Michael S. Arnold

Incoherent population mixing creates a large background in ultrafast, action-detected pump–probe and 2D spectroscopies. This background often obscures the desired coherent responses and leads to incorrect assignment of spectral features. In a previous study, we postulated three classes of incoherent population mixing but could only measure one. In this work, we report spectra collected using a two-quantum (2Q) pulse sequence generated by a fully phase stable spectrometer, enabling spectra containing all three classes of incoherent responses to be measured. We test the postulated formalism describing incoherent population mixing by collecting and simulating spectra measured using a 2Q pulse sequence. A 2Q pulse sequence reported here also provides a means of determining the magnitude of these backgrounds relative to the desired coherent response. Using this technique, we obtain results consistent with the existence of three incoherent population mixing responses, reveal their spectroscopic signatures, and establish a formalism for their description, enabling future methods for their identification and elimination.

Many conclusions about energy conversion in next-generation photovoltaic devices are gleaned indirectly from optical measurements of exciton dynamics, not directly from photocurrent itself. This method is problematic because optical measurements report on all excitons, not just productive ones. Using a new ultrafast photocurrent spectrometer, we compare exciton dynamics of semiconducting carbon nanotubes measured in films to those measured in devices using photoabsorption- and photocurrent-detected transient and two-dimensional spectroscopies. We find that photoabsorption detection greatly overestimates the importance of long-lived excitons for photovoltaic device performance. Excitons diffuse across nanotubes for picoseconds, but we find that the photocurrent is mostly created by excitons that diffuse little before dissociating at the electron transfer interface within 30 femtoseconds of being created. Thus, scientific conclusions reached from optical-only studies bear little importance to the performance of these devices, calling into question the processes thought critical for efficient photoconversion. This study points to the necessity for directly measuring photocurrent-generating exciton dynamics, not surmising them from optical spectroscopy alone.

Graphene nanoribbons (GNRs) offer the possibility of combining the exceptional carrier mobility of graphene with a width- and edge-tunable bandgap, enabling semiconducting behavior essential for nanoelectronic and optoelectronic devices. Anisotropic chemical vapor deposition of GNRs on Ge(001) has emerged as a promising avenue for synthesizing high-aspect ratio, aligned GNRs with faceted armchair edges on technologically relevant substrates, yet the influence of carbon precursor on growth outcomes remains unexplored. Here, we present a systematic comparison of GNR synthesis on Ge(001) using methane (CH4), ethylene (C2H4), and benzene (C6H6) as feedstocks. The choice of precursor significantly impacts GNR growth rates, anisotropy, and structural quality. CH4 yields the highest-aspect ratio ribbons, while C2H4 and C6H6 produce lower-aspect ratio ribbons presumably due to the multi-carbon active species they generate, lowering the barrier for nucleating new atomic rows when being incorporated into the growing edges of GNRs. These findings establish direct links between precursor chemistry and GNR growth characteristics, offering opportunities for further optimization and tailored synthesis.

Carbon nanotubes (CNTs) must be ordered into densely aligned arrays to fully exploit their electronic properties in next-generation integrated circuits. Recent advances have shown that CNTs can accumulate and self-order at liquid-liquid interfaces, from which the CNTs can be transferred onto a substrate to create dense CNT arrays with remarkable electronic characteristics. Here, by leveraging in situ polarized optical microscopy, we investigate the self-assembly of CNTs at organic solvent-water interfaces and answer key questions about CNT assembly structure and formation kinetics. We find that CNTs spontaneously form liquid crystalline (LC) phases at the liquid-liquid interface with a density strongly dependent on the concentration of CNTs in the organic solvent ink. This LC behavior is robust across a range of polymer wrappers, including polyfluorenes, triblock copolymers, and polycarbazole (PCz). Polarized microscopy reveals that the resulting LC domains are polycrystalline in nature with domain size governed by the kinetics of LC formation. Additives can alter interfacial dynamics─either by promoting Marangoni flow or by enhancing CNT transport─offering an avenue to tune domain characteristics. We find that the LC domain structure formed at the interface is largely preserved upon transfer to a solid substrate, indicating that optimizing interfacial ordering is key to achieving high-quality CNT arrays for electronic applications. In cases where distortions occur during transfer, they often arise from a mismatch between the substrate translation speed and the transport velocity of the LC to the solid surface.

Semiconductor doping has yielded performance enhancements in organic thermoelectrics, light-emitting diodes, transistors, and photovoltaics. In analogous devices using semiconducting single-walled carbon nanotubes (s-SWCNTs), similar doping strategies have been studied for the enhancement of thermoelectrics, light-emitting diodes, and transistors, but virtually no studies have examined the impact of doping on photovoltaic behavior. Here, we investigate chemically p-doped (6,5) s-SWCNTs as light-absorbing donor materials in photovoltaic donor–acceptor heterojunctions with C₆₀. Although doping reduces absorption, the photocurrent generated per absorbed photon remains constant up to a dopant density of ∼80 μm^–1. This unexpected outcome contrasts with the sharp quenching of photoluminescence observed at similar doping levels. We attribute this divergence to the ultrafast electron transfer from s-SWCNTs to C₆₀, and we quantitatively describe this system by using a diffusion-limited contact quenching model. We also demonstrate that doping sensitizes s-SWCNTs to generate photocurrent from trions near 1.07 eV. While trion photocurrent is weakerlikely due to reduced diffusivity and a lower driving force for dissociationit provides a direct route to electrically probe trionic states. These findings establish that doped s-SWCNTs can operate efficiently in photovoltaic devices and serve as a platform for studying trion dynamics in excitonic semiconductors.

Excitons diffusing to a charge-separating interface is a necessary step to convert energy into current in next-generation photovoltaics. In this report, made possible by a new ultrafast spectrometer design, we compare exciton dynamics measured using both photoabsorption- and photocurrent-detected transient and 2D spectroscopies. For a device with semiconducting carbon nanotubes as the exciton transport material, we find that photoabsorption detection greatly overestimates the importance of long-lived excitons for device performance. Excitons diffuse and transfer between nanotubes for several picoseconds, but the large majority of photocurrent is created within 30fs by excitons that diffuse little to the C60 electron transfer material. These results change our understanding of the material features most important for these photovoltaics. Photoabsorption detection measures all excitons, but not all photogenerated excitons generate current. To understand device efficiency, this study points to the necessity for directly measuring the exciton dynamics responsible for photocurrent.

The wafer-scale synthesis of single-crystal graphene on technologically important substrates is a major challenge inhibiting the development of next-generation devices that harness the exceptional electronic, thermal, and mechanical properties of single-crystal graphene. Here, the factors controlling the domain orientation of graphene grown epitaxially on Ge(110) are elucidated, and this insight is utilized to produce graphene with minimal polycrystallinity. In the early stages of growth, most graphene islands have unidirectionally aligned lattices. However, we discover a secondary nucleation phenomenon in which misoriented graphene domains nucleate near/from the island edges, introducing defective grain boundaries and significantly increasing polycrystallinity throughout growth. We find that secondary nucleation occurs when islands grow over Ge steps, which form because of an interplay between the island growth and Ge surface topography evolution. Strategies for suppressing secondary nucleation are developed, enabling the synthesis of graphene in which the predominant crystal orientation has high coverage >99% and low rotational spread <0.6°. This work overcomes the irreproducibility of graphene epitaxy on Ge(110) reported in the literature, providing a route toward the large-area synthesis of single-crystal graphene on technologically useful semiconductors.