Ultra-Low Coverage Spontaneous Etching and Hyperthermal Desorption of Aluminum Chlorides from Cl 2 /Al(111)

1. Ultra-Low Coverage Spontaneous Etching and Hyperthermal Desorption of Aluminum Chlorides from Cl2/Al(111) Tyler J. Grassman, Gary C. Poon, and Andrew C. Kummel University of California, San Diego Gordon Research Conferences: Dynamics at Surfaces – August 10-15, 2003 1.0 0.9 0.8 1.0 1000 0.7 0.9 900 1.0 800 1000 0.8 0.9 0.6 0.8 700 900 0.7 0.7 0.6 600 0.5 800 0.5 500 700 0.4 0.6 0.3 0.4 400 600 0.2 0.5 0.1 300 500 0.0 0.3 0 1 2 3 4 5 200 400 0.4 100 0.2 300 0.3 0 200 0.1 0.2 100 0 0.1 0.0 535 850 1327 0 10 20 30 40 50 60 70 80 90 100 0.0 0 10 20 30 40 50 60 70 80 0.7 0.6 0.7 0.5 0.6 0.4 0.5 0.3 0.4 0.2 0.3 0.1 0.2 0.0 0.1 Abstract DFT: Bonding to Al Adatoms Results 0.0 0.11 0.27 0.65 Non-resonant multiphoton ionization (MPI) and time-of-flight mass spectrometry (TOF-MS) have been used to monitor the desorption of aluminum chloride (AlxCly) etch products from the Al(111) surface at 100 K and 500 K during low-coverage (< 5% monolayer) monoenergetic Cl2 (0.11 eV to 0.65 eV) dosing. The desorption products in this low-coverage range show predominantly hyperthermal exit velocities under all dosing conditions. For example, with 0.27 eV incident Cl2, the etch product was found to have a most-probable velocity of 517 ± 22 m/s at an Al(111) surface temperature of 100 K. This corresponds to 22 times the expected thermal desorption translational energy for AlCl3. Sticking probability measurements and AlxCly etch rate measurements show etching even at Cl2 coverages of less than 5% monolayer at surface temperatures between 100 K and 500 K. These experimental results are consistent with a combination of fast-time-scale and low Cl surface coverage formation of aluminum chlorides and the presence of activated aluminum chloride chemisorption states having potential energies above the vacuum level, and are contrary to the standard high-coverage picture of aluminum chloride etch requiring high surface coverage (>1 monolayer) and subsurface Cl diffusion.Density functional theory calculations yield results that are consistent with both our experimental findings and mechanistic descriptions. Measured velocities, translational energies, and peak-to-width ratios for desorbing aluminum chlorides at all incident Cl2 translational energies and Al(111) surface temperatures studied. Non-bonded Bonded E (eV)b vp:w vp (m/s) Einc (eV) Ts (K) E (eV)a Dose Species AlCl3 Al2Cl6 612.3 ± 21.2 ― 0.522 ± 0.037 0.65 ± 0.06 Cl2/He: lowc 0.65 100 DE = -2 eV 0.65 100 Cl2/He: highd 255.5 ± 9.2 ― 0.092 ± 0.007 0.84 ± 0.08 Cl2/He 0.65 500 653.1 ± 15.5 0.296 ± 0.014 ― 0.56 ± 0.02 Cl2/Ne 0.27 100 517.7 ± 22.0 ― 0.378 ± 0.032 0.82 ± 0.03 Cl2/Ne 0.27 500 552.9 ± 14.2 0.213 ± 0.011 ― 0.76 ± 0.03 0.11 100 460.3 ± 22.3 ― 0.303 ± 0.029 0.69 ± 0.06 pure Cl2 pure Cl2 0.11 500 533.3 ± 9.8 0.197 ± 0.007 ― 0.71 ± 0.05 Al Al Cl Cl Al(ad.) DFT slab calculations of Cl adsorbates on the Al(111) surface near an Al adatom indicate a strong energetic preference for Cl atoms to bond to Al adatoms rather than remain dispersed on the surface (out of bonding range). This is likely to be the case for regrowth islands and step edges, as well. The surface reaction, 3Cl(ad.) + Al(ad.) → AlCl3(ad.), is computationally found to be about 2 eV thermodynamically favorable. These AlCl3 adsorbates are also likely be highly mobile on the surface, as the surface binding potential for the molecules (with respect to the different possible adsorption sites) was found to have a maximum corrugation of only 0.1 eV. a: Translational energy calculated from mass of AlCl3 etch product and measured vp b:Translational energy calculated from mass of Al2Cl6 etch product and measured vp c: Total dosing time ≤ 10 sec d: Total dosing time > 10 sec Experimental Methods • Very low Cl2 flux (~ 2x1014 cm-2 sec-1) pulsed molecular beam • MPI (210.2 nm) and TOF-MS → photodissociate AlxCly, detect Al+ • King & Wells type sticking measurements with QMS • DFT-GGA with Vanderbuilt ultrasoft pseudopotentials and plane-wave basis, 7×6×1 21 k-point Monkhorst-Pack, 250 eV plane-wave cut-off for Cl/Al(111) adsorbate system (Vienna Ab-initio Simulation Package, VASP) Exit vs. Incident Velocities/Energies DFT: Bonding on Terraces (a) (c) (Ts = 500 K) (Ts = 100 K) Experimental Experimental Clustered AlCl3 (thermal) AlCl (thermal) Dispersed Al2Cl6 (thermal) AlCl3 (thermal) AES Mechanical Chopper (7 ms) 535 850 1327 Exit Velocity, vexit (m/s) Exit Velocity, vexit (m/s) Skimmer Ionizer (210.2 nm) QMS Pulsed Nozzle (10Hz) Al(111) Cl2 Incident Velocity, vinc (m/s) Cl2 Incident Velocity, vinc (m/s) Collimator TOF-MS (Cl2/He, Cl2/Ne pure Cl2) LEED (b) (d) Thermal Thermal (Ts = 500 K) (Ts = 100 K) AlCl (exp) AlCl3 (exp) AlCl3 (exp) Al2Cl6 (exp) 0.11 0.27 0.65 Exit Energy, Eexit (eV) Exit Energy, Eexit (eV) DFT slab calculations of the Cl/Al(111) adsorbate/terrace system meant to examine the possibility of adsorbate clustering do not show a thermodynamic preference for clustering geometries. However, the calculations do indicate a strong preference for ontop adsorption sites, with differences in total energies of -0.4 to -0.9 eV compared to other sites. Ontop-site Cl adsorbates are also found to pull the Al terrace atom to which they were bonded out of the surface plane by 0.4 Å, thereby likely making them more vulnerable to attack by other atomic or molecular adsorbates, and helping to replenish nucleation sites. 1.0 0.9 Cl2 Incident Energy, Einc (eV) Cl2 Incident Energy, Einc (eV) 0.8 Etch product exit velocities and energies plotted against incident velocities and energies, respectively, for surface temperatures of 100 K and 500 K. The open symbols (squares, circles, triangles, diamonds) represent experimental data, while the filled symbols represent the expected values from a purely thermal desorption mechanism. The data clearly shows that the etch products are exiting the surface at hyperthermal velocities. 0.7 Time-of-Flight Distributions 0.6 0.5 0.4 0.3 0.2 Values extracted from experimental time-of-flight spectra compared against values taken from expected thermal distributions (flux-weighted Maxwell-Boltzmann for density detector): 0.1 (a) 0.0 (Inc. Cl2/Ne Ts = 100 K) Conclusion Etch Rate Profiles 1.0 0.9 0.8 “most-probable time of flight” 0.7 Normalized Al+ Intensity (arb.) (a) (b) 0.6 Observed Al2Cl6 TOF (a) (b) 1.0 Standard Activated Chemisorption Model Activated Chemisorption State Model 0.5 0.9 500 m/s Al2Cl6 component (Ts = 100 K) Expected 100 K Al2Cl6 TOF 0.8 0.7 0.4 Energy (arb.) Energy (arb.) 0.6 “most-probable velocity” 0.5 0.3 0.4 30 60 90 120 150 180 210 240 270 300 0 activated chemisorption state 0.3 0.2 Time (msec) 0.2 0.1 exit kinetic energy exit kinetic energy 0.1 (b) 0.0 Etch Rate (arb.) Etch Rate (arb.) 0 1 2 3 4 5 vacuum level vacuum level (Inc. Cl2/Ne Ts = 500 K) 0.0 “most-probable translational energy” 550 m/s AlCl3 component (Ts = 500 K) 0 10 20 30 40 50 60 70 80 Begin surface Cl2 exposure Thermal distribution should exhibit most-probably velocity to width ratio (vp:w) of about 1. An experimental ratio smaller than unity indicates a wider distribution than would be expected for a purely thermal desorption mechanism. Begin surface Cl2 exposure chemisorption well chemisorption well Distance From Surface (arb.) Distance From Surface (arb.) Normalized Al+ Intensity (arb.) The unusual desorption phenomena observed in this work is consistent with a model consisting of a combination of fast surface agglomeration of Cl and AlxCly adsorbates (as seen in the etch rate profiles, as well as the computational data) and the existence of activated aluminum chloride chemisorption states, with potential wells above the vacuum level. The activated chemisorption state model is diagrammed in the figure above (b), and is compared with the standard activated chemisorption model (a) in which the chemisorption well is below the vacuum level and desorbing species must surmount an activation barrier. Time (sec) Time (sec) Observed AlCl3 TOF Etch rate profiles of the 0.27 eV incident energy Cl2 on (a) the 100 K and (b) the 500 K Al(111) surface. The insets are blow-ups of the shaded regions and show the first 5 seconds of data. As seen in the figures, etching begins immediately upon exposure of the Al(111) surface to the low-flux Cl2 molecular beam, at surface coverages of < 5% monolayer. Such results indicate fast time-scale surface agglomeration of adsorbed Cl atoms/molecules and submolecular aluminum chlorides. Expected 500 K AlCl3 TOF Maxwell-Boltzmann-like time-of-flight distribution curves for the etch products from the 0.27 eV incident Cl2 beam on the (a) 100 K and (b) 500 K Al(111) surface. The solid curves show the experimentally observed desorption distribution, and the dashed curves show the expected thermal desorption time-of-flight distribution for the etch product mass and surface temperature of interest, as indicated in the figures. The most probable time-of-flights are indicated by the vertical, single-headed arrows. The full-width half-max of the experimental distributions are indicated by the horizontal, double-headed arrows. Time (msec)