Number of ALD publications per year between 1972 and 2004. Search made with ISI Web of Science (see Ref. 4).
Schematic illustration of one ALD reaction cycle.
Overview of the materials grown by ALD. Classification according to Reactant , with details of the investigations in Table III. Growth of pure elements as well as compounds with oxygen, nitrogen, sulphur, selenium, tellurium, and other compounds grouped together are indicated through shadings of different types at different positions. The elements are named according to the recommendations of The International Union of Pure and Applied Chemistry (IUPAC, http://www.iupac.org/reports/periodic̱table/, dated 1 November 2004).
Typical ligands of the reactants used in ALD. represents the central atom, most often a metal. The ALD reactants can be divided into two main groups, inorganic and metalorganic; organometallic reactants with a direct metal–carbon bond (Ref. 1126) form a subgroup of the latter. In this work, eight main ligand groups are further distinguished: elements [no ligands], halides [F, Cl, Br, and I], alkyls [Me, Et, , Ay, ,,, and Np], cyclopentadienyls [Cp, CpMe, , CpEt, , and ], alkoxides [OMe, OEt, , , , mmp, and dmae], -diketonates [acac, thd, hfac, od, and methd], alkylamides and silylamides [, NEtMe, , and ], and amidinates [ and ].
Overview of the different types of metal reactants used in ALD (based on the list of Reactants in Table III). The dark background indicates that metal reactants containing the particular type of ligand have been used for ALD: (a) elements [no ligands], (b) halides [F, Cl, Br, and I], (c) alkyls [Me, Et, , Ay, ,, , and Np], (d) cyclopentadienyls [Cp, CpMe, , CpEt, , and ], (e) alkoxides [OMe, OEt, , , , mmp, and dmae], (f) -diketonates [acac, thd, hfac, od, and methd], (g) alkylamides and silylamides [, NEtMe, , and ], and (h) amidinates [ and ].
Three different types of monolayers relevant to ALD: (a) a chemisorbed monolayer (the substrate before chemisorption indicated above, with reactive sites shown), (b) a physisorbed monolayer, and (c) a monolayer of the ALD-grown material.
Examples of how the amount of material adsorbed can vary with time: (a) irreversible saturating adsorption (i.e., self-terminating reaction), (b) reversible saturating adsorption, (c) combined irreversible and reversible saturating adsorption, (d) irreversible nonsaturating adsorption (deposition), and (e) irreversible saturating adsorption not allowed to saturate. The vertical dashed line marks the end of the reactant supply and the beginning of a purge or evacuation.
Effect of the reactant partial pressure on the amount of material chemisorbed in a gas–solid reaction: (a) the equilibrium chemisorption coverage in reversible adsorption (equilibrium constants ) and (b) the chemisorption coverage after saturation in irreversible adsorption.
Schematic representation of five reaction cycles, assuming irreversible adsorption: (a) chemisorption coverage as a function of time (solid line: species adsorbed in the reaction of the Reactant assumed to be of type , dashed line: species adsorbed in the reaction of Reactant ; the beginning and end of a reaction cycle and Steps 1–4 are indicated), (b) the amount of atoms adsorbed as a function of time , (c) the deposition rate of atoms as a function of time [obtained as the time derivative of the curve in panel (b)], (d) amount of material deposited as a function of the number of reaction cycles , and (e) the GPC as a function of the number of reaction cycles .
Chemisorption mechanisms identified for ALD: (a) ligand exchange reaction of the reactant with surface “” groups, releasing gaseous , (b) dissociation of the in surface sites, and (c) association of the species onto the surface. In this scheme, .
Factors identified to cause saturation of irreversible chemisorption: (a) steric hindrance of the ligands and (b) number of reactive surface sites.
Schematic illustration for analyzing sterically hindered chemisorption on the basis of the size of the reactant [Model I by Ritala et al. (Refs. 462 and 468) and Morozov et al. (Ref. 133)], the size and geometry of the chemisorbed species [Model II by Ylilammi (Ref. 1127)], and the size and number of ligands [Model III by Siimon and Aarik (Ref. 432) and Puurunen (Ref. 1128)]. Left: side view, right: top view.
Variation of the GPC with the ALD processing temperature in the ALD window: (a) the GPC decreases with temperature, (b) the GPC is constant with temperature (possible with different values at different temperature ranges, as shown by the dashed line), (c) the GPC increases with temperature, and (d) the GPC first increases and then decreases with temperature.
Dependency of the GPC on the number of reaction cycles in different types of ALD processes (Ref. 247): (a) linear growth, (b) substrate-enhanced growth, (c) substrate-inhibited growth of Type 1, and (d) substrate-inhibited growth of Type 2.
Schematic illustration with increasing number of reaction cycles of selected growth modes possible in ALD: (a) two-dimensional growth, (b) island growth and, (c) random deposition.
Self-termination in the process at in the steady-growth regime: effect on the measured GPC (labeled “deposition per cycle”) of (a) the pulse and the following purge times (Steps 1 and 2) and (b) the pulse and the following purge times (Steps 3 and 4). (Reprinted from Sneh et al. (Ref. 222) with permission. Copyright 2002, Elsevier.)
Self-terminating reactions in the ALD process in the steady-growth regime: effect of reactant pressures to the GPC (labeled “growth rate”) at (Ref. 267). [Reprinted from Kumagai et al. (Ref. 267) with permission. Copyright 1994, Institute of Pure and Applied Physics.]
GPC in the process to grow amorphous aluminum oxide on flat substrates in the steady-growth regime (Refs. 228, 199, 212, 222, 229, 179, 234, and 246). The GPC is expressed as the amount of aluminum atoms attached per cycle per square nanometer of surface , which converts to the thickness increment per cycle through (Refs. 237 and 1128). A line was fitted to the data on flat substrates, except the points of Kuse et al., which probably represent unsaturated conditions. Confidence limit of one standard deviation is shown. Data points obtained on high-surface-area alumina substrates are shown for reference (Ref. 216). The results (Ref. 216) were calculated through the mass balance [Eq. (23) for reaction at with alumina heat treated at the temperature indicated in the axis. The results for reaction at can be used for the comparison, because increasing the temperature to does not affect the amount of adsorbed species (Fig. 20).
Effect of the reaction temperature on the amount of aluminum adsorbed in the reaction on alumina heat treated at , according to Puurunen et al. (Ref. 216). The confidence limits represent one standard deviation.
Effect of the substrate heat-treatment temperature on the amount of aluminum adsorbed in the reaction (a) on alumina at (엯 Ref. 216) and (b) on silica at or (◻ Refs. 209 and 210). The OH surface concentrations are shown as crosses for reference (alumina: × Ref. 216, silica: × Ref. 209, and + Ref. 1190). The confidence limits represent one standard deviation.
Effect of the surface OH group concentration on the amount of material chemisorbed in the reaction on alumina and silica substrates (Ref. 262): (a) amount of aluminum atoms adsorbed in the reaction , (b) methyl group concentration after the reaction , and (c) the average methyl/aluminum (Me/Al) ratio in the adsorbed species . Data from Ref. 216 for alumina and Refs. 209 and 210 for silica, for reaction at the indicated temperatures. The OH surface concentrations are from Ref. 216 for alumina and Refs. 209 and 1190 for silica. In panels (a) and (b), the lines have been fitted to the data. The confidence limits represent one standard deviation. In panel (b), the dashed line indicates the maximum theoretical methyl group concentration of according to Model III (Refs. 237 and 1128).
Illustration of the physical situation corresponding to Eq. (24): number of aluminum atoms and methyl groups adsorbed in the self-terminating reaction on surfaces with different OH group concentrations , resulting in average Me/Al ratios of (a) 3, (b) 2, and (c) 1.5. The squares represent an area of ; the methyl groups are drawn in scale . The small dependency that the number of adsorbed methyl groups has on the surface OH group concentration [Fig. 22(b)] has been neglected for simplicity.
The correlation of Eq. (24), marked as a solid line, is in line with the GPC in the process (data points as in Fig. 19). The GPC values for the correlation were calculated assuming that decreases linearly with temperature from at to at (Ref. 1195).
Summary of the in situ MS results for the process of Juppo et al. (Ref. 211) (open symbols, left axis) and Rahtu et al. (Ref. 217) (black squares, right axis): (a) total amount of methane produced during a reaction cycle divided by three, proportional to the GPC , (b) the amount of methane produced during the reaction, proportional to the surface OH group concentration before the reaction , (c) the amount of methane produced during the reaction, proportional to the methyl group concentration after the reaction , and (d) Me/Al ratio on the surface after the reaction, obtained by dividing [panel (c)] with [panel (a)]. The results of Juppo et al. obtained for the shortest water pulses are not shown, because in them, the methane-producing reactions had not saturated (Ref. 211). All results are in arbitrary units (a.u.). Although the exact dependency is not known, for plotting the results in the same graphs, 25 (unnormalized) units of Juppo et al. (left axis) were assumed to correspond to 1.5 (normalized) units of Rahtu et al. (right axis).
Effect of and pulse times to the average Me/Al ratio after reaction with alumina, as reported by Rahtu et al. (Ref. 217). The and reactions were saturated after a pulse time of about .
GPC in the process at as a function of the number of ALD reaction cycles when a chemical rich in OH groups and a hydrogen-terminated silicon are used as substrates. Data from Zhao et al. (Ref. 1176).
Illustration of the importance of taking into account the mass of the deposited layer when calculating the GPC on high-surface-area substrates. Difference between Eqs. (25) [full symbols] and (26) [open symbols] for a substrate with a specific surface area of : (a) , assuming a constant GPC ; (b) , assuming ; and (c) , assuming . indicates the number of ALD reaction cycles.
Some Soviet–Russian ALD investigations.
Different names of ALD.
Overview of ALD processes based on two reactants (Source: ISI Web of Science, status in February 2005). Description of the ligands in Fig. 4.
Computational chemistry investigations on ALD processes.
Typical ALD processes of the type Reactant , reported for element reactants (references in Table III).
Typical ALD processes of the type Reactant , reported for metal halides that contain one type of ligand (references in Table III).
Typical ALD processes of the type Reactant , reported for metal alkyls that contain one type of ligand (references in Table III).
Typical ALD process of the type Reactant , reported for metal cyclopentadienyls that contain one type of ligand (references in Table III).
Typical ALD processes of the type Reactant , reported for metal alkoxides that contain one type of ligand (references in Table III).
Typical ALD processes of the type Reactant , reported for metal -diketonates that contain one type of ligand (references in Table III).
Typical ALD process of the type Reactant , reported for metal amides that contain one type of ligand (references in Table III).
Typical ALD processes of the type Reactant , reported for metal amidinates that contain one type of ligand (references in Table III).
Article metrics loading...
Full text loading...