measurements were then performed with gaseous TMP adsorbed on maghemite films employing an in situ gas reaction cell connected to the same gas generators as in the DRIFT experiments. This gives a relation between liquid transmission data for and transmission data on films. In particular, it shows that ≈3 1020 TMP molecules m2 is adsorbed after 40 min gas dosing (typically the longest dosing time employed). (iii) Calibration of DRIFT spectra in units of log(1/R) of TMP adsorbed on maghemite powder against the (calibrated) transmission FTIR spectra of TMP adsorbed on the maghemite film at each point of time was performed, yielding a relation between log(1/R) units and AU units, and hence TMP 8950
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A
ARTICLE
Figure 1. XRD diffractograms of (a) 31 nm hematite, (b) 16 nm maghemite, and (c) 8 32 nm goethite nanoparticles.
Table 1. Physical Properties of the Iron Mineral Particles Used in the Present Study material hematite
d (nm)
d (nm) Fe:O ratio surface area BJH av pore
(TEM)
(XRD)
(XPS)
(m2 g1)
width (nm)
30.74d
2:3.031
28.98
22.2
16e
2:2.931
90.03
7.9
1:1.981
100.06
13.6
∼30
maghemite ∼20 (∼35)
a
goethite
∼11b
8.06f
∼62c
31.83g
a
Value in parentheses shows the average d determined for the (sedimented) large particles. b Average calculated as an average of height and width. c Average calculated from the length of the long axis. d Average calculated from the Æ012æ, Æ104æ, Æ110æ, and Æ116æ reflections. e Average calculated from the Æ220æ, Æ311æ, and Æ400æ reflections. f Average calculated from the Æ110æ reflection. g Average calculated from the Æ002æ reflection.
concentration. This calibration showed that 1 log(1/R) unit corresponded to 2.36 1021 and 1.08 1021 TMP molecules m2 using the ν(PdO) and ν(C—O—P) bands, respectively.
3. RESULTS AND DISCUSSION 3.1. Materials Properties. Figure 1 shows XRD spectra of powders of hematite, maghemite, and goethite, respectively. The major Bragg reflections shown in Figure 1 are in good agreement with published data for each polymorph23 and proves the phase purity of the nanoparticles. Using the Scherrer equation, dhkl = kλ/Bhkl cos θ, the particle size, dhkl, of the (hkl) reflection planes were estimated by taking into account the instrumental resolution in the full-width at half-maximum, Bhkl, of diffraction peaks at diffraction angle, θ, and using a shape factor k = 0.9. The average diameter was found to be d = 30.7 and d = 15.8 nm for hematite and maghemite, respectively (Table 1). For goethite d002 = 31.8 nm and d110 = 8.1 nm, demonstrating the asymmetric particle morphology with a longer dimension along the Æ010æ directions compared to the Æ100æ and Æ001æ directions (Table 1). Figure 2a shows TEM images of maghemite nanoparticles. The particles appear spherical with a size distribution exhibiting an apparent bimodal distribution with a smaller fraction of larger
Figure 2. TEM micrograph images of (a) 16 nm maghemite, (b) 8 32 nm goethite, and (c) 31 nm hematite nanoparticles. The particle morphology with majority facets is shown in the inset. In (c) a few particles exhibiting pronounced rhombohedral morphology are indicated by dotted, black circles.
particles mixed with small particles. This is attributed to contamination of coarse particles from the sediment during the refinement process described above as proved by calculating the average particle diameter of the sedimented particles (coarse particles; see Experimental Section). The size of two sets of maghemite particles (small-medium and medium-large, respectively) were determined to d = 20 and d = 35 nm, respectively (Table 1). The goethite particles appear as rods in TEM (Figure 2b) consistent with the asymmetric d002 and d110 values obtained from the XRD analysis. The average particle size is d = 62 nm and d = 11 nm measured parallel and perpendicular to the long-axis of the particles, respectively. A morphology analysis of the particles, using Wulff construction25 by aid of the XRD, TEM, and literature data, suggests that the goethite particles expose predominantly {010} and {101} facets in good 8951
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A
ARTICLE
agreement with previous reports,19 while maghemite particles expose {100}, {111}, and {110} facets.26 From the TEM micrographs in Figure 2c it is seen that hematite powder contains aggregated particles, of which many have coalesced during the calcination procedure described in the Experimental Section. This prohibits a detailed morphology analysis similar to maghemite and goethite above. The morphology of the hematite particles may be compared with previous studies which report a variety of different particle shapes and depends on detailed preparation procedures.3234 Cornell and Schwertmann reported that in the absence of additives (pertinent in our case), hexagonal plates which are often rounded and rhombohedral in shape predominate.1 Hematite nanoparticles prepared by thermolysis of acidic ferric chloride have been shown to have rhombohedral shape terminated by {104} facets due to energetic and structural reasons.35 Inspection of our TEM images reveals isolated particles that are consistent with a rhombohedral morphology (a few particles are indicated by dotted black circles in Figure 2c). We therefore tentatively propose that the predominant shape of our hematite nanoparticles is rhombohedral. From TEM an average diameter of d ≈ 30 nm was estimated for isolated particles that do not appear coalesced and agrees well with the XRD value for hematite (Table 1). Raman spectroscopy is known to be sensitive to phase purities of metal oxides and readily allows for in situ measurements as a
function of, e.g., temperature and reaction environment. Figure 3 shows Raman spectra obtained at room temperature after annealing at 400 C for 20 min (hematite), 300 C for 80 min (maghemite), and 200 C for 20 min (goethite), respectively, in synthetic air, thus corresponding to the experimental pretreatment conditions in DRIFT experiments. Goethite dehydroxylation and the kinetic phase transition to hematite is reported to start at ca. 210 C,36 while maghemite to hematite transformation starts at 300 C at a low rate.37 Higher temperatures result in facile phase transition of goethite and maghemite to hematite. Therefore careful control of laser power is necessary in Raman to avoid laser-induced phase transformations (cf. Experimental Section). It is evident from Figure 3 that mainly absorption bands due to the pure phases of each sample is observed and agree well with previously reported data for corresponding bulk phases (Table 2 and refs 1, 38, and 39). On goethite small traces of hematite may be discerned as a weak band at ∼223 cm1. This is not due to laser heating as checked by analogous experiments with nonannealed samples conducted at varying laser powers and thus attributed to the 15 min annealing at 200 C, which is close to the reported temperature for goethite to hematite phase transition. However, we consider the small amount of hematite phase present in Raman (which is not seen in XRD) as negligible. To effectively remove residual surface bonded impurities from the oxides that otherwise affect TMP adsorption properties, we have here adopted this annealing procedure as the best
Figure 3. Raman spectra displaying dry nanoparticles of (a) 31 nm hematite, (b) 16 nm maghemite, (c) 31 nm maghemite, and (d) 8 32 nm goethite nanoparticles.
Figure 4. Fe 2p XPS spectra of (a) 31 nm hematite, (b) 16 nm maghemite, and (c) 8 32 nm goethite nanoparticles. The spectra are shifted along the ordinate axis for clarity.
Table 2. Compilation of Raman Bands and Vibrational Mode Assignments of Hematite, Maghemite, and Goethite and Hydroxide Nanoparticles hematite 1
ν (cm )
goethite 1,3841
assignment
1
ν (cm )
maghemite 1,38,39
assignment
224
A1 g FeO sym str
223
A1g FeO sym str
243
Eg FeO sym. bend
240
Eg
290
Eg FeO sym bend
298
Eg FeOH sym bend
297 408
Eg FeO sym bend
386 397
Eg FeOFe/OH sym str
495
A1g FeO sym str
477
A1g FeOH asym str
609
Eg FeO sym bend
548
FeOH asym str
656
Eu
678
FeO sym str
1
ν (cm ) 192 ∼358
8952
assignment1,38,39,41 A1g FeO sym str Eg FeO sym str
505
T2g FeO asym bend
708
A1g sym str
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A
ARTICLE
Figure 6. ATR-spectrum of spectroscopically pure liquid TMP.
Table 3. Saturation Coverage and Adsorption Rate Constants for TMP Adsorption Using First-Order Adsorption Kinetics
Figure 5. In situ DRIFT spectra showing gas phase adsorption of TMP onto the mineral oxide surfaces, after 0, 2, 4, 6, 10, 19, and 29 min (bottom to top). The measurements were performed in synthetic air at 20 C.
compromise between sample cleanliness and phase purity. Adopting a too low annealing temperature (or neglecting annealing all together) yield DRIFT spectra that exhibited impurity-induced artifacts. With the present procedure we remove adventitious carbon as checked by DRIFT spectroscopy. Overall, we conclude that the (hydr)oxide nanoparticles maintain their structure during the calcination pretreatment procedure employed here to remove residual organic impurities prior to the TMP adsorption experiments. Figure 4 shows XPS spectra of hematite, maghemite, and goethite nanoparticles, respectively. In the Fe 2p XPS spectrum
Nsat (molecules g m2) 1019
ka (min1)
hematite
1.2
0.020
maghemite
0.6
0.026
goethite
0.7
0.025
from each sample, the four photoelectron peaks 2p3/2, 2p1/2, 2pSAT1, and 2pSAT2 are positioned at 711, 725, 720, and 733 eV, respectively, which is consistent with Fe3+.42,43 The atomic concentration deduced from the XPS data confirm the 2:3 stoichiometry of hematite (R-Fe2O3) and maghemite (γ-Fe2O3) and the 1:2 Fe:O stoichiometry of goethite (R-FeOOH) (Table 1) with no significant binding energy difference compared to tabulated bulk data, or between different particle sizes.44 3.2. TMP Adsorption. Figure 5 shows in situ DRIFT spectra of TMP adsorbed on hematite, maghemite, and goethite particles, respectively, as a function of dosing time. The spectrum of neat liquid TMP is shown in Figure 6 for comparison. The vibrational frequencies and peak assignments of TMP adsorbed on the different (hydr)oxides during the first minutes of dosing are compiled in Table 3. The DRIFT spectra in Figure 5 acquired after 10 min, displaced OH groups condense and adsorbed H2O builds up on the surface. At the same time the relative intensity of hydrogen bonded surface OH bands increases (the 3666 cm1 band). Similar bands at 3687 and 3656 cm1 due to isolated surface OH groups are observed on goethite. In particular, the lower 3656 cm1 band has previously been assigned to stretching vibrations of singly coordinated surface hydroxyl groups.1 The low frequency OH band at 3490 cm1 on goethite has been attributed to the stretch vibration of doubly/triply coordinated surface hydroxyl groups1 or a H—O—H stretching of nonstoichiometric hydroxyl units (excess water) in the goethite structure.60 We note that it has been reported that the distribution and reactivity of surface hydroxyl groups are dependent on both sample preparation methods and predominantly exposed crystal facets.59 Thus OH groups on hematite that complete the coordination of the low coordinated Fe atoms on the {104} planes are expected to be more acidic than OH groups on the other surfaces. Figure 8 shows the absolute value of the absorbance of the hydroxyl groups normalized to the TMP ν(PdO) bands between 1220 and 1250 cm1 (log(|1/ROH|)/log(1/RTMP)) as a function of dosing time. The amount of depleted surface hydroxyl groups (negative bands) compared to adsorbed TMP molecules increases in the order goethite < maghemite , hematite. This correlates with the higher reactivity toward TMP for hematite and supports a stronger interaction with surface OH groups. It is evident from Figure 8 that the OH depletion on maghemite and goethite follows the TMP adsorption kinetics at t > 8 min (flat curves). The declining initial ratio suggests that excess OH groups are displaced or consumed during an initial adsorption period. On hematite the (log(1/R|OH|)/log(1/RTMP)) ratio decreases also at t > 8 min, suggesting a kinetically activated surface reaction between OH and TMP with a higher rate than on the other particles. The analysis above reveals that both the interactions with Fe Lewis acid sites (PdO—Fe) and hydrogen bonding (PdO 3 3 3 H) contributes to the observed DRIFT spectra; the former is dominant on the oxides, and the latter, on the hydroxide. The latter is explicitly shown in Figure 9, which shows the mode-resolved adsorption kinetics on goethite and should be compared to Figure 7. On the hydroxide the large dips of the 8956
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A 3687 and 3656 cm1 bands correlate with the main ν(PdO) band at 1253 cm1, while on the oxides the depletion of isolated surface groups correlate with the main ν(PdO) bands at 12211225 cm1. The latter is manifested in a large decrease of log(1/R|OH|)/log(1/RTMP) for hematite in Figure 8 and supports the hypothesis above of strong interaction between methoxy groups and surface OH groups. Thus we attribute the 1253 cm1 band observed on goethite to hydrogen bonding of the phosphoryl oxygen to surface OH. A smaller downshift of the ν(PdO) band for hydrogen bonded PdO groups compared to cation coordinated surface sites has also been reported by others.3,53 The weaker 1232 cm1 band in Figure 5c is accordingly attributed to coordination with Lewis acid sites. The smaller shift of this band compared to corresponding bands on the oxides and the considerable lower intensity of the ν(PdO) mode at 1232 compared to 1253 cm1 on goethite indicate a substantially lower fraction of accessible Lewis acid cation sites on the hydroxide. In Figure 10a we depict the adsorption structures on the oxides, and in particular hematite, deduced from the DRIFTS experiments. In Figure 10b the predominant adsorption structure on goethite is shown. 3.3. TMP Dissociation. In Figure 5 new absorption bands that are not associated with TMP appear in the spectra after ca. 46 min (see Table 4). An absorption bands at ∼1080 cm1, which is particularly apparent on the oxides band is attributed to the νs(PO) combination stretching vibration in coordinated dimethyl phosphate (DMP), in good agreement with previous studies.55 The formation of DMP is also qualitatively supported by the reported decomposition (hydrolysis) pathway of DMMP on TiO2.61 Methoxy formation is indicated by the typical ν(CH) bands around 2820 cm1 previously reported on other metal oxides during DMMP adsorption.3,50,52,62 The 2820 cm1 band is present on all oxide surfaces with similar intensity. On hydroxide the relative intensity of the methoxy bands is lower. Instead, a peak at 2806 cm1 appears that partly obscures the former band, indicating two types of methoxy groups. The strong absorption bands due at 1579 and 1561 cm1 on R-Fe2O3 are due to the typical asymmetric νas(O—C—O) modes in formate with concomitant bands at 1380 and 1355 cm1 due to the corresponding ν(CH) and νs(O—C—O) modes due to oxidation of surface methoxy.52,62,63 These bands are weaker on γ-Fe2O3 and not discernible on R-FeOOH. The TMP dissociation with formation of surface methoxy and formate intermediates follows naturally from the interaction with PdO moiety with Lewis acid Fe sites discussed above due to charge compensation of methyl group of the C—O bond and the weakening of the O—P bond (cf. Figure 10a) and further points to the higher reactivity of the oxides, and in particular hematite, compared to goethite. Figure 11a shows the absorbance (log(1/R)) vs dosing time of the νs(CH) methoxy bands at 2820 cm1 normalized to the ν(PdO) bands in TMP to obtain the relative surface concentrations and thus to be able to make intersample comparisons. Several things can be noted from Figure 11a. First, the relative concentration of methoxy increases as R-FeOOH , γ-Fe2O3 < R-Fe2O3. The same order is also found for formate using the νs(O— C—O) bands at ca. 1350 cm1 during the first 35 min of dosing. Second, the initial rate of methoxy formation as deduced from the slope of the curves in Figure 11a is less than half on goethite compared to hematite. This is attributed to the two different types of adsorption sites on goethite (Lewis acid and Brønstedt acid), where only the Lewis acid site coordination (Figure 10b)
ARTICLE
Figure 12. Schematic picture of the dominant reaction mechanisms for adsorption and reaction of TMP on (a) hematite/maghemite (surface oxidation pathway 1) and (b) goethite (hydrolysis pathway 2).
leads to surface methoxy formation (pathway 1). In Figure 11b the ratio of the concentration of DMP (∼1080 cm1 band) and TMP (12201250 cm1 bands) as a function of adsorption time is shown. From Figure 11b it is evident that DMP successively accumulates on the surface and that the relative surface concentration of DMP is lower on goethite than the other surfaces. Moreover, the relative concentration of DMP is higher on maghemite than hematite. This is consistent with the higher surface methoxy formation rate on hematite compared to maghemite shown in Figure 11a, and shows that DMP is rapidly further decomposed on hematite. On goethite, a weak band situated at ∼1146 cm1 appears after ca. 20 min of dosing, which may be assigned to the totally hydrolyzed product phosphoric acid.52 Thus a picture emerges that differentiates between two dominant TMP decomposition mechanisms on the oxides and hydroxide, respectively; namely, surface mediated oxidation and 8957
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A hydrolysis of TMP. The former decomposition pathway is faster than the latter. On the oxides, TMP decomposition occurs through interaction with Lewis acid surface sites with formation of coordinated (possibly bidentate)54 TMP and DMP as critical intermediates (pathway 1 in Figures 10 and 11). This mechanism is particularly pronounced on hematite, which we argue expose the highest density of reactive Lewis acid sites. Dissociation of methyl groups in TMP (and DMP) results in surface methoxy and formate formation. In contrast, on goethite, interaction with Brønstedt acid sites dominates. These TMP molecules eventually become hydrolyzed by residual water present on the surface, which gradually accumulates in the course of the experiments (the reaction gas inevitably contains a trace amount of water). This is evidenced by the appearance of bands due to phosphoric acid in the DRIFT spectra.54 Figure 12 outlines the dominant TMP decomposition pathways on the various mineral surfaces.
4. CONCLUSIONS TMP adsorption on well-characterized hematite, maghemite, and goethite nanoparticles was studied by in situ DRIFT spectroscopy as a model system for OP sorption on iron minerals. It is found that TMP bonds mainly to Lewis acid Fe sites through the O phosphoryl atom (PdO—Fe) on hematite and maghemite. On goethite, most TMP molecules bond to Brønstedt acid surface OH groups and form hydrogen bonded surface complexes. It is found that the number of accessible (Lewis acid) adsorption sites is about twice as high on hematite compared to maghemite and goethite. The observed trends of the reactivity suggest an intimate relation between the surface chemistry of exposed crystal facets, presumably due to the existence of open {104} facets on hematite. The vibrational mode analysis and uptake kinetics suggest two main reasons for the observed trend of reactivity toward TMP (hematite > maghemite > goethite): (i) larger number of accessible Lewis acid adsorption sites on hematite and (ii) stronger interaction between the Lewis acid Fe sites and the phosphoryl O atom on TMP for hematite and maghemite compared to goethite (as evidence by the ν(PdO) frequency shifts). On the oxides a surface oxidation pathways dominates during the initial adsorption, which results in the formation of surface methoxy and formate as the OP methyl fragments dissociate. In contrast, on goethite a slower hydrolysis pathway is identified, which eventually yields phosphoric acid. These results are important to understand OP surface chemistry on iron minerals. ’ AUTHOR INFORMATION Corresponding Author
*E-mail: lars.osterlund@foi.se and lars.osterlund@angstrom.uu.se.
’ ACKNOWLEDGMENT This work is supported by grants from the Centre for Environmental Research, CMF (no. 0632299), and the Swedish Research Council (no. 621-2006-5152). We thank Andrey Shchukarev for XPS measurements. ’ REFERENCES (1) Cornell, R. M.; Schwertmann, U. The Iron Oxides: Structure, Properties, Reactions, Occurrence and Uses; VCH: New York, 1996.
ARTICLE
(2) Ekerdt, J. G.; Klabunde, K. J.; Shapley, J. R.; White, J. M.; Yates, J. T. J. Phys. Chem. 1988, 92, 6182. (3) Mitchell, M. B.; Sheinker, V. N.; Mintz, E. A. J. Phys. Chem. B 1997, 101, 11192. (4) Garade, A. C.; Bharadwaj, M.; Bhagwat, S. V.; Athawale, A. A.; Rode, C. V. Catal. Commun. 2009, 10, 485. (5) Vayssieres, L. Appl. Phys. A: Mater. Sci. Process. 2007, 89, 1. (6) Dannenberg, A.; Pehkonen, S. O. J. Agric. Food Chem. 1998, 46, 325. (7) Henderson, M. A.; Jin, T.; White, J. M. J. Phys. Chem. 1986, 90, 4607. (8) Schnurer, Y.; Persson, P.; Nilsson, M.; Nordgren, A.; Giesler, R. Environ. Sci. Technol. 2006, 40, 4145. (9) Stumm, W.; Sulzberger, B. Geochim. Cosmochim. Acta 1992, 56, 3233. (10) Casey, W. H.; Swaddle, T. W. Rev. Geophys. 2003, 41. (11) Jones, D. L.; Dennis, P. G.; Owen, A. G.; van Hees, P. A. W. Plant Soil 2003, 248, 31. (12) Dahl, R. C. Adv. Argon. 1977, 28, 83. (13) Olsson, R.; Giesler, R.; Loring, J. S.; Persson, P. Langmuir 2010, 26, 18760. (14) Ikaev, A. M.; Mingalyov, P. G.; Lisichkin, G. V. Colloid J. 2007, 69, 741. (15) Nanocatalysis; Heiz, U., Landmann, U., Eds.; Springer-Verlag: Heidelberg, 2007. (16) Venema, P.; Hiemstra, T.; Weidler, P. G.; van Riemsdijk, W. H. J. Colloid Interface Sci. 1998, 198, 282. (17) Gaboriaud, F.; Ehrhardt, J. Geochim. Cosmochim. Acta 2003, 67, 967. (18) Eggleston, C. M.; Stack, A. G.; Rosso, K. M.; Higgins, S. R.; Bice, A. M.; Boese, S. W.; Pribyl, R. D.; Nichols, J. J. Geochim. Cosmochim. Acta 2003, 67, 985. (19) Prelot, B.; Villieras, F. e. a. J. Colloid Interface Sci. 2003, 261, 244. (20) Boily, J. F.; Lutzenkirchen, J.; Balmes, O.; Beattie, J.; Sjoberg, S. Colloids Surf., A 2001, 179, 11. (21) Ekstrand, A.; Jansson, K.; Westin, G. Chem. Mater. 2005, 17, 199. (22) Lundholm, K.; Bostrom, D.; Nordin, A.; Shchukarev, A. Environ. Sci. Technol. 2007, 41, 6534. (23) ICDD, The Powder Diffraction File, PDF-2; International Center for Diffraction Data, 2004. (24) Pettersson, M.; Lundell, J.; Khriachtchev, L.; Rasanen, M. J. Am. Chem. Soc. 1997, 119, 11715. (25) Kaminsky, W. J. Appl. Crystallogr. 2007, 40, 382. (26) Zhao, N.; Ma, W.; Cui, Z.; Song, W.; Xu, C.; Gao, M. ACS NANO 2009, 3, 1775. (27) Wang, L.; Gao, L. J. Phys. Chem. C 2009, 113, 15914. (28) Bengtsson, A.; Shchukarev, A.; Persson, P.; Sjoberg, S. Geochim. Cosmochim. Acta 2009, 73, 257. (29) Mattsson, A.; Lejon, C.; Stengl, V.; Bakardjieva, S.; Oplustil, F.; € Andersson, P. O.; Osterlund, L. Appl. Catal., B 2009, 92, 401. € (30) Osterlund, L.; Stengl, V.; Mattsson, A.; Bakardjieva, S.; Andersson, P. O.; Oplustil, F. Appl. Catal., B 2009, 88, 194. (31) Harrick, N. J. Internal Reflection Spectroscopy; Harrick Scientiffic Corp.: New York, 1987. (32) Matijevic, E.; Scheiner, P. J. Colloid Interface Sci. 1978, 63, 509. (33) Ocana, M.; Rodriguezclemente, R.; Serna, C. J. Adv. Mater. 1995, 7, 212. (34) Bodker, F.; Morup, S. Europhys. Lett. 2000, 52, 217. (35) Rodriguez, R. D.; Demaille, D.; Lacaze, E.; Jupille, J.; Chaneac, C.; Jolivet, J. P. J. Phys. Chem. C 2007, 111, 16866. (36) Przepiera, K.; Przepiera, A. J. Therm. Anal. Calorim. 2003, 74, 659. (37) Schimanke, G.; Martin, M. Solid State Ionics 2000, 136, 1235. (38) Legodi, M. A.; de Waal, D. Dyes Pigm. 2007, 74, 161. (39) Oh, S. J.; Cook, D. C.; Townsend, H. E. Hyperfine Interact. 1998, 112, 59. 8958
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959
The Journal of Physical Chemistry A
ARTICLE
(40) Kustova, G. N.; Burgina, E. B.; Sadykov, V. A.; Poryvaev, S. G. Phys. Chem. Miner. 1992, 18, 379. (41) Chamritski, I.; Burns, G. J. Phys. Chem. B 2005, 109, 4965. (42) NIST X-ray Photoelectron Spectroscopy Database, 2003; Vol. 3.5, V. (43) Yamashita, T.; Hayes, P. Appl. Surf. Sci. 2008, 254, 2441. (44) Aronniemi, M.; Lahtinen, J.; Hautojarvi, P. Surf. Interface Anal. 2004, 36, 1004. (45) Henderson, M. A.; White, J. M. J. Am. Chem. Soc. 1988, 110, 6939. (46) Li, Y. X.; Klabunde, K. J. Langmuir 1991, 7, 1388. (47) Li, Y. X.; Schlup, J. R.; Klabunde, K. J. Langmuir 1991, 7, 1394. (48) Kanan, S. M.; Tripp, C. P. Langmuir 2001, 17, 2213. (49) Kim, C. S.; Lad, R. J.; Tripp, C. P. Sens. Actuators, B 2001, 76, 442. (50) Sokrates, G. In Infrared and Raman Characteristic Group Frequencies; Tables and Charts, 3rd ed.; Wiley: Chichester, U.K., 2001; pp 94. (51) Rasko, J.; Kecskes, T.; Kiss, J. J. Catal. 2004, 224, 261. (52) Trubitsyn, D. A.; Vorontsov, A. V. J. Phys. Chem. B 2005, 109, 21884. (53) Waghe, A.; Kanan, S. M.; Abu-Yousef, I.; Jensen, B.; Tripp, C. P. Res. Chem. Intermed. 2006, 32, 613. (54) Lefevre, G. Adv. Colloid Interface Sci. 2004, 107, 109. (55) Florian, J.; Baumruk, V.; Strajbl, M.; Bednarova, L.; Stepanek, J. J. Phys. Chem. 1996, 100, 1559. (56) Reva, I.; Simao, A.; Fausto, R. Chem. Phys. Lett. 2005, 406, 126. (57) Segal, S. R.; Cao, L. X.; Suib, S. L.; Tang, X.; Satyapal, S. J. Catal. 2001, 198, 66. (58) Davydov, A. A. Molecular Spectroscopy of Oxide Catalyst Surfaces; John Wiley & Sons: Chichester, U.K., 2003. (59) Ferretto, L.; Glisenti, A. J. Mol. Catal. A: Chem. 2002, 187, 119. (60) Ruan, H. D.; Frost, R. L.; Kloprogge, J. T.; Duong, L. Spectrochim. Acta Part A 2002, 58, 967. (61) Moss, J. A.; Szczepankiewicz, S. H.; Park, E.; Hoffmann, M. R. J. Phys. Chem. B 2005, 109, 19779. (62) Rusu, C.; Yates, J. T., Jr. J. Phys. Chem. B 2000, 104, 12292. (63) Busca, G.; Lorenzelli, V. Mater. Chem. 1982, 7, 89.
8959
dx.doi.org/10.1021/jp201065w |J. Phys. Chem. A 2011, 115, 8948–8959