Determining the Structure and Defects of Manganese Oxides using X-Ray Absorption Spectroscopy

Determining the Structure and Defects of Manganese Oxides using X-Ray Absorption Spectroscopy Stanley Quan Office of Science, Science Undergraduate Laboratory Internship Program University of California, Berkeley Stanford Linear Accelerator Center Menlo Park, CA August 15, 2008 Prepared in partial fulfillment of the requirement of the Office of Science, Department of Energy’s Science Undergraduate Laboratory Internship under the direction of John Bargar and Apurva Mehta in the Environmental Remediation Science group at the Stanford Synchrotron Radiation Laboratory at Stanford Linear Accelerator Center Participant: _ Signature Research Advisor: _ Signature TABLE OF CONTENTS Abstract iii Introduction Materials and Methods Results Discussion and Conclusion Acknowledgements References 10 Tables 11 Figures 12 ii ABSTRACT Determining the Structure and Defects of Manganese Oxides using X-Ray Absorption Spectroscopy STANLEY QUAN (University of California, Berkeley, CA 94720) JOHN BARGAR (Stanford Linear Accelerator Center, Menlo Park, 94025) APURVA MEHTA (Stanford Linear Accelerator Center, Menlo Park, 94025) Since manganese oxides are stable over time, and studies suggest that their biologically and abiotically formed states are distinguishable, they possess the key properties to be useful as biosignatures Biosignatures are biological indicators of life and may be used to search for extraterrestrial life Because bacteria form poorly crystallized manganese oxides in nature, we must refine detailed crystal structures before being able to utilize them as biosignatures After performing X-ray Absorption Spectroscopy (XAS) on various manganese oxides and fitting to a single scattering model with Mn-O and Mn-Mn shells, the Extended X-ray Absorption Fine Structure (EXAFS)derived fit data was compared to X-Ray Diffraction (XRD)-predicted local structure results In particular, we were able to explain the significantly lower EXAFS coordination numbers for the O and Mn shells compared to the XRD results Also, by constraining to XRD parameters for fitting, we found increases in disorder and reduced chi square that indicate defects and that EXAFS is a better representation of the structures We were able to rank the manganese oxides by ideal structure as predicted by XRD, from the least ideal to the most ideal: small tunnel (pyrolusite, ramsdellite), tunnel (coronadite, cryptomelane), layer (chalcophanite, lithiophorite, birnessite), to layer/tunnel (todorokite) iii INTRODUCTION Biosignatures are biological indicators for the presence of life Identifying them is significant because they allow us to detect signs of life and retrieve biochemical information from Earth’s geological records Since they are stable over time, and studies suggest that their biologically and abiotically formed states are distinguishable by Electron Paramagnetic Resonance (EPR), manganese oxides possess the key properties to be useful as biosignatures [1] For example, the presence of considerable manganese oxide deposits at 2.22 billion years ago indicates the transformation of the atmosphere from an anoxic state to an oxic state Finding biogenic manganese oxides on the surface of other planets, such as Mars, would provide evidence that suggests the existence of life In nature, manganese is primarily found in two forms today: the reduced, soluble Mn(II), occurring mostly as Mn2+(aq) ions, and the oxidized, insoluble Mn(IV) as manganese oxides In aquatic environments, various bacteria convert Mn(II) to Mn(IV) in an energy-yielding reaction that forms manganese oxides [1] With over 30 known manganese oxide minerals, they are found in most soils and sediments, making them useful as geological time scales Since bacteria form poorly crystallized manganese oxides in nature, mostly as desert varnish, which is a thin layer of manganese oxides, iron oxides, and clay, we must examine and refine their structures before being able to use them as biosignatures X-ray absorption spectroscopy (XAS) and XRay Diffraction (XRD) are two different, but complimentary x-ray interactions often performed to accomplish this XAS is a useful technique in determining the electronic and molecular structure of the manganese oxides It is a particularly powerful technique because of its high sensitivity and elemental specificity, effectively probing the local structure XAS seems to be better suited for studying manganese oxides because of the poorly crystallized and defective biogenic forms produced by bacteria It includes both X-ray Absorption Near Edge Structure (XANES) and Extended X-ray Absorption Fine Structure (EXAFS) and involves shining high energy x-rays (from synchrotron radiation) on the sample so that an atom will absorb the energy At threshold frequency, a bound electron will be ionized and emit a photoelectron If we measure the absorption of light, the threshold energy shows up as a sharp increase (“edge”) in the absorption coefficient [2] This is known as a K-edge if a 1s electron is ionized We can determine the local structure surrounding the atom by looking at the absorption as energy increases From about 50-100 eV past the x-ray absorption edge to 1000 eV, we see damped oscillations, the sum of sinusoidal waves from each neighboring atom, which make up the EXAFS After it is emitted from the absorbing atom at threshold energy, the photoelectron is backscattered from nearby electron density surrounding the neighboring atoms The EXAFS oscillations are a result of the interference pattern observed between the outgoing and backscattered photoelectron waves, with constructive interference producing maxima and destructive interference producing minima [2] Since the backscattered wave’s amplitude and phase is dependent on the type of backscattering atom and the distance from the central atom, we are able to derive information about the absorbing atom’s coordination environment, including the number and nature (approximate atomic number) of neighboring atoms [2] Complimentary to XAS, XRD probes crystal structure with a longer length scale While XAS explores the immediate environment of the selected element, XRD assumes periodicity in order to observe a larger range Comparing data between XRD and XAS will yield a more complete picture of the structure of manganese oxides Because biogenic manganese oxides tend to be weakly diffracting, traditional techniques have been unsuccessful in illuminating the molecular-scale structures [1] But since synchrotron radiation is of very high intensity and collimated, we can measure x-ray absorption data for dilute, hydrated, and radioactive samples We seek to use this method to verify identities and refine detailed structures of manganese oxides MATERIALS AND METHODS The x-ray absorption spectroscopy and data analysis on manganese oxides were conducted at the Stanford Synchrotron Radiation Laboratory (SSRL) Various manganese oxide samples were obtained from Sam Kim of the NASA Jet Propulsion Laboratory and prepared at the SSRL Each sample was ground with mortar and pestle, homogeneously mixed with a 3mg manganese oxide to 14mg boron nitride ratio, and placed in an aluminum sample holder with kapton windows Kapton windows don’t suffer from radiation damage and are used because of their high mechanical and thermal stability as well as their high transmittance to x-rays The manganese oxide samples were diluted because they were too concentrated This eliminates the problem of self-absorbance, when the absorption of beam at the front is so strong that it corrupts and reduces amplitudes from the rear of the sample The experimental setup can be seen in Figure XAS measurements were collected at beamline 11-2 at the SSRL The x-ray beam from the synchrotron source goes through a monochromator to become monochromatic light of a single energy before hitting the sample and detectors Absorbance and transmittance data were collected by ion chamber detectors, and fluorescence measurements were taken by a Lytle detector after the beam struck the sample and passed through Soller filters The energy was calibrated with permanganate foil and assigning the pre-peak point to 6543.34 eV The measurements were taken with a data scan range of 6300-7200eV The experiment was run at room temperature, with 3mm horizontal and 1mm vertical slits The XAS data represent 1- and 2-scan averages with scan time of 30 minutes per scan Data analysis, including background subtraction and EXAFS fitting, was done with the program SixPACK [3] In background removal, the XANES spectrum is edge normalized The pre-edge region is fitted with a linear first-order polynomial to subtract the baseline from the EXAFS, and the post-edge region is fitted with a second-order polynomial to draw a smooth background through the EXAFS oscillations, removing any background absorption due to neighboring atomic interference This leaves us with a flat “zero” background pre-edge and a unit-step edge with a constant post-edge jump of one [4] All absorption data are now relative energy compared to the edge value The edge energy is taken as the origin of wavenumber k, and the EXAFS is k3-weighted to enhance the oscillations at high k The EXAFS is then broken down into components by fitting it with FEFF paths based on the scattering path of a photo-electron [4] Each path is taken from reference compounds and represents one of the atomic interactions that make up the full multiple scattering model The parameters of radial distance (R), the Debye-Waller factor for disorder (σ2) and the experimental energy threshold (E0) are allowed to float as Mn-O and Mn-Mn shells are fitted to the χ(k) plot and Fourier Transform with FEFF paths The amplitude reduction factor (S02) is set at 0.9 and E0 is constrained to the same value for all paths These theoretical fit models are then compared to related known structures to examine the relevance of any similarities and discrepancies The Webmineral Mineralogy Database [5] and the MINCRYST Crystallographic Mineral Database [6] were used to reference known properties and crystal structures of the manganese oxides Theoretical crystal structures were generated using CrystalMaker software [7] RESULTS A sample background subtraction can be seen in Figure and a sample EXAFS fit is shown in Figure The EXAFS χ(k) plot and Fourier Transform plot can be seen in Figure and 5, respectively The EXAFS can be described as a sum of many sine waves, with each Fourier component characterizing an amplitude and phase that depend on the distance between the absorber atom and the coordinating atoms, and the type of coordinating atoms The Fourier Transform gives us peaks from scattering by the nearby atomic neighborhood The edge is at about 6552 eV The graphs are stacked for comparison, with a color corresponding to each manganese oxide Transmission data was taken for all of the manganese oxides and used for fitting except for two Fluorescence measurements were used for coronadite, because of its significant Pb content, and for pyrolusite, because of Fe impurity in the sample X-rays don't transmit very well through lead, and thus the data quality of the transmission won’t be as good as the fluorescence The theoretical crystal structure for each of the manganese oxides can be seen in Figures through 13 Table is a summary of EXAFS parameters from the Mn K-edge using the single scattering manganese oxide model, which includes the Mn-O, Mn-Mn edge-sharing, and Mn-Mn corner-sharing paths, for reference compounds and our samples The EXAFS parameters were determined experimentally, while the XRD parameters were taken from known crystal structures Table includes data on XRD, EXAFS fits, and constrained-to-XRD EXAFS fits DISCUSSION AND CONCLUSION As seen in Table 1, when the XANES data was constrained to the XRD parameters for fitting (N=6), disorder σ2 and the reduced chi square increased significantly for most of the manganese oxides, while radial distance R remained relatively the same Only birnessite and todorokite showed reductions in σ2 and reduced chi square With XRD constraints, birnessite has an increased reduced chi square value, but no negligible change in disorder σ2 Todorokite shows a decrease in reduced chi square value and also a decrease in σ2, suggesting ideal structure with N=6, as predicted by XRD for an ideal structure The notable increase of value in the fitted parameters for the other manganese oxides gives a strong indication that there are vacancies in the structures which were not recognized in the XRD technique Therefore, defects and the EXAFS provide a better representation of the structures In Figures and 5, we notice a trend relating to defects As you move from top to bottom in the EXAFS χ(k) and Fourier Transform plots, lattice disorder increases, with minimal defects in todorokite and birnessite In the EXAFS χ(k) plot (Figure 4), this is most evident at 6.8 k(Å -1), where we observe an increasingly broader feature as we move down the y-axis To a lesser degree, the trend is also seen at 8.0 and 9.0 k(Å-1) We can employ a similar ranking order in the Fourier Transform plot (Figure 5), with the crystal structure becoming less ideal as we move down the manganese oxides from todorokite to pyrolusite From the Fourier Transform plot presented in Figure 3, we can see the single scattering model, with the first two large peaks representing Mn-O and Mn-Mn shells The third peak seen in the Fourier Transform of pyrolusite is from a second Mn-Mn shell Besides showing clear Mn-edge and Mn-corner shells, it also appears to have more significant scattering shells that would be seen in the full multiple scattering manganese oxide model, as evidenced by relatively larger FT peaks compared to other manganese oxides between 4.0 and 6.0Å We observe larger oscillations as we move down the series from todorokite to pyrolusite, indicating defects and less ideal structure Defects include possible vacancies, cations, and bending in the crystal structure Furthermore, after constraining to XRD data and then fitting (Table 1), the change in disorder (Δσ2) progressively increases percentage-wise as we move down the series, from todorokite to pyrolusite This further reinforces the ideal structure ranking of the manganese oxides by lattice order and defects Also in Table 1, we can observe that none of the coordination numbers (N) from the EXAFS are equal to the expected value of for both Mn-Mn and Mn-O shells obtained from XRD Using the single scattering model, nearly all of the experimentally calculated N’s are significantly lower than the expected value, with only birnessite’s Mn-Mn shell and todorokite’s Mn-O and Mn-Mn shells exceeding N=6 The lower coordination numbers from the EXAFS suggest that both layer and tunnel manganese oxide structures cause amplitude loss, which may arise from splitting of the O distances caused by Jahn-Teller distortions from structural Mn 3+ in the layers [8] The splitting of the O shells produces a net destructive interference in the overall Mn-O shell and thus decreases the observed EXAFS amplitude The lower coordination numbers for the Mn-Mn shells are likely due to vacancies in the layer structures and misfitting because of scattering from O atoms around the second shell Mn octahedra [8] The amplitudes are highly sensitive to bending of the Mn octahedral layer, as seen in layer structures, and to manganese site vacancies, as seen in tunnel structures [8] Both of these structural qualities reduce the single scattering intensities In contrast to the differing coordination number measurements between EXAFS and XRD data, the radial distances (R) of Mn-O and Mn-Mn shells are very similar between the two experimental methods Furthermore, there are EXAFS patterns that allow us to characterize the manganese oxide structures as layered and tunneling Looking at the EXAFS stack plot (Figure 4) and the structures from previous studies [9], it seems as though layered manganese oxides exhibit more ideal structure than tunnel manganese oxides The samples can be ordered by type of structure from least ideal to most: small tunnel (pyrolusite, ramsdellite), tunnel (coronadite, cryptomelane), layer (chalcophanite, lithiophorite, birnessite), to layer/tunnel (todorokite) In comparison to the other samples, the least ideal structure is seen in pyrolusite and ramsdellite, which are known to exhibit small tunnel structures with little or no atoms in the tunnel vacancies This is due to them having a greater degree of corner-sharing with respect to edge-sharing octahedra [8] Pyrolusite tunnels are too small to accommodate other chemical species, while ramsdellite tunnels only contain minor amounts of other atoms [9] Furthermore, the next least ideal structure is seen in coronadite and cryptomelane, which have tunnel structures The defects are likely due to their respective predominant tunnel cations, K and Pb [9] The layered structures of chalcophanite, lithiophorite, and birnessite show more ideal structure in the EXAFS, possibly due to their interlayer cations being in between layers, rather than tunnel cations encircled by the Mn and O in tunnel structures Lastly, todorokite, with its unique layer/tunnel structure, displays the most ideal structure Reference structures confirm a tunnel structure constructed of triple chains of MnO6 octahedra, which share corners with each other to form large tunnels with square cross sections [8] The EXAFS and Fourier Transform for todorokite seem to support both tunnel and layer structure In the EXAFS (Figure 4), todorokite displays the strong features at 6.8, 8.0, and 9.0 k(Å-1) that are characteristic of layered structures Todorokite also shows a larger reduction in amplitude in the Mn-Mn shell in the Fourier Transform (Figure 5), which would seem to indicate less ideal structure seen in tunnel structures In conclusion, with data analysis of the XAS experiment results on various manganese oxides, we were able to locate distinct features that allow us to characterize them by their layered and tunnel structures The observed differences between the EXAFS-derived fit and XRDpredicted local structure results, along with the structural drawings of the different Mn oxides, give us a better general understanding of their similar, yet unique structures In particular, we were able to explain the significantly lower EXAFS-derived coordination numbers for the O and Mn shells compared to the XRD results Also, by constraining to XRD parameters for fitting, we found increases in disorder and reduced chi square that indicate EXAFS provides a better representation of the structures We were able to rank the manganese oxides by ideal structure, from the least ideal to the most ideal: small tunnel (pyrolusite, ramsdellite), tunnel (coronadite, cryptomelane), layer (chalcophanite, lithiophorite, birnessite), to layer/tunnel (todorokite) Layered structure manganese oxides seem to better fit the ideal structures predicted by XRD than tunnel structure ones Future work should include fitting manganese oxide data with the full multiple scattering model This analysis would produce more structural information about Mn oxides and hopefully lead to the ability of distinguishing abiogenic from biogenic Mn oxides Conceivably, this would make it possible to use manganese oxides as bio-signatures of life ACKNOWLEDGEMENTS This research was accomplished at the Stanford Synchrotron Radiation Laboratory at Stanford Linear Accelerator Center I would like to thank the U.S Department of Energy, Office of Science and Stanford Linear Accelerator Center for funding my research and giving me the opportunity to participate in the SULI program I also want to thank my mentors John Bargar and Apurva Mehta for their invaluable support and guidance, Eleanor Schofield and Sam Webb for their helpful assistance, and Susan Schultz, Farah Rahbar, and Steve Rock for organizing the SULI program REFERENCES [1] Perry, R and Kolb, V., “Biochemical markers in rock coatings”, in Perspectives in Astrobiology, Hoover, R., Rozanov, A., and Paepe, R., Eds Amsterdam, Netherlands: IOS Press, 2005 pp.120-125 [2] Scott, R., "X-Ray Absorption Spectroscopy," in Structural and Resonance Techniques in Biological Research, Denis L Rousseau, Ed., Orlando, FL: Academic Press, 1984, pp 295358 [3] Webb, S., SixPACK (Sam's Interface for XAS Package), Stanford Synchrotron Radiation Laboratory, Stanford Linear Accelerator Center, Stanford, CA, 2008 [4] Webb, S., "SixPACK Documentation", [Online document], Jan 2003, Available at HTTP: http://www-ssrl.slac.stanford.edu/~swebb/sixpack.htm [5] Webmineral Mineralogy Database, Barthelmy, D., Jan 2008, Available at HTTP: http://webmineral.com/ [6] MINCRYST: Crystallographic and Crystallochemical Database for Mineral and their Structural Analogues, Chichagov, A., Apr 2008, Available at HTTP: http://database.iem.ac.ru/mincryst/ [7] CrystalMaker Software Ltd, CrystalMaker, Oxfordshire, UK, version 1.0.4 [8] Webb, S., "Structural characterization of biogenic Mn oxides produced in seawater by the marine bacillus sp strain SG-1," American Mineralogist, vol 90, 2005, pp 1342-1357 [9] Post, J., "Manganese oxide minerals: Crystal structures and economic and environmental significance," Proceedings of the National Academy of Sciences of the United States of America, vol 96, no 7, pp 3447–3454 10 TABLES Sample Todorokite Shell Mn-O Mn-Mn Birnessite Mn-O Mn-Mn Lithiophorite Mn-O Mn-Mn Chalcophanite Mn-O Mn-Mn Cryptomelane Mn-O Mn-Mn Coronadite Mn-O Mn-Mn Ramsdellite Mn-O Mn-Mn Pyrolusite Mn-O Mn-Mn edge Mn-Mn corner Parameter N R (Å) σ2 N R (Å) σ2 Red Chi Sq EXAFS 7.2 1.89 0.010 7.7 2.87 0.015 234.44 XRD 1.77-1.82 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 4.7 1.91 0.004 8.8 2.83 0.002 1735.60 1.93-2.00 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 4.2 1.91 0.005 4.5 2.92 0.008 204.16 1.95 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 1.89 0.006 2.87 0.007 584.36 1.86-1.97 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 4.6 1.90 0.005 2.9 2.87 0.005 563.78 1.88-1.90 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 4.1 1.89 0.005 2.5 2.89 0.006 271.36 1.95-2.06 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 4.1 1.89 0.004 2.50 2.87 0.005 1051.90 1.87-2.13 N R (Å) σ2 N R (Å) σ2 N R (Å) σ2 Red Chi Sq 4.5 1.89 0.004 2.9 2.88 0.005 2.7 3.44 0.002 138.96 1.88-1.89 2.94 2.85-2.95 2.93 2.81-2.91 2.85-3.454 2.70-3.54 2.86-3.44 2.87 3.43 Constrained XRD fit 1.89 0.008 2.88 0.013 221.11 Δ σ2= -13.3% Δ Red Chi Sq.= -13.33 1.90 0.004 2.83 0.002 1904.78 Δ σ2= 0% Δ Red Chi Sq.= 169.18 1.92 0.007 2.92 0.010 275.19 Δ σ2= +25% Δ Red Chi Sq.= 71.03 1.89 0.007 2.87 0.012 675.20 Δ σ2= +71.4% Δ Red Chi Sq.= 90.84 1.90 0.007 2.87 0.011 676.30 Δ σ2= +120% Δ Red Chi Sq.= 112.52 1.89 0.007 2.90 0.013 333.27 Δ σ2= +116.7% Δ Red Chi Sq.= 61.91 1.89 0.007 2.87 0.011 1248.01 Δ σ2= +120% Δ Red Chi Sq.= 196.11 1.89 0.006 2.88 0.010 3.44 0.006 244.68 Δ σ2= +300% Δ Red Chi Sq.= 105.72 Table 1: EXAFS Parameters N is coordination number, R is radial distance, and σ is disorder Observe that Δσ2 increases when going down the series Also, Δ(Red Chi Sq.) increases for all samples except for birnessite, which is also the only manganese oxide to have a positive Δσ 2, indicating its ideal structure 11 FIGURES Figure 1: Diagram of Experimental Setup Absorption and transmission data is collected by the ion chamber detectors and fluorescence measurements are taken by the Lytle detector Figure 2: Sample background subtraction (birnessite) The red line is the pre-edge function and the blue line is the post-edge function to which the background is removed and edge normalized The edge is at about 6552 eV 12 chi(k) plot Fourier Transform Figure 3: Sample EXAFS fit (birnessite) The red line is the fitted curve to the blue raw data with the single scattering manganese oxide model Figure 4: EXAFS χ(k) plot Notice a trend at 6.8Å-1 that shows an increasingly broader feature in the EXAFS when going down the y-axis from todorokite to pyrolusite The EXAFS indicate todorkite and birnessite have the most ideal crystal structures as predicted by XRD, with more defects and lattice disorder in the other manganese oxides 13 Figure 5: Fourier Transform plot The structures tend to get increasingly less ideal when going down the manganese oxides from todorokite to pyrolusite The first two peaks represent the first Mn-O and MnMn shells Figure 6: Birnessite Figure 7: Chalcophanite 14 Figure 8: Coronadite Figure 9: Cryptomelane Figure 10: Lithiophorite Figure 11: Pyrolusite Figure 12: Ramsdellite Figure 13: Todorokite 15 ... verify identities and refine detailed structures of manganese oxides MATERIALS AND METHODS The x-ray absorption spectroscopy and data analysis on manganese oxides were conducted at the Stanford Synchrotron... thin layer of manganese oxides, iron oxides, and clay, we must examine and refine their structures before being able to use them as biosignatures X-ray absorption spectroscopy (XAS) and XRay Diffraction... good as the fluorescence The theoretical crystal structure for each of the manganese oxides can be seen in Figures through 13 Table is a summary of EXAFS parameters from the Mn K-edge using the single