Human-machine collaboration for improving semiconductor process development.

One of the bottlenecks to building semiconductor chips is the increasing cost required to develop chemical plasma processes that form the transistors and memory storage cells1,2. These processes are still developed manually using highly trained engineers searching for a combination of tool parameters that produces an acceptable result on the silicon wafer3. The challenge for computer algorithms is the availability of limited experimental data owing to the high cost of acquisition, making it difficult to form a predictive model with accuracy to the atomic scale. Here we study Bayesian optimization algorithms to investigate how artificial intelligence (AI) might decrease the cost of developing complex semiconductor chip processes. In particular, we create a controlled virtual process game to systematically benchmark the performance of humans and computers for the design of a semiconductor fabrication process. We find that human engineers excel in the early stages of development, whereas the algorithms are far more cost-efficient near the tight tolerances of the target. Furthermore, we show that a strategy using both human designers with high expertise and algorithms in a human first-computer last strategy can reduce the cost-to-target by half compared with only human designers. Finally, we highlight cultural challenges in partnering humans with computers that need to be addressed when introducing artificial intelligence in developing semiconductor processes.

Engineering gold-platinum core-shell nanoparticles by self-limitation in solution.

Core-shell particles with thin noble metal shells represent an attractive material class with potential for various applications ranging from catalysis to biomedical and pharmaceutical applications to optical crystals. The synthesis of well-defined core-shell architectures remains, however, highly challenging. Here, we demonstrate that atomically-thin and homogeneous platinum shells can be grown via a colloidal synthesis method on a variety of gold nanostructures ranging from spherical nanoparticles to nanorods and nanocubes. The synthesis is based on the exchange of low binding citrate ligands on gold, the reduction of platinum and the subsequent kinetically hindered growth by carbon monoxide as strong binding ligand. The prerequisites for homogeneous growth are low core-binding ligands with moderate fast ligand exchange in solution, a mild reducing agent to mitigate homonucleation and a strong affinity of a second ligand system that can bind to the shell's surface. The simplicity of the described synthetic route can potentially be adapted to various other material libraries to obtain atomically smooth core-shell systems.

Self-Limiting Shell Formation in Cu@Ag Core-Shell Nanocrystals during Galvanic Replacement.

The understanding of synthetic pathways of bimetallic nanocrystals remains limited due to the complex energy landscapes and dynamics involved. In this work, we investigate the formation of self-limiting Cu@Ag core-shell nanoparticles starting from Cu nanocrystals followed by galvanic replacement with Ag ions. Bulk quantification with atomic emission spectroscopy and spatially resolved elemental mapping with electron microscopy reveal distinct nucleation regimes that produce nanoparticles with a tunable Ag shell thickness, but only up to a certain limiting thickness. We develop a quantitative transport model that explains this observed self-limiting structure as arising from the balance between entropy-driven interdiffusion and a positive mixing enthalpy. The proposed model depends only on the intrinsic physical properties of the system such as diffusivity and mixing energy and directly yields a high level of agreement with the elemental mapping profiles without requiring additional fit parameters.

Manipulating the Transition Dipole Moment of CsPbBr3 Perovskite Nanocrystals for Superior Optical Properties.

Colloidal cesium lead halide perovskite nanocrystals exhibit unique photophysical properties including high quantum yields, tunable emission colors, and narrow photoluminescence spectra that have marked them as promising light emitters for applications in diverse photonic devices. Randomly oriented transition dipole moments have limited the light outcoupling efficiency of all isotropic light sources, including perovskites. In this report we design and synthesize deep blue emitting, quantum confined, perovskite nanoplates and analyze their optical properties by combining angular emission measurements with back focal plane imaging and correlating the results with physical characterization. By reducing the dimensions of the nanocrystals and depositing them face down onto a substrate by spin coating, we orient the average transition dipole moment of films into the plane of the substrate and improve the emission properties for light emitting applications. We then exploit the sensitivity of the perovskite electronic transitions to the dielectric environment at the interface between the crystal and their surroundings to reduce the angle between the average transition dipole moment and the surface to only 14° and maximize potential light emission efficiency. This tunability of the electronic transition that governs light emission in perovskites is unique and, coupled with their excellent photophysical properties, introduces a valuable method to extend the efficiencies and applications of perovskite based photonic devices beyond those based on current materials.

Design Principles for Trap-Free CsPbX3 Nanocrystals: Enumerating and Eliminating Surface Halide Vacancies with Softer Lewis Bases.

We introduce a general surface passivation mechanism for cesium lead halide perovskite materials (CsPbX3, X = Cl, Br, I) that is supported by a combined experimental and theoretical study of the nanocrystal surface chemistry. A variety of spectroscopic methods are employed together with ab initio calculations to identify surface halide vacancies as the predominant source of charge trapping. The number of surface traps per nanocrystal is quantified by 1H NMR spectroscopy, and that number is consistent with a simple trapping model in which surface halide vacancies create deleterious under-coordinated lead atoms. These halide vacancies exhibit trapping behavior that differs among CsPbCl3, CsPbBr3, and CsPbI3. Ab initio calculations suggest that introduction of anionic X-type ligands can produce trap-free band gaps by altering the energetics of lead-based defect levels. General rules for selecting effective passivating ligand pairs are introduced by considering established principles of coordination chemistry. Introducing softer, anionic, X-type Lewis bases that target under-coordinated lead atoms results in absolute quantum yields approaching unity and monoexponential luminescence decay kinetics, thereby indicating full trap passivation. This work provides a systematic framework for preparing highly luminescent CsPbX3 nanocrystals with variable compositions and dimensionalities, thereby improving the fundamental understanding of these materials and informing future synthetic and post-synthetic efforts toward trap-free CsPbX3 nanocrystals.

Tailoring Morphology of Cu-Ag Nanocrescents and Core-Shell Nanocrystals Guided by a Thermodynamic Model.

The ability to predict and control the formation of bimetallic heterogeneous nanocrystals is desirable for many applications in plasmonics and catalysis. Here, we report the synthesis and characterization of stable, monodisperse, and solution-processed Cu-Ag bimetallic nanoparticles with specific but unusual elemental arrangements that are consistent with a recently developed thermodynamic model. Using air-free scanning transmission electron microscopy with energy-dispersive X-ray spectroscopy, the distribution of Cu and Ag positions was unambiguously identified within individual nanocrystals (NCs), leading to the discovery of a Cu-Ag nanocrescent shape. A simple yet versatile thermodynamic model was applied to illustrate how the interplay between surface and interface energies determines the particle morphology. It is found that there exists a range of surface-to-interface energy ratios under which crescent-shaped nanocrystals are the thermodynamically favored products, with the morphology tunable by adjusting the Ag content. We further show the conversion of Cu-Ag nanocrescents into Ag@Cu2O upon mild oxidation, whereas fully core-shell Cu@Ag NCs are robust against oxidation up to 100 °C. The plasmonic and interband absorptions of Cu-Ag NCs depend on the composition and the degree of Cu oxidation, which may find application in light-driven catalysis.

The Making and Breaking of Lead-Free Double Perovskite Nanocrystals of Cesium Silver-Bismuth Halide Compositions.

Replacing lead in halide perovskites is of great interest due to concerns about stability and toxicity. Recently, lead free double perovskites in which the unit cell is doubled and two divalent lead cations are substituted by a combination of mono- and trivalent cations have been synthesized as bulk single crystals and as thin films. Here, we study stability and optical properties of all-inorganic cesium silver(I) bismuth(III) chloride and bromide nanocrystals with the double perovskite crystal structure. The cube-shaped nanocrystals are monodisperse in size with typical side lengths of 8 to 15 nm. The absorption spectrum of the nanocrystals presents a sharp peak, which we assign to a direct bismuth s-p transition and not to a quantum confined excitonic transition. Using this spectroscopic handle combined with high-resolution transmission electron microscopy (TEM) based elemental analysis, we conduct stoichiometric studies at the single nanocrystal level as well as decomposition assays in solution and observe that Ag+ diffusion and coalescence is one of the pathways by which this material degrades. Drying the nanocrystals leads to self-assembly into ordered nanocrystal solids, and these exhibit less degradation than nanocrystals in solution. Our results demonstrate that Cs2AgBiX6 (X = Cl, Br) nanocrystals are a useful model system to study structure-function relationships in the search for stable nontoxic halide perovskites.

Carbon Dioxide Dimer Radical Anion as Surface Intermediate of Photoinduced CO2 Reduction at Aqueous Cu and CdSe Nanoparticle Catalysts by Rapid-Scan FT-IR Spectroscopy.

Monitoring of visible light sensitized reduction of CO2 at Cu nanoparticles in aqueous solution by rapid-scan ATR FT-IR spectroscopy on the time scale of seconds allowed structural identification of a one-electron intermediate and demonstrated its kinetic relevancy for the first time. Isotopic labeling (12C: 1632, 1358, 1346 cm-1; 13C: 1588, 1326, 1316 cm-1) revealed a species of carbon dioxide dimer radical anion structure, most likely bound to the catalyst surface through carbon. Intermediacy of Cu-C(═O)OCO2- surface species is in agreement with a recently proposed mechanism for electrocatalytic CO2 reduction at Cu metal nanoparticles based on Tafel slope analysis. Spontaneous decrease of the intermediate after termination of the photosensitization pulse (Sn porphyrin excited at 405 nm) was accompanied by the growth of HCO3-. CO was produced as well, but sensitive detection required photolysis for tens of minutes. A direct kinetic link between a C2O4- surface intermediate and the CO product was also demonstrated for photocatalyzed CO2 reduction at aqueous CdSe nanoparticles, where first order growth of a Cd-C(═O)OCO2- species was accompanied by rise of CO (monitored by a fast Ni complex trap) and HCO3- showing a distinct induction period. The detection of the one-electron surface intermediate and confirmation of its catalytic relevancy was enabled by the delivery of electrons one-by-one by the photosensitization method. The observation of carbon dioxide dimer radical anion points to approaches for rate enhancements of heterogeneous CO2 reduction by creating catalytic environments that favor formation of this intermediate.

Surfactant-mediated electrodeposition of a water-oxidizing manganese oxide.

Splitting water into hydrogen and oxygen is one of the most promising ways of storing energy from intermittent, renewable sources in the future. Toward this goal, development of inexpensive, stable, and non-toxic catalysts for water oxidation is crucial. We report that the electrodeposition of manganese oxide in the presence of sodium dodecyl sulfate (SDS) produces a material that is highly active for electrocatalytic water oxidation at pH near 7 and remains stable for over 24 hours of sustained electrolysis. Clark electrode measurements demonstrate more than 95% Faradaic efficiency for oxygen evolution after an initial charging period. We found that catalytic performance was optimized in films prepared by electrodeposition using a precursor solution containing moderate concentration of substrates, namely 25 mM Mn(2+) and 25 mM SDS. Microstructure and elemental analyses revealed that the deposited material, a mixed-phase manganese oxide, is structurally similar to materials used for electrochemical capacitors and batteries, drawing a parallel between highly studied cathode materials for rechargeable batteries and heterogeneous catalysts for water oxidation.