Surface Chemistry and Device Response of AlGaN/GaN Sensors
- 1. Surface Chemistry and Device Response on AlGaN/GaN surfaces Jeremy Gillbanks – September 2015 Supervised by Prof. Giacinta Parish and Prof. Brett Nener
- 2. Sensor Context 1 Semiconductor Doping High Electron Mobility Transistors Substrate Design Field Effect Transistors Chemical Sensors Chemical Sensors Field Effect Transistors CHEMFETs ISFETs Silicon-based devices Heterostructure- based devices HEMTs AlGaN/GaN AlGaAs/GaAs BioFETs Other Sensors
- 3. Our Sensors 2 Semiconductor Doping High Electron Mobility Transistors Substrate Design Field Effect Transistors Chemical Sensors Chemical Sensors Field Effect Transistors CHEMFETs ISFETs Silicon-based devices Heterostructure- based devices HEMTs AlGaN/GaN AlGaAs/GaAs BioFETs Other Sensors
- 4. AlGaN/GaN Sensors Advantages over traditional ISFET Sensors – Stability – Low cost – No reference electrode Applications – Recycled water monitoring – Lab-on-a-chip sensor arrays 3 AlGaN capped transistor Ren 2008 Sensor array design Asadnia 2015 Active area
- 5. Research Gap Previous research completed by the Microelectronics Research Group at UWA 4 Demonstrating ionic concentration, regardless of pH (2010) Dipolar molecule orientation and sensor response Sensor selectivity toward negative ions (2010) GaN cap has greater affinity to Cl- ions than AlGaN (2014) 2DEG conductivity increase with positive charge build up (2014)
- 6. Project Objectives Aim: Molecular contact angle vs. device response 5 Glycine Benzil (non-polar) 6-Amino-2-Naphthoic Acid Hypothesis: • Adhesion via negatively charged carboxyl group • Dipolar molecules will affect device response via molecular orientation This is the first time dipolar molecular orientation has been investigated on a GaN capped device.
- 7. Molecule Selection Glycine 6-Amino-2-Naphthoic Acid NEXAFS conducted at AS C N Background Correction Choose Step Edge Gaussian Peak Fitting Spectral Subtraction Bond Angle Calculation Molecular Orientation Compare to Device Response O Benzil Experimental Procedure 6 6-Amino-2-Naphthoic Acid only
- 8. Molecule Selection Glycine 6-Amino-2-Naphthoic Acid NEXAFS conducted at AS C N Background Correction Choose Step Edge Gaussian Peak Fitting Spectral Subtraction Bond Angle Calculation Molecular Orientation Compare to Device Response O Benzil Project Scope 7 6-Amino-2-Naphthoic Acid only
- 9. Molecule Selection Glycine 6-Amino-2-Naphthoic Acid NEXAFS conducted at AS C N Background Correction Choose Step Edge Gaussian Peak Fitting Spectral Subtraction Bond Angle Calculation Molecular Orientation Compare to Device Response O Benzil Seminar Scope 8 6-Amino-2-Naphthoic Acid only
- 10. NEXAFS: How it works • Near Edge X-ray Absorption Fine Structure 9 • Incident photon energy is near the edge of the ionisation potential of the scanned atom • Ammeter allows replacement current to be recorded from photoelectron loss • Allows measurement of individual molecular orbitals for C, N and O atoms Experimental Setup Mennell 2015 This is the first time a NEXAFS study has been conducted on a GaN substrate
- 11. Non-linear curve fitting 10 6-Amino-2-Naphthoic Acid Nitrogen K-edge scan
- 12. XPS X-ray Photoelectron Spectroscopy • Composition, thickness 11 6-Amino-2-Naphthoic Acid Nitrogen XPS K-edge scan 6-Amino-2-Naphthoic Acid Gallium XPS K-edge scan Before deposition After deposition After deposition Before deposition
- 14. Spectral Subtraction 13 6-Amino-2-Naphthoic Acid Curve Fitting
- 15. Identifying the peak 14 Measured angle from nitrogen scan: 43.7˚ ± 10˚ Measured angle from carbon scan: 46˚ ± 2˚ (Home 2015) Naphthoic Acid Peak fit at 404 eV (corresponds to C-N σ* bond)
- 16. Angle of naphthoic acid to surface I can corroborate Michael Home’s finding that 6- amino-2-naphthoic acid lies at 44˚ to the device surface. 15 44˚ Device Surface
- 17. Future Work • Ensure adequate coverage • Normalise on the device surface • Test simple alcohols/acids – Methanol – Formic acid – Benzoic acid • Test simple amine groups – Methylamine – Aniline • Test simple amino acids with benzene rings – Meta-, ortho-, or para-amino benzoic acid • Later: test larger molecules – Tyrosine 16 Tyrosine Formic acid Aniline
- 18. Key Points • We have been the first to successfully orientate glycine and 6-amino-2-naphthoic acid on a GaN capped device – Every molecule to be sensed has a specific angle at which it adheres to the surface – The orientation effects the device response • Future work has been successfully identified • Special thanks to – Prof. Giacinta Parish & Prof. Brett Nener – Farah Khir, Matt Myers, Murray Baker – Michael Home, Chris Mennell, Ben Sutton – The III-N research group 17
- 20. Background Correction • Remove oscillations in incident photon intensity over time and energy • Au leaf used (300 eV – 1000 eV) 19 Device setup at the Australian Synchrotron Courtesy: F. Khir
- 22. Benzil 21
- 23. Sources of Error & Biases (1/2) • Noise – Using peak areas instead of peak heights decreases effect of noise on local regression • Local regression formula inadequately smoothed • Back scattered electrons – Reduced to insignificance due to multiple incident angles – Highly energetic photons (with adequate coverage) shouldn’t penetrate the adsorbate • Photoelectrons generated from surrounding atoms • Thermal motion ineffectively averaged between scans 22
- 24. … (2/2) • Monochromator’s linearly polarised light – K-shell spectra are highly polarisation-dependent – Linear polarisation simplifies the dipole matrix element • Replacement current efficiency (resistance, etc…) • Inconsistent incident photon intensity in excess of what is corrected for using the reference foil • Substrate does not display three-fold or higher symmetry • Adsorbate not a homogenous layer • Hydrogen bonds effect spectra in a measurable way • Ineffective spectral subtraction • Adsorbate damaged during x-ray scan 23
- 25. Limitations • Building block model – Used when you have a new molecule that has not been scanned using NEXAFS before • E.g. 6-amino-2-naphthoic acid or benzil – Limitations: • Conjugated molecular orbitals are difficult to identify during deconvolution • More of a problem for carbon K-edge NEXAFS scans 24
- 26. Why not more samples? • AS has high resolution – Resolution: 0.1 eV – Energy Range: ~40 eV – 0.25% steps • Trends successfully identified => conclusions are valid • Common practice is to use best spectra, not to average • Experiment cost: ~$600k 25
- 27. References 1. Title Slide: Substratehttp://pubs.rsc.org/services/images/RSCpubs.ePlatform.Service.FreeContent.ImageService. svc/ImageService/Articleimage/2006/DT/b515727g/b515727g-f5.gif 2. Title Slide: Australian Synchrotron logo https://events.synchrotron.org.au/event/1/picture/10.jpg 3. Title Slide: Microelectronics Research Group http://mrg.ee.uwa.edu.au/images/microelectonicsResearchGrou.gif 4. Slide 5: Glycine http://www.actgene.com/images/Glycine.jpg 5. Slide 5: 6-Amino-2-Naphthoic Acid http://www.sigmaaldrich.com/content/dam/sigma- aldrich/structure6/165/mfcd01861831.eps/_jcr_content/renditions/mfcd01861831-medium.png 6. Slide 5: Benzil http://www.sigmaaldrich.com/content/dam/sigma- aldrich/structure3/116/mfcd00003080.eps/_jcr_content/renditions/mfcd00003080-medium.png 7. Slide 15: 6-Amino-2-Naphthoic Acid http://pubchem.ncbi.nlm.nih.gov/image/img3d.cgi?cid=2733954 8. Slide 16: Formic Acid http://chem-tracking.de/onewebstatic/ed4ba8c401-Ameisensäure.jpg 9. Slide 16: Aniline http://chemwiki.ucdavis.edu/@api/deki/files/9113/aniline.png 10. Slide 16: Tyrosine http://img1.wikia.nocookie.net/__cb20140401122944/resscientiae/images/2/29/Tyrosine.jpg All other slides are of the author’s creation unless otherwise cited. 26
Editor's Notes
- #5: Stability: less drift & non-toxic, even at high power applications or hostile conditions (pH, temperatures, etc…) Low cost: uses current MOSFET (common) manufacturing methods No ref. electrode: smaller sizes => sensor arrays Applications: recycled water monitoring, air pollution monitoring, biomonitoring (anything that’s real-time) Lab on a chip sensor arrays (each sensor can be individually functionalised to detect a particular ion or molecule)
- #6: Sensor selectivity: negative end of molecules sticks to surface GaN cap: better due to better affinity to ions Ionic concentration: suitable for use in an array of sensors 2DEG concentration: changes with molecules impacting the surface Unknown: dipolar molecule orientation and how it impacts sensor response
- #7: Include reasons for molecules Explain Synchrotron Explain how NEXAFS works
- #11: Why the inner shell electrons? Why not the outer ones? – The outer ones vary to widely. Inner ones detail atomic structure.
- #12: Mention what it is physically Why it is important Identify it on the graph Mention where is must be Near the ionisation potential Or at least near to it Explain why there may be a difference (physisorbed) How wide the initial guess must be from literature How high the guess must be Why choose Gaussian peaks How to guess initial locations based on physical bonds How to guess heights/widths based on LOESS operation How it works (assumptions) Limitations of building block method
- #14: N-scan Ga XPS Get substrate spectral locations Subtract from 6A2NA Identify final peak Identify angle
- #15: Get substrate spectral locations Subtract from 6A2NA
- #16: How the peak was chosen Why peak areas rather than peak height? With noisy signal data (like ours) smoothing is required to fit a local regression fit Smoothing distorts the signal and peak height => Use peak area where smoothing is not required 0.3% different from literature value
- #19: From Veronica: - work out where you can drink water - don’t give comments about what people will/won’t understand - easy to listen to - add a UWA border to make it seem less empty - have three key points for the end of the speech - limitations of studies?