Zoltán
Németh
*,
Éva G.
Bajnóczi†
,
Bogdán
Csilla
and
György
Vankó
Wigner Research Centre for Physics, Hungarian Academy of Sciences, Budapest, Hungary. E-mail: nemeth.z@wigner.mta.hu
First published on 11th April 2019
Aqueous solutions of the ternary system Ni(II)–EDTA–CN− are investigated with X-ray Absorption Spectroscopy (XAS) as a function of cyanide concentration with an enhanced laboratory von Hámos X-ray spectrometer. The near-edge structure of the spectra identifies unambiguously the formation of the pentacyanidonickel(II) complex at excess CN− concentrations. An analysis of the extended energy range of the XAS spectra reveals the molecular structure of the distinct molecular components present and provides a detailed description of the barely detectable mixed ligand [NiEDTA(CN)]3− complex. This thorough Extended X-ray Absorption Fine Structure (EXAFS) study demonstrates the potential of the emerging laboratory XAS spectrometers to become routine probes in various areas of chemistry, materials science, physics and related disciplines.
In recent years, however, attempts have been made to adapt state-of-the-art detector and analyser technologies to standard laboratory X-ray sources with the aim to build tabletop high energy resolution spectrometers.2–8 While these spectrometers cannot compete with the brilliance of large scale facilities, they proved to be quite applicable to routine X-ray Absorption Near Edge Structure (XANES)2,8,9 or X-ray emission spectroscopic5,7,8 investigations on condensed samples. These promising first results encourage us to widen the field of laboratory spectroscopies to include one of the most used X-ray techniques: Extended X-ray Absorption Fine Structure (EXAFS).1,10 EXAFS became a most fruitful method to describe molecular structure, providing atomic distances with the precision of an X-ray diffraction measurement but with the advantage of being an element selective local technique which is suitable for investigating non-crystalline samples including liquids. These properties make it an ideal laboratory characterization tool for researchers in chemistry, physics, environmental sciences, etc. and even for industrial applications. The demand for such a laboratory tool is indeed high and urgent: a first validation of laboratory EXAFS on nickel foil with a Johann-type spectrometer has just been published by Jahrman et al.8
The present work on a ternary Ni–EDTA–CN− solution series demonstrates that a thorough EXAFS study can indeed be performed with a laboratory X-ray source. The variation of the complex equilibria with the cyanide concentration was studied previously via laboratory XANES,9 and the results revealed the speciation of the Ni. However, the structure of the complexes could not be determined in the energy range available in a single spectrometer geometry. Also, the pure XANES signal from two intermediate complexes was not obtained with good enough accuracy. Now the system has been revisited using an improved laboratory spectrometer; two more concentration points were added, and the structural parameters of the species occurring in solution were evaluated. In particular, at moderate cyanide concentrations (0 ≤ cCN/cNi ≤ 4) the structure of the intermediate complex was determined to have a 1:
1 EDTA
:
CN− ligand ratio, where the in-bonding cyanide ligand displaces an EDTA oxygen donor atom from the nickel. Moreover, it is shown unambiguously that by increasing the cyanide ion concentration in the solution above a 4
:
1 ratio (which occurs at cCN = 1.0 M in our case) between the cyanide and the nickel cations, a pentacoordinated nickel cyanide compound starts to form alongside the co-existing tetracyanidonickel(II). Complemented with evaluated laboratory EXAFS spectra of some reference samples, this paper demonstrates clearly that by exploiting novel detectors and analysers, a reasonable fraction of common EXAFS experiments can be performed without the need to apply for synchrotron beamtime. As concentrated cyanide solutions are very poisonous, the local laboratory venue is better suited for this experiment from a safety point of view, since it allows us to eliminate the dangers that can occur during long distance transport of such materials, as well as accidental human exposure in the busy sample preparation laboratories and beamlines of large scale facilities. Working in the familiar environment of the home institution makes enforcing the necessary safety measures easier, and decreases the associated dangers.
![]() | ||
Fig. 1 EXAFS spectra of Ni metal foil (blue) and cc. NiSO4 solution (orange). Respective sample acquisition times: 6.5 h and 9 h. Left and right insets show the corresponding k- and R-space spectra, respectively, where the experimental data is plotted with dots and the fit is shown with lines. A synchrotron reference EXAFS scan is also plotted in light blue above the laboratory Ni metal spectrum.8 |
Parameter | Ni metal | cc. NiSO4 | ||
---|---|---|---|---|
Exp. | Ref. | Exp. | Ref. | |
Amplitude reduction factor | 0.54(7) | 0.51(4) | ||
Edge energy (eV) | 8332(1) | 8333 | 8343(2) | 8340.9823 |
First neighbour distance (Å) | 2.442(8) | 2.492 | 2.07(2) | 2.035 |
Debye–Waller factor (Å2) | 0.005(1) | 0.004(1) |
![]() | ||
Fig. 2 Molar distribution of the different nickel compounds in a Ni–EDTA–CN ternary solution as a function of cyanide concentration. The Ni2+ and EDTA concentrations are set to 0.25 and 0.3 M, respectively. The hereby used CN− concentrations are indicated with vertical grey lines. Data re-plotted from ref. 9. |
Fig. 3 shows the recorded EXAFS spectra for all samples of the studied ternary Ni–EDTA–CN− system with cyanide concentrations between 0 and 3 M. Typical total data acquisition times for these 250 mM Ni2+ solutions were a full day, except the last sample with cCN = 3 M, where a red layer precipitated on the sample holder's window in a few hours, making the transmission data unreliable. Thus, for this sample the measurement was repeated with a fresh sample (and a clean sample holder) three times with a total of 9 h of acquisition. Even with this strategy, it was not possible to reach good enough quality for EXAFS analysis, although the XANES part is satisfactory. This exemplifies well one of the limits of laboratory EXAFS experiments, arising due to their relatively long spectral acquisition times. However, we are in the process of developing appropriate flow cells to circumvent this obstacle.
The XANES spectral features (the spectra are replotted in the bottom panel of Fig. 3, while some parameters are shown in Fig. 4) follow nicely those reported in ref. 9. Thanks to the higher data quality here, even the shift of the edge position can be clearly tracked, and it matches excellently the evolution of the pre-edge peak. As discussed in ref. 9, the pre-edge peak arises almost exclusively from the square planar [Ni(CN)4]2− component, making it a potent indicator of this species. Apparently the edge positions follow very closely the trend set by the pre-edge intensities, which suggests that the electronic changes reflected by the edge shift are most dominantly connected to the tetracyanidonickel(II) component, too. However, for further discussion of this feature as well as for evaluation of the EXAFS data the contributions of the unique species have to be separated.
This is demonstrated in Fig. 5 for the most challenging case (due to precipitation, vide supra), where the XANES part of the spectrum of [Ni(CN)4]3− is shown. This spectrum has been reconstructed from the original data of solutions with cCN = 2 M and 3 M, where the presence of this species is non-negligible. Based on the area of the pre-edge, the ratio between the [Ni(CN)4]2− and [Ni(CN)5]3− complexes was determined, and a respectively weighted tetracyanido-spectrum was subtracted from the solution's spectrum, the remaining signal being that of the pentacyanidonickel(II) complex. The XANES data of the sample with cCN = 1 M (which is supposed to have almost exclusively the [Ni(CN)4]2− molecule) and the reconstructed data for the pentacyanido complex are then contrasted with the theoretically modelled spectra. The pre-edge region was modelled with time-dependent DFT (TD-DFT) calculations, while the full XANES spectra were calculated with FEFF (details of the calculations can be found in ref. 9, where a similar comparison has also been made; however, the improved data quality makes it worth reporting a new version of this figure). It is apparent in Fig. 5 that the experimental spectra follow nicely the calculated ones for both components, and their difference. Consequently, in addition to demonstrating the spectral separation method, these spectra also verify the formation of the pentacyanidonickel(II) complex at these high cyanide concentrations. The XAS spectrum of the pure [NiEDTA(CN)]3− species was obtained similarly from the data with lower cyanide concentration (0 M < cCN ≤ 0.5 M), where the [NiEDTA(CN)]3− component has about a 50% mole fraction.
The reconstructed spectra of the two species which do not occur solely (namely [NiEDTA(CN)]3− and [Ni(CN5)]3−) enable us to determine their absorption edge positions, as well. The main absorption edge of a 3d element stems from the excited electronic 1s–4p transition and it shifts in energy with the changes of the 3d valence electron orbital occupation due to its screening effect. Thus it is usually used to determine the oxidation state. Compared to the base [NiEDTA]2− molecule, the [NiEDTA(CN)]3− intermediate complex doesn't show any relevant variation in the edge position (both being at 8345.0 eV, derived from the maximum of the spectra's first derivative), while the homoleptic cyanido complexes, especially tetracyanidonickel(II), have considerably higher edge energies (8348.8 eV and 8347.0 eV for [Ni(CN4)]2− and [Ni(CN5)]3−, respectively). This is in good accordance with the assumed shorter metal–ligand distances in the case of the cyanido complexes, which will be quantitatively analysed in the following section.
In contrast to the reference Ni metal and NiSO4 samples, the lower signal-to-noise ratio due to the lower nickel concentration in the Ni–EDTA–CN system hinders the use of data at high wave numbers. Hence the Fourier-transform window was k = 3–10 Å−1. Still, the R-space EXAFS fits on both the first and second shell with an R window of 1–3 Å show good agreement with the DFT calculated structures for all three species (Fig. 6B). Amplitude reduction factors were set to 1, while the Debye–Waller factors vary between 0.003 and 0.006 Å2 for the single scattering paths. Coordination numbers were fixed. E0 values were fitted independently and were −2(1) eV, −2(1) eV and −4(2) eV for [NiEDTA]2−, [NiEDTA(CN)]3− and [Ni(CN)4]2−, respectively. The deviation of the atomic distances from the FEFF calculated values (ΔR) and the Debye–Waller factors (σ2) were fitted with one ΔR and one σ2 parameter for the first shell Ni–OEDTA and Ni–NEDTA bonds, one pair of ΔR and σ2 for the second shell Ni–CEDTA bonds and two individual ΔR and σ2 parameters for the Ni–CCN and Ni–NCN bonds. The resultant fitted bond lengths are summarized in Table 2. Multiple scattering paths with a FEFF calculated rank above 20 were included in the fits. Their path lengths were deduced from the parameters of the single scattering paths, while their Debye–Waller factors were approximated to be twice the corresponding single scattering path's σ2.
Distances/Å | [NiEDTA]2− | [NiEDTA(CN)]3− | [Ni(CN)4]2− | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Exp. | BP86 | B3LYP | Ref. | Exp. | BP86 | B3LYP | Exp. | BP86 | B3LYP | Ref. | |
Ni–OEDTA | 2.05(1) | 2.080 | 2.078 | 2.0527 | 2.04(1) | 2.118 | 2.106 | ||||
2.096 | 2.093 | 2.06529 | 2.08(1) | 2.165 | 2.148 | ||||||
2.05630 | |||||||||||
Ni–NEDTA | 2.07(1) | 2.103 | 2.129 | 2.07229 | 2.07(1) | 2.148 | 2.166 | ||||
2.07230 | 2.12(1) | 2.202 | 2.255 | ||||||||
Ni–CEDTA | 2.81(2) | 2.838 | 2.844 | 2.84(2) | 2.881 | 2.884 | |||||
2.85(2) | 2.876 | 2.871 | 2.8627 | 2.88(2) | 2.919 | 2.936 | |||||
2.90(2) | 2.929 | 2.943 | 2.91(2) | 2.953 | 2.990 | ||||||
Ni–CCN | 2.08(5) | 1.977 | 2.051 | 1.86(1) | 1.863 | 1.882 | 1.86(1)28 | ||||
Ni–NCN | 3.25(7) | 3.150 | 3.211 | 3.04(1) | 3.038 | 3.047 | 3.03(1)28 |
The [NiEDTA]2− data reflect quasi octahedrally coordinating 4 oxygen and 2 nitrogen atoms as first neighbours (first blue peak around 1–2 Å in the R-space spectrum in Fig. 6B), while the second shell consists of EDTA carbon atoms at three slightly different positions, as well as the multiple scattering paths (second blue peak above 2 Å in Fig. 6B). The reduced chi-square of this fit is 77. As for the [Ni(CN)4]2− molecule, the electron density distribution reflected by the spectrum shows very clearly the well known bonding structure of the cyanide ions: the four carbon atoms coordinate the nickel cation and the nitrogen atoms follow linearly. In order to gain an appropriate fit for the amplitudes, as well, the Ni–C–N–Ni forward scattering and Ni–C–N–C–Ni double forward scattering paths (the significant multiple scattering paths from the FEFF calculation) had to be taken into account. The reduced χ2 of this fit was found to be 39.7. The resulting distances agree fairly well with those from DFT modelling, particularly with the values obtained with the BP86 functional, although the B3LYP results are also satisfactory (see Table 2).
The most compelling result of this study is the structural description of the reconstructed intermediate complex between the more stable [NiEDTA]2− and [Ni(CN)4]2− molecules, which has never been directly determined before. As seen from the absence of any significant pre-edge features for this species, the coordination number of Ni must remain 6. The R-space spectrum resembles that of the [NiEDTA]2− complex, the only obvious difference being the increased amplitude of the second peak and its shift to higher R values. As described in ref. 31 and 9, the most likely scenario is that one cyanide ion bonds to the nickel thereby releasing one of the oxygen atoms of the EDTA ligand, resulting in the 1:
1 mixed ligand [NiEDTA(CN)]3− complex. According to this, a DFT optimized structure was calculated (Fig. 7) and used as an FEFF input file to determine the corresponding scattering paths. This model was fitted to the reconstructed [NiEDTA(CN)]3− EXAFS spectrum (green dots and line in Fig. 6B). The reconstructed spectrum shown with green dots in Fig. 6 is the average of the reproduced [NiEDTA(CN)]3− species signal from the recorded spectra with 0.15 ≤ cCN ≤ 0.75. The fit reproduces the spectrum satisfactorily (χred2 = 184), and provides accurate structural parameters (Table 2). As a consequence of the cyanide–OEDTA exchange, the remaining bonding of 3 oxygen and 2 nitrogen atoms of the EDTA ligand is best described with 2 sets of 2 different bond distances, as the octahedral symmetry is slightly distorted, while the carbon atoms at the second coordination shell are less affected. From the cyanide point of view, this ligand cannot bond so closely to the nickel as in the tetracyanidonickel(II) complex, as the spin multiplicity of the nickel ion is still triplet. It shows 0.2 Å longer Ni–CCN and corresponding Ni–NCN bond lengths compared to [Ni(CN)4]2−. This is in perfect agreement with the generally significantly larger ionic radius of Ni(II) in a six-coordinated environment compared to the square planar four-coordinated one, the Shannon effective ionic radii for Ni(II) in the two environments being 0.69 Å and 0.49 Å, respectively.32 DFT predicts a few percent elongation for the Ni–O and Ni–N bonds, which is observed only to a small extent in the experiment. However, the calculated Ni–CCN bonds are significantly larger than those in [Ni(CN)4]2−, and the B3LYP values agree rather well with the experiment. The DFT description of the second shell is again satisfactory.
![]() | ||
Fig. 7 DFT calculated structure of the [NiEDTA(CN)]3− molecular ion (Ni: green, N: blue, O: red, C: grey, H: white). |
The present work demonstrates the power of the recently developed laboratory XAS spectrometer prototypes as being capable of measuring synchrotron grade high quality hard X-ray XANES as well as EXAFS spectra of both solid and liquid samples. Laboratory XAS methods show incredible potential for extending the toolbox of researchers in the fields of physical chemistry and chemical physics or even industrial quality control by utilizing the power of high energy resolution X-ray absorption spectroscopies, while bypassing the necessity of applying for synchrotron beamtime and traveling to the synchrotron. They are also excellent when dealing with samples with safety concerns, as safety measures are easier to ensure inside an institute, where a smaller number of staff members need to be involved and made familiar with the possible dangers and necessary preventive procedures. Such instruments could be installed at any laboratory and used to study the electronic-, spin- and structural properties of transition metal based compounds with element selectivity in a condensed state.
Footnote |
† Present address: Department of Molecular Sciences, Swedish University of Agricultural Sciences, Uppsala, Sweden. |
This journal is © the Owner Societies 2019 |