Carbonate Complexation of Mn²⁺ in Aqueous Phase

Abstract

The chemical speciation of Mn²⁺ within cells is critical for its transport, availability and redox properties. Herein we investigate the redox behavior and complexation equilibria of Mn²⁺ in aqueous solutions of bicarbonate by voltametry and electron paramagnetic resonance (EPR) spectroscopy, and discuss the implications for the uptake of Mn²⁺ by mangano-cluster enzymes like photosystem II (PSII). Both the electrochemical reduction of Mn²⁺ to Mn⁰ at an Hg electrode and EPR (in the absence of a polarizing electrode), revealed formation of 1:1 and 1:2 Mn-(bi)carbonate complexes as a function of Mn²⁺ and bicarbonate concentrations. Pulsed EPR spectroscopy, including ENDOR, ESEEM and 2D-HYSCORE, were used to probe the hyperfine couplings to ¹H and ¹³C nuclei of the ligand(s) bound to Mn²⁺. For the 1:2 complex the complete ¹³C hyperfine tensor for one of the (bi)carbonate ligands was determined and it was established that this ligand coordinates to Mn²⁺ in bidentate mode with ¹³C-Mn distance of 2.85 ± 0.1 Å. The second (bi)carbonate ligand in the 1:2 complex coordinates possibly in monodentate mode, which is structurally less defined, and its ¹³C signal is broad and unobservable. ¹H ENDOR reveals that 1-2 water ligands are lost upon binding of one bicarbonate ion in the 1:1 complex while 3-4 water ligands are lost upon forming the 1:2 complex. Thus, we deduce that the dominant species above 0.1 M bicarbonate concentration is the 1:2 complex, [Mn(CO₃)(HCO₃)(OH₂)₃]^-.

I. Introduction

The bio-availability of essential trace metal ions (Fe²⁺, Mn²⁺, Cu²⁺, Zn²⁺, Cd²⁺ and Ni²⁺) in ecosystems is governed by their complexation equilibria with naturally occurring ligands that exist in open oceans and at soil-water interfaces¹. Inorganic ligands like carbonate/bicarbonate ions, derived from CO₂ and soluble carbonate minerals, play a dominant role in metal ion speciation in marine ecosystems. This speciation leads to significant changes in the redox potentials of metal ions and thus, influences the bio-geochemical cycles of elements on a macroscopic scale. The carbonate system (CO₂/HCO₃^-/CO₃^2-) is particularly versatile in this regard for several reasons. First, it readily chelates metal ions and also forms μ-carbonato-bridges leading to oligomeric clusters M_x(CO₃)_y in solution that are in equilibrium with insoluble mineral forms for example siderite (FeCO₃), rhodochrosite (MnCO₃) etc². Secondly, the amphoteric properties of bicarbonate (pK_a = 6.3, 10.5) straddle the pH of the oceans which enables bicarbonate to drive the metal ion speciation by serving as either an acid or a base. The metal-ligand speciation go together with the changes in redox potential which is exploited by organisms for the selective uptake, intracellular delivery and redox poising of metal ions within functional sites in macromolecules³.

An example of such a macromolecule that uses bicarbonate for metal cluster assembly is photosystem II (PSII), a multi-subunit membrane-spanning protein complex required for oxygenic photosynthesis^4,5. A domain within the PSII, known as the Water Oxidizing Complex (WOC), harbors an inorganic Mn-based catalyst that oxidizes water to dioxygen. Early biochemical studies (reviewed in ⁶) had indicated that bicarbonate is involved in functioning and/or assembly of the WOC inorganic core. In last decade, several types of studies have indicated a specific role for bicarbonate in functioning of the WOC. The rate of electron transfer from Mn²⁺ ion to apo-PSII is accelerated in presence of bicarbonate⁷. Bicarbonate increases the affinity of the Mn²⁺ to apo-PSII by forming a ternary complex, and thus accelerates the rate of light driven assembly of the inorganic core during its biogenesis or repair, a process known as photoactivation⁸. A recent crystal structure of PSII-WOC from T. elongutus at 3.5 Å resolution has proposed the atomic structure to consist of an Mn₄CaO₄ core, having four oxo/hydroxo/aquo ligands bridging the metals to form a cubical Mn₃CaO₄ subcore that is bridged to the fourth external Mn atom at one of the corner oxos⁴, although this model remains debated^9,10. The XRD data provides evidence for the presence of a planar ligand in the active site, suggested to be HCO₃^-/CO₃^2- or NO₃^- that is positioned between Ca and the external Mn atom. Site directed mutagenesis of the protein residue that defines the binding site has provided some support for a possible functional bicarbonate at this location.^4,11 Geochemical evidence has led to the proposal that Mn-bicarbonate clusters may have played the seminal role in the evolution of the first oxygenic phototroph from an ancestral anoxygenic bacterial precursor^12,13. This hypothesis has been supported by evidence from site-directed mutants of non-oxygenic purple bacterial reaction center which are engineered to photo-oxidize Mn²⁺ when bicarbonate is added to the media^14,15.

The stoichiometry and stability constants for the formation of Mn²⁺-bicarbonate complexes have been studied previously by pH potentiometric titrations as a function of bicarbonate concentration. In these measurements the molecularity of the complexes were assumed based on simple chemical principles of charge balance. Mn-bicarbonate complexes have also been the subject of scrutiny by biologists owing to their efficiency in catalyzing the dismutation of hydrogen peroxide, also known as its ‘pseudo-catalase’ activity^16,17. The observed pseudo-catalase rates have been used as an indirect probe for the molecularity of these complexes. On this basis it has been reported that these complexes can form dimers in solution with bicarbonate, although no direct structural evidence has been presented^18-20. Previously we have reported direct electrochemical measurements of the oxidation potential for the conversion of Mn²⁺ to Mn³⁺ in the presence of bicarbonate²¹. This suggested the presence of di-manganese or oligomeric forms in solution. In none of the mentioned studies, structure sensitive spectroscopic methods were used to determine the speciation of Mn²⁺ in bicarbonate solution.

Herein, we characterize the equilibrium constants and chemical structures of Mn²⁺-bicarbonate complexes that form in aqueous solution using EPR and potentiometric titration. Cyclic voltammetry on reduction of Mn²⁺ at varied bicarbonate concentrations is used to determine metal:ligand stoichiometry of the Mn²⁺-bicarbonate complexes. EPR spectroscopy is used to confirm the ligand stoichiometry and to measure the ligand field symmetry of the Mn²⁺ complexes^22,23. Hyperfine couplings to ligand nuclei (¹³C and ¹H) obtained from ENDOR and ESEEM provides us the information about coordination geometry and distances to the ligands^24-26.

II. Experimental Section

Solutions of Mn²⁺ and NaHCO₃ were prepared by mixing stock solutions of 10 mM MnClO₄·6H₂O (Sigma) and freshly prepared 10-300 mM NaHCO₃ (Sigma-Aldrich). NaH¹³CO₃ (99% pure) powder from Cambridge Isotopes Laboratory Inc. was used to prepare samples with ¹³C-labeled bicarbonate. In all the preparations, the pH of the solution was 8.3 (self buffered by NaHCO₃).

Mn²⁺-bicarbonate solutions are quasi-stable at room temperature, with a white precipitate slowly forming upon standing due to formation of insoluble MnCO_3(s). At low Mn²⁺ concentrations (<0.25 mM) used in our voltammetry experiments, a visible precipitate developed within 2-3 hours, time sufficient to perform the experiment. Freshly mixed solutions were always used in these experiments. At high concentrations of Mn²⁺ (1 mM) and HCO₃^- (200-300 mM) as used in room temperature EPR experiments, the precipitate forms within minutes, i.e. on the time scale of the experiment. Thus, the Mn²⁺ EPR signal was recorded at different time points after sample mixing and, the signal at time point zero (i.e. immediately after mixing and before precipitation starts) was reconstructed by back extrapolation.

For frozen solution EPR experiments, the samples were prepared in (1:3) CH₃OH/H₂O solvent mixture²⁷, with methanol added as a glass-forming agent.^a Stock solutions of MnClO₄.6H₂O and NaHCO₃ in (1:3) CH₃OH/H₂O were mixed and immediately frozen in liquid N₂ to avoid any loss of Mn²⁺ due to precipitation. KHCO₃ (in 1:3 CH₃OH/H₂O) was also tested instead of NaHCO₃ to prepare samples, although due to poor glass formation upon freezing it could not be used for any quantitative measurements. Calibrated EPR tubes and identical volumes of the samples were used to allow quantitative comparison. In separate titration experiments (see Figure S3 in Supporting Information) we confirmed that addition of methanol and freezing to low temperatures do not significantly alter the Mn²⁺ speciation equilibrium from that observed at room temperature.

Voltammetry experiments were done with a standard potentiostat (polarographic analyzer model PA-3; LP, Praha) using a three-electrode cell as described earlier²⁸. In Mn²⁺ reduction experiments, the working electrode was a hanging mercury drop with a diameter of 0.72 mm which was renewed before each measurement. A Pt plate (1×1 cm) was used as a counter electrode, while a saturated calomel electrode formed the reference electrode. The measured wave potentials were then recalculated versus the normal hydrogen electrode (NHE). Scan rate was 50 mV/sec.

CW EPR spectra were collected using a Bruker ESP300e spectrometer. Room temperature aqueous solution spectra were measured using a flat cell in a standard TM102 cavity at microwave (MW) frequency 9.69 GHz. Frozen solution experiments (77 K) were done using a liquid N₂ finger dewar in a TE011 cavity at 9.14 GHz. The MW power setting was 0.2 mW, modulation amplitude 5-10 G, and modulation frequency 100 KHz.

Pulsed EPR experiments were performed using an X-band Bruker Elexsys580 spectrometer. The sample temperature was set in the range 5-10 K using a liquid helium flow cryostat CF935 from Oxford Inst. In all the Electron Spin Echo (ESE) based experiments durations of the π/2 and π pulses were set to 16 and 32 ns, respectively. Field-sweep ESE-detected EPR spectra were collected using 2-pulse echo sequence, π/2 - τ - π - τ - echo, with fixed inter-pulse delay τ = 200 ns; the integrated echo intensity was detected while sweeping the magnetic field. For 2-pulse Electron Spin Echo Envelope Modulation (ESEEM) experiments, π/2 - τ - π - τ - echo, the echo intensity was measured as a function of τ, starting from initial 100 ns and incrementing with 16 ns step. 2D HYSCORE experiment, π/2 - τ - π/2 - t₁ - π - t₂ - π/2 - τ - echo, was recorded at fixed τ = 176 ns and with t₁ and t₂ incremented independently from 200 ns by 24 ns steps. For Mims ENDOR experiments, the stimulated echo pulse sequence, π/2 - τ - π/2 - T - π/2 - τ - echo, was used with non-selective microwave pulses (16 ns) and with τ = 200 ns. A selective RF pulse (50 μs) of variable frequency was applied during a time interval T. Whenever necessary, phase cycling was done to eliminate unwanted echo signals.

Data processing and analysis was done with WINEPR software (Bruker BioSpin) and MATLAB 6.0 (MathWorks Inc.). Prior to Fourier Transformation the 1D ESEEM time-domains were normalized to unity background by fitting the relaxation decays with a smoothing spline function followed by division of the experimental time-domain by the fitted baseline decay. The 2D HYSCORE time-domain traces were baseline corrected in two dimensions with a suitable polynomial function. Appodization with a Hamming window and zero filling was performed prior to Fourier Transformation (FT). All the 1D frequency-domain ESEEM spectra shown are the phase-corrected cosine part of the FT. The 2D HYSCORE spectrum is presented in absolute mode after 2D-FT. The ENDOR spectra shown were normalized to the echo signal intensity in the absence of radio frequency.

EPR theory and simulations

Mn²⁺ has 5 unpaired electrons (3d⁵) and its electronic ground state in octahedral ligand field has the high spin configuration (⁶S) with electron spin S=5/2. The spin-Hamiltonian for Mn²⁺ is represented in the form

$$H=g\beta B_0{S}_Z+\overrightarrow{S}\cdot \widehat{D}\cdot \overrightarrow{S}+A_{Mn}\cdot \overrightarrow{S}\cdot \overrightarrow{I}+\sum _i(\overrightarrow{S}\cdot {\widehat{A}}_{\left(i\right)}\cdot {\overrightarrow{I}}^{\left(i\right)}+g_n{\beta }_n{B}_0{I}_Z^{\left(i\right)}).$$ (1)

Here the first term describes the electron Zeeman interaction; g is electron g-factor, β the Bohr magneton, B₀ the magnetic field applied along Z-axis in the laboratory coordinate system. The g-factor is assumed to be isotropic which is a good approximation for Mn²⁺ in octahedral ligand field at X-band frequency. The second term is a zero-field splitting (ZFS) interaction described by a traceless tensor $\widehat{D}$. It results from spin-orbit interaction of the ⁶S ground state with excited electronic states, and its magnitude and symmetry depends on both the strength and symmetry of the ligand field around Mn²⁺. The tensor D is usually described by two parameters D and E representing the axial and rhombic ZFS components, respectively. The third term is the magnetic hyperfine interaction with the ⁵⁵Mn nuclear spin (I=5/2); since Mn²⁺ has half-filled 3d⁵ configuration, the hyperfine coupling A_Mn is isotropic to a good approximation. The last term represents the hyperfine interaction ${\widehat{A}}_{\left(i\right)}$ and the nuclear Zeeman interaction (g_n is nuclear g-factor, β_n the nuclear magneton) summed over all magnetic nuclei (i) in the coordination sphere of Mn²⁺. In our case, the possible ligands to Mn²⁺ are HCO₃^-, CO₃^2- and H₂O.

All EPR experiments described in this work were analyzed using the spin-Hamiltonian (1), however not all its terms are equally important in describing each experiment. Since the electron Zeeman interaction ≫ ($\widehat{D}$ and A_Mn) ≫ (${\widehat{A}}_{\left(i\right)}$ and nuclear Zeeman interactions), a perturbation approach can be conveniently applied. The field-sweep EPR spectra can be fully described by the first three terms of the spin-Hamiltonian, and accordingly only the electron g-tensor, ZFS coupling and ⁵⁵Mn hyperfine parameters are obtainable from these spectra. The weak hyperfine couplings with ligand nuclei are usually unresolved in the EPR spectra and contribute only to the EPR linewidth.

ESEEM and ENDOR experiments were performed to resolve weak hyperfine couplings from ligand nuclei. Recent work by Astashkin et al. has considered the case of high spin systems interacting with a nuclear spin I=1/2²⁹. It has been demonstrated that the position and intensity of the nuclear transitions in ESEEM/ENDOR spectra depend mostly on the ZFS interaction and the ligand hyperfine coupling. The position of the sum combination line (ν_σ) in ESEEM spectra has been found to be most sensitive and described by the product of two parameters: the ZFS coupling D and the anisotropic part of hyperfine coupling tensor T. Thus, D and T can in principle be derived from the observed shift in ν_σ, and the isotropic part of the hyperfine tensor can then be estimated from the position of other lines in the spectra²⁹. We found in practice that situation is more complex, and we were only able to extract these parameters after analyzing the correlation peaks in the 2D HYSCORE spectra. In this analysis and spectral simulations we used the analytical expressions published by Astashkin et al. for nuclear transition frequencies in high-spin electron system. All spectral simulation programs for ESEEM, ENDOR and HYSCORE were written using MATLAB® (the programs are available on request). ESEEM and HYSCORE were first simulated in the time domain and then after introducing appropriate dead times (to simulate experimental conditions) were Fourier transformed to form frequency domain spectra. Since in our pulsed EPR experiments, the excitation bandwidth of the pulses (10G) was comparable to the linewidth of the ±1/2 transition, in simulations we assume that complexes in all orientations with respect to the magnetic field contribute equally to the calculated spectrum.

III. Results

(A) Electrochemistry

The voltage-current curves corresponding to the electrochemical reduction of ${\mathrm{Mn}}_{aq}^{2+}$ to Mn⁰ in the voltammetry experiments are shown in Figure 1A. In each curve, the current reaches a maximum at a potential designated by E_p, the peak potential. For one- or two-electron reversible processes, the difference between the peak potential (E_p) and the mid-point potential (E_1/2) is equal to 28 mV and 14 mV, respectively³⁰. In the discussion below, we use E_p (instead of E_1/2) for data plotting and analysis.

Figure 1.

(A) Voltage-current curves of Mn²⁺ (0.25 mM MnSO₄) reduction in aqueous solutions of 0.1 M LiClO₄ at different concentrations of added NaHCO₃: 0 mM (curve 0); 2.91mM (curve 1); 5.81mM (curve 2); 8.72 mM (curve 3); 17.4 mM (curve 4); 26.2 mM (curve 5); 37.8 mM (curve 6); 49.4 mM (curve 7); 78.5 mM (curve 8); 113 mM (curve 9); 157 mM (curve 10); 209 mM (curve 11); 296 mM (curve 12); 381mM (curve 13). The curves are shifted on the x-axis for clarity. The peak potentials are indicated at the top of each curve. (B) Peak potentials (E_p) from (A) are plotted as a logarithmic function of bicarbonate concentration (see Fig 1A for experimental conditions). The data fits are shown using the Lingane (solid line) and DeFord-Hume (dashed line) equations, in the later case the speciation model of Eqn. (7) was assumed. Also shown are the major species at each concentration range. The arrow shows the bicarbonate concentration at which the pulsed EPR measurements were carried out.

In the absence of added bicarbonate (curve 0 in Figure 1A) the peak potential E_p is observed at - 1.200 V, which corresponds to a mid-point potential E_1/2 = -1.186 V for the two-electron reduction of Mn²⁺ (Mn²⁺/Mn⁰ couple), in agreement with the literature values³¹. It is important to note that E_p does not depend on the concentration of Mn²⁺ (data not shown), although the height of the observed reduction wave is directly proportional to [Mn²⁺] concentration (for [Mn²⁺] = 55 μM to 1 mM). The difference between the potentials the peak potential (E_p) and the mid-point potential (E_1/2) is equal to 30 mV. These facts indicate a diffusion controlled and reversible transfer of two electrons. The cyclic voltage-current curves (0-13) do not show an anodic peak upon reversing the voltage sweep, indicating irreversibility of the process. This irreversible behavior is typically observed for aqueous cations of metals that are weakly soluble or insoluble in mercury (Fe, Co, Ni, Mo, W, Mn)³². At present, it is generally accepted that cathodic reduction of Mn²⁺ ions on mercury electrode is quasireversible and the absence of anodic peaks upon reversing the voltage sweep (Figure 1A, curve 0) results from dissolution of reduced Mn⁰ within the mercury drop and the additional energy spent in forming of a manganese/mercury amalgam^33,34. Thus, Mn²⁺ reduction at the Hg electrode can be expressed by the following two-step process:

$$M{n}_{aq}^{2+}+2\cdot \stackrel{‒}{e}\leftrightarrow M{n}_{\text{surface}}^0$$ (2)

$$M{n}_{\text{surface}}^0\to M{n}_{\left(Hg\right)}$$ (3)

where the first reaction (2) describes reversible reduction of the ion at the electrode surface, and the second reaction (3) is irreversible dissolution of the metal to form an amalgam. The standard rate constant of an electrode process (k_S) for reduction of Mn²⁺ on a Hg electrode in 1 M NaClO₄ or NaCl, is equal to 5·10^-4 cm·s^-1, and the transfer coefficient α is equal to 0.66 at 30 °C. These values characterize the process of Mn²⁺ reduction as quasi-reversible according to Bond³⁰. Therefore, in spite of the formal irreversibility of the reduction and re-oxidation voltamograms, we can consider the reduction process as quasi-reversible and thus use standard equations to determine the stability constants and the stoichiometry of complexation.

In the presence of coordinating anions like bicarbonate, Mn²⁺ may form a coordination complex with bicarbonate by displacing coordinated water molecule(s):

$$M{n}^{2+}+q\cdot HC{O}_3^{-}\underset{\rightleftharpoons }{K}{\left[Mn{\left(HC{O}_3\right)}_q\right]}^{(2-q)}.$$ (4)

Here q defines the number of bicarbonate ions in the coordination shell of Mn²⁺, and K is the stability constant of the complex (water ligands are not shown). Since bicarbonate is a negatively charged ligand, the formed complex [Mn(HCO₃)_q]^(2-q) is expected to have a more negative reduction potential compared to ${\mathrm{Mn}}_{aq}^{2+}$. Accordingly, we find that the peak potential in the Mn²⁺-bicarbonate solutions shifts to more negative potentials at increasing bicarbonate concentrations (curves 4-13 in Figure 1A). Also, the peak current I_p decreases and the peak broadens due to complexation as more bicarbonate is titrated in solution. It is expected that complex formation affects only the reversible reduction step of eq. (2), while not affecting the irreversible dissolution of Mn⁰ shown in eq. (3). In a simple case when Mn²⁺ speciation in bicarbonate solution is dominated by only one complex, eg. [Mn(HCO₃)_q]^(2-q), the dependence of Mn²⁺ reduction potential (E_p) on bicarbonate concentration can be described by the Lingane equation³²

$$E_p-E_p^{\text{free}}=-\frac{0.059}{n}\log K-q\frac{0.059}{n}\log \left[HC{O}_3^{-}\right],$$ (5)

where potentials E are in volts, and n = 2 for two-electron reduction of Mn²⁺ to Mn⁰. In this case, plotting the peak potential E_p versus log of bicarbonate concentration will result in a linear plot with a slope defining q and an intercept defining K. $E_p^{\text{free}}$ designates the peak potential corresponding to the Mn²⁺ aqua ion.

The dependence of E_p on bicarbonate concentration is shown in Figure 1B where three distinct regions of bicarbonate concentrations can be distinguished by their different slopes. Assuming that the Lingane approximation is applicable (ie. the slope of each linear region corresponds to one dominant [Mn(HCO₃)_q]^(2-q) species), the following three species can be deduced. At low bicarbonate concentrations (0-10 mM), the Mn²⁺ peak potential is insensitive to added bicarbonate (q=0) which reveals that free ${\mathrm{Mn}}_{aq}^{2+}$ ion is the dominant species in solution. At the highest bicarbonate concentrations (50 to 200 mM), a linear phase of the potential yields a slope of 60 mV/log [HCO₃^-], corresponding to q ≅ 2 at n = 2. Thus, [Mn(HCO₃)₂] is the most abundant species at these high concentrations. Extrapolation of the fitted line to log [HCO₃^-] = 0 allows evaluation of the standard reduction potential E⁰ = -1.275 V and the stability constant K₂ = 349 M^-2 (at 22°C) for this species.

At intermediate bicarbonate concentrations (10-60 mM) a slope of 14 mV/log[HCO₃^-] is observed which formally corresponds to q = 0.5 and thus indicates a complex consisting of two Mn²⁺ per one HCO₃^-, i.e. [Mn₂(HCO₃)]³⁺. This stoichiometry is unreasonable in the context of the Lingane equation, as it deals only with complexes where q is integer and ≥ 1 ³⁵. To verify whether dimeric Mn₂ clusters occur in solution, as suggested by the slope q = 0.5, we examined the dependence of E_p on Mn²⁺ concentration between 55 μM to 1 mM. The results show that E_p does not depend on the Mn²⁺ concentration. Also, the reduction current I_p depends linearly on [Mn²⁺] which clearly shows that the stoichiometry of 2:1 is not feasible (Figure S1 in Supporting Information). Additional support for the absence of dimeric Mn²⁺ clusters in solution speciation is provided below by analyzing the EPR data. We conclude that the slope with q = 0.5 is an artifact of the use of the Lingane equation and its underlying assumptions of non-overlapping speciation³⁵.

The titration data were therefore re-examined using the DeFord-Hume method eqn. (6) which explicitly accounts for two or more species present in solution at equal or substantial fractions³⁵. The DeFord-Hume equation:

$$E_p-E_p^{\text{free}}=-\frac{0.059}{n}\log (1+\sum _q{K_i}^{\ast }{L}^i),$$ (6)

where K_i denotes the stability constant for each complex involved in speciation, was used to fit the data (shown with a dashed curve in Figure 1B). The fit is equally good as the Lingane equation although the new model shown in eqn. (7) does not involve any dimeric Mn₂ complexes and only involves monomeric Mn²⁺ species in solution. The stability constants derived from the fit are shown in Table 1, together with other data available from the literature and obtained through other techniques³⁶.

$$M{n}_{aq}^{2+}\underset{K_1}{\overset{HC{O}_3^{-}}{\rightleftharpoons }}{\left[Mn\left(HC{O}_3\right)\right]}^{+}\underset{K_{21}=K_2∕K_1}{\overset{HC{O}_3^{-}}{\rightleftharpoons }}\left[Mn{\left(HC{O}_3\right)}_2\right]$$ (7)

Table 1.

Mn2⁺ stability constants (K) with HCO₃^-/CO₃^2- from the electrochemistry and EPR data at room temperature

Mn:HCO3-	Literature Data19,34 K (μ)*	This Work K (μ)*
1:1	63.1 M-1 (0) 18.6 M-1 (0.3) 2.82 M-1 (3)	22 ± 5 M-1 (0.1) 18 ± 2 M-1 (0.3)
1:2	n.a	550 ± 200 M-2 (0.1) 568 ± 50 M-2 (0.3)

^*‘μ’: ionic strength of solution (in M).

The stability constant K₂ of the 1:2 complex Mn(HCO₃)₂ can be viewed as a product K₂ = K₁·K₂₁ which involves two reversible steps, i.e. binding of the first (K₁) and the second (K₂₁) bicarbonate ligands. The binding constant of the second bicarbonate ligand can then be estimated as K₂₁ = K₂/K₁ = 568/18.5 = 30.7 M^-1. The fact that K₂₁ and K₁ are comparable makes it evident that at intermediate bicarbonate concentrations (10-60 mM in Figure 1B) two complexes with stoichiometries 1:1 and 1:2 are present at comparable fractions in solution. The Lingane equation (5) is clearly not applicable in this situation and the misleading slope q = 0.5 results from the transient region where more than two distinct Mn²⁺ species are present in the equilibrium. This analysis serves as a caveat for using the Lingane equation when interpreting changes of redox potentials in metal-ligand solution equilibria.

(B) EPR Spectroscopy

1. Room Temperature EPR Measurements

In liquid solutions, Mn²⁺ is subject to fast rotational and tumbling motion which causes all anisotropic interactions in the spin-Hamiltonian eqn. (1) to average out, resulting in an EPR spectrum (shown as inset in Figure 2A) which can be described by only using isotropic interactions. The spectrum consists of six lines centered at g = 2.0 which are split by isotropic hyperfine coupling with the ⁵⁵Mn nucleus (I=5/2). Each of the six hyperfine lines consists of five overlapping, electron spin transitions within the S=5/2 electron spin multiplet of Mn²⁺. These five transitions occur between adjacent M_S sublevels (M_S = ±5/2, ±3/2, ±1/2, where Ms is the electron spin projection) and are weighted by the respective transition probabilities, S(S+1)-M_S(M_S+1), to contribute non-equally to the EPR signal intensity³⁷. In addition, incomplete motional averaging of the anisotropic interactions (like ZFS and anisotropic hyperfine interactions) can result in M_S-dependent broadening of the transitions, and thus those transitions which involve higher M_S states are usually broader than transition between the inner states M_S = +1/2 and -1/2.³⁸ This nonequivalent broadening results in additional weighting of the transitions, so that broader transitions (higher M_S) contribute smaller peak intensity into the EPR signal.

Figure 2.

(A) Dependence of the Mn²⁺ EPR signal intensity on bicarbonate concentration in aqueous solution at room temperature (Mn²⁺ concentration 0.5 mM, pH 8.3). Fit to an equilibrium model described in Eqn. 8 is shown with solid line. Inset shows typical Mn²⁺ EPR signal in aqueous solution centered at g 2.0. (B) Dependence of the Mn²⁺ EPR signal intensity (normalized to total amount of Mn²⁺ in the sample) as a function of Mn²⁺ concentration at fixed 50 mM NaHCO₃ in aqueous solution at room temperature. The straight line shows the invariance on Mn²⁺ concentration, clearly eliminating the presence of any Mn²⁺ clusters in speciation.

The Mn²⁺ aqua ion has small but finite ZFS (∼280-300 MHz) arising from small distortions to the octahedral symmetry of the six water ligands²².Complex formation with bicarbonate further reduces the symmetry of the ligand field around Mn²⁺ ion and thereby causes the ZFS to increase. Larger ZFS in the complex results in broader electron spin transitions and thus weaker intensity of the six-line EPR signal compared to more symmetric ${\mathrm{Mn}}_{aq}^{2+}$ ion. The transitions that involve higher spin states M_S = ±3/2 and ±5/2 are mostly affected and can be broadened to contribute little or no intensity in the six-line EPR signal. However, the inner spin transition between M_S = +1/2 and -1/2 (which contributes approximately 25% of the total signal intensity) is least affected and broadened only to the second order by ZFS. Thus, the lower symmetry Mn²⁺-bicarbonate complexes might be expected to show fewer than 25% of the EPR intensity of more symmetric ${\mathrm{Mn}}_{aq}^{2+}$ ion based on transition probabilities.

Upon addition of anions like bicarbonate, three Mn²⁺ complexes of different ligand field symmetry co-exist in solution as represented by eqn. (7). If these complexes are assumed to be in slow exchange equilibrium (10^-7-10^-8 s)³⁹, then the EPR spectrum at any concentration of bicarbonate can be represented as a weighted sum of the individual Mn²⁺ signals from each involved complex. At small bicarbonate concentrations the ${\mathrm{Mn}}_{aq}^{2+}$ ion prevails in solution and therefore the EPR intensity is expected high. By increasing the bicarbonate concentration, the low-symmetry Mn²⁺-bicarbonate complexes become abundant in solution and thus the EPR signal reduces in intensity.

Figure 2A describes the experimentally observed changes in intensity of the six-line Mn²⁺ EPR spectrum (normalized to the intensity of ${\mathrm{Mn}}_{aq}^{2+}$ signal) as a function of bicarbonate concentration. Since linewidth of the hyperfine lines in the spectrum remains invariant with ligand concentration, the peak-to-peak intensity was used to plot the spectral intensity. The normalized intensity decreases at higher bicarbonate concentrations, indicating the equilibrium shifts from the free ${\mathrm{Mn}}_{aq}^{2+}$ towards the lower symmetry bicarbonate complex(es). The drop of intensity approaches saturation at about 30-40% of the initial Mn²⁺ intensity: This residual intensity corresponds to the lower symmetry complex(es) whose intensity comes primarily from the ±1/2 transition, as discussed earlier. The experimental points were fitted using a model shown in eqn. (8):

$$I\left(c\right)=\sum _{i=0}^2{n}_i\left(c\right)\cdot I_i$$ (8)

where n_i(c) and I_i are the concentration dependent statistical weights and individual spectral intensities, respectively, of the Mn²⁺ complexes with i bicarbonate ligands; n_i(c) was calculated using the equilibrium model eqn. (7) and assuming the equilibrium constants K_i to be close to those derived from the electrochemical titration data (Figure 1B). The good fits (the dashed line in Figure 2A) were obtained at K₁ = 22 ± 5 M^-1 and K₂ = 550 ± 200 M^-2 and the intrinsic intensity coefficients I₁ = 0.3 ± 0.1 and I₂ = 0.45 ± 0.15 for complexes with one and two bicarbonate ligands, respectively (the spectral intensity I₀ of the ${\mathrm{Mn}}_{aq}^{2+}$ ion was assumed to be unity). The derived values for I₁ and I₂ are consistent with expected lower symmetry of the ligand field in the Mn²⁺- bicarbonate complexes (and thus larger ZFS and lower EPR intensity) as compared to ${\mathrm{Mn}}_{aq}^{2+}$. The fact that I₁ < I₂ indicates that symmetry of the ligand field is higher in case of two bicarbonate ligands bound to Mn²⁺.

In order to test for the presence of cluster-like species in solution (i.e. Mn²⁺-dimer as might be inferred from the slope q = 0.5 in the electrochemistry titration data), the EPR intensity was examined at a fixed bicarbonate concentration (50 mM) and over the range 0.2-1 mM of Mn²⁺ concentrations, as shown in Figure 2B. The observed zero-order dependence of the fractional Mn²⁺ intensity (i.e. weighted by the total amount of Mn²⁺ in solution) as seen in Figure 2B rules out the possibility of oligomeric forms of Mn²⁺ in solution speciation.

2. Frozen Solution EPR Measurements

Concentration dependence of EPR intensity

The perpendicular-mode EPR spectra on Mn²⁺ in 25% CH₃OH/water frozen glass at 8-77 K shows a characteristic six-line signal centered around g = 2 and a low intensity feature at g = 4 (Figure S2A in Supporting Information). The signal at g = 2 arises from monomeric Mn²⁺ complexes. Its peak-to-trough intensity when plotted as a function of bicarbonate concentration (Figure S3 in Supporting Information) can be used to evaluate the stability constants of [Mn(HCO₃)⁺] and [Mn(HCO₃)₂] complexes formed in solution, in a fashion similar to described above for the room temperature EPR data. The stability constants derived from the fit in frozen solutions are similar to those extracted from the room temperature experiments. This agreement indicates that there are no appreciable changes in Mn²⁺-bicarbonate speciation equilibrium upon addition of 25% methanol and/or upon freezing to low temperatures.

Also extracted from the fit are individual spectral intensities (I_i) of the contributing complexes. The values I₁ = 0.15 ± 0.05 and I₂ = 0.38 ± 0.1 in frozen solution are lower as compared to room temperature results. This difference is not unexpected because there is no motional averaging at frozen solution and thus the transitions with higher M_S are broadened more significantly by ZFS and therefore contribute smaller intensity than at room temperature.

The weak g = 4 signal observable in the EPR spectra at low temperatures is assignable to either a double-quantum transition of monomeric Mn²⁺ or to a transition within the excited state S=1 of an anti-ferromagnetically coupled di-manganese complex Mn²⁺-L-Mn^{2+ 40}. To resolve this uncertainty, the intensity of the g = 4 signal was examined while varying the Mn²⁺ concentration (Figure S2B in Supporting Information). The intensity is found to scale linearly with Mn²⁺ concentration proving that the g = 4 feature belongs to a monomeric form(s) of Mn²⁺. This assignment is in line with the room temperature EPR and electrochemistry data showing that the ‘dimer’-like species (or oligomeric Mn²⁺) are absent in the bicarbonate solutions.

Field-sweep pulsed EPR spectra

All pulsed EPR experiments reported below were done on samples with 0.5 mM Mn²⁺ and with no added or 100 mM bicarbonate in solution. As can be estimated using the equilibrium constants in Table 1, at 100 mM bicarbonate the total Mn²⁺ in solution is represented by a mixture of 70% [Mn(HCO₃)₂], about 20% [Mn(HCO₃)]⁺ and 10% ${\mathrm{Mn}}_{aq}^{2+}$.

Figure 3 shows the absorption EPR lineshapes of Mn²⁺ in the absence and presence of added bicarbonate, recorded using field-sweep electron spin echo (ESE)-detected EPR. In addition to the resolved six-line hyperfine pattern at the center of the spectrum (also observed in the first-derivative CW EPR spectra, Figure S2A in Supporting Information), the absorption lineshapes in Figure 3 also reveal a broad underlying signal which extends for hundreds of gauss on either side of the central g = 2 feature. A detailed interpretation of the Mn²⁺ absorption spectrum, in the case of Mn²⁺ aqua ion, has been given elsewhere.²² It was shown that the Mn²⁺ spin-Hamiltonian (1), involving two isotropic terms (electron Zeeman and ⁵⁵Mn hyperfine interactions) and one anisotropic ZFS coupling, is sufficient to fully describe the absorption spectrum. Non-zero ZFS (estimated as D = 280-300 MHz for ${\mathrm{Mn}}_{aq}^{2+}$)²² both broadens and shifts four out of five electron spin transitions, ±1/2 ↔ ±3/2 and ±3/2 ↔ ±5/2, away from g = 2 and thus spreads them around the center to produce the broad signal. The central transition, +1/2 ↔ -1/2, is only susceptible to the second-order broadening by ZFS and forms the resolved six-line hyperfine pattern that sits on top of the broad signal. Additional broadening effects come from significant D- and A-strains (caused by heterogeneity in the ligand field) that affect the outer transitions to a larger extent than the inner transitions²².

Figure 3.

EPR absorption signal of Mn²⁺ (0.5 mM) detected using 2-pulse ESE field-sweep technique in frozen glass solution (H₂O/CH₃OH) at 10 K: (solid black) with no added bicarbonate; (dashed red) with 100 mM NaH¹²CO₃; (dashed green) with 100 mM NaH¹³CO₃. Vertical line indicates the magnetic field position (3490 G) at which ESEEM, ENDOR and HYSCORE measurements were carried out. Interpulse delay τ = 400 ns was used in the field-sweep experiments.

Upon addition of 100 mM bicarbonate, the absorption EPR spectrum transforms to show less intensity at the center and stronger intensity at the spectrum wings (Figure 3). This spectral transformation is expected for [Mn(HCO₃)₂] which has a lower ligand field symmetry and a larger ZFS as compared to ${\mathrm{Mn}}_{aq}^{2+}$. The larger ZFS forces the outer transitions, ±1/2 ↔ ±3/2 and ±3/2 ↔ ±5/2, to spread further around the spectrum center, and thus to produce even broader underlying signal and an increased intensity at the spectral wings. In addition, the increased ZFS causes a second-order broadening of the inner transition (+1/2 ↔ -1/2), resulting in lower signal intensity at the center. This spectral transformation can be qualitatively reproduced in spectral simulations using D ∼ 700 MHz. However, we have not attempted a detailed spectral simulation owing to spectral complexity associated with the presence of a background signal from 20% Mn(HCO₃)⁺ and also to lack of exact knowledge of spin relaxation times (T₂ and its orientation dependence) which are essential for accurate simulation of the ESE-detected spectra. Finally, we note that the Mn²⁺ spectral transformation is identical upon addition of ¹²C or ¹³C bicarbonate. This lack of resolution is expected because the ligand ¹³C hyperfine couplings are small and therefore unresolved in the broad EPR linewidth. ESEEM/ENDOR methods are applied to resolve this small coupling, as described below.

¹³C hyperfine coupling for bicarbonate ligands

Two-pulse ESEEM spectra were recorded at magnetic field 3490 G (marked by an arrow in Figure 3), the center of one of the resolved ⁵⁵Mn hyperfine lines. This field corresponds mostly to the +1/2 ↔ -1/2 transition, however, the other four transitions also contribute appreciably (about 30-40 % of the spectral intensity) at this field. The cosine FT ESEEM traces recorded for Mn²⁺ frozen solutions in the absence and presence of ¹²C or ¹³C-bicarbonate are shown in Figure 4A. In the absence of bicarbonate, only two peaks are seen in the spectrum: a positive peak at proton Zeeman frequency 14.96 MHz (ν_H) and a negative peak at 29.8 MHz (∼ 2ν_H). These peaks, also known as the matrix lines, arise from weakly coupled ¹H nuclei surrounding the Mn²⁺ aqua ion⁴¹. The low frequency feature at around ∼0.5 MHz is an artifact of the FT procedure.

Figure 4.

(A) Cosine FT traces of 2-pulse ESEEM of Mn²⁺ (0.5 mM) in frozen glass solution (H₂O/CH₃OH) recorded at 3490 G and 10 K. Trace (1) no added bicarbonate, trace (2) 100mM NaH¹²CO₃, and trace (3) 100mM NaH¹³CO₃. Trace (2)-(3) at the bottom is the difference spectrum of (2) and (3) and shows the nuclear spin transitions arising only from ¹³C nuclei (this difference spectrum is also shown in (B)). The vertical lines show the positions of Zeeman frequencies for ¹H, ²³Na and ¹³C nuclei at 3490 G. The arrow indicates the sum combination line of ¹³C. (B) Comparison of experimental (solid line) and simulated (dash-dot line) ¹³C ESEEM spectra. Simulation parameters are shown in Table 2.

Upon addition of¹²C (I = 0) bicarbonate (NaHCO₃), the spectrum (trace 2 in Figure 4A) shows the new peak at 3.93 MHz which is assigned to the matrix frequency of ²³Na (I=3/2). Upon addition of ¹³C (I = 1/2) bicarbonate (NaH¹³CO₃), the spectrum (trace 3) shows additional lines in the spectral region of 2-5 MHz, overlapping with the ²³Na line. These additional features can be clearly revealed by subtracting the normalized spectra of ¹³C and ¹²C-bicarbonate that eliminates the lines from ¹H and ²³Na nuclei (bottom trace in Figure 4A and also zoomed in Figure 4B). The prominent features in this difference spectrum are the two positive-amplitude peaks at 2.9 and 4.26 MHz and also their sum combination at 7.34 MHz which has negative amplitude. The two positive peaks are positioned around the Zeeman frequency of ¹³C (ν_I = 3.72 MHz at 3490 G) and correspond to ¹³C nuclear transition frequencies within two electron spin manifolds M_S = ±1/2.^b Their splitting of 1.4 MHz gives an approximate measure of isotropic part (a_iso) of the ¹³C hyperfine coupling. The individual linewidth of each line ∼1 MHz gives a measure of anisotropic part (T) of the ¹³C hyperfine coupling²⁹. The significant values of a_iso and T indicate that this ¹³C belongs to a bicarbonate ligand in the first coordination shell of Mn²⁺.

The sum combination line at 7.34 MHz is downshifted from the value 2ν_I (= 7.44 MHz), the double Zeeman frequency of ¹³C at 3490 G. It has been shown that the combination line can shift from 2ν_I to higher or lower frequencies depending on the magnitude and relative signs of dipolar part (T) of ¹³C hyperfine tensor and the axial ZFS parameter (D), and also on the relative orientation of these two tensors²⁹. The negative shift in our case indicates either the same sign of T and D at 0 deg orientation of the two tensors or their opposite sign at 90 deg orientation. Although the shift of the combination line can be useful in estimating T and D, more accurate values can be extracted from the correlation 2D HYSCORE spectra (see below).

Mims ENDOR was recorded at 3490 G, and the difference spectrum (¹³C-bicarbonate - ¹²C-bicarbonate) is shown in Figure 5 in the region around ¹³C Zeeman frequency (see Figure S5 in Supporting Information for the experimental raw data). The two broad lines at 3.0 and 4.2 MHz are the same as seen in the ESEEM spectrum, however their anisotropic lineshapes are much better resolved in the ENDOR spectrum. These anisotropic lineshapes arise from orientation dependence of the ¹³C nuclear transition frequencies in the Mn²⁺ complex. At 3490 G many Mn²⁺ complexes with different orientations are excited by microwave pulses, and their weighted contributions sum up to produce the broad anisotropic lines. Each line shows two turning points corresponding to minimum and maximum nuclear transition frequencies in their orientation dependence. These turning points are useful for accurate determination of the ¹³C hyperfine parameters in spectral simulations.

Figure 5.

Experimental (top trace) and simulated (bottom trace) Mims ¹³C ENDOR spectra of Mn²⁺ (0.5 mM) in frozen glass solution (H₂O/CH₃OH) at 3490 G and 10 K. The experimental spectrum was obtained by subtraction of the spectra recorded using ¹³C and ¹²C-labeled bicarbonate. The vertical arrow labeled ν_I indicates the position of ¹³C Zeeman frequency at 3490 G. Simulation parameters are given in Table 2.

Two-dimensional (2D) HYSCORE was done at 3490 G in order to correlate the ¹³C nuclear transition frequencies in the two M_S = ±1/2 sublevels. Figure 6A shows the contour plot of the 2D spectrum in the low-frequency region where lines from ¹³C are expected. Only the (+ +) quadrant of the full 2D spectrum is shown, since the other (+ -) quadrant shows no additional features. The peaks lying along the main diagonal of the 2D spectrum correspond to the matrix lines of weakly-coupled ¹³C and ²³Na. The two off-diagonal cross-peaks are also seen and correlate the ¹³C nuclear transitions in the two electron spin sublevels. These two cross-peaks are mirror images of each other with respect to the main diagonal of the 2D spectrum. Each cross-peak has a pyramidal shape with a triangular base that is outlined for clarity in Figure 6A. The edges of the triangular base, while not resolved clearly, can be outlined qualitatively from analyzing the cross-peak slices along the x- and y-directions in the 2D plot (e.g. some slices are seen in a 3D surface plot presentation of this spectrum in Figure S7 in Supporting Information and show a rectangular lineshape with clearly resolved low- and high-frequency turning points). The top of the pyramid is shifted toward one corner of the triangular base and peaks at 2D coordinates (3.45, 4.05) MHz.

Figure 6.

(A) Contour plots of experimental (A) and simulated (B) 2D HYSCORE spectra (shown in the range of 2-6 MHz) of Mn²⁺ in frozen solution (H₂O/CH₃OH) in the presence of 100 mM NaH CO₃, recorded at 3490 G and 10 K. The position of the ¹³C cross-peaks are bordered by triangles for clarity. Matrix lines from distant ¹³C and ²³Na nuclei are also seen at (3.72, 3.72) MHz and (3.92, 3.92) MHz. The dashed line is shown perpendicular to the main diagonal of the 2D spectrum and passing through (ν_I, ν_I). The entire body of the ¹³C cross-peaks is positioned below this line. Simulation is done using parameters from Table 2.

These pyramidal cross-peak shapes arise from superposition of many orientations of the Mn²⁺ complex in frozen solution samples and result from orientation-dependent correlation of the ¹³C nuclear transition frequencies in two electron spin manifolds M_S = ±1/2. The triangular cross-peak shape indicates that the spin-Hamiltonian of Mn²⁺ has rhombic symmetry which can arise either from rhombicity in ¹³C hyperfine or ZFS tensor, or (in case of axial tensors) from non-collinear orientation of the two tensors ⁴². The corners of the triangular base represent three turning points in the anisotropic cross-peak lineshape. This feature contrasts with only two turning points found in spectral lineshapes of the 1D ENDOR spectrum in Figure 4. However, when the 2D spectrum in Figure 6A is projected onto the horizontal (or vertical) axis, two of the three turning points coincide resulting in a lineshape with only two turning points, consistent with the 1D ENDOR spectrum. We also point out that the triangular cross-peaks are shifted below the line (shown dashed in Figure 6A) that is perpendicular to the main diagonal and passes through coordinates (ν_I, ν_I) in the 2D spectrum. This shift to lower frequencies has the same origin as the negative shift of the sum combination line observed in the 1D ESEEM spectrum (Figure 4) and it can only arise for specific values of T and D and at specific orientations of the two tensors.

Simulations of the 2D HYSCORE spectrum were performed to extract the ZFS coupling and the ¹³C hyperfine parameters associated with the Mn-bicarbonate complexes. Since our goal was to reproduce the lineshapes of the cross-peaks and not their relative intensities, only the inner transition (-1/2 ↔ +1/2) was taken into account for the HYSCORE simulations. The extracted parameters from HYSCORE simulations were then refined in simulations of 1D ESEEM and ENDOR spectra. In this HYSCORE simulation our criteria for the best fit were (1) the pyramidal shape of the ¹³C cross-peak with triangular base and three turning points, (2) the peak intensity of the cross-peak which is shifted towards one (inner) corner of the triangular base, (3) the entire body of the cross-peak shifted to low frequencies, e.g. below the dashed line in Figure 6A. This set of constraints allowed us to uniquely determine the simulation parameters as explained below.

Six fitting parameters were involved in the simulation: the isotropic (a_iso) and anisotropic (T) components of the ¹³C hyperfine tensor (axial symmetry of the tensor was assumed), the coupling (D) and rhombicity (E) of the ZFS tensor, and two Euler angles (β and γ) to define orientation of the hyperfine tensor in the ZFS tensor frame (only two Euler rotations are needed because we assumed the hyperfine tensor to be axial). The Euler angles β and γ are defined as two subsequent rotations of the hyperfine tensor eigen-frame around X and then Z axes in the eigen-frame of the ZFS tensor. The initial settings for three out of six parameters were estimated from the above analysis of the ESEEM/ENDOR spectra (e.g. a_iso = 1.4 MHz and T = 1 MHz) and from the qualitative simulations of the absorption EPR spectrum (e.g. D ∼ 700 MHz). The remaining parameters (E, β and γ) were allowed to span over all possible values. The simulated HYSCORE spectrum was highly sensitive to the relative signs of D and T, and also to the angle β between the principal axes of the two tensors (e.g. see some selected simulations for different values of D and β in Figure S6 in Supporting Information). Therefore, these three parameters were optimized first. Considering the case of T and D having the same sign, only the narrow range of β from 0 to 10 degrees produced the triangular cross-peak shapes positioned below the dashed diagonal line in Figure 6A (which thus satisfies the criteria 1 and 3). However, the peak intensity was shifted to the centroid of the triangular shape unlike in the experimental spectrum where the peak intensity is at the corner of the triangular base (criteria 2). When β was increased close to 45 degrees, the triangular shape shrinked to stripes with no sharp edges positioned along the dashed line in Figure 6A, and for β > 45 degrees the cross-peaks were positioned above this line. Thus, fixing the signs of T and D to be the same and at any angle β is insufficient to simulate the experimental spectrum.

Considering unlike signs of T and D, angles β from 0-45 degrees resulted in cross-peaks above the dashed line, while 45 < β ≤ 70 degrees produced the “bell”-shaped peaks (Figure S6C in Supporting Information). However, for values 70 < β ≤ 90 degrees, the cross-peaks are triangular shaped and below the dashed line. Also at β close to 90 degrees, the peak intensity of the cross-peak shifts to the inner corner of the triangular base as is observed in the experimental spectrum (Figure S6E in Supporting Information). Thus, all three fitting criteria are satisfied for T and D of the opposite signs and β = 90 ± 10 degrees. Fine tuning of the simulated spectrum was performed next by varying γ, E and a_iso. We found that introducing non-zero E sharpened the turning points of the triangular shaped cross-peak making it closer to the experimental lineshape. The closest fit was obtained at D/6 < E < D/3 and angles 45 < γ ≤ 90 degrees. In the final step, the accurate position of the peaks was adjusted by varying a_iso in the range of 1.0-1.3 MHz. The best fit parameters are summarized in Table 2.

Table 2.

Spin-hamiltonian parameters for the 1:2 complex, Mn²⁺(CO₃^2-)(HCO₃^-)(OH₂)₃, as derived from the 2D HYSCORE simulations

aiso (MHz)	T (MHz)	D (MHz)	E (MHz)	Euler Angles (β,γ) in degrees
-1.1(±0.1)	0.8(±0.1)	-700(±50)	±115(±25)	90(±10) 90(±45)

The best fit parameters obtained from HYSCORE simulations were then used to simulate the 1D ESEEM and ENDOR spectra and an excellent agreement was found between the simulated and experimental spectra as demonstrated in Figures 4 and 5. Both lineshapes and line positions, including two lineshape turning points, are reproduced in the simulated spectra. To obtain the correct relative intensities of the ¹³C transitions in the simulated ESEEM spectrum it was important to include the contributions from electron spin transitions which involve higher M_S states. Assuming 70-80% of the ESEEM intensity from the +1/2 ↔ -1/2 transition and the rest 20-30% from other transitions produces the best fit to the experimental intensities as shown in Figure 4B (the simulations with other fractional contribution from the high spin state transitions are shown for comparison in Figure S4B in Supporting Information).

In order to deduce the absolute number of ¹³C nuclei bound to Mn²⁺, the absolute peak intensities in the normalized simulated ESEEM spectra were compared to the experimental ESEEM (Figure S4A in Supporting Information). The ESEEM simulations for two equivalent ¹³C nuclei were done using an approximation of “uncorrelated orientations of the nuclei”⁴³ and accordingly the time-domain for two nuclei was calculated as a simple square of the time-domain of a single ¹³C. The validity of this approximation was examined by extending the simulations to account for specific relative orientations of the two ¹³C nuclei in the complex. It was found that for any orientation of the two nuclei (including the special case of “trans” orientation of the ligands where suppression of modulation depth is observed in S=1/2 and I=1/2 systems⁴⁴) yielded closely comparable result to the approximation of “uncorrelated orientations”. By comparing the simulations for one and two nuclei, it was concluded that only one ¹³C nucleus contributes to the observed experimental ESEEM spectra (see Figure S4A). The second ¹³C of the second bicarbonate ligand in [Mn(HCO₃)₂] is probably coordinated in different mode, therefore has different hyperfine coupling and is not observed in the ESEEM spectra.

Hyperfine interactions to Mn²⁺ ligand protons

Mims ¹H ENDOR was used to probe the hyperfine couplings to ligand protons in the coordination sphere of Mn²⁺. The spectra recorded at three different concentrations of added bicarbonate in solution are shown in Figure 7A. All three spectra are quite similar and show the same set of resolved peaks (labeled with arrows) with only their relative intensities varying on bicarbonate concentration. The most intense peak at 14.9 MHz (labeled as ν_p) and also the small features that are partially resolved on this peak correspond to interactions to remote protons from the 2^nd and farther coordination shells of Mn²⁺. All other resolved peaks are due to protons of the water ligands from the 1^st coordination sphere of Mn²⁺ (labeled with ±1/2 and ±3/2) and correspond to nuclear spin transitions from the complexes orientated perpendicular to the static magnetic field arising from the electron spin manifolds M_S = ±1/2 and ±3/2, respectively²². As reported previously by Tan et.al. ²² and Manikandan et. al.⁴⁵, there are additional overlapping features that arise from the parallel orientations of the complex which are hidden underneath the more pronounced perpendicular features (e.g. weak parallel features from the M_S = ±1/2 sublevels overlap with the perpendicular features from the M_S = ±3/2 sublevel).

Figure 7.

(A) Mims ¹H ENDOR spectra of Mn²⁺ (0.5 mM) in frozen glass solution (H₂O/CH₃OH) at 3490 G and 10 K, and at different concentration of added NaHCO₃: (black) 0 mM, (red) 50mM and (green) 100mM. The ¹H matrix line is shown by the arrow marked by ν_p. The lines marked with ±1/2 (perpendicular feature) and ±3/2 (perpendicular feature, although a small contribution from the ±1/2 parallel feature is also present) are from protons of water ligands and arise from electron spin manifolds ±1/2 and ±3/2, respectively. Their splitting gives a measure of the perpendicular component of ¹H hyperfine coupling (A_⊥). Each ENDOR spectrum was normalized to the respective echo intensity in the absence of RF pulse. (B) Normalized intensities of the ¹H ENDOR lines from (A) are plotted as a function of bicarbonate concentration. These lines are from the water ligands to Mn²⁺ and correspond to the four electron spin sublevels: M_S = -3/2 , -1/2 (■), +1/2 and +3/2 . The line intensities of each transitions were normalized by the respective line intensities in the ${\mathrm{Mn}}_{aq}^{2+}$ spectra (no added bicarbonate) and 12 protons were assumed in the first coordination shell of ${\mathrm{Mn}}_{aq}^{2+}$. The upper and lower horizontal dashed lines mark the normalized ENDOR intensities for ${\mathrm{Mn}}_{aq}^{2+}$ and for the 1:2 complex, [Mn(CO₃)(HCO₃)(H₂O)₃]^-, respectively. The solid line represents the fit to the model shown in Eqn. (9).

Upon addition of bicarbonate, no new peaks appear in the ENDOR spectra. At 100 mM bicarbonate, 80% of Mn²⁺ has two bicarbonate ligands but no new peaks assignable to bicarbonate protons could be resolved. Protons of the bicarbonate ligands are probably weakly coupled to the Mn²⁺ ion (because of the longer proton-Mn²⁺ distance) and their transitions must be hidden unresolved in the intense matrix ¹H line. We further notice that the resolved peaks from the water ligands (labeled with ±1/2 and ±3/2 in Figure 7) do not change their position and shape but only their intensity decreases upon addition of bicarbonate and formation of the Mn²⁺-bicarbonate complex. The peaks become less intense as greater fraction of the complexes form at higher bicarbonate concentrations. This suggests that bicarbonate ligands bind to Mn²⁺ by replacing some of the water ligands (therefore the decrease in the peaks intensity) but coordination geometry of the remaining water ligands is not affected (therefore the peak frequencies are unchanged).

Reduction of the peak intensity of the M_S = ±1/2 and ±3/2 transitions as a function of added bicarbonate can be used to assess the number of the remaining water ligands in the coordination shell of the Mn²⁺ complexes. Figure 7B shows the normalized intensity of the ENDOR peaks plotted as a function of added bicarbonate. Since the M_S = ±1/2 transitions overlap with the matrix line (ν_P) their intensities were determined as peak-to-trough amplitudes in the first derivative plots (calculated numerically) of the ENDOR spectra. This approach enables extraction of accurate intensities of these transitions without requiring spectral deconvolution. The peak intensities for the ±3/2 transitions were measured directly from the ENDOR spectra after baseline correction. For Figure 7B, the intensities for each of the four transitions measured at different concentrations of bicarbonate were normalized to the intensity of the respective peak in the ${\mathrm{Mn}}_{aq}^{2+}$ spectra (no added bicarbonate). Because ${\mathrm{Mn}}_{aq}^{2+}$ has six water ligands (i.e. 12 protons) in the first coordination shell³⁹, then the normalized intensities in Figure 7B give a quantitative measure of the average number of water ligands (protons) per one Mn²⁺ in bicarbonate solution. We notice that the four ENDOR transitions show nearly identical dependence on bicarbonate concentration within experimental error. We interpret this as clear indication that the relative contributions from the five EPR transitions of Mn²⁺ to the ENDOR spectrum at 3490 G do not change appreciably with bicarbonate concentration. Furthermore, these relative contributions are comparable for all three complexes involved in the bicarbonate equilibrium (Eqn. 7), namely ${\mathrm{Mn}}_{aq}^{2+}$ and the 1:1 and 1:2 complexes. These conclusions together with the assumption that geometry of the remaining water ligands in the first shell of Mn²⁺ is not appreciably affected upon bicarbonate complexation, allow us to formulate a simple model to describe the dependence of the ENDOR peak intensity on bicarbonate concentration:

$$p_{av}\left(c\right)=\sum _{i=0}^2{n}_i\left(c\right)\cdot p_i$$ (9)

where, p_i is the number of water ligand protons in each complex involved in the equilibrium (p₀ = 12 for ${\mathrm{Mn}}_{aq}^{2+}$), and n_i(c) is the mole fraction for each complex as a function of bicarbonate concentration c. To calculate n_i(c) we used the equilibrium model of Eqn. (7) and the binding constants K₁ and K₂ obtained from the room temperature EPR and electrochemistry data (Table 1). This leaves only two fitting parameters p₁ and p₂. Fitting of the four dependences in Figure 7B using the same set of parameters results in p₁ = 8-10 protons (4-5 water ligands) for the 1:1 complex and p₁ = 4-6 protons (2-3 water ligands) for the 1:2 complex. Based on these numbers we formulate the 1:1 complex as [Mn²⁺(HCO₃^-)(OH₂)_4-5] and the 1:2 complex as [Mn²⁺(HCO₃^-)₂(OH₂)_2-3].

Discussion

Stability constants of the Mn²⁺ complexes

The observed changes in redox potential (Figure 1A and 1B) of the Mn²⁺/Mn⁰ couple in the presence of bicarbonate is rationalized by the formation of two electro-active Mn-bicarbonate complexes with metal:ligand stochiometries of 1:1 and 1:2, respectively. The 1:1 complex forms at bicarbonate concentrations between 10 and 75 mM, and a 1:2 complex dominates in solution above 100 mM bicarbonate. This speciation was also confirmed in room temperature EPR measurements (Figure 2A). Both methods resulted in stability constants in agreement with each other (Table 1) with K₁=18 M^-1 and K₂=568 M^-2. This constants are in good agreement with K₁ = 18.6 M^-1 for the 1:1 complex [Mn(HCO₃)]⁺ determined by potentiometric titrations at 298 K¹⁹, and with K₁ = 63 and 2.8 M^-1 measured at 298 K at different ionic strengths of solution equal to 3.0 and 0.0, respectively³⁶. The value of K₂ has never been reported before.

Binding modes and local structure

The ¹³C-hyperfine tensor resolved from ESEEM, ENDOR and HYSCORE spectra unequivocally show that Mn²⁺ forms first coordination shell complexes with bicarbonate/carbonate. The simulations parameters (Table 2) can be used to infer the local structure and specifically the binding modes of carbonate ligand in the coordination sphere of Mn²⁺.

The anisotropic part of the ¹³C hyperfine tensor (T = 0.8 MHz) may involve two contributions: a direct dipole-dipole hyperfine interaction (T_dd) between the ¹³C nucleus and the unpaired electron spin residing on Mn²⁺, and also the anisotropic term (T_p) arising from the finite electron spin density localized on the p-orbital of the bicarbonate ligand. The isotropic coupling extracted from our simulations a_iso = -1.1 MHz corresponds to about ∼3% of the unpaired electron spin density on the p-orbital of the bicarbonate ligand as can be estimated based on a_iso ∼ 33 MHz reported for 100% spin density localized on the p-orbital of the (bi)carbonate radical in solutions⁴⁶. This 3% of p-orbital spin density translates into T_p = 0.09 MHz as estimated using the value T_p = 3 MHz reported for carbonate radical anion⁴⁷. Assuming that the main axes of the two anisotropic tensors (T_dd and T_p) are oriented by 90 degrees to each other (as might be expected for the complex geometry shown in Scheme 1), the actual value of the dipole-dipole term can then be estimated T_dd = 0.85 MHz. This later value can be used in a point-dipole approximation to estimate distance (R) between the electron spin density localized on Mn²⁺ ion and the ¹³C nucleus of the bound (bi)carbonate ligand:

$$T_{dd}=\frac{g_e{\beta }_e{g}_n{\beta }_n}{h\cdot R^3}$$ (10)

Here all parameters are as defined above for Eq. (1). In case of ¹³C, this equation reduces to T = 19.9/R³(in MHz), where R is in Å. Thus, T_dd = 0.85 ± 0.1 MHz derived from the spectral simulations corresponds to a distance of 2.85 ± 0.1 Å between the Mn²⁺ and ¹³C nuclei.

Scheme I.

Deduced structure of the 1:2 complex, Mn²⁺(CO₃^2-)(HCO₃^-)(OH₂)₃, formed in solution at high bicarbonate concentrations (> 50 mM) at pH 8.3. The arrows (D_ZZ and A_ZZ) show orientations of the principal axes of ZFS and ¹³C hyperfine coupling tensors. Their relative angle 90° corresponds to angle β derived from ¹³C HYSCORE simulations.

Carbonate complexes of 3d-transition metal ions have previously been synthesized and their X-ray structures have been resolved (reviewed in ⁴⁸). It has been shown that (bi)carbonate can bind in either mono-dentate or bi-dentate modes and the average distances (Mn-C) for both types of coordination are summarized in Table 3. No distances have been reported for Mn²⁺-bicarbonate complexes although a distance of 3.048 Å for a bridging bicarbonate in polymeric MnCO_3(s) has been observed⁴⁹. For other transition metal ions, e.g. Cu²⁺ and Co²⁺/Co³⁺, the metal-carbon distances are known for both mono-dentate and bi-dentate coordination modes (Table 3). Co²⁺ usually exhibits a mono-dentate coordination while Co³⁺ prefers a bi-dentate coordination. Cu²⁺ was observed in both mono- and bi-dentate coordination modes. Limited EPR data are available on Cu²⁺-bicarbonate complexes in solutions and demonstrate comparable metal-carbon distances to what found in single crystals⁵⁰. Based on the Cu²⁺-C and Co²⁺/Co³⁺-C distances and renormalizing for a greater ionic radius of Mn²⁺ (see a footnote for Table 3), we estimated the expected Mn-C distances of 2.75-2.9 Å for a bi-dentate coordination of bicarbonate to Mn²⁺ and 3.3-3.35 Å for a mono-dentate coordination. It is concluded that the distance 2.85 ± 0.1 Å derived from our EPR data for one of the (bi)carbonate ligands in the 1:2 complex corresponds to the bi-dentate coordination. The bi-dentate mode of coordination invokes a π-type overlap between 2p_π orbital of carbonate ligand and two d_π orbitals (d_zx and d_zy) of Mn²⁺. Thus a significant transfer of unpaired electron spin density from Mn²⁺ onto (bi)carbonate ligand can be expected that explains the substantial isotropic hyperfine coupling, a_iso = 1.1 MHz, found on the ¹³C nucleus. To our best knowledge the bi-dentate coordination has never been previously identified in aqueous solution speciation of Mn²⁺.

Table 3.

X-ray distances for metal-(bi)carbonate complexes. Both mono- and bi-dentate coordination modes are included

r(Metal-13C) in carbonate modes	(Cu-13C) Å	(Co-13C) Å	(Mn-13C) Å
Mono-dentate	2.9a	3.1b	3.35c
Bi-dentate	2.4-2.5a	2.3-2.4b	2.75-2.9c
			2.75-2.95 (this work)

^aBased on X-ray diffraction on single crystals of copper complexes: [Cu(HCO₃)]_(aq), [Cu(CO₃)]_(aq), [Cu(CO₃)₂]^2-_(aq) and Na₂[Cu(CO₃)₂]_(s)⁵⁰

^bBased on X-ray diffraction on single crystal of cobalt complexes: K₂[Co^II(H₂O)₄(CO₃)₂]_(s) for mono-dentate, [Co^III(NH₃)₄CO₃]Br_(s) for bi-dentate⁴⁸

^cEstimated based on ionic radii (Co³⁺ = 0.63 Å, Co²⁺ = 0.74 Å, Cu²⁺ = 0.69 Å and Mn²⁺ = 0.80 Å)

Upon comparing the experimental and simulated ESEEM intensities (Figure S4A) we concluded that the resolved ¹³C coupling belongs to one (bi)carbonate ligand in the 1:2 complex and that the second bicarbonate ligand is EPR-invisible. We thus speculate that this second ligand coordinates probably in a different (mono-dentate) mode and its poor spectral resolution is explained by either a small hyperfine coupling or by a large dispersion (strain) in the hyperfine tensor rendering the nuclear spin transitions to be broad and unobservable (e.g. the disappearance of the proton combination line was reported in ESEEM spectra of ${\mathrm{Mn}}_{aq}^{2+}$)²⁹. The ¹H ENDOR data supports this model by revealing that there are three water ligands in the first coordination shell of Mn²⁺ in 1:2 complex. Having three water ligands leaves only three coordination sites on Mn²⁺ to bind two (bi)carbonate ligands. Two of these sites are occupied by the first carbonate ligand bound in bi-dentate mode. The second bicarbonate thus binds to the remaining vacant binding site in a mono-dentate coordination mode.

Our suggested structure for the 1:2 complex is shown in Scheme I. This involves two (bi)carbonate ligands, one in bi-dentate and the second in mono-dentate mode, and also three water ligands. We assume a trans relative position for the two (bi)carbonate ligands (not known directly from our simulations), yielding a facially ligated pseudo-octahedral complex. The arrows show orientations of the principal axes of ZFS tensor (D_ZZ) and also the ¹³C hyperfine tensor (A_ZZ) for the bi-dentate bicarbonate ligand in the molecular frame of the complex. The hyperfine tensor arises from dipole electron-nuclear spin interaction and its principal axis (A_ZZ) lies along a vector connecting Mn²⁺ to ¹³C nucleus of the ligand. The ZFS tensor arises from non-cubic distortions to the Mn²⁺ ligand field and therefore the ZFS principal axis (D_ZZ) should lie oriented along the main symmetry axis of the ligand field. (Bi)carbonate ions are known to be a stronger ligand than H₂O molecules, and therefore for the structure shown in Scheme I the two (bi)carbonate ligands define the plain of the strongest distortion to the ligand field symmetry of Mn²⁺. The symmetry axis of the ligand field (and thus the orientation of the principal axis of ZFS) should be perpendicular to the plain as is shown with arrow in Scheme I. With this arrangement the principal axes of ZFS tensor (D_ZZ) and the ¹³C hyperfine tensor (A_ZZ) form an angle 90° which is consistent to β = 90° derived from the HYSCORE spectral simulations.

Mn²⁺ speciation model and its significance

Scheme II shows the chemical speciation of Mn²⁺ in aqueous bicarbonate solutions as inferred from the data presented in this work. The room temperature electrochemistry and EPR data reveal the presence of 1:1 and 1:2 complexes in solution. The 1:1 species formed in the concentration range of 10-50 mM bicarbonate is represented in Scheme II by two species [Mn(HCO₃)]⁺ and [MnCO₃], e.g. bicarbonate and carbonate forms of the complex. The existence of two (protonated and deprotonated) forms can be rationalized as follows. The pK_a of free bicarbonate ion in solution is 10.5. The pK_a of bicarbonate bound to Mn²⁺ is unknown but expected to be significantly lower because of the electropositive nature of Mn²⁺. Upon binding to Mn²⁺ the pK_a of water molecule is decreased by 5 units, from 15.6 to 10.5. Taking into account the negative charge on bicarbonate molecule one expects a smaller effect (decrease by 2-3 units) as compared to a water molecule. The stability constants for the 1:1 Mn-bicarbonate and Mn-carbonate complexes have been measured⁴⁸ and can be used to estimate pK_a∼8.5 for bicarbonate bound to Mn²⁺. Thus at pH = 8.3 used in our electrochemistry and EPR experiments both bicarbonate and carbonate complexes are expected at comparable fractions. Similarly, the 1:2 species in Scheme II is also represented by the equilibrium of [Mn(HCO₃)₂] and [Mn(CO₃)(HCO₃)]^- considering the deprotonation of one of the bound bicarbonate ligands. Thus the overall Mn²⁺ speciation is represented by equilibrium of five species and their population depends on bicarbonate concentration and pH of the solution.

Scheme II.

Model describing speciation equilibrium of Mn²⁺ in bicarbonate solutions at pH 8.3.

The speciation shown in Scheme II is important for understanding the redox chemistry of Mn²⁺ in aqueous solutions. The oxidation potentials (in Mn²⁺/Mn³⁺ redox couple) for the 1:1 and 1:2 Mn²⁺-bicarbonate complexes have been found to decrease significantly from E⁰ = 1.18 V for ${\mathrm{Mn}}_{aq}^{2+}$ to 0.67 V and 0.52 V, respectively²¹. These low oxidation potentials bring these bicarbonate complexes well within the range of oxidation capability of many metal enzymes, like photosystem II. Thus formation of the Mn²⁺-bicarbonate complexes might have important implication for biological oxidation pathways of Mn²⁺. An important example is found in the field of photosynthesis. The photo-pigments of the native reaction centers from Rhodobacter sphaeroides, a class of anoxygenic phototrophs, have redox potential close to 0.5 V which can be raised slightly to 0.6-0.7 V by selective mutation of the residues near the reaction center¹⁴. These potentials are not high enough to oxidize ${\mathrm{Mn}}_{aq}^{2+}$ (1.18 V). However, it has been shown that the redox potential of the mutant is already sufficiently high to allow oxidation of Mn²⁺-bicarbonate complexes which have potentials 0.55-0.6 V. This work has demonstrated unequivocally that upon formation of Mn²⁺-bicarbonate complexes, Mn²⁺ becomes available for a broader range of biological reactions.

Another important example is the photo-assembly process of the Mn₄-cluster in PSII-WOC. Previous study has shown that bicarbonate anion plays a selective role in enhancing the rate of assembly of the inorganic Mn₄ core responsible for oxygen evolution activity during biogenesis and repair of PSII-WOC. In vitro studies indicate that presence of bicarbonate (∼ 1 mM) leads to a 4-fold acceleration of the net rate of light-induced photo-oxidation of the first Mn²⁺ by apoWOC-PSII and an increase in the yield of reconstituted centers owing to reduction of photoinhibition⁸. Formation of ternary Mn²⁺-bicarbonate-apoPSII complex (with a low oxidation potential) appears to be the origin of the increased assembly rate of the cluster in the presence of bicarbonate⁵¹.

The Mn²⁺-bicarbonate complexes have also been hypothesized to play important role in the evolutionary origin of oxygenic photosynthesis in the archean era^12,52. At highly elevated bicarbonate concentration in archean ocean the speciation of many polyvalent metal ions was largely shifted in favor of metal_x-(bi)carbonate_y complexes. The low pH of the archean ocean prevented from precipitation of insoluble metal-carbonate/hydroxide minerals, thus leading to increased concentration of the metal-bicarbonate complexes in solution. The favorable oxidation potentials of the resulting Mn-bicarbonate species would have enabled them to serve as possible electron donors to early photosynthetic prokaryotes.

The present work on Mn²⁺ bicarbonate speciation clearly rules out the earlier speculations of Mn²⁺ dimeric complexes in bicarbonate solutions^20,53. Although the pseudo-catalase activity rate was observed to be proportional to square of Mn²⁺ concentration, this could probably be explained by either a transient dimeric complex or an outer-sphere interaction of two monomeric species in the rate-determining step of peroxide degradation. Our spectroscopic study does not support the existence of a pre-formed dimanganese (II, II) complex in solution.

V. Conclusions

Electroreduction and EPR spectroscopy have provided a self-consistent description of the chemical speciation of Mn²⁺ with bicarbonate in aqueous solution. The speciation involves two complexes with Mn²⁺:bicarbonate stoichiometries of 1:1 and 1:2; their stability constants have been determined. The molecular structure of ¹³C-labelled complexes has been studied using ESEEM, ENDOR and 2D HYSCORE. The derived structures are Mn²⁺(HCO₃^-)(OH₂)_4-5 for the 1:1 complex and Mn²⁺(CO₃^2-)(HCO₃^-)(OH₂)₃ for the 1:2 complex.

ABBREVIATIONS

CW

continuous wave

EPR

electron paramagnetic resonance

ENDOR

electron-nuclear double resonance

ESEEM

electron spin echo envelope modulation

HYSCORE

hyperfine sublevel correlation spectroscopy

RT

room temperature

XRD

X-ray diffraction

ZFS

zero field splitting.