Local Anodic Oxidation of GaAs: A Nanometer-Scale Spectroscopic Study with PEEM

Local Anodic Oxidation of GaAs: A Nanometer-Scale Spectroscopic Study with PEEM S. Heun, G. Mori, M. Lazzarino, D. Ercolani, G. Biasiol, and L. Sorba Laboratorio TASC-INFM, 34012 Basovizza, Trieste A. Locatelli Sincrotrone Trieste, ELETTRA, 34012 Basovizza, Trieste

Outline A brief introduction to spectromicroscopy The spectroscopic photoemission and low energy electron microscope (SPELEEM) Local Anodic Oxidation (LAO) of GaAs

Motivation Why XPS? chemical state information surface sensitive ease of quantification (in general) nondestructive

Motivation Why XPS? chemical state information surface sensitive ease of quantification (in general) nondestructive Why spectromicroscopy? semicond. nanostructures: self-organized islands (dots) F. Ratto et al.: Appl. Phys. Lett. 84 (2004) 4526.

Motivation Why XPS? chemical state information surface sensitive ease of quantification (in general) nondestructive Why spectromicroscopy? semicond. nanostructures: self-organized islands (dots) surface faceting (wires) F.-J. Meyer zu Heringdorf et al.: Phys. Rev. Lett. 86 (2001) 5088.

Motivation Why XPS? chemical state information surface sensitive ease of quantification (in general) nondestructive Why spectromicroscopy? semicond. nanostructures: self-organized islands (dots) surface faceting (wires) carbon nanotubes S. Suzuki et al.: Appl. Phys. Lett. 85 (2004) 127.

Motivation Why XPS? chemical state information surface sensitive ease of quantification (in general) nondestructive Why spectromicroscopy? semicond. nanostructures: self-organized islands (dots) surface faceting (wires) carbon nanotubes catalysis, chemical waves A. Locatelli et al.: J. Am. Chem. Soc., in press.

Motivation Why XPS? chemical state information surface sensitive ease of quantification (in general) nondestructive Why spectromicroscopy? semicond. nanostructures: self-organized islands (dots) surface faceting (wires) carbon nanotubes catalysis, chemical waves surface magnetism (XMCD) A. M. Mulders et al.: Phys. Rev. B, submitted.

Spectromicroscopy at Elettra 6 1: Spectromicroscopy 16 3 5 2: ESCA Microscopy 3: Nanospectroscopy 6 4: TwinMic 5: Microfluorescence 1 6: IR Microscopy 2 3 4 2.0 GeV / 320 ma 2.4 GeV / 140 ma

Scanning vs. direct imaging type Monochromatic light Sample (scanned) Photo-emission detector Detector Photon optics is demagnifying the beam: Scanning Instrument Whole power of XPS in a small spot mode Flexibility for adding different detectors Rough surfaces can be measured Limited use for fast dynamic processes Lower lateral resolution than imaging instruments Sample hv Stigmator Intermediate lens Projective lens MCP Screen Object plane Object lens Backfocal plane Intermediate image Intermediate image Image plane CCD camera Electron optics to magnify irradiated area: Imaging Instrument High lateral resolution (20 nm) Multi-method instrument (XPEEM/PED) Excellent for monitoring dynamic processes Poorer spectroscopic ability Sensitive to rough surfaces e -

The SPELEEM Spectroscopic photoemission and low energy electron microscope

The SPELEEM at ELETTRA

The Energy Filter R = 100 mm Pass Energy = 900 V

XPEEM: Spectroscopic Microscopy Images from a Field Effect Transistor (FET) at different binding energies. Photon energy 131.3 ev. Sample from M. Lazzarino, L. Sorba, and F. Beltram

XPEEM: Core Level Spectroscopy Pb/W(110), Pb 5d core level, hv = 80 ev Best energy resolution: 250 mev

Imaging of Dispersive Plane W{110} clean surface W 4f core level hv = 131 ev Resolution 130 mev Residuals Bulk DS Fit Bulk Surface Experimental Data intensity (a.u.) Surface 32.0 31.5 31.0 30.5 binding energy (ev)

Lateral resolution in LEEM mode Pb/W(110), E = 11 ev Lateral resolution: 10 nm 0 10 Rel. distance (nm) 20 30 40 100% 9.6nm 85% Intensity (arb. units) Experimental Data Fit 1pixel = 2.9nm 300nm 15% 0% 3.3 pixels 16 18 20 22 Pixels 24 26 28 30

Lateral resolution in XPEEM mode Pb/W(110), Pb 5d 5/2, hv = 80 ev, KE = 58.2 ev Lateral resolution: 27 nm Rel. distance (nm) 0 20 40 60 80 100 120 140 160 300 nm Intensity (arb. units) Experimental Data Fit 1 (Worst case) Fit 2 (Best case) 1 pixel = 5.7nm (5.1pixels) Best case 25.2nm (4.4pixels) Worst case: 29.1nm 10 15 20 25 Pixels 30 35 40

Outline A brief introduction to spectromicroscopy The spectroscopic photoemission and low energy electron microscope (SPELEEM) Local Anodic Oxidation (LAO) of GaAs a b 1.0 a b Intensity (a.u.) 0.8 0.6 0.4 native oxide AFM oxide 0.2 0.0 11010910810710610510410310210110099 98 97 96 95 binding energy (ev) M. Lazzarino, S. Heun, B. Ressel, K. C. Prince, P. Pingue, and C. Ascoli: Appl. Phys. Lett. 81 (2002) 2842.

Local Anodic Oxidation (LAO) Commonly accepted model: Water electrolysis H 2 O H + + OH -. OH - groups (or O - ) migrate towards the substrate-oxide interface. Oxide penetration induced by the intense local electric field (>10 7 V/cm). Versatile tool at relatively low cost High lateral resolution but small area

Local Anodic Oxidation (LAO) After oxidation. Oxide removal with HF 10%, 30 s. The penetration depth is 1.0-1.5 times the oxide height.

LAO on GaAs/AlGaAs Before oxidation: Z After oxidation: Z Z Z GaAs AlGaAs GaAs

Devices made with LAO K. Matsumoto et al., APL 68 (1996) 34. A. Fuhrer et al., Nature 413 (2001) 822. G. Mori et al., JVB 22 (2004) 570. R. Held et al., APL 73 (1998) 262.

Setup for Lithography on GaAs voltage source Hygrometer AFM camera Plexiglass hood AFM N 2 +H 2 O Thermomicroscope Microcope CP-Resource water bottle

GaAs LAO-Oxide: Desorption Before photon exposure After hours of exposure 14V 12V 12V 16V 10V hv = 130eV 14V 12V 1µm D. Ercolani et al.: Adv. Funct. Mater., in press.

Height reduction vs. exposure time hv = 130 ev = 9.5 nm We observe a linear relation between exposure time and height reduction. A dependence on other oxidation parameters (bias, writing speed) could not be detected. D. Ercolani et al.: Adv. Funct. Mater., in press.

The Knotek-Feibelman mechanism E E E Valence electrons Valence electrons Valence electrons 3d 3d 3d E Ga or As Initial state Valence electrons 3d Ga or As Ga or As This Auger decay leads to a final state with two vacancies in the valence band weakening the bond between Ga or As and O. Ga or As Final state D. Ercolani et al.: Adv. Funct. Mater., in press.

Spectra From GaAs LAO-Oxide Time resolved spectroscopy with SPELEEM using Dispersive Plane (hv = 130 ev) Sample S03B Hole (3,2) Writing voltage 15 V Structure height 3 nm Image taken with secondary electrons: Photon energy: 130 ev Kinetic energy: 0.3 ev Field of view: 10 µm G. Mori et al.: J. Appl. Phys., submitted.

DP-Spectra From GaAs LAO-Oxide Ga 3d remains unchanged, As-oxides signal disappeares with time. Ga-oxides mainly Ga 2 O with small contribution of Ga 2 O 3. G. Mori et al.: J. Appl. Phys., submitted.

Chemistry of the GaAs LAO-Oxide Photon assisted partial desorption of the AFM-grown oxide was observed. The AFM-oxide is mainly composed of Ga 2 O, with a small fraction of Ga 2 O 3 and As-oxides. The shape of the Ga peak does not change with exposure time (early stage Ga and As-oxides of desorption), however, Ga-oxides desorb. All As-oxides desorb completely. No Ga oxides As-oxides detected after some hours of exposure. GaAs substrate The As-oxides are located only at the surface. Evidence for the presence of unoxidized GaAs in the LAO-oxide. Chemical composition does not depend on writing bias. G. Mori et al.: J. Appl. Phys., submitted.

LAO on III-V Heterostructures Sample I

AFM-LAO on Sample I

XPEEM Core Level Spectra Aluminium observed at the surface of the LAO oxide Unoxidized regions: No Al 2p signal Main component: Al 2 O No chemical difference between different bias

Shallow Oxidations A B Ratio Al / Ga: Area A: 1 / 3 Area B: 1 / 13 Average oxide height: Area A: 1.5 ± 0.6 nm Area B: 1.0 ± 0.2 nm Surface Roughness might be important!

Variation in Marker Thickness 2 nm AlAs (h = 9.1 ± 1 nm) 5 nm AlAs (h = 8.9 ± 0.9 nm) Ratio Al / Ga: 0.57 Ratio Al / Ga: 1.12

Refined Model h OH - O - Ga + Al + d GaAs d AlAs Diffusion of oxygen-rich ions plus substrate ions Homogeneous mixing of components Ratio Al / Ga: 0 for h < d GaAs (h - d GaAs ) / d GaAs for d GaAs < h < d GaAs + d AlAs d AlAs / (h - d AlAs ) for h > d GaAs + d AlAs Surface and interface roughness have to be considered!

Comparison to Experiment Al intensity too high. Normalization Factor. Average RMS roughness of the LAO-oxide: 1 nm. Convolution with Gaussian curve.

Summary Presence of Al-oxides in the topmost layers of GaAs/AlAs/GaAs heterostructures detected. Classical model of LAO process has to be revised. Alternative model includes diffusion of ionized substrate atoms to the surface. Good agreement between model and experiment. Al diffusion might be faster than Ga diffusion.