Speciation Studies of Metals in Trace Concentrations - ACS Publications

Subscriber access provided by NEW YORK UNIV

Article

Speciation studies of metals in trace concentrations: The mononuclear uranyl(VI) hydroxo complexes Björn Drobot, Anne Bauer, Robin Steudtner, Satoru Tsushima, Frank Bok, Michael Patzschke, Johannes Raff, and Vinzenz Brendler Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.5b03958 • Publication Date (Web): 15 Mar 2016 Downloaded from http://pubs.acs.org on March 21, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Speciation studies of metals in trace concentrations: The mononuclear uranyl(VI) hydroxo complexes Björn Drobot,∗,† Anne Bauer,† Robin Steudtner,† Satoru Tsushima,† Frank Bok,† Michael Patzschke,† Johannes Ra,†,‡ and Vinzenz Brendler∗,† †Helmholtz-Zentrum Dresden-Rossendorf, Institute of Resource Ecology, Dresden ‡Helmholtz-Zentrum Dresden-Rossendorf, Helmholtz Institute Freiberg for Resource Technology,

Dresden

E-mail: [email protected]; [email protected]

Abstract A direct luminescence spectroscopic experimental setup for the determination of complex stability constants of mononuclear uranyl(VI) hydrolysis species is presented. The occurrence of polynuclear species is prevented using a low uranyl(VI) concentration of 10 -8 M (2.4 ppb). Time-resolved laser-induced uorescence spectra were recorded in the pH range from 3 to 10.5. Deconvolution with parallel factor analysis (PARAFAC) resulted in three hydrolysis complexes. A tentative assignment was based on thermodynamic calculations: UO 22+, UO 2(OH) +, UO 2(OH) 2, UO 2(OH) 3. An implementation of a Newton-Raphson algorithm into PARAFAC allowed a direct extraction of complex stability constants during deconvolution yielding log( β1 M, 1 ◦ C )1:1 = -4.6, log( β1 M, 1 ◦ C )1:2 = -12.2, log( β1 M, 1 ◦ C )1:3 = -22.3. Extrapolation to standard conditions gave: log( β 0 )1:1 = -3.9, log( β 0 )1:2 = -10.9 and log( β 0 )1:3 = -20.7. Luminescence characteristics (band position, lifetime) of the individual mononuclear hydroxo species were derived to serve as a reference data set for further investigations. A correlation of luminescence spectroscopic features with Raman frequencies was demonstrated for the mononuclear uranyl(VI) hydroxo complexes for the rst time. Thereby a signal-to-structure correlation was achieved and the complex assignment validated.

Introduction

necessary to predict migration patterns, namely for long term safety predictions and reliable risk assessment. In the present work we focus on uranyl(VI) hydrolysis as the fundamental system for aqueous uranyl(VI) chemistry. A precise knowledge of this system is indispensable for the understanding of more complex multiligand systems. The condence of any predictive calculations of geochemical speciation and associated reactive transport correlates with the accuracy of complex stability constants. In the case of uranium these constants usually have been derived from potentiometric and calorimetric titrations or liquid-liquid-extraction. 17,18 Metal concentrations above 10-5 M are needed to perform accurate titrations, imposing limitations by solubility, e.g. formation of meta-schoepite. 19 Furthermore, polynuclear species occur in this concentration range. 20,21 Up to now, the number, stoichiometry and structure of relevant hydrolysis species is still under discussion. 2224 Unfortunately, most of the methods for structure elucidation also require relatively high concentrations. 25 Electrospray ionization with time-of-ight mass spectrometry is one of the promising tools for determination of stoichiometry coecients as previously shown. 2631 However, this method cannot detect neutral species and it is impossible to measure positively and negatively charged complexes in the same experiment. Further speciation studies have been performed using vibrational spec-

The focus on uranyl(VI) chemistry originates not only from its rich speciation patterns but also from the potential hazards associated to uranium. In the past decades handling of uranium for military and civilian purpose has left huge uranium mining piles 1 and contaminated sites 24 and raises the question of nuclear waste disposal. Eorts are being made to deal with those legacies worldwide. The risk emanating from uranium is linked to its radioactivity as well as chemotoxicity. For the most abundant natural isotope ( 238 U) this chemotoxicity is much more detrimental than its radiotoxicity. 5,6 Thus, uranyl(VI) speciation is of high relevance not only with respect to nuclear power 7 but also to health care, as it governs its migration as well as bioavailability. Even if precipitation of uranyl(VI) limits its mobility, in diluted aqueous solutions it can possibly penetrate geothechnological barriers of a wast disposal 810 and reach the biosphere. Interaction studies of uranyl(VI) with biomolecules has demonstrated a high anity for respective ligands, 1115 and complexation with proteins can interfere or change their functionality. 16 Therefore, even submillimolar concentrations of uranyl(VI) may have a huge environmental impact. Accurate thermodynamic constants based on appropriate experiments in the low concentration range are therefore

ACS Paragon Plus Environment

1

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

troscopy (e.g. infrared 32 and Raman spectroscopy 33 or infrared multiphoton dissociation 34 ) as well as Xray spectroscopies (e.g. high-energy X-ray scattering 35 and soft X-ray photoelectron spectroscopy 36 ). However, trace metal concentrations below 10 -6 M are not accessible for any of these methods. Luminescence spectroscopy is another widely used tool for uranyl(VI) hydrolysis 3739 and uranyl(VI) speciation in general. 4042 Quenching by ligands can limit its eld of application. Therefore, cryo measurements are often used, to overcome this drawback. Its high sensitivity enables luminescence studies below saturation; however, data deconvolution as well as signal-tostructure correlation is not straightforward. Parallel factor analysis (PARAFAC) was demonstrated to be a useful tool for luminescence data deconvolution. 39,43,44 We have therefore decided to study the uranyl(VI) hydrolysis by combining time-resolved laser-induced uorescence spectroscopy (TRLFS) and PARAFAC. The experiments were performed at a uranyl(VI) concentration of 10-8 M where the calculated model (data from the OECD/NEA Thermochemical Database, the NEATDB 18 ) predicts the absence of polynuclear species (see Figure S1). In this simplest system, extraction of thermodynamic constants of mononuclear uranyl(VI) hydrolysis species can be achieved. Furthermore, luminescence and Raman (as a directly complementary technique) spectra have been analyzed aiming at a structure-to-signal correlation (linking bond strength) as they both are sensitive to symmetrical vibrational levels of uranyl(VI) complexes.

Page 2 of 11

uum). The wavelength was chosen to eectively excite the uranyl(VI) species via equatorial ligand-to-metal charge transfer. 39 The laser energy was tuned to an average of 1 mJ per 5 ns pulse. Small variations (< 2%) were readjusted between all measurements. An ICCDcamera (ICCD- 3000, HORIBA) was used to record time-resolved uorescence spectra. The slit width of the monochromator (iHR 550, Horiba, 100 lines/mm grid) front entrance was set to 2000 µm to increase the signal (by accepting sparse spectral smearing). The spectral resolution of the data is 0.46 nm. A transformation into the wavenumber space yields a resolution of ∼22 cm-1 (at 460 nm) to ∼13 cm-1 (at 600 nm). Appropriate weighting of fast and slow luminescence decays within single TRLFS experiments was ensured by using a dynamic step size for the delay time shifts. ti = t0 + tmin · i +

i4 30000

(1)

t0 =initial oset to laser pulse, set to 200 ns

tmin =minimum delay shift, set to 5 ns i =number of step in series

To guarantee equal weighting in the deconvolution single TRLFS records were normalized to the same number of photons. A recorded background was used for baseline correction. Stacking of TRLFS data gave a 22 x 30 x 360 data array (pH x time x wavelength)

Data processing

Material

Parallel factor analysis was performed using the N-way toolbox 47,48 for Matlab R2013a with some modications (see ow chart in Figure S2). A monoexponential constraint was implemented (using the Optimization Toolbox) for the extracted lifetimes as previously described. 39 Furthermore, component distribution was constrained to conrm speciation. For that purpose an adopted Newton-Raphson algorithm, proposed by Carrayrou et al., 49 was used. The basic algorithm was implemented for Matlab by Smith. 50 This le is intended for solving chemical equilibrium problems for known stoichiometries and stability constants. We modied it in a way that it ts stability constants to a given distribution. PARAFAC is an iterative algorithm; after 25 iterations optimized PARAFAC distributions were passed on to the Newton-Raphson algorithm and complex stability constants were tted. Afterwards, the corresponding speciation problem was solved. These results were then passed back to PARAFAC for the next iteration step until the convergence criterion was reached. Based on core consistency (PARAFAC, Figure S3) and thermodynamics, a four component model was chosen: 1:0 aquo ion (UO22+), 1:1 complex (UO2(OH)+), 1:2 complex (UO2(OH)2), 1:3 complex (UO2(OH)3), where for the sake of clarity, all species formulas are written without equatorial coordinating water (Figure 1). Conver-

Uranyl(VI) stock solutions were prepared as previously described. 45 Sample preparation was carried out in an inert gas glove box (1 atm N 2 ) to avoid carbonate in the system. Uranyl(VI) concentrations were xed to 1.0×10−8 M and a constant background electrolyte concentration of 1 M NaClO 4 was used in all experiments. The pH (3 to 10.5) was adjusted (0.1 M NaOH / 0.1 M HClO4) using a Metrohm Titrator (Titrando) equipped with a Metrohm double junction pH electrode (3 M KCl / 1 M NaClO4) to avoid chloride ux in the samples. A three-point calibration (at 1 ◦ C) with NIST buers was routinely performed before pH adjustment. A correction of the measured pH was performed using a factor of 0.23. 46 Samples were measured at 1 ◦ C in a capped quartz glass cuvette to preserve the nitrogen atmosphere.

Methods Time-resolved laser-induced uorescence spectroscopy (TRLFS) TRLFS measurements were performed using the fourth harmonic of a Nd:YAG laser (266 nm, Minilite, Contin-

ACS Paragon Plus Environment

2

Page 3 of 11

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

1: 0-aquoi on

1: 1-hydr oxo compl ex Figure 2: Franck Condon scheme. Shown are electronic transitions ( T1 ν0 → S0 νn and T1 ν1 → S0 νn ) corresponding to luminescence. Transitions from the vibrational ground state (green) are responsible for the main luminescence features. Occupation of excited state higher vibrational levels is described by the Boltzmann equation and thus depends on the temperature (see Figure S7). Transition from those levels to the ground state is responsible for hot bands (red). This illustrates that luminescence features are separated by energies equal to the vibrational frequencies of a molecule.

1: 2-hydr oxo compl ex

1: 3-hydr oxo compl ex

signal-to-noise ratio in the spectra and can be adopted by the user (Figure S6). Spectra were transformed into wavenumber space to account for the correlation of energy and luminescence wavelength. Peak positions were estimated using the third derivative of the spectra and selected by intensity (> 2.5 % of maximum intensity) and distances (gap between peaks > 300 cm -1 ). Peak amplitude as well as hot-band ratio (emission from the rst vibrational level of the excited state, 51 see Figure 2) were estimated from spectra intensities at the corresponding peak position. The ratio of Lorentzian of the pseudo-Voigt prole was initially set to 0.9 (Gaussian 0.1) and the prole width to 230 cm -1 . Those parameters (initial guesses) were optimized using the Levenberg-Marquardt solver from the Optimization toolbox for Matlab. Some additional constrains were implemented. The gaps between two peaks were restricted to decrease with energy (smaller gaps between peaks of lower energy). This reects smaller energetic dierences between higher vibrational levels (anharmonic oscillator; see Figure 2). The hot band gaps are the same as the main peak gaps (in cm -1 ). No parameter was xed and even initial guesses were automatically estimated (to ensure objectivity). Peak positions (energies of emissive electronic transitions) as well as peak gaps (energy dierences of vibrational ground state levels / Raman frequencies) were thereby determined.

Figure 1: Structures of the uranyl(VI) aquo ion (1:0) and mononuclear hydroxo complexes with formulas (UO 2)1(OH)n 2n (n=1,2,3). Assuming an coordination number of 5, structures were optimized using density functional theory (DFT) according to Ref. 39.

gence criteria of 10−6 for PARAFAC and 10−11 for the speciation constrain were used. After deconvolution 98.7 % of the data variance is explained. The remaining error (dierence between raw data and reconstructed data) was below 15 % and unsystematic (Figure S4). Thus, PARAFAC data processing delivered the following information: Speciation (including complex stability constants), individual emission spectra and luminescence decays.

Luminescence spectra tting After PARAFAC deconvolution, in a separate step, luminescence spectra tting followed, i.e. the deconvoluP tion of several transitions ( T1 ν0 → P of a sum P spectrumP S0 νn + T1 ν1 → S0 νn ) into signals of individual transitions ( T ν → S 1 0 P P P 0 ν0 , ..., T1 ν0 → S0 νn , T1 ν1 → S0 ν0 , ..., T1 ν1 → S0 νn ), see Figure 2). For that purpose an automated algorithm was newly developed (see ow chart in Figure S5). Initial guesses were derived by the following procedure. First raw emission spectra were denoised using fast Fourier transformation. The best degree of smoothing depends on the

Speciation modeling and parameter extrapolation to standard conditions Before usage in common geochemical speciation codes, e.g. PHREEQC, 52 EQ3/6 53 or Geochemist's Work-

ACS Paragon Plus Environment

3

Analytical Chemistry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

bench, 54 log(β ) values obtained in this work were converted to standard state (T = 298.15 K, I = 0 M). To be consistent with the NEA-TDB, 17,18 the specic ioninteraction Theory (SIT) 55 was used to calculate the log(β ) values at innite dilution. However, there is hardly any data published so far to describe the temperature dependency of the SIT ion-ion-interaction parameters (). Therefore, the experimentally determined stability constants were rst extrapolated to 298.15 K before converting them to innite dilution. For the temperature correction, the integrated Van't-Ho expression (2) was found to be adequate within the considered temperature range.

the conclusion from Moriyasu et al. 58 who proposed a dependence of uranyl(VI) luminescence decay on pH but not on ionic strength. Instead, a clear correlation of quantum yield as well as lifetime of the uranyl(VI) aquo ion on ionic strength is proved while the lifetime itself is unaected by the pH. Thus temperature and ionic strength were both optimized to improve the signal quality. Species distribution as a function of pH was derived using the PARAFAC deconvolution algorithm described above. In Figure 3, the results from the last PARAFAC iteration (symbols) are compared with species calculation (lines), using the best t log( β1 M,1 ◦ C ). The deviation is marginal which correlates with the explained raw data variance (98.7 %). Multiple deconvolution runs with randomized starting vectors result in slightly dierent constants. The declared uncertainty is the standard deviation of 15 individual PARAFAC runs. Stability constants were extrapolated to standard conditions (T = 25 ◦ C, innite dilution) and summarized in Table 1. Notably the temperature correction had the greatest impact on the extrapolation. Such dependence of log( β ) on temperature was previously described for the uranyl(VI) hydrolysis. 59,60 To our surprise, the values for the 1:1 and 1:2 hydrolysis species are remarkably higher than given in the NEA-TDB update issued 2003. 18 The eect of this deviation is shown in Figure S10. To verify our results, ve independent experiments (one UV-vis, see Figure S11 and four luminescence measurements, see Figure S12) were analysed using the method described here. The uranyl(VI) concentration diers in these studies which accounts for a speciation with a multitude of possible polynuclear complexes. Therefore, we only focused on the range up to pH 6.5. In this range the aquo ion, 1:1 and 3:5 complex are present in the samples with 10 −5 M U(VI) 39 while the NEA predict the dominance of the dimeric 2:2 complex over the 1:1 complex for Uranyl(VI) concentrations ≥ 10−4 M. Respective complex stability constants extrapolated to standard conditions gave average values of log(β 0 )1:1 = -4.2 ± 0.3, log(β 0 )2:2 = -4.7 and log(β 0 )3:5 = −13.5 ± 0.3. The raw values are given in the SI. The log(β 0 )1:1 derived from these experiments is within the error margins of the value of this work (see Table 1 log(β 0 )-this work) and therefore consistent. Moreover, the extrapolated values of the 2:2 and 3:5 complex clearly dier from the NEA values. Our data were well explained with the mentioned set of complexes. This limited set might be one reason for the discrepancies, since additional complexes would eect the values of the detected species. A dierent chemical equilibrium in the excited state could be an other explanation for the divergence of our values and the NEA-TDB values. Although the results from UV-vis are coherent with the luminescence spectroscopic ndings, the comparability of both is limited by the experimental conditions. Studies are in progress to further clarify this. Aside from goodness of t, the quality of thermo-

log(βT1 ) − log(βT2 ) =−

1 1 ∆r H 0 ( − ) R · ln(10) T1 T2

(2)

R =gas constant

T =temperature (K)

∆r H 0 =standard molar enthalpy of reaction

Within the small temperature shift (24 K) the standard molar enthalpy of reaction was considered to be constant. 56 ∆r H 0 values were taken from Eliet et al.. 37 To conrm the NEA-TDB, a conversion from molarity to molality according to NEA 17,18 was performed. Then, the obtained log(β ) values for the hydrolysis complex formation reactions (3) were converted to zero ionic strength using the SIT equation (4) UO22+ + nH2O ↔ UO2(OH)2−n + nH+ n

(3)

log(β 0 ) =log(βI ) − ∆(z 2 ) · D + ∆ · Im − nH + · log(aH2 O )

Page 4 of 11

(4)

z =charge number

D =Debye-Hückel term

 =ion-ion interaction parameters

Im =ionic strength (mol/kg)

nH + =number of released protons

aH2 O =water activity

The calculated log(β ) values at standard state are given in Table 1. All , ∆r H 0 and the Debye-Hückel parameter values used for the conversion to standard state are given in the supporting information (Table S1).

Results and discussion To increase the quantum yield, all TRLFS measurements were carried out at a lower temperature of 1 ◦ C 39,57 (see Figure S8) and a high ionic strength of 1 M NaClO4 (see Figure S9). Our results did not reproduce

ACS Paragon Plus Environment

4

Page 5 of 11

Analytical Chemistry 2+

+

UO

UO (OH)

hydrolysis species

(M)

2

Concentration of uranyl(VI)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

UO (OH)

2

2

2

UO (OH) 2

-

3

-8

1,0x10

-9

5,0x10

0,0 2

4

6

8

10

pH

Figure 3: Experimental species distribution (symbols) and calculated speciation with the extracted stability constants (lines). Conditions: 10 −8 M U(VI), 1 M NaClO 4, 1 ◦ C, nitrogen atmosphere. Table 1: Comparison of determined, extrapolated and previously reported complex stability constants. A graphical illustration of the dierences is given in Figure S10. ∗ no error for ∆r H01:2 available for consideration.

log(β 0 )

log(β 0 )

NEA-TDB

NEA-TDB

2003 18

1992 17

log(β1 M,1 ◦ C )

log(β )

this work

this work

UO22+ + H2O ↔ (UO2)(OH)+ + H+

-4.6 ± 0.3

-3.9 ± 0.5

-5.25 ± 0.24

-5.2 ± 0.3

UO22+ + 2H2O ↔ (UO2)(OH)2 + 2H+

-12.2 ± 0.5

-10.9 ± 0.5∗

-12.15 ± 0.07

≤-10.3

UO22+ + 3H2O ↔ (UO2)(OH)3 + 3H+

-22.3 ± 0.6

-20.7 ± 0.7

-20.25 ± 0.42

-19.2 ± 0.4

Reaction

0

solid samples. 6267 A transfer of this knowledge to systems at moderate temperatures is challenging and is therefore still missing. The approach proposed here attempts to improve this situation. Emission spectra extracted from PARAFAC were tted using a pseudoVoigt peak shape and included hot bands. The underlying assumption of a hot band transition is supported by the dierences of uranyl(VI) aquo ion emission spectra recorded at 25 ◦ C and -120 ◦ C, respectively (Figure S7). The individual spectra were compared to deconvoluted (single pseudo-Voigt shaped peaks, hot bands, and the sum of them) in Figure 4. The minor deviations illustrate the validity of this approach. The position of the rst main peak shifts to lower energies with increasing hydrolysis level (Figure 5, for a summary of spectra characteristics see Table 2). This indicates that complexation with hydroxide ions lowers the energy gap between the triplet and ground state. Moulin et al. proposed the same trend, but their data are contradictory for the 1:2 complex. 68 Figure 5 additionally illustrates a decreasing energy gap between the main peaks with increasing hydrolysis. According to Figure 2, this gap corresponds to the symmetric stretching vibrational energies and pro-

dynamic constants strongly depends on the chosen model. Therefore, an independent proof of estimated stoichiometries is indispensable. Unfortunately there is no straightforward way provided by uranyl(VI) luminescence spectroscopy. Since luminescence lifetime is inuenced by the experimental conditions (ionic strength / temperature) it is only useful for relative comparison and data deconvolution. The lifetimes extracted here are: τ1:0 = 10.2 ± 0.2 µs, τ1:1 = 44.9 ± 2.4 µs, τ1:2 = 78.9 ± 6.8 µs and τ1:3 = 3.4 ± 0.2 µs. A direct comparison with literature is not possible, as similar setups were not described. With increasing hydrolysis we previously reported an initial increase (aquo ion compared to 1:1 complex) followed by a decrease (1:1 compared to 1:3) of the lifetimes. 39 The same trend is reproduced here. We nd the longest lifetime for the uncharged 1:2 complex. Therefore extracted luminescence lifetimes could be seen as a rst indication of correct stoichiometry. With their sensitivity towards complexation, it is clear that emission spectra should reect structural information. For spectra recorded at low temperature (