The thermal atomic layer etching (ALE) of Al 2 O 3 can be achieved using sequential fluorination and ligand-exchange reactions. Although previous investigations have characterized the etch rates and surface chemistry, no reports have identified the volatile etch products. This study explored the volatile etch species during thermal Al 2 O 3 ALE at 300 °C using quadrupole mass spectrometry (QMS). HF was the fluorination reactant; Al(CH 3 ) 3 (trimethylaluminum (TMA)) and AlCl(CH 3 ) 2 (dimethylaluminum chloride, (DMAC)) were the metal precursors for ligand exchange. When TMA was used as the metal precursor after the fluorination of Al 2 O 3 powder, the QMS measurements revealed that the main ion species were consistent with dimers of AlF(CH 3 ) 2 (dimethylaluminum fluoride (DMAF)) with itself (DMAF + DMAF) or with TMA (DMAF + TMA). These ion species were observed after loss of a methyl group as Al 2 F 2 (CH 3 ) 3 + at m/z = 137 and Al 2 F(CH 3 ) 4 + at m/z = 133, respectively. In addition, an ion species consistent with a trimer was also observed as Al 3 F 3 (CH 3 ) 5 + at m/z = 213. Very similar results were observed for TMA exposures on AlF 3 powder. Comparable results were also obtained using DMAC as the metal precursor for ligand exchange. In contrast, SiCl 4 and TiCl 4 are not successful metal precursors because they do not lead to thermal Al 2 O 3 ALE at 300 °C. QMS measurements revealed no Alcontaining etch species after SiCl 4 and TiCl 4 exposures on AlF 3 powder. However, SiF x Cl y + and TiF x Cl y + species were observed which suggested that ligand-exchange reactions can occur without the release of Al-containing etch species. Density functional theory (DFT) and coupled cluster singles, doubles, and perturbative triples (CCSD(T)) calculations were performed to support the preference for dimer products. The theoretical results confirmed the stability of the dimer products and showed that dimers with two Al−F−Al bridging bonds are the most stable and dimers with two Al−CH 3 −Al bridging bonds are the least stable. In addition, the calculations suggested that dimers with terminal CH 3 ligands are most able to desorb from the surface because these dimers need to break weak Al−CH 3 −Al bridging bonds. Transmission electron microscopy (TEM) studies confirmed the thermal Al 2 O 3 ALE of Al 2 O 3 films on W powders. The TEM images revealed that the etch process was uniform and conformal after various numbers of thermal Al 2 O 3 ALE cycles using HF and TMA as the reactants.

Platinum (Pt) atomic layer deposition (ALD) usually yields Pt nanoparticles during initial film growth. In contrast, deposition of continuous and ultrathin Pt films is needed for many important applications, such as the oxygen reduction reaction in polymer electrolyte membrane (PEM) fuel cells. A continuous and high radius of curvature Pt film is more stable and has a higher area-specific activity than the Pt nanoparticles commonly used in PEM fuel cells. However, the Pt film must be ultrathin and have a large surface area to be cost effective. In this paper, a review of earlier Pt ALD studies on flat substrates is presented that demonstrates that tungsten, with a higher surface energy than platinum, can serve as an adhesion layer to achieve Pt ALD films that are continuous at ultrathin thicknesses of ∼1.5 nm. This work utilized MeCpPtMe3 and H2 plasma as the Pt ALD reactants. The deposition of continuous and ultrathin Pt ALD films using MeCpPtMe3 and H2 plasma as the reactants is then studied on two high surface area substrate materials: TiO2 nanoparticles and 3M nanostructured thin film (NSTF). Transmission electron microscopy (TEM) showed uniform and continuous Pt films with thicknesses of ∼4 nm on the TiO2 nanoparticles. TEM with electron energy loss spectroscopy analysis revealed W ALD and Pt ALD films with thicknesses of ∼3 nm that were continuous and conformal on the high aspect ratio NSTF substrates. These results demonstrate that cost effective use of Pt ALD on high surface area substrates is possible for PEM fuel cells.

Pt nanoparticles were grown on titanium oxide and tungsten oxide at 200 °C by Pt atomic layer deposition (ALD) using platinum(II) hexafluoroacetylacetonate [Pt(hfac) 2 ] and formalin as the reactants. The Pt ALD surface chemistry and Pt nanoparticles were examined using in situ Fourier transform infrared (FTIR) vibrational spectroscopy and ex situ transmission electron microscopy (TEM). The FTIR spectra identified the surface species after the Pt(hfac) 2 and formalin exposures on TiO 2 . An infrared feature at ∼2100 cm −1 in the FTIR spectrum after Pt(hfac) 2 and formalin exposures on TiO 2 was consistent with CO on Pt, revealing that Pt(hfac) 2 and formalin exposures led to the formation of Pt nanoparticles. The FTIR spectrum of Pt(hfac) 2 on TiO 2 was very similar to the FTIR spectrum of hexafluoroacetylacetone (hfacH) on TiO 2 . The FTIR spectra also revealed that hfacH blocked the adsorption of Pt(hfac) 2 on TiO 2 . The coverage of the Pt nanoparticles could be reduced by preadsorbing hfacH on TiO 2 prior to Pt(hfac) 2 adsorption. Time-dependent FTIR spectra showed that the coverage of hfacH and its adsorption products were reduced versus time following hfacH exposure. Pt ALD on WO x at 200 °C led to the growth of Pt nanoparticles that were fairly similar to the Pt nanoparticles from Pt ALD on TiO 2 . The TEM images revealed that the size of the Pt nanoparticles on WO x could be adjusted by varying the number of Pt ALD cycles. Because of site-blocking by the hfac ligands, the Pt(hfac) 2 and formalin reactants required many more ALD cycles for nucleation and growth compared with other Pt ALD surface chemistries.

Atomic layer processing such as atomic layer deposition (ALD) and thermal atomic layer etching (ALE) is usually described in terms of sequential, self-limiting surface reactions. This picture for ALD and thermal ALE leaves out the possibility that the metal precursor in ALD and thermal ALE can also convert the surface material to another new material. This perspective introduces the previous evidence for conversion reactions in atomic layer processing based on a variety of studies, including Al2O3 ALD on ZnO, growth of Zn(O,S) alloys, “self-cleaning” of III-V semiconductor surfaces, and thermal ALE of ZnO and SiO2. The paper then focuses on the reaction of Al(CH3)3 [trimethylaluminum (TMA)] on ZnO as a model conversion system. A variety of techniques are utilized to monitor ZnO conversion to Al2O3 using TMA at 150 °C. These techniques include FTIR spectroscopy, quadrupole mass spectrometry (QMS), x-ray reflectivity (XRR), gravimetric analysis, x-ray photoelectron spectroscopy (XPS), and quartz crystal microbalance (QCM) measurements. The various studies focus on ZnO conversion to Al2O3 for both hydroxyl-terminated and ethyl-terminated ZnO substrates. FTIR studies observed the conversion of ZnO to Al2O3 and provided evidence that the conversion is self-limiting at higher TMA exposures. QMS studies identified the volatile reaction products during the TMA reaction with ZnO as CH4, C2H4, C2H6, and Zn(CH3)2. The CH4 reaction product preceded the appearance of the Zn(CH3)2 reaction product. XRR investigations determined that the thickness of the Al2O3 conversion layer on ZnO limits at ∼1.0 nm at 150 °C after larger TMA exposures. A gravimetric analysis of the conversion reaction on ZnO nanoparticles with a diameter of 10 nm displayed a percent mass loss of ∼49%. This mass loss is consistent with an Al2O3 shell of ∼1 nm on a ZnO core with a diameter of ∼6 nm. XPS studies revealed that ZnO ALD films with a thickness of 2 nm were almost completely converted to Al2O3 by large TMA exposures at 150 °C. QCM investigations then measured the mass changes for lower TMA exposures on hydroxyl-terminated and ethyl-terminated ZnO films. More mass loss was observed on ethyl-terminated ZnO films compared with hydroxyl-terminated films, because TMA does not have the possibility of reacting with hydroxyl groups on ethyl-terminated ZnO films. The mass losses also increased progressively with temperatures ranging from 100 to 225 °C on both hydroxyl-terminated and ethyl-terminated ZnO films. The perspective concludes with a discussion of the generality of conversion reactions in atomic layer processing.

Platinum (Pt) decorated carbon nanotubes (CNTs) nanocatalysts (Pt/CNTs) were successfully prepared by an atomic layer deposition (ALD) method. Increased ratios of D to G band with increasing ALD cycles in Raman spectra suggested an interaction between Pt nanoparticles (NPs) and CNTs as well as increased Pt loadings. The absence of the Pt peaks in XRD before and after thermal annealing treatment demonstrated an ultra-small Pt NPs size. TEM revealed a perfect distribution of the ultra-small Pt NPs on the tube wall surface with different ALD cycles and an agglomeration of Pt NPs was observed when increasing the annealing temperature. Increased peak current density (jp) values in CV (25 cycles, denoted as Pt/CNTs-25C, 40.89 mA/mgPt, 50 cycles, Pt/CNTs-50C, 667.6 mA/mgPt and 100 cycles, Pt/CNTs-100C, 1205.7 mA/mgPt) were obtained toward the methanol oxidation reaction (MOR) with increasing the ALD cycles. 100 and 300°C hydrogen annealed Pt/CNTs-50C (denoted as Pt/CNTs-50C-100A and Pt/CNTs-50C-300A) gave increased and decreased jp values (1200 and 800 mA/mgPt) due to the positive hydrogen reduction of Pt(II)/Pt(IV) species and the negative Pt NPs agglomeration. Pt/CNTs-100C-100A exhibited the highest jp value (2000 mA/mgPt) after applying the optimal 100°C annealing treatment, enabling it an excellent candidate for MOR application.