ARTICLE pubs.acs.org/Langmuir
Surface Complexation of 2,5-Dihydroxybenzoic Acid (Gentisic Acid) at the Nanosized Hematite-Water Interface: An ATR-FTIR Study and Modeling Approach K. Hanna* and F. Quiles Laboratoire de Chimie Physique et Microbiologie pour l’Environnement, LCPME, UMR 7564 CNRS-Nancy Universite, 405 rue de Vandoeuvre, 54600 Villers-les-Nancy, France
bS Supporting Information ABSTRACT: In this study, characteristic interactions of 2,5-dihydroxybenzoic acid (or gentisic acid, GA) with the surface of 15-nm-sized hematite (R-Fe2O3) were studied by combining batch macroscopic experiments, in situ attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopic investigations, and surface complexation modeling. A correlation between the pH, the amount of adsorbed GA, and the amount of Fe(III) released from the hematite surface was observed, whereas the dissolution of hematite nanoparticles became significant only at low pH and high ligand loading. From the ATR-FTIR results, two aqueous complex structures have been identified depending on pH. At the hematite-water interface, the occurrence of one deprotonated inner-sphere “bidentate” complex and one outer-sphere complex was suggested through all of the investigated pH range. At high surface coverage, variations of vibrational band intensities were observed, suggesting the occurrence of nonspecific molecular interactions. The macroscopic results (i.e., GA batch sorption and the ligand-promoted dissolution of hematite) obtained under a wide range of experimental conditions corroborated the ATR-FTIR microscopic findings. GA adsorption was described by a surface complexation model fitted to pH-adsorption curves with 1 mM sorbate concentration in the pH range of 3-9. Two surface complexes (one outer-sphere species (tFeOH2)2 3 3 3 H2L(1þ,1-) and one inner-sphere species (tFe)2H2L) were proposed using the three-plane model. The inner-sphere complexes were predominant at low pH values, and the relative concentrations of the outer-sphere species increased with the pH increase. The formation of inner-sphere complexes at acidic pH values can promote the dissolution of nanosized hematite. At high solute loading, GA oxidation into carboxybenzoquinone compounds by ferric species was suspected, suggesting the occurrence of a redox reaction analogous to that of hydroquinone compounds.
1. INTRODUCTION The transport and mobility of organic species in soils and groundwater are strongly related to the nature and relative abundance of the reactive mineral phases.1 Iron oxides and oxyhydroxides are ubiquitous in soils and aquifer sediments. They can form the primary reactive interface for trace metals, oxyanions, and organic ligands containing carboxylic and hydroxy functional groups.2,3 In environmental systems, where crystal growth is inhibited by chemically heterogeneous conditions, nanoscale particles are predominant.4 Ferrihydrite is then the typical low-crystalline nano-iron oxyhydroxide. Well-structured iron phases such as hematite or goethite can also occur as very small particles of nanometric size.5 In particular, hematite (R-Fe2O3) is one of the most thermodynamically stable iron oxides at ambient temperature, and it is quite abundant in natural settings.4,6,7 Because of their unique size-dependent properties (large specific surface area and high surface reactivity), the nanosized particles play a preeminent role in sorption processes, especially when clay minerals are absent. 2,5-Dihydroxybenzoic acid, also called gentisic acid (GA) or monocarboxyhydroquinone, is representative of the aromatic carboxylate r 2011 American Chemical Society
compounds found in soils and groundwaters.8 It is also a typical end product of the metabolic degradation of various kinds of polycyclic aromatic hydrocarbons (PAH), such as the degradation of naphthalene via salicylic acid.9 Because of their high mobility in porous media, interactions of carboxylated compounds with oxide particles are of great importance in several colloid and soil science processes such as the aqueous and particular transport of contaminants. In general, the interaction with organic anions occurs via exchange of the surface hydroxyls of metal oxides with oxygens of the functional groups of organic ligands.10-12 Infrared spectroscopic and theoretical investigations suggest that organic acids bind to oxide surfaces simultaneously as inner- and outer-sphere complexes and that their relative abundance varies with the pH value, oxide type, and ligand structure.13-24 In most of these published studies, interactions of lowmolecular-weight organic acids with metal oxide surfaces are addressed. However, the sorption modes of carboxylated hydroquinone Received: October 21, 2010 Revised: December 20, 2010 Published: February 18, 2011 2492
dx.doi.org/10.1021/la104239x | Langmuir 2011, 27, 2492–2500
Langmuir compounds at the oxide-water interface were not explored to a significant extent. In addition, these aromatic compounds containing one carboxylic group and two phenolic hydroxyl groups may exhibit unusual sorption behavior. In this work, characteristic interactions of 2,5-dihydroxybenzoic acid or gentisic acid (GA) with the surface of a synthetic nanohematite are studied by combining batch macroscopic and in situ spectroscopic investigations. GA sorption was studied under varying chemical conditions (exposure time, solute concentration, pH, and ionic strength). Desorption experiments using three inorganic ligands and one organic solvent were also performed for a better understanding of the sorption behavior. UV-visible and liquid chromatography (LC)/UV measurements were performed to evaluate the possible oxidation of GA by ferric species. Fourier transform infrared spectroscopy in attenuated total reflection mode (ATRFTIR) was performed to identify the main surface complex structures at different pH values and surface coverages. Predictions of sorption versus pH are made from surface complexation modeling using the three-plane model (TPM) and the surface parameters of the nanosized hematite.
2. MATERIALS AND METHODS 2.1. Chemicals. 2,5-Dihydroxybenzoic acid (gentisic acid, GA, purity greater than 99%) was obtained from Sigma-Aldrich. GA has three ionizable hydrogen ions (H3L), one from carboxylic (pK1 = 2.96) and two from phenolic functional groups (pK2 = 10.46 and pK3 = 13.41).12 The chemical properties of gentisic acid are reported elsewhere.25 FeCl3 3 6H2O (98%) was purchased from Aldrich and used as received. 2.2. Synthesis and Characterization of Hematite Nanoparticles. Hematite nanoparticles (R-Fe2O3) were synthesized as
described in our previous work.26 The obtained particles were characterized by grain size measurements, N2 adsorption measurements, chemical analysis, scanning electron microscopy (SEM), transmission electron microscopy (TEM), and X-ray diffraction (XRD). The mineralogical nature of the iron phase was confirmed by Raman spectroscopy and XRD. Detailed results are reported in the Supporting Information. 2.3. Batch Experiments and ATR-FTIR Spectroscopy. Batch sorption experiments of GA on the surface of the synthesized solid were performed under various experimental conditions (exposure time, GA concentration, pH, and ionic strength). All batch experiments were performed in triplicate. They are detailed in the Supporting Information. Briefly, solutions of GA were mixed with a hematite suspension to give a final concentration of 1 g/L of hematite. The pH was adjusted to the desired value with NaOH (0.1 M) or HCl (0.1 M), and NaCl was added as the background electrolyte (10 mM). The flasks were shaken at 20 ( 1 °C in a rotary shaker. The supernatants were then centrifuged at 10 000g for 60 min and filtered through 0.22-μm-pore-size membrane filters (Millipore Corp., Bedford, MA) prior to the measurements of GA by UV-visible spectrophotometry18 and dissolved Fe by ICP-AES (inductively coupled plasma atomic emission spectrometry). ATR-FTIR spectra were recorded between 4000 and 780 cm-1 on a Bruker Vector 22 spectrometer equipped with a KBr beam splitter and a deuterated triglycine sulfate (DTGS) thermal detector. Spectra recording, data storage, and data processing were performed using the Bruker OPUS 3.1 software. A nine-reflection diamond ATR accessory (DurasamplIR, SensIR Technologies) was used to acquire spectra of wet samples. The angle of incidence was 45°, and the refractive index of the crystal was 2.4. One hundred bidirectional double-sided interferogram scans were collected per spectrum, which corresponds to a 1 min accumulation time. The resolution of the single-beam spectra was 4 cm-1. All interferograms were Fourier processed using the Mertz phase-correction mode and a Blackman-Harris three-term apodization function. No ATR correction was performed. ATRFTIR spectra are shown on an absorbance scale corresponding to log(Rreference/Rsample), where R is the internal reflectance of the device.
ARTICLE
For aqueous solution analysis, a stock solution of 0.1 M GA was prepared in 10-2 M NaCl. The GA-Fe(III) (1:2 and 1:5 molar ratios) solutions were obtained by the addition of FeCl3 3 6 H2O to the GA stock solution. The pH value of the solution was adjusted by adding either NaOH (0.1 M) or HCl (0.1 M). Before ATR-FTIR spectra recording, solution samples of Fe(III)-GA at a fixed pH were filtered through 0.22-μm-pore-size membrane filters (Millipore Corp., Bedford, MA) and directly applied to the diamond ATR crystal. A glass lid was placed over the crystal during infrared spectrum recording to prevent water evaporation. Prior to ATR-FTIR analysis, tubes from batch sorption experiments were centrifuged at 10 000g for 60 min. A wet sample paste was directly and uniformly applied to the diamond ATR crystal. The sample-holding region was covered with a glass lid to prevent water evaporation during measurements, and ATR-FTIR spectra were then recorded immediately. Appropriate spectra were used as a reference: a water spectrum and the filtered supernatant spectrum for GA and GA/FeCl3 solutions, respectively. Water vapor subtraction was performed when necessary. ATRFTIR spectra are presented in the useful region of 1800-780 cm-1. 2.4. Surface Complexation Modeling. As for goethite, different types of surface sites can exist on hematite depending of the proportion of crystal faces, which should have different binding affinities.27-30 The hematite used in this study was synthesized by the forced hydrolysis of ferric chloride without an aging procedure. This fast nucleation process generally produces small spherical iron oxide nanoparticles with poorly defined morphology.31 These features make the identification of the different crystal planes difficult and motivated the use of a simplified surface complexation model (SCM). In the present SCM, only singly coordinated sites are considered to participate in surface complexation reactions. This simplification cancels out the contributions of the doubly and triply coordinated groups, which are not reactive under our defined conditions. The estimated average value of singly coordinated groups for hematite proposed by Baron and Torrent27 was used (5 sites/nm2 with a total site density of about 8 μmol/m2). It is close to the value determined by solid titrations in the work of Jarlbring et al.32 (i.e., 4.5 reactive sites/nm2). The proton affinity constants of hematite were calculated from potentiometric titrations, as reported in previous work.26 The three-plane model (TPM) recently implemented in the PHREEQC2 code was used.33 In this model, three planes are defined: the surface plane or o plane (0), the β plane (1), and the diffuse plane or d plane (2). “Strongly” bound ions adsorb on the surface plane (0 plane) as inner-sphere complexes. The β plane, placed away from the surface, allows the sorption of “weakly” binding ions or electrolyte ions. In addition to the electrolyte binding constants, the TPM requires two capacitances for the surface and the β plane, frequently referred to as inner- and outer-layer capacitance, C1 and C2, respectively. The electrical double layer (EDL) function was used to determine the values of charge and electric potential of the 0, 1, and 2 planes. The charges of the adsorbates were spatially distributed among the 0, 1, and 2 planes. All calculations were carried out using the charge-modeling parameters and the constants listed in Table 1.
3. RESULTS AND DISCUSSION 3.1. Hematite Nanoparticles. TEM and SEM images show that the hematite nanoparticles (NH) are quasi-spherical and highly aggregated (Figure S1). The average size of the hematite aggregated particles varied between 10 and 25 nm, which is in agreement with earlier published work using similar synthesis methods (12 nm34 and 10.5 nm35). The PZC of the nanosized hematite was found to be around 8.1, which is within the range of the literature data (7.8-9.1).34,35 The surface area of NH measured by the N2 adsorption isotherm was 205 m2/g. The diameter of the spherical particles (d) can be related to the surface area (A) as A = 6/(Fd) = 205 m2/g 2493
dx.doi.org/10.1021/la104239x |Langmuir 2011, 27, 2492–2500
Langmuir
ARTICLE
Table 1. Formation Reactions of Surface Species in the Three-Plane Model with Only a Singly Coordinated Sitea surface group
Δz0
Δz1
Δz2
0.5-
log K
GA concentration (mM)
0
tFeOH
tFeOH20.5þ tFeOH20.5þ-Cl0.5þ
þ1 þ1 0
þ1
0
-1
(FeOH2)2 3 3 3 H2L(1þ,1-) (tFe)2H2L
þ1
-1
0
22.5
tFeOH
Table 2. First-Order Rate Constant of Adsorption (Min-1) (Correlation Coefficient, r2) at Two GA Concentrations and Three pH Values
-Na
0 -1
0 0
8.1 7.7
24.3
a
TPM with C1 = 1.5 F/m2, C2 = 2.0 F/m2, and [tFeOH]tot= 1.7 mM for 205 m2/L. The proton charge is attributed to the surface (Δz0), the charge of electrolyte ion, or the ligand to the 1 plane (Δz1), and no charge is added to the 2 plane (Δz2 = 0).
Figure 1. Sorption of GA on hematite vs time at three pH values (3, 6, and 8). [Hematite] = 1 g/L; T = 20 ( 1 °C, 10 mM NaCl as the supporting electrolyte. [GA] = 0.5 mM (a) and 1 mM (b). Lines represent model fits.
(where F is the hematite density, F = 5.24 106 g/m3). Therefore, the calculated average diameter of the supposedly spherical particle was around 5 nm. The discrepancy between the experimental and calculated values is probably due to the extent of aggregation of the nanoparticles, as shown in the SEM images. The hematite nanoparticles may likely form aggregated colloids in water, which have a different reactivity from that of individual hematite particles. Previous studies reported that both the particle size and aggregation state must be considered when evaluating the reactivity of nanoparticle suspensions.36 Maden and Hochella37 also reported that reactivity differences as a function of particle
pH 3
pH 6
pH 9
0.5
0.040 (0.97)
0.047 (0.98)
0.049 (0.98)
1
0.030 (0.97)
0.032 (0.98)
0.034 (0.99)
size are likely related to the changing electronic and geometric structure of the hematite surfaces. 3.2. Sorption and Dissolution Batch Experiments. 3.2.1. At 0.5 to 1 mM GA. The kinetic data obtained under batch conditions revealed that solute uptake reached a steady state at 90 min with 1 g/L as the solid loading and for both GA concentrations (0.5 and 1 mM, Figure.1). Different kinetic models, namely, a pseudo-first-order, a pseudo-second-order, and an intraparticle diffusion model were applied to the data as explained in previous work.25 The pseudo-first-order expression provided the best fit, whereas the pseudo-second-order and the intraparticle diffusion models did not fit the data well. Adsorption data over a 90 min period were therefore treated according to the first-order kinetics by plotting ln(Q/(Q - Qt)) as a function of time, t, where Q and Qt (μmol/m2) are the amount of adsorption at equilibrium and at time t (min). Kinetic constants k were obtained by applying linear regression analysis for the two GA concentrations at three pH values (Table 2). The sorption rate constant decreases slightly with GA concentration regardless of the pH value tested. The concentration-dependent kinetics may result from the existence of a chemical potential gradient in the sorbent.38 At low surface coverage, a limited number of sites seem to be easily accessible whereas sorption slows down at high loading because of the decreasing number of available free sites. This behavior may be related to the availability of reactive sites within the interstices of hematite aggregates and to the mass-transfer resistance including interaggregate diffusion or sorption-retarded diffusion through pores.38 In this case, the size of hematite aggregates may be important, constituting the rate-limiting step. However, the aim of this work was not intended to highlight the aggregation effect on the sorption processes. Nevertheless, the effect of GA addition on the aggregation process of a hematite colloidal suspension was tested at pH 6. Suspensions of 0.1 g/L of NH and 10 mM NaCl were added to a 1-cm-path-length cuvette, and the change in transmitted light at 560 nm was monitored as a function of time.39 First, the relative stability of NH suspensions results from the electrostatic repulsion between particles at pH 6 ( fluoride > sulfate.2 About 60% of the sorbed GA was removed from the surface by fluoride or sulfate addition, whereas more than 90% of GA was removed from the hematite surface with the phosphate treatment. This suggests that all of these ligands can competitively displace the sorbed GA and that the interactions between hematite and GA were weaker than the fluoride- sulfate-, or phosphate-iron interaction. 2495
dx.doi.org/10.1021/la104239x |Langmuir 2011, 27, 2492–2500
Langmuir To determine the role of hydrophobicity in GA sorption, desorption experiments were also performed using methanol as cosolvent. Less than 10% of GA was removed, suggesting that the hydrophobic contribution for GA sorption with iron oxides was of less significance. 3.2.2. At 10 mM GA. A strong increase in the sorbed amount of GA on the hematite surface was observed at a higher solute concentration (10 mM, with a total ligand loading of about 49 μmol/m2) over the pH range tested (Figure S3). However, this sorption behavior is atypical of carboxylated compounds, where adsorption is decreased with increasing pH and instead reaches an average surface coverage of ∼2.4 μmol/m2 regardless of the pH value. This strong sorption increase at high ligand loading could be attributed to the intermolecular interactions between the molecules in solution and sorbed molecules. This mechanism may be explained by outer-sphere complexation via nonspecific interactions such as van der Waals forces and/or hydrogen bonding.12,18,20 However, pH independence of sorption throughout the investigated pH range (3-9) is unusual, suggesting the occurrence of other removal processes of GA. For instance, the molecular transformation of GA may lead to an overestimation of the sorbed amount determined by the depletion method of the parent compound. In line with high sorption, a 10 mM organic ligand concentration was also shown to induce a higher extent of Fe dissolution (Figure S3). However, there is no direct correlation between the released amount of Fe(III) and the sorbed amount of GA. Because gentisic acid is a benzene-based quinone with a high reduction potential (E° = 0.769 V),40 one can imagine that the oxidative transformation of GA by ferric species may occur as previously reported for 1,4-dihydroquinone.41 Because the redox potential of gentisic acid is pH-dependent,42 the redox transformation process could also be variable as a function of pH. Previous work reported that gentisic acid can be chemically oxidized using common chemical oxidants leading to the formation of unstable compounds such as carboxybenzoquinone.43,44 However, Holmes et al.45 reported that FeCl3 failed to oxidize GA in aqueous solution. In the present study, both UV-visible and LC/UV measurements were performed to test the aqueous transformation of GA at high loading (10 or 100 mM) with hematite or with a chemical oxidant (i.e., hydrogen peroxide) at pH 6. The GA/oxidant = 1 (stoichiometry 1:1) molar ratio was chosen to ensure oxidation of the parent compound. For all of the experiments, an additional UV peak at 270 nm has appeared along the reaction time, together with an HPLC peak in the corresponding chromatogram. According to Uchimya and Stone,40 this peak could be attributed to the carboxybenzoquinone compound formed when GA reacts with a chemical oxidant. However, the relative abundance of this peak was very low in comparison with that of the parent compound. Soluble Fe2þ was also measured for the hematite/GA system at both GA loadings (10 and 100 mM) and at pH 6. Only trace levels of Fe2þ were detected, which is consistent with the findings of McBride46 for catechol and hydroquinone oxidation by iron oxides. Indeed, the electron transfer between the organic compound and Fe(III) generated Fe2þ, which was rapidly reoxidized by O2. As suggested by McBride46 for hydroquinone, coordination of the organic ligand at the hematite surface is a prerequisite for the electron transfer, underscoring the role of inner-sphere complexation in the GA oxidation. Finally, these macroscopic measurements show that there are additional removal mechanisms at a higher GA concentration. These suggestions will be discussed further on the basis of infrared spectra.
ARTICLE
Figure 4. ATR-FTIR spectra of 100 mM GA in 10 mM NaCl aqueous solutions at different pH values: (a) pH 1.6, (b) pH 8.5, and (c) pH 11.5. Offsets of spectra are used for clarity.
3.3. ATR-FTIR Spectroscopic Investigations. 3.3.1. Gentisic Acid in Aqueous Solution at Different pH Values. Because the
absorption bands of the functional groups and the aromatic ring are influenced both by the degree of protonation and coordination to metals, assignments of the main bands were attempted for the free nonionized and ionized molecules. Toward this aim, ATR-FTIR spectra of gentisic acid solutions containing 10 mM NaCl at different pH values were recorded (Figure 4). Depending on the pKa values, the pH was selected to have each species (nonionized form (H3L), monoanion (H2L-), and bianion (HL2-)) as the dominant species in solution. Tentative assignments of the principal bands between 1800 and 800 cm-1 are based on assignments made previously for polycrystalline gentisic acid47-49 and for analogous aromatic carboxylic acids50,51 (Table 3). Assignments of the more interesting bands against GA complexation are discussed below. The bands at 1670 and 1310 cm-1 disappear from the spectra with increasing pH. They are assigned to the CdO and carboxylic COH stretching modes of the H3L species, respectively. The pairs of bands at {1572, 1382} and {1556, 1369} cm-1 that are not present in the spectrum of the nonionized form are assigned to the asymmetric and symmetric stretching bands of the carboxylate group of H2L- and HL2-, respectively. The carboxylate symmetric stretching mode is probably coupled with mode 14 of the benzene ring with Wilson’s numbering. Ring modes {19a and 19b} are usually intense in the infrared spectra. The corresponding bands are located at {1453 and 1490}, {1458 and 1490}, and {1423 and 1478} cm-1 for H3L, H2L-, and HL2- species, respectively. The first ionization does not significantly change the wavenumbers of both 19 modes. The second ionization leads to high red shifts of both bands. For all GA species, the intensity of band 19b is greater than that of band 19a. The 1300-1200 cm-1 region where phenolic and aromatic ring bands absorb usually exhibits three intense bands. By comparison with salicylic and salicylate derivatives,50 the bands at 1275 and 1282 cm-1 are assigned to benzene ring mode 3. The band that shifts from 1217 to 1232 cm-1 when the pH value increases from 1.6 to 3.3 is assigned to the phenolic stretching modes in position 2 (νC2ΦOH). The bands at 1246 cm-1 at pH 1.6 and 3.3 and at 1259 cm-1 at pH 11.5 are assigned to the νC5Φ-OH and νC5Φ-Omodes, respectively. 2496
dx.doi.org/10.1021/la104239x |Langmuir 2011, 27, 2492–2500
Langmuir
ARTICLE
Table 3. Principal Wavenumbers (cm-1) and Tentative Assignment Modes of Gentisic Acid, Sodium Gentisate, and Its Ferric Complexes in Aqueous Solutions and on the Hematite Surface between 1800 and 800 cm-1 under Different Experimental Conditionsa gentisic acid (0.1 M) þ FeCl3 at varying
hematite (1 g/L) þ gentisic acid at varying
concentrations
concentrations
gentisic acid 0.1 M
0.2 M
0.2 M
0.5 M
pH 1.6 pH 8.5
pH 11.5
pH 0.8
pH 5.2
pH 0.4
(H3L) (H2L-)
(HL2-)
monodentate
bidentate
monodentate
1670
1760
10-3 M
10-2 M
10-1 M
10-2 M
10-1 M
pH 3.2
pH 3.2
pH 3.2
pH 7.9
pH 7.9
tentative νCdO
1764
∼1639 ∼1646 1612 1595
(1582 sh) (1535 sh) 1572
1504
δH2O 1612 1597
1556
1504
1625
1609 1596
1549 1505
1504 1482
1490
1490
1478
1453
1458
1423
1382
1369
1345
1345
assignments
ν8a ν8b
1610 1593 1574
1553
1569 νaCOO-
1538
1538 νaCOO-
1552
1553
1540
1538
1504
1505
1505
1505
1482
1482
1486
1479
νCC aromatic ring 1481 ν19b complex ν19b ν19a
1470 1364
1454 1381
1469 1365
1345
1454 1385
1454 1377
1463 1389
1454 1371
1452 ν19a; νsCOO 1380 ν14 þ νsCOO(coupled)
1342
1342
1370
1340
1347 νsCOO- þ δC2Φ -OH (?)
1310 1292 1275
νC-OH (carboxylic)
1310 1275
1282
1257
νC-O
1292 1244
1249
1245
1245
1266
ν3
1251
νC5Φ-O-
1259 1246
1246
1217
1232(sh)1232
1232
1231
1204
1213
1179 1139
1134
1131
1229 (sh)
1228
1228
1218
1229
1259 νC5Φ-OH 1243
1204
1215
1215
1210
1213
1218 νC2Φ-OH νC2Φ-O- ?
1181 1138
1128
1136
1136
1141
1136
1136 ν9b
1087
1087
1094
1087
1085 ν15
946
946
946 879
947
946
1105 1079
1083
1079
ωCH aromatic ring
1014
a
941 886
941 886
941 886
834
834
834
825
825
945
997
ωCH aromatic ring ωCH aromatic ring
838
842
ωCH aromatic ring
830
834
ωCH aromatic ring
Key: ν, stretching; δ, bending; F, rocking; ω, wagging; a, asymmetric; s, symmetric; sh, shoulder. pH (0.2.
3.3.2. Aqueous Complexes of Gentisic Acid with Fe3þ. Figure 5 shows the ATR-FTIR spectra of the complexed species obtained by mixing GA (100 mM) with FeCl3 (200 or 500 mM) at different pH values. Because of iron species precipitation, the spectra recorded at pH 8 did not allow the identification of complexed species. Note that the Fe(III)-GA stability constants are not known, which prevent us from building a speciation diagram. Similar features are observed in the ATR-FTIR spectra recorded for GA-Fe solutions at 1:5 and 1:2 molar ratios at pH
