European Network on New Sensing Technologies for Air Pollution Control and Environmental Sustainability - EuNetAir COST Action TD1105 2 nd International Workshop EuNetAir on New Sensing Technologies for Indoor and Outdoor Air Quality Control ENEA - Brindisi Research Center, Brindisi, Italy, 25-26 March 2014 An account of our efforts towards air quality monitoring in epitaxial graphene on SiC Jens Eriksson jenser@ifm.liu.se Applied Sensor Science, Linköping University/ Sweden COST is supported by the EU Framework Programme ESF provides the COST Office through a European Commission contract
Outline Why graphene sensors? Epitaxial graphene on SiC Effect of graphene layer thickness on gas sensitivity and selectivity Understanding unintentional doping in epitaxial graphene Controlling graphene layer uniformity Tuning sensor properties by surface modifications 2
Why Graphene sensors? Unique band structure of graphene leads to a low density of states near the Dirac point (E D ) small changes in the number of charge carriers results in large changes in the electronic state Every atom at the surface ultimate surface to volume ratio Low noise, chemically stable (in non-oxidizing environment) p p* e <1eV Graphene is highly sensitive to chemical gating due to its linear energy dispersion and vanishing density of states near the Dirac point and therefore has potential as a low noise, ultra-sensitive transducer. e Graphene sensors are normally highly sensitive, but suffer from poor reproducibility, selectivity, and speed of response... d Reproducibility is an issue that partly arises from the graphene synthesis 3
manufactures and supplies Graphene on SiC 4 Sublimation of Si from SiC in Ar at 2000ºC Scalable, wafer-scale films compatible with standard semiconductor processing High thickness uniformity (> 90% ML, rest 2 ML) Thickness controlled by temperature 2013-11-20 Jens Eriksson Linköping 2013 Spin off from Linköping University, Sweden 22.11.2011
Silicon carbide Ceramic High chemical inertness Oxidation resistant Stable at high temperature Hardness Melting point ~ 2700 C Light-weight Semiconductor Wide band gap High electron drift velocity High breakdown field High thermal conductivity Polytypism: > 200 chemically identical polytypes E g = 3.0 ev E g = 2.4 ev E g = 3.2 ev 5
Graphene production Graphene layers sit on a buffer or interfacial layer The buffer layer is covalently bound to the underlying SiC Electronic coupling between SiC and graphene C. Riedl et al. PRL 103 (2009) Ideal Epitaxial Graphene on SiC has E F pinned above the Dirac point Causes electron doping! Hall measurements show that our graphene has Ns 10 12 cm -2 A. Tzalenchuk, et al, Nat Nano, 5, 186-189, (2010) S. Sonde et al., Physical Review B 80, 241406 (R) (2009) ARPES: E F 0.4 ev above E D S. Y. Zhou, et al., Nat. Mater. 6, 770 (2007) 6
Sensor response to environmental gating NO 2 strongly electron withdrawing 1.80 Single layer 100 Multi layer Resistance change R/R 0 1.60 1.40 1.20 1.00 80 60 40 20 NO 2 Concentration [ppb] in N 2 0.80 0 5 10 15 20 25 30 35 Time [h] 0 Large n-type response to ppb concentration NO 2 Small p-type response to ppm concentration NO 2 Why is single layer more sensitive? Current flow through all layers gas adsorption only on top layer Different band structure leads to different responsivity; change resistivity with carrier density Or difference in sticking coefficients of gases on single and multi layer graphene R. Pearce et al. Sens. and actuators B. Chem., 155(2): 451-455, 2011 7
NO 2 sensing, single or double layer graphene?
V (mv) Counts (%) Scanning Kelvin probe microscopy work function mapping Nanoscale mapping of graphene thickness uniformity and doping Topography is mapped in 1 st pass Surface Potential is mapped in 2 nd pass Maps difference in work function Morphology Δ V pot 5 nm 100 mv Potential distribution 10 10 µm 2 Shifts from single to bilayer domains normally occur at terrace edges 50 40 30 20 10 0-10 2L Δv pot 50 mv 1L 2L 0 300 600 900 1200 1500 Distance (nm) 20 18 16 14 12 1L 88 % 1L 10 8 6 4 2L 2 0-20 -10 0 10 20 30 40 50 60 V (mv) ΔΦ between 1LG and 2LG allows nanoscale mapping of graphene thickness Controllable environment allows observing changes in 1LG and 2LG upon gas interaction Eriksson et al.,applied Physics Letters 100 (2012) 24160 9
SKPM in controlled environment NO 2 : Electron withdrawing gas increases ΔV CPD, 2L-1L NO 2 on Ambient 2LG 1LG NO 2 off V CPD (1LG) and V CPD (2LG) decrease, but V CPD (1LG) decreases more In N 2 : after vacuum Φ 1LG = Φ 2LG 1LG 2LG Corrugations in 2LG upon repeated gas exposure and vacuum cleaning 10
Different shifts for 1LG and 2LG? Response to < 1 ppm NO 2 vs. time Different energy dispersions Linear for 1LG Parabolic for 2LG From 1-2L ΔV CPD : Non-invasive estimation of carrier concentration R. Pearce, J. Eriksson, T. Iakimov, L. Hultman, A. Lloyd Spetz and R. Yakimova, ACS Nano 7 (5), pp 4647 4656 (2013) Calculated change in carrier concentration not the same for 1 and 2LG Different responsivity for 1 and 2LG doesn t account for all difference in sensitivity Different sticking coefficients also important 11
Effect of humidity on surface potential Environment affects the surface potential V CPD (1LG) decreases V CPD (2LG) constant 12
Controlling graphene layer uniformity and unintentional doping Large spread observed in N S for samples grown under identical conditions There is strong indication that a correlation exists between the substrate surface morphology and the electronic properties of the epitaxial graphene. Yakes et al., Nano Lett. 10, 1559 1562 (2010) 13
1LG (%) Mono layer coverage depends on terrace width 3C-SiC 100 90 80 70 60 50 10 Terrace width < 300 nm no 1LG 7 5 2 11 1 40 6 30 200 400 600 800 1000 1200 1400 1600 1800 2000 Terrace width (nm) As the terrace width increases, the area covered by 1LG increases Graphene growth starts at step edges; many step edges many nucleation sites Terrace width > 1200 nm gradual decrease of 1LG - Island growth in the absence of steps Substrate polytype and doping for hexagonal SiC (n-type 6H-SiC or SI 4H-SiC) do not significantly influence uniformity 3C-SiC higher 1LG % for lower terrace width, 1LG % independent on terrace width 3 4 8 12 9 14
Carrier concentration depends on SiC surface Φ (so N S ) depends strongly on terrace width 300 250 3C-SiC N S 3C-SiC: Lower doping, independent on terrace width 1LG (mev) 200 150 100 50 0 400 600 800 1000 1200 1400 1600 1800 Terrace width (nm) Unintentional doping Unintentional doping For 1LG: For 1LG: N Variations in Φφ 1LG follow vs. terrace variations width in E F : D 2 0 F E 2 Eriksson et al., Applied Physics Letters 100 24160 (2012) Variations in φ 1LG follow variations in E F : Scattering indicates that also other factors affect E F ΔN D,Max 8 10 12 cm -2 ΔN D,Max 8 10 12 cm -2 σ ΔN D 6 1011 cm -2 σ ΔN D 6 1011 cm -2 15
Height (nm) Height (1nm) Height (5nm) Height (pm) Surface restructuring during Si sublimation 4H-SiC SiC substrate 5 5 µm 2 3C-SiC SiC substrate 10 10 µm 2 800 600 400 SiC Graphene 200 0-200 -400-600 -800 0,00,51,01,52,02,53,03,54,0 Distance ( m) All substrates undergo significant restructuring during graphene growth Graphene Graphene 5 5 µm 2 10 10 µm 2 Step profile -2.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Distance ( m) Differing restructuring of different nominally on-axis SiC substrates 1.5 1.0 0.5 0.0-0.5-1.0-1.5 No correlation seen between SiC step distance before growth and how much the SiC restructures upon graphene growth SiC EG Terrace width ( m) Change in terrace width 1,4 1,2 1,0 0,8 0,6 0,4 0,2 0,0-0,2 0 200 400 600 800 Substrate terrace width (nm) 3C-SiC 3C-SiC restructures less, and even a reduction of the terrace width is possible 16
1LG (%) Effects of surface restructuring 90 80 70 60 50 a ) -200 0 200 400 600 800 1000 1200 Terrace width (nm) More significant restructuring leads to less uniform graphene (mev) 40 30 20 10 0-10 -200 0 200 400 600 800 Terrace width (nm) Work function decreases (n-doping increases) with amount of restructuring Minimize the restructuring Use 3C-SiC substrates Due to less step-bunching, 3C-SiC better lends itself to a well-controlled surface morphology and better control of the electronic properties of the graphene Eriksson et al., Mater. Sci. Forum 740-742 (2013) 153-156 17
Step bunching in 4H, 6H, and 3C-SiC 4H-SiC 4H-SiC 6H-SiC 3C-SiC 3C-SiC Yazdi et al., Carbon 57 477 484 ( 2 0 1 3 ) 18
Normalized Resistance Uniform 1LG leads to very reproducible sensor characteristics ΔS depends on thickness due to differing band structures for 1LG, 2LG... MLG Ambient 1 ppm NO 2 1,25 Uniform 1LG leads to very reproducible sensor characteristics 1,20 1,15 NO 2 sensor 1 sensor 3 sensor 2 1,10 1,05 Air 100 ppb 50 ppb NO 2 withdraws electrons Same change in charge carriers causes larger shift of the Fermi energy for 1LG R. Pearce, J. Eriksson, T. Iakimov, L. Hultman, A. Lloyd Spetz and R. Yakimova, ACS Nano 7 (5), pp 4647 4656 (2013) 1,00 60 70 80 90 100 110 120 130 140 Time (hours) NO 2 sensing interesting for: Emission control (few ppm) Air quality control (few ppb) 1LG is more sensitive to NOx than 2LG or MLG Uniform 1LG required for maximum sensitivity and reproducibility Different sensors fabricated on 100% 1LG show identical response Epitaxial graphene on SiC enables highly reproducible sensor fabrication 19
Graphene sensors issues: selectivity, response/recovery time, reproducibility Obstacles: sensitivity, selectivity, response/recovery time, reproducibility Sensitivity UV cleaning : ppt and ppq level detection surpassing specially trained dogs! o O-functionalizes graphene Selectivity Chen et al., Applied Physics Letters 101, 053119 (2012) Sensors and Actuators B 166 167, 172 176 (2012) Surface functionalizations by e.g. oxygen, nanoparticles, defect engineering, smart operation and smart analysis Response/recovery times J. Mater. Chem. 22, 11009 (2012) Functionalization, current-, bias- or temperature cycling. Integration of UV-LED Reproducibility Sensors on epitaxial graphene Variations in Ns can be compensated by the use of FET sensor SenSiC AB patent application 20
Functionalization with metal and metal oxide nanostructures for selectivity tuning Graphene growth Sputtering nanopouros metal Contact deposition Sensor fabrication Before metallization AFM, SKPM, Kelvin Probe After metallization AFM, SKPM, Kelvin Probe Gas testing Aim: To develop a reproducible method for functionalization with nano structures Thin layers of Au and Pt DC sputtered onto EG/SiC at elevated pressure Ideally we want islands or nanoparticles to maximize metal-graphene-gas boundaries Au, Pt Epitaxial graphene SI 4H-SiC on-axis 21
Functionalization with metal and metal oxide nanostructures for selectivity tuning Scanning Kelvin probe microscopy: Maps surface morphology and surface potential ΔΦ between 1LG and 2LG allows nanoscale mapping of graphene thickness (and doping) As-grown As-grown 5 nm otential Morphology Morphology Surface potential 100 mv 5 5 nm nm Surface potential Morphology Thin porous metallization Thin porous metallization 100 100 mv mv Surface potential Surface potential Surface potential Surface potential Morphology Morphology Morphology Surface potential Morphology 1LG 1LG 1 µm 2 1 1 µm µm 2 2 1 µm 2 1 µm 1 1 µm µm 2 2 2 1 1 µm µm 2 2 1LG 1LG 2LG 1LG 1 1 nm nm 2LG 2LG 10 10 µm µm 10 10 µm µm As-grown Au 55 nm Au 2 nm Pt Pt 22 nm Pt Pt 0-1 nm Au 5 nm Au 2 nm Pt 2 nm Morphology shows deposition of continuous porous metal ideally: islands 1LG/2LG potential contrast: surface retains the electronic properties of graphene Pt wets the surface better than Au screens the graphene for thick depositions 22
Effect of decoration on sensor response As-grown graphene Au decorated graphene Resistance change R/R 0 1,7 As grown graphene 140 1,6 100 C 120 1,5 100 1,4 80 1,3 60 1,2 40 1,1 20 1,0 0 0 5 10 15 20 25 30 35 Time (hours) Effects of metallization: Improved speed of response Improved detection limit More stable base line Response to ppb concentrations of NO 2 NO 2 concentration (ppb) in N 2 Resistance change R/R 0 2,0 1,8 1,6 1,4 1,2 Au on graphene 100 C 1,0 0 0 2 4 6 8 10 12 14 16 18 20 22 Time (hours) Suppressed response to H 2 /CO while maintaining NO 2 response (Au < 5 nm) RT (b) 1000 800 600 400 200 NO 2 concentration (ppb) in N 2 Resistance change R/R 0 Selectivity 0 2 4 6 8 10 12 14 16 18 Time (hours) Response % Response Time (min), 50 ppb NO 2 Recovery Time (min) As-grown Au, 5 nm Pt, 5 nm As-grown Au, 5 nm Pt, 2 nm 30% 6 1.5 2.3 316 14 14,8 60% 23 9 10.9 834 47 49 90% 99 74 41.7 2136 135 175,5 1,6 1,5 1,4 1,3 1,2 1,1 1,0 0,9 0,8 0,7 0,6 0,5 500 ppb NO 2 40 ppm NH 3 400 ppb NO 2 50 ppm NH 3 100 ppb NO 2 250 ppm H 2 RT 50 ppb NO 2 100 C 500 ppm CO 30 ppb NO 2 J. Eriksson, D. Puglisi, Y. H. Kang, R. Yakimova, A. Lloyd Spetz, Physica B 439, 105 108 (2014) 23
Sensor response(%) Increased sensitivity 45 40 Metal decoration leads to increased interaction due to 35 30 25 20 15 10 5 nm Au on graphene 0 100 200 300 400 X Increased spillover zones Metal-graphene e - transfer Increased O ad O - e - e - e - e - e - O - Graphene SiC X NO 2 concentration (ppb) in N 2 / 20% O 2 Detection limit < 1 ppb Porous metal grains or nanoparticles increase the probability of interaction between the graphene surface and adsorbates 24
Concentration (ppb) Designed Nanoparticles by Pulsed Plasma It is expected that decoration with different metals or metal-oxide nanostructures will allow careful targeting of selectivity to specific molecules Plasma-based nanoparticle (NP) synthesis process Highly reproducible thin film deposition technique Preliminary show that TiO 2 NPs allow enhanced sensitivity towards formaldehyde and benzene The effect depends on the size of the deposited NPs (< 5 nm, sensitive to benzene, > 50 nm, sensitive to formaldehyde) 1812 1810 125 C 724 30 Resistance 1808 1806 1804 1802 1800 1798 500 ppb formaldehyde Annealed at 250 C before test 20% r.h. Resistance 722 720 CH 2 O C 10 H 8 C 6 H 6 20 10 1796 120 150 180 210 240 270 300 25 Time (min) 718 5 10 15 Time (h) 0
CONCLUSIONS Sensing with epitaxial graphene promising, ppb level NO 2 detection Obstacles (selectivity and speed) are being overcome Thin (0.5 5 nm), porous decoration can result in improved selectivity, sensitivity, stability, and response/recovery times The effect depends on the choice, thickness, and nanostructure of the decoration Air quality control: ppb level detection limit required, a likely application Emerging interest in detection of VOCs in living environments ppb level detection crucial. Graphene is an excellent candidate It is expected that decoration with different metals or metal-oxide nanostructures will allow careful targeting of selectivity to specific molecules 26