REU in Physics at Howard University Raman Spectroscopy, COMSOL Multiphysics and Molecular Dynamics Simulation Studies of Tungsten Oxide (WO3) as a Potential Metal-Oxide Gas Sensor Larkin Sayre

About my project About me Rising sophomore Interested in majoring in physics and mechanical engineering My project is in Professor Misra’s Spectroscopy Lab Working with Raul Garcia and Daniel Casimir Project focused on tungsten oxide and its interaction with NOx

Metal-Oxide Gas Sensors (MOGS) The basic principle The conductivities of metal oxides change when they undergo reversible reactions with the gases we are trying to detect This conductivity change can be measured and used to identify the gases present 4 components of MOGS: gas sensing material, substrate, electrodes, heater. Applications: Environmental – gases associated with climate change Safety – sensing harmful gases - NOx

Overview of the project Main goal: Look at behavior of WO3 and its interaction with NOx 3 main aspects of my project: Raman Spectroscopy – the molecular structure of WO3 COMSOL modelling – the macro side LAMMPS simulations – the nano side

What is Raman Spectroscopy? The basic principle: A laser is directed towards the molecule and the scattered light is detected and interpreted. Key points: Rayleigh Scattering Raman Scattering Equipment Thermo-Scientific DXR SmartRaman Spectrometer Interpretation of the spectra produced

Using the Equipment - Procedure Silicon substrate The sensors must first be calibrated The sample is placed in a plastic holder Short test iterations to ensure laser is hitting the sample Top view WO3 deposit Laser

Analyzing the Spectrum Sample of polystyrene used Analyzing the Spectrum Examples of peak assignments: Peaks at 1002, 1602, 1583 and 620 cm-1 correspond to benzene ring vibrations 1002 – “ring breathing mode” 2800-3100 – C-H stretching vibrations Units are “wavenumber” – 1/wavelength

Effect of heating on the Raman Spectrum of WO3 Raman spectra increasing temperature from 30 Celcius to 190 Celcius. Raman spectra decreasing temperature from 190 Celcius to 30 Celcius

Raman Spectrum of WO3 WO3 is the most used metal oxide for sensing NOx molecules Investigating the effects of exposure to NO on the bonding in WO3 using Raman Spectroscopy WO3 is layered on a silicon substrate and exposed (for a set amount of time) to NO Exposure time must be adjusted and controlled to attain results

Exposure of WO3 to NO The hypothesis is that the reaction between WO3 and NO is NON- REVERSIBLE and therefore will produce a permanent change in the Raman spectrum of the WO3 when NOx is absorbed NOx – Environmental pollutants Why is WO3 a good NOx sensor? Large range of operational temperatures (200-500 degrees Celcius) N-type, transition

Using COMSOL Multiphysics to model Metal Oxide on Silicon Substrate

Results My model outputs plots for: Temperature Electric Potential Isothermal Contours

LAMMPS and Molecular Dynamics Simulation LAMMPS Citation: S. Plimpton, Fast Parallel Algorithms for Short-Range Molecular Dynamics, J Comp Phys, 117, 1-19 (1995), http://lammps.sandia.gov/ LAMMPS and Molecular Dynamics Simulation Large-scale Atomic/Molecular Massively Parallel Simulator LAMMPS is a program that carries out molecular dynamics simulations It predicts how the system of atoms will behave using classical mechanics approximations (Newton’s Equations of Motion) How does molecular dynamics relate to research using Raman Spectroscopy? Simulating the vibrational modes of the molecules Using trajectories to model Raman spectrum

LAMMPS Output lmp_serial.exe < wo3_attempt_larkin.txt Produced 250 atoms Information on computational cost

Visualizing the results VMD – Visual Molecular Dynamics

Conclusion Where do I go from here? Continue to improve my LAMMPS and COMSOL models COMSOL workshop in Greenbelt July 8th Carry out exposures of WO3 to NO and investigate effect on Raman Spectrum Continue to Investigate behavior of WO3

Acknowledgements Raul Garcia and Daniel Casimir Professor Misra NSF for REU funding

Infrared vs. Raman Spectroscopy Both detect photons emitted Elastic vs. inelastic scattering Rayleigh and Raman scattering IR is absorption spectroscopy – photons absorbed have the SAME wavelength as those emitted Why do we use Raman spectroscopy for these experiments? Raman active transition – change in polarizeability of the molecule IR active transition – dipole moment change This is why IR and Raman spectra are often complimentary

Polarizability For a vibration to be Raman-active there must be a change in polarizability Polarizability is the ease with which an electron cloud is distorted by an external electric field. The electric field of the incident laser acts interacts with the electron cloud

Polystyrene To help me learn how to use the equipment and interpret the data I took the Raman Spectrum of polystyrene which has a polymer structure of: Phenyl and CH2 groups – The bonds within the molecule dictate how the spectrum will look Picture source: http://www.pslc.ws/macrog/tact.htm

What does the Raman Spectrum of a molecule show? Some photons strike the molecule inelastically These interactions cause vibrations in the bonds of the molecules (e.g. stretching and rotation) and changes in energy levels The Raman shift is the signature of these vibrations detected by the machine

The importance of laser intensity and frequency By varying laser wavelength different spectra can appear for the same molecule For instance: Raman Spectrum of glucose is studied at two different laser wavelengths and the resulting spectrum has peaks in different places even though the sample is unchanged http://web.mit.edu/spectroscopy/doc/papers/2007/Noninvasive%20glucose%20_Shih_07.pdf This is because different laser intensities will excite different vibrational modes within the molecule and give different spectra The choice of laser wavelength has an important impact on experimental capabilities: Sensitivity - Raman scattering intensity is directly related to wavelength Spatial resolution

Analogy – Hooke’s Law When visualizing the vibrations in the molecule caused by the laser, classical mechanics analogies can be used Hooke’s Law F = -kx Lighter portions of the molecule – higher frequency vibrations – and vice versa Examples: Comparing C=C (put sample wavenumber here) and C-H (and here) Ring-breathing mode of benzene ring

Analyzing the spectrum continued… The peaks I found (in cm-1): 620.93 – v(6b) radial ring stretching mode 795.84 1002.54 – ring breathing mode 1031.66 1155.48 1182.52 1450.48 – CH2 stretching mode 1583.37 – ring vibrational mode 1602.55 – ring vibrational mode The peaks in the 3000cm-1 range are characteristic of the C-H vibrational modes in the polystyrene. Polystyrene has repeated CH2 groups. Heavier portions of the polymer have lower frequencies while the lighter portions (C-H) have higher frequencies.

Comparison to standard Raman Spectrum of Polystyrene (courtesy of Thermo-Fisher) Spectrum from Thermo-Scientific library contained in the DXR SmartRaman Machine Spectrum that I collected The ThermoScientific DXR SmartRaman equipment – built-in libraries of standard spectra The minor differences in wave numbers is most likely due to variation in resolution and frequency of the incident light and artifacts from impurities on the sample Similarity between spectra gives confidence that my technique was correct

Comparison to polystyrene spectrum found online Source: http://www.chem.ualberta.ca/~mccreery/ramanmaterials.html The internet source gives a spectrum with three fewer peaks than my spectrum. They are circled in blue. This may be simply due to the resolution of my spectrum being better. It is easier to see the peaks especially in the approximately 3000cm-1 range and therefore easier to assign them.

Daily Spectrum Quiz This spectrum shows CO2 The top left spectrum is of CCl4 and the bottom is of C2Cl4. The difference is the C=C double bond that gives the shift in the spectrum and the peak at 1576. The peak at 1576 is weak because the C=C bond is strong. These two spectra are closely related. The major difference is the peak at 1576 on the right hand spectrum. What does this represent?

Detailed Analysis of CCl4 Spectrum 3N – 6 normal modes 15 - 6 = 9 normal modes Peak at 454cm-1 is from the symmetric stretching of the C-Cl bonds From literature (http://www.physics.rutgers.edu/ugrad/3 89/raman/raman.pdf) I found that the 770 peak in fact contains two peaks at approximately 760 and 790cm-1 The resolution of this spectrum was not sufficient to distinguish the two Source: http://ed.augie.edu/~viste/Raman/RamanQuantum.html

Mystery Spectrum of 3 DIFFERENT Molecules Solution My thought process: All three spectra MUST be related (peaks correspond closely) - indicates trend in molecular structure. I recognized some of the characteristics of phenyl groups that were present in polystyrene. Therefore, top spectrum is benzene. The second and third spectra therefore definitely have phenyl groups plus extra groups attached Simplest addition to the benzene ring is successive H and CH3 groups. Third spectrum is of toluene Image source: http://pubs.rsc.org/en/content/articlehtml/2008/cp/b713965a

H H H C C O H H H Mystery Spectrum Molecular Structure of Ethanol: How I figured out what molecule the spectrum shows: 1460 peak is characteristic of CH2 stretching 1055 is an in-plane bending in the CH3 group and 1280 is a C-C stretch Therefore it must have CH3 and multiple carbons I guessed that it has two carbons based on the relatively few peaks present Next I considered functional groups to add to CH3-CH2 Adding OH to make ethanol - works as a solution H H

Using COMSOL Tetrahedron and helix difference Created using toroid and multiple cone difference and union functions Tetrahedron and helix difference Using Geometry tools to create more complex shapes: Tools such as union and difference under Boolean operations can build complex geometries from simple primitives.

Using COMSOL COMSOL is a CAD modelling software that creates simulations of real-world systems. It is heavily used by researchers and academics and it is a valuable skill for me to pick up during my REU. The classic simulation example is the busbar with DC current running through it producing Joule heating. This heating can be mapped by COMSOL and displayed as below. The bar section is copper while the pins attached are titanium.

Results of busbar joule heating simulation

Mapping of electric potential of busbar Demonstrates charge distribution of the busbar

Example Geometry using a workplane Credit: Raul Garcia

3D results showing Electric Displacement Field Norm Credit: Raul Garcia

Metal oxide deposited onto substrate via CVD

Iron electrodes

Heated cell Heated cell with silicon substrate placed on top for heating

Tutorial with MEMS Joule Heating

Tutorial with Heating Circuit

Input Text File

LAMMPS Commands

Aspect I found most interesting: Maryland NanoCenter Talks by scientists working in nanotechnology Poster sessions for researchers to present their investigations Aspect I found most interesting: Researchers talking about piezoelectric materials – applications of nanotechnology http://www.nanocenter.umd.edu/nanoday/

Planetarium and Telescope Observatory

Technical Writing Workshop