Larkin Sayre
MIT Class of 2017
REU in Physics at Howard University May 27th – August 1st 2014
Raman Spectroscopy and COMSOL Multiphysics Simulation Studies of Tungsten Oxide (WO3) as a Potential Metal-Oxide Gas Sensor (MOGS)
Abstract:
This paper analyzes the Raman spectrum of tungsten oxide (WO3) under a variety of temperatures to give insight into its application as a Metal-Oxide Gas Sensor (MOGS) which is a branch of sensors used to detect a wide range of environmental and hazardous gases.By interpreting the change in conductivity of metal oxides when they come into contact with gases, the identity of the gases and their concentrations can be measured. WO3 is a promising metal oxide for gas sensing and therefore the study of its behavior under temperatures up to and within operational temperatures is important to the production, efficiency and life-span of metal oxide gas sensors. This investigation builds on past research by R. Garcia-Sanchez et al.1 that detected a Raman peak at approximately 1500cm-1when WO3 on a silicon substrate was heated to 160ᵒC. This analysis looks at how this peak is altered by cooling from 190ᵒCdown to room temperature. This study finds that the peak appears at approximately 110ᵒC and is initially unaltered by cooling but fades over a period of days when left at room temperature. The peak forms due to W-OH bond formation as a result of the hygroscopic properties of WO3. The approach of using WO3 samples to inform MOGS design is also critically analyzed.
1
Introduction:
Raman spectroscopy of WO3deposits on silicon substrates allows for the analysis of vibrational modes within the molecules of WO3. When the laser interacts with the molecule of WO3some of the light is backscattered inelastically and the energy shift it has undergone can be measured. Each peak of the Raman spectrum produced by this backscattered light corresponds to a vibrational mode within the molecule which gives information on its structure and bonding. Therefore, the spectrum of WO3under different temperature conditions can be used to draw conclusions about the behavior of the metal-oxide and hence make predictions for its use as a MOGS. As part of the project this approach is critically analyzed and conclusions are drawn about the usefulness of Raman spectroscopy in the development of MOGS.
MOGS are used to detect many different environmental and hazardous gases. One of the main MOGS uses of WO3 is in the detection of NOx2. NOx is produced in car exhaust and can be a respiratory irritant. A valuable extension of this work would be to analyze how the exposure of WO3 to NOx affects the Raman spectrum. In this way Raman spectroscopy could give information on the gas-metal-oxide interaction between WO3 and NOx and show whether or not this is a reversible reaction. This project finds that the heating of WO3 has a long-lasting effect on its Raman spectrum and it is possible that NOx interaction is also long-lasting.
Procedure:
6 identical deposits of WO3 were used to repeat the experiment decreasing temperature from 190ᵒCto 30ᵒC at intervals of 20ᵒC. First, Sample 1 was heated to 190ᵒCto verify previous work1 that discovered a peak at approximately 1500cm-1. The temperature-induced peak appeared as predicted between 1500cm-1 and 1600cm-1. This investigation looks at the effect of decreasing temperature from 190 to 30ᵒCto analyze the effect on the peak at 1500cm-1 and the other peaks associated with the Raman spectrum of WO3.
The WO3 used was deposited onto silicon substrates by drop-coating and they were then furnace-fired1. The deposit is about 2.5mm in diameter and 4 µm in thickness. Its structure is a monoclinic3 lattice at the temperatures analyzed. The COMSOL Multiphysics models were created to demonstrate the geometry of the heated cell and substrate and investigate the heat transfer from the heated cell to the WO3. The COMSOL Multiphysics Modeling software was also critically assessed as a possible tool for designing and optimizing MOGS.
The silicon substrates are approximately2.55cmx1.15cm with a thickness of 0.1cm. The circular deposits of WO3 have radius of approximately 0.25cm. The substrates are fastened to a heated cell by a small plate. The heater allows for temperatures from room temperature to 200ᵒC. The temperatures analyzed in this paper are from 30-190ᵒCin intervals of 20ᵒC. Some Raman Spectrometers have the capability to heat samples built into the stage where samples are placed but the spectrometer used for this investigation, the ThermoScientific DXR SmartRaman Spectrometer, does not have this capability. Therefore the silicon substrates were clamped to the heated cell in Figures 1-3 to be brought up to the desired temperatures. The heated cell remained attached while the Raman spectra were collected.
COMSOL Modelling:
Figure 1
Figure 2
Figure 3
Figures 1-3 show COMSOL models of the heated cell and silicon substrate with WO3 deposit
During this project COMSOL was assessed as a possible software package to optimize design of MOGS4. Figures 1-3 show COMSOL models that demonstrate the geometry of the heated cell and the silicon substrate with the circular deposit of WO3. COMSOL allows each section of the geometry to be assigned a material and if the material is not available in the built-in library, the coefficients of the material can be used to create a new material. In this way, WO3 can be modeled by COMSOL even though it is not present in the material library. COMSOL can couple multiple physics phenomena during modeling which makes it ideal for MOGS design. Heat transfer of heaters, gas interactions and structural mechanics modules can be used to optimize the sensor design.
Results:
Figure 4 – The Raman spectrum of monoclinic WO3decreasing temperature from 190ᵒC (top spectrum) to 30ᵒC(bottom spectrum) at intervals of 20ᵒC. Blue arrow indicates heat induced peak.
The most significant change in the Raman spectrum from room temperature up to 190ᵒC is the appearance of the peak at approximately 1500cm-1. The peak appeared at approximately 110ᵒC. Preliminary results indicate that time elapsed at a certain temperature does not affect the height of the new peak. However, increasing the temperature does increase the peak size when increased from 110ᵒCto 190ᵒC. Only prolonged time at room temperature allows for the peak to slowly diminish. (See Figure 4 below). All other characteristic peaks of monoclinic WO3 appear to be unaltered by cooling and heating.
There does appear to be some decrease in peak height in Figure 4 (below) when the sample was cooled from 190ᵒCto 30ᵒC but this may be due to experimental error and it would need to be repeated to verify.
Figure 5 (overleaf) demonstrates the peak trend over 72 hours. The top spectrum is directly after cooling to room temperature from 190ᵒC. The subsequent spectrums show the peaks every 24 hours. As demonstrated, the heat-induced peak does not diminish a significant amount over this period. Further studies are needed to investigate how long it takes the peak to disappear.
Figure 5 - The Raman spectrum of monoclinic WO3 from 0 to 72 hours after heating to 190ᵒCat 24 hour intervals. Blue arrow indicates heat induced peak.
Conclusion:
The added peak formed on heating is due to W-OH bonds interacting more strongly with the laser at increased temperatures. The reason why the peak does not disappear immediately once returned to room temperature is unknown.
Limitations:
MOGS are created with very specific properties and therefore research towards their more effective design and application requires much more faithful imitation of the actual sensor design5. The substrates used in this project, somewhat resemble the substrates used in sensors but not closely enough to directly inform MOGS development. For example, the addition of electrodes across the WO3is one of many factors that would alter the results obtained by this research. This research, therefore, gives insight into the effects of temperature on monoclinic WO3 below operational temperatures of MOGS but more ‘operando’5 spectroscopy techniques would be needed to apply to MOGS development.
Possible Further Studies:
The effect of cooling below room temperature could provide insight into the mechanism behind the peak formation and disappearance. This would show whether the extra decrease in temperature would speed up or slow down the disappearance of the peak and could help in understanding the peak’s formation.
The analysis of dampened WO3 deposits would also provide further insight into the mechanism of temperature-affected peaks and the effect of the presence of water on the Raman spectrum of the WO3.
Possible Further Studies with LAMMPS:
LAMMPS6 is an open source code that performs Molecular Dynamics simulations. By creating an input text file that specifies lattice spacing, atom type, atom interactions and other parameters of the system we are trying to model, it outputs a ‘dump’ file. This ‘dump’ file can be run by a Molecular Dynamics visualization package such as Ovito or VMD. With LAMMPS, the trajectories of atoms can be calculated and complicated interactions can be simulated. During this project the structure of a single sheet of graphene was produced using LAMMPS. This relatively simple example could be extended to model the more phenomena, for example water interacting with the sheet of graphene. This simple lattice example illustrates the possible uses of LAMMPS to model the lattice of WO3 and its interactions with other molecules.
LAMMPS has four main components that need to be specified in each input text file. They are: initialization, atom definition, settings and running simulations. The ‘initialization’ dictates units, dimensions, atom styles etc. The next step, atom definition, dictates the lattice spacing of the system. The third section, settings, includes properties such as pair and bond coefficients. Most simulations require pair interactions under ‘pair_style’ and ‘pair_coefficients’. Finally a simulation can be set up for the molecules produced. The main command used here is ‘run’ followed by the number of time-steps, keywords and values that define the specific simulation.
LAMMPS would be a valuable extension of this research and could create useful simulations of WO3 and NOx interactions which could more directly inform gas sensor design by giving information on the nano scale. The combination of Raman spectroscopy and LAMMPS (for the nano scale) along with COMSOL (for the macro aspects) provides a powerful study of WO3 as a potential MOGS.
Author Information:
Email: Tel: 206-356-5328
Acknowledgements:
The author would like to acknowledge the NSF for financial support of the Research Experience for Undergraduates (REU) summer program and the Howard University REU in Physics Site (NSF Grant PHY-1358727).
Mentoring and instruction from Professor Prabhakar Misra, Raul Garcia-Sanchez and Daniel Casimir of the Howard University Department of Physics and Astronomy is gratefully acknowledged.
COMSOL Multiphysics is also recognized for module trials, workshops and webinars as well as use of the COMSOL Modeling software.
References:
(1) Garcia-Sanchez, R.; Ahmido, T.; Casimir, D.; Baliga, S.; Misra, P. Thermal Effects Associated with the Raman Spectroscopy of WO3 Gas-Sensor Materials. J. Phys. Chem. A 2013, 117, 13825-13831
(2) Penza, M.; Martucci, C.; Cassano, G. NOx Gas Sensing Characteristics of WO3 Thin Films Activated By Noble Metals (Pd, Pt, Au) Layers. Sensors and Actuators B, 1998, 52-59
(3) Wang, F,; Di Valentin, C.; Pacchioni, G. Electronic and Structural Properties of WO3: A Systematic Hybrid DFT Study. J. Phys. Chem. C 2011, 115, 8345-8353
(4)Ahuja, M.; Talwar, R.; Prasad, B. Simulation of MEMS Microhotplate For Metal Oxide Gas Sensors: Review: International Journal of Electronics and Electrical Engineering, Vol. 1, Spl. Issue 2, pp. 1-4, May 2014
(5) Gurlo, A,; Riedel, R. In Situ and Operando Spectroscopy for Assessing Mechanisms of Gas Sensing. Angew. Chem. Int. Ed. 2007, 46, 3826-3848
(6) S. Plimpton,Fast Parallel Algorithms for Short-Range Molecular Dynamics, J. Comp. Phys. 1995, 117, 1-19
1