Characterizations of two-dimensional materials with cryogenic ultrahigh vacuum near-field optical microscopy in the visible range

The ability to understand the behavior of materials hinges on the ability to measure lattice structures and properties with relevant spatial resolution. Improvements to measurement methods and the development of new instrumentation have led to studies of materials with respect to increasingly local interactions or phenomena.¹ At the surface or interface of materials, dimensional confinement can lead to new or exotic properties that can be highly localized. This becomes especially relevant in the case of atomically thin layers,² or two-dimensional (2D) materials.³ As a result, a thorough understanding of this class of materials requires a highly sensitive characterization method with nanoscale or angstrom-scale spatial resolution.

Since the discovery of graphene in 2004,⁴ 2D materials have attracted considerable attention due to their unique electronic, optical, and mechanical properties that arise within atomically thin sheets.^5,6 The ever-increasing library of 2D materials and more recently van der Waals heterostructures⁷ requires the ability to observe relevant phenomena with a high level of sensitivity with respect to atomic-scale landscapes.⁸ In this manner, the ongoing development of new characterization methods is essential to realizing the full potential of both new and established materials.⁹ The situation becomes further complicated when systems exhibit nanoscale heterogeneity or local disorder and phenomena,¹⁰ which can only be captured by measurements with a high degree of spatial resolution.

The invention of the scanning probe microscope (SPM), specifically the scanning tunneling microscope (STM), enabled the ability to image the electronic structure and atomic lattice of surfaces in real-space.¹¹ As a result, the SPM quickly came to revolutionize our understanding of surfaces and interfaces. Within the STM, an atomically sharp metallic probe or tip is brought very close to a conductive or semiconductive sample. When the probe is terminated in a single apex atom, it becomes possible to image a surface with atomic resolution. A voltage bias is applied between the sample and the tip resulting in the flow of electrons (tunneling current). This tunneling current is exceedingly sensitive to the local density of states of the surface at the position of the tip.¹² Subsequently, STM images can capture a mixture of topographical and electronic information, and so, the tip-sample junction becomes a rich source of local information that can be used to image materials by raster scanning the probe tip across the surface. However, stability of the tip-sample junction is essential, and so, many STMs rely upon ultrahigh vacuum (UHV) and cryogenic temperatures. Such implementations have resulted in studies that provide unprecedented images of molecules and materials at the atomic scale, resolving individual molecules, atoms, and even bonds.¹³ 

By contrast, traditional imaging and spectroscopic methods that rely on light are diffraction-limited according to the Abbe diffraction limit, where the resolving distance is roughly proportional to $λ2$⁠, where $λ$ is the wavelength of light used to image a sample.¹⁴ SPM circumvents this limit by relying on the interaction between an ideally atomically sharp probe tip and a sample to yield images with subnanoscale resolution. Some efforts have found success in using the SPM probe to either confine light into the near-field or alternatively locally induce photon emission or measure the effects of light absorption.¹⁵ Although initially proposed in 1928 by Synge,¹⁶ near-field optical microscopy was fundamentally limited by contemporaneous technology and has only recently become a viable method to characterize low-dimensional materials at the nanoscale and beyond due to significant technical developments. In 1972, Ash and Nicholls realized super-resolution aperture scanning microscopy with a microwave experiment where they were able to resolve a grating with a linewidth of $λ/60$ by relying on a subwavelength-sized aperture in a diaphragm.¹⁷ Slightly later efforts by Pohl in 1984 would use illumination through a hole in a probe tip to image a grating with 488 nm radiation demonstrating a resolving power of at least $λ/20$⁠.¹⁸ Based on these early demonstrations, many early efforts focused on illumination through a probe with an aperture to overcome the diffraction limit in optical microscopy. Pohl, a pioneer in scanning near-field optical microscopy (SNOM), has previously published some remarks¹⁹ and a review²⁰ that provide a fascinating account of the early history of a technique that has since found widespread adoption in countless forms. Pohl recounted clear inspiration from the scanning tunneling microscope,¹⁹ and so from its outset, SNOM has used light-matter interactions within the SPM tip-sample junction to circumvent the diffraction limit and study materials.²¹ 

While both SPM and SNOM measurements were initially demonstrated in ambient conditions, implementations in UHV and at cryogenic temperatures have benefited from improved instrumental and tip-sample junction stability, leading to the ability to use light-matter interactions at the atomic scale to study materials.^22,23 At the same time, the combination of UHV and low temperatures also enables the observation of properties or phenomena that only occur or become detectable under these conditions.^1,24 These include structural, electronic, and optoelectronic properties, as well as highly sensitive phonon modes and hybrid light-matter states.⁹ Importantly, the energy and spatial resolution found in these conditions permit the ability to define these phenomena with respect to atomic landscapes. In this Review, we primarily focus on cryogenic UHV studies of materials performed with one type of apertureless scattering-type SNOM (s-SNOM), tip-enhanced Raman spectroscopy (TERS). Methods such as these use far-field excitation coupled into the near-field of the SPM tip along with the detection of inelastically scattered photons in the far-field, as shown for STM-TERS in Fig. 1(a). To frame STM-TERS within the larger field of methods that couple light-based processes into the tip-sample junction of an SPM in cryogenic UHV conditions, we also introduce another technique that relies on the detection of visible light emitted from the tip-sample junction alongside a few recent applications. In scanning tunneling microscope induced luminescence (STML), light emission is locally stimulated by electrons tunneling inelastically within the junction of the surface and tip. These photons are then detected in the far-field as shown in Fig. 1(b). For more thorough general reviews of these other techniques as well as underlying theory and technical background, we refer the interested reader to review articles that focus on these methods in each relevant section.²⁵ Ultimately, these methods provide new tools to study the properties of materials [Fig. 1(c)]. While we necessarily limit this review to methods that rely on far-field detection of visible light, for completeness, we mention other techniques that use alternative excitation sources or detection schemes in the conclusion, as other phenomena relevant to 2D materials occur at different time and energy scales.

As previously mentioned, many of the early experiments demonstrating s-SNOM relied on light transmission through an aperture with a diameter smaller than the wavelength of light.^18,26–31 In fact, aperture-based s-SNOM remains quite active as technological advancements drive improvements in the fabrication of the crucial probe tip.^32,33 Light excitation and collection can both be accomplished through a nanoaperture at the apex of the otherwise metal-coated tapered optical fiber that serves as the probe.³⁴ While this method has been previously implemented into cryogenic UHV SPM systems,^35,36 here we will instead focus on apertureless methods.

The implementation of s-SNOM into a cryogenic UHV chamber can be challenging due to the required optical access to the tip-sample junction for both collection and excitation. This can potentially compromise the ability to reach and maintain low temperatures due to the necessity to introduce ports into the thermal shielding that typically surrounds the SPM head in low-temperature UHV systems. Some early efforts kept the final focusing and collection optics outside of the UHV chamber so that their alignments could be finely tuned easily.^37–39 However, more recently, these optical elements (lenses^40–42 or a parabolic mirror^43–45) have been moved into the UHV chamber as close to the tip-sample junction as possible on piezoelectric drivers to maximize their collection efficiency, while still maintaining the ability to adjust their alignment. The other optical elements typical of Raman spectroscopy are necessarily kept outside of the UHV chamber, including the optical excitation source (laser) and detector (spectrometer). As a result, optical alignment requires the adjustment of both in vacuo and more easily accessible optical elements. Overall, the implementation of s-SNOM into a cryogenic UHV chamber requires careful consideration of several factors while simultaneously striking a balance between optimizing the SPM function and optical excitation and detection.

TERS falls within the broader category of s-SNOM, drawing clear initial inspiration from surface-enhanced Raman spectroscopy. Surface-enhanced Raman spectroscopy was initially observed in electrochemistry research in the mid to late 1970s when submonolayer coverages of molecules, such as pyridine, yielded unexpectedly strong Raman spectra on roughened silver electrode surfaces.^46–48 At the time, this was attributed to two possible mechanisms, but over time, the principal contribution was found to arise from an electromagnetic enhancement, where the interaction of light with particles much smaller than the incident wavelength drives the coherent oscillation of surface conduction electrons, which is known as a localized surface plasmon.⁴⁹ This near-field strongly enhances the Raman scattering process⁵⁰ and has resulted in the ability to obtain optical vibrational spectra of a single molecule.^51–54 

In 1985, shortly after the demonstration of the STM and acknowledging the relatively recent discovery of surface-enhanced Raman spectroscopy, Wessel proposed surface-enhanced optical microscopy (SEOM).⁵⁵ This may be considered a reinvention of the method proposed by Synge much earlier¹⁶ and a precursor to TERS. When Synge initially described his idea of near-field optical microscopy, he focused on two of the most important experimental limitations: (1) a very intense source of illumination and (2) the ability to make very small adjustments in the position of a sample, on the order of $10−7cm$⁠. The widespread adoption of lasers addressed the first issue, while the invention of the STM and the ability to use piezoelectric translators to precisely raster scan a probe or sample addressed the second. In fact, when Wessel proposed SEOM, he acknowledged clear inspiration from the STM. In SEOM, a laser beam is incident on a submicrometer-sized plasmonic particle that is kept extremely close to the surface, yielding a confined and enhanced electric field. By scanning the surface across the probe particle, Wessel expected the ability to obtain excellent signal for measurements with 5 nm spatial resolution based on the confinement of light in the near-field.⁵⁵ Over time, the spatial resolution of TERS measurements would surpass this expectation and come to provide subnanoscale spatial resolution through improvements to the instrumentation and technique, as well as implementation into cryogenic UHV conditions. While the first UHV-TERS studies focused on thin films or submonolayer coverages of organic molecules on a surface,^37,56,57 the ubiquitous use of Raman spectroscopy to characterize low-dimensional materials,^58,59 along with the ability to fabricate new materials in UHV, quickly led to efforts to develop instrumentation capable of cryogenic UHV-TERS. These works will now become our focus, while a broader review of TERS studies of 2D materials performed in ambient conditions is available elsewhere.⁶⁰ 

In one of the first demonstrations of UHV-TERS of a low-dimensional material, Shiotari et al. investigated graphene nanoribbons (GNRs) on an Au(111) substrate at room temperature.⁶¹ They used an on-surface polymerization technique to fabricate GNRs on the surface, beginning from halogenated aryl precursor molecules via an Ullmann-like coupling reaction, shown in Fig. 2(a), that has been extensively used to fabricate carbon nanostructures on surfaces.^62–65 Thermal annealing resulted in the formation of seven-armchair GNRs (7-AGNRs),^66,67 which were first visualized with STM imaging shown in Fig. 2(b). By positioning the tip over a 7-AGNR, they were able to obtain strongly enhanced near-field Raman spectra that exhibited the radial breathing-like phonon mode, D band, and G band characteristic of GNRs as shown in Fig. 2(c) in the spectrum labeled (iii). Notably, TERS spectra acquired of the clean Au(111) surface (i) and polyanthrylene/Au(111) (ii) lacked any characteristic spectral peak. Additionally, they noticed strong intensity fluctuations or blinking behavior when they acquired a time series of near-field spectra as shown in Fig. 2(d). They attributed this blinking to thermally and laser-induced fluctuations in the tip apex geometry by comparing the experimentally derived intensity distribution with theoretical models. Ultimately, although they were able to capture near-field Raman spectra of the 7-AGNRs, the seemingly unstable tip-sample junction prevented defining the spatial resolution more stringently.

Even though the previously mentioned study was performed in UHV, benefiting from the pristine environment and well-defined sample, the tip-sample junction was found to exhibit significant instability. By cooling the system down to cryogenic temperatures, it becomes possible to significantly improve the mechanical stability of both the tip and sample, as well as minimize thermal drift and improve the energy resolution of spectra. This has led to unprecedented spatial resolution, approaching the atomic scale, as well as the ability to identify subtle effects.^41,68–71 This includes the ability to probe phonon modes near the atomic scale to identify highly localized strain within heterostructures,⁷² define atom-resolved images of ionic lattices,⁷³ and image single atomic adsorbates,⁷⁴ among others.⁷⁵ The improved stability leads to the ability to perform longer duration measurements, such as TERS mapping or line scan experiments, which provide the means to generate real-space spectroscopic images and estimate the lateral resolution of the enhanced near-field. While TERS mapping and line scan measurements have been extensively used for nanoscale studies of materials in ambient and room temperature conditions,^76–79 the spatial resolution has been found to substantially improve at cryogenic temperatures in UHV.⁸⁰ By acquiring sequential TERS spectra with the tip positioned according to a defined pattern, such as at discrete points in a grid or line, a full Raman spectrum is acquired at each pixel. This essentially enables hyperspectral imaging, where the evolution of specific vibrational modes can be tracked in real-space with a TERS map or line scan.⁸¹ Importantly, the acquisition of TERS maps or line scans can take significant amounts of time, even a few or tens of seconds per pixel or point. As a result, the mechanical stability found at low temperatures in UHV becomes critical to achieve the highest degree of spatial resolution. These methods have come to represent an essential aspect within TERS measurements and have led to the realization that the supreme stability of the tip-sample junction enables new approaches that demonstrate subnanoscale spatial resolution.

Within this context, recent TERS studies have focused on low-dimensional allotropes of carbon group elements that can be fabricated in vacuo on a supporting substrate. Specifically, graphene analogs composed of silicon or boron have been investigated. In 2012, silicene, a 2D honeycomb lattice of silicon was first fabricated on an Ag(111) surface and studied with STM.⁸² Following this work, silicene was studied with far-field Raman spectroscopy with both experimental and theoretical approaches.^83–86 However, from the beginning, STM imaging revealed the presence of complex and diverse phases within silicene, therefore requiring spectroscopic characterization with a high degree of spatial resolution. In 2017, silicene on Ag(111) was investigated with UHV-STM-TERS performed at 77 K.⁸⁷ The multiple phases, T, and $13×13$ phases, observed in STM images, shown in Fig. 3(a), could be spectroscopically identified based on characteristic TERS spectra, shown in Fig. 3(b), acquired with the tip positioned over each phase. Furthermore, defects, edges, and even local strain between domain boundaries [Fig. 3(c)] were identified based on the phonon modes observed in TERS spectra [Fig. 3(d)] with a reported enhancement factor as high as $109$⁠, compared to far-field Raman spectra acquired with the tip retracted from the surface, and 0.5 nm spatial resolution. While silicon atoms form a 2D honeycomb lattice on Ag(111), silicon atoms have been found to form unstable clusters or nanoribbons on the anisotropic Ag(110) surface.^88,89 However, the arrangement of atoms within these structures could only be poorly defined with STM requiring spectroscopic methods that lacked spatial resolution.^90,91 With the high spatial resolution of TERS, Sheng et al. acquired vibrational fingerprints of Si clusters and single-strand nanoribbons (SNRs) directly supplementing STM images with spectroscopy. Based on the phonon modes observed in TERS spectra of a nanocluster [Fig. 3(e)], with a spatial resolution near STM, they defined a pentagonal atomic structure as shown in the inset of Fig. 3(f). Furthermore, they found that the observed TERS intensity of specific phonon modes of single-strand silicon nanoribbons [Fig. 3(g)] are sensitive to the tip-sample distance as maintained by manipulating the feedback parameters of the STM, in this case the setpoint tunneling current [Fig. 3(h)].

Following their study of silicene, the same group used TERS to study the vibrational properties of borophene,⁹³ a relatively recently discovered 2D allotrope of boron that can be fabricated on supporting metal substrates.^94–96 The complicated chemistry of boron observed in the bulk material was found to have similar effects in 2D structures, resulting in polymorphic diversity in the monoelemental system. Boron atoms form highly delocalized bonds by sharing electron pairs among three or more atoms.⁹⁷ This results in remarkable chemical and mechanical properties including extremely high bending flexibility and in-plane elasticity.^98,99 When fabricated on the Ag(111) surface, borophene has been found to form two distinct phases that can be defined by a triangular lattice with different densities of periodic hole arrays.¹⁰⁰ The STM image shown in Fig. 4(a) and the atomic lattice shown in Fig. 4(b) illustrate what is referred to as $β12$ or $ν1/6$ borophene.⁹³ As shown in Fig. 4(c), Sheng et al. acquired TERS spectra of $β12$ borophene that showed a strongly enhanced peak at $189cm−1$ that corresponds to the $B3g2$ phonon mode that consists of nearly entirely vertical atomic displacements. They found that they were able to spectroscopically identify the phase of a borophene sheet with $1nm$ spatial resolution.

When integrated into van der Waals heterostructures, the lattice mismatch between materials can result in localized strain that strongly affects mechanical, optical, and electronic properties.^101,102 In ambient conditions, TERS has been used to study highly localized strain in graphene,¹⁰³ MoS₂/Au heterostructures,¹⁰⁴ and nanostructures,^105,106 among others.¹⁰⁷ Based on the expected mechanical properties of borophene, efforts have been made to integrate it as a supporting substrate in a vertical heterostructure to realize its potential within devices. Instead of vertical heterostructures, these early works found that borophene formed abrupt lateral heterojunctions with organic molecules,¹⁰⁸ graphene,¹⁰⁹ or graphene nanoribbons.¹¹⁰ These observations were attributed to preferential adsorption of the secondary species onto the supporting Ag(111) substrate instead of borophene. To address this and probe the effects of molecular adsorption on the lattice of borophene, Li et al. deposited a monolayer (ML) of tetraphenyldibenzoperiflanthene (DBP) onto an Ag(111) surface that was already covered by borophene. They found that in this case, DBP would adsorb onto the borophene and considered this mixed vertical heterostructure with STM and TERS as shown in Fig. 5(a). As previously mentioned, borophene can form two distinct phases on the Ag(111) surface, which can be differentiated based on characteristic phonon modes. Despite being covered by DBP, the authors identified these phases with TERS mapping of these specific vibrational modes that were found to shift due to molecular adsorption as demonstrated in Fig. 5(b). However, beyond the simple identification of buried borophene phases, by comparing experimentally acquired TERS spectra with simulations [Fig. 5(c)], they were able to identify that molecular adsorption induced $∼0.6%$ tensile strain [Fig. 5(d)] within the atomic lattice of underlying borophene. Furthermore, by deliberately using the STM tip to remove individual molecules, they found that the molecule-induced strain could be reversibly released as confirmed with TERS line scan measurements that demonstrated angstrom-scale spatial resolution as shown in Figs. 5(e) and 5(f). The high spatial resolution also revealed a delicate strain spillover effect propagated within the borophene lattice $∼1nm$ beyond the location of molecular adsorption.

At cryogenic temperatures, it becomes possible to reliably manipulate atoms or molecules within the tip-sample junction.^111,112 While the previously mentioned study removed molecules with the STM tip to prove the origin of strain within an underlying substrate as shown in Figs. 5(e) and 5(f), it is also possible to deliberately absorb a molecule or atom onto the apex of the tip, therefore functionalizing the tip. This enables the ability to use interactions between the functionalized tip apex and the surface to probe the properties, such as the atomic structure and electrostatic potential, of a material in unique ways.^113–117 In this manner, Crampton et al. used a cobalt tetraphenylporphyrin (CoTPP)-terminated silver tip to achieve atomically resolved ion selective imaging of an insulating Cu₂N monolayer fabricated on a Cu(100) surface.⁷³ Figure 6(a) shows the correlation between TERS spectra for a CoTPP molecule adsorbed onto the Ag tip and a CoTPP molecule lying on the Cu(100) surface. Then, with the CoTPP-functionalized tip, they found that by tracking the TERS intensity of the $1568cm−1$ mode, they could image the atomic lattice of a monolayer of the ionic insulator, Cu₂N, as can be seen in Fig. 6(b). In comparison with STM imaging [Fig. 6(c)], they found that the spatial resolution of TERS imaging matched the resolution found in the STM topography as represented in the extracted line scans in Fig. 6(d). Bias-dependent STM imaging was used to confirm the location of N atoms within the lattice [Figs. 6(e) and 6(f)]. Furthermore, they asserted that by tracking the spectral shift of specific vibrational modes, they could define ion selective images based on Stark shift maps. While in this case, they used an aromatic polyatomic molecule to probe the surface, they have also reported similar studies of molecular systems with a tip functionalized with a single CO molecule.^115–117 The use of a probe molecule enables the study of electronic and structural properties through mapping electrostatic forces that may not otherwise be observable with Raman spectroscopy.^118,119

The most recent studies of molecules with TERS that have demonstrated intramolecular spatial resolution have relied on increasingly small tip-sample gap distances. Over time, experiments and theory have come to suggest that the spatial resolution of TERS measurements approaches the angstrom scale when light is confined to the atomistic near-field and the tip-sample gap distance becomes typically <2 Å.^70,120,121 Importantly, this approach necessitates the supremely stable conditions found in UHV and low temperatures, as otherwise the tip-sample junction becomes unstable. Following their characterization of the electronic structure of ultrathin ZnO layers epitaxially grown on the Ag(111) surface with TERS,⁴⁴ Liu et al. investigated TERS spectra of a material when the tip-sample gap vanishes and atomic point contact (APC) is made.¹²² They first considered ultrathin layers of ZnO on Ag(111) as shown in the STM image in Fig 7(a). TERS spectra acquired with the tip positioned over two-ML ZnO at different tip-sample gap distances revealed that spectra change significantly in the vicinity of the APC. They attributed this to a combination of electromagnetic and chemical enhancement (via electronic resonance) based on their previous characterization of the electronic structure of the ZnO layers. Additionally, they found that the APC could be formed and broken in a reversible manner as illustrated in the waterfall plot of TERS spectra shown in Fig. 7(b), where the spectral evolution about the APC regime was observed to be symmetric. They specifically considered the vibrational mode at $395cm−1$ that was defined to be the result of contact between the tip and ZnO and was identified specifically as a local out-of-plane stretching mode of an oxygen atom in the second layer based on density functional theory calculations as illustrated in the model in Fig 7(d). The authors suggested that interactions between the Ag tip and oxygen atoms in the first layer resulted in charge rearrangement and led to stiffening in the interlayer bond within ZnO, therefore, resulting in the observed $395cm−1$ mode. Furthermore, they attributed the dramatic enhancement of TERS intensity upon APC formation to hybridization between the tip and sample, mentioning that this approach could increase the sensitivity and capability of TERS to consider other materials.

Most TERS studies rely on a plasmonic supporting substrate in addition to the plasmonic probe, i.e., silver tips are used with silver substrates. This results in the so-called gap mode configuration where plasmonic modes couple between the tip and substrate and tunnel across the junction.^123,124 By contrast, based on their observation of a significant increase in TERS intensity when the tip is brought into contact with even a nonplasmonic sample, Liu et al. used APC to study the phonons of the Si(111)-7 × 7 surface.¹²⁵ They first visualized the surface with STM while illuminated with the 633 nm laser that they also used as the source for TERS measurements. As can be seen in Fig. 7(e), illumination during scanning led to the formation of atomic defects that occur only in faulted half unit cells. And so, they used STM imaging to identify unfaulted half unit cells where they brought the Ag tip into atomic point contact with the surface and recorded TERS spectra [Fig. 7(f)]. Aside from the constant observation of the optical phonon mode of bulk Si at $520cm−1$⁠, they noted the emergence of characteristic Raman peaks that they assigned to surface phonons within the APC regime. Based on this, they suggested that, in principle, APC could be used to study other nonplasmonic substrates and understand light-matter interactions at the atomic scale in metal-semiconductor heterostructures.

The continuing realization of new 2D materials as well as increasing efforts to fabricate heterostructures composed of disparate materials requires the ongoing development of new characterization methods. The previously discussed TERS studies of low-dimensional materials have demonstrated the ability to use near-field Raman spectroscopy to probe and define phonon modes that can be used to identify phases, as well as highly localized phenomena. This includes defects and even interactions between layers that may manifest in the form of lattice strain. Since many TERS studies focus on molecular systems, perhaps due to lasting influences from its inspiration from surface-enhanced Raman spectroscopy, its application to studies of materials remains comparatively unexplored. The studies mentioned here demonstrate the powerful potential of this technique to characterize phonons of materials with angstrom-scale spatial resolution.

In a manner that predates STML, but is conceptually similar, Young et al. used a field emission probe brought close to a conducting surface in 1972 and measured secondary electron emission with an electron multiplier detector pointed toward the sample surface.^126,127 In this manner, they were able to generate topographic maps of a 180 line/mm diffraction grating with 30 Å vertical and 4000 Å lateral resolution with an instrument that they coined the Topografiner. Later, based on observations of photon emission from metal-oxide-metal tunnel junctions¹²⁸ and surfaces exposed to low-energy electron beams¹²⁹ through what was attributed to inelastic electron tunneling processes, Gimzewski et al. placed a photon detector near the tip-sample junction of an STM in 1988.¹³⁰ They observed light emission from the tunneling gap of an STM. Since the tip becomes the source of excitation in STML, the instrumental setup becomes relatively simpler compared to apertureless s-SNOM. The optical elements within the UHV chamber are only used for the collection of emitted photons that are generated by tunneling electrons. This results in the use of optical elements with high numerical apertures, such as parabolic mirrors or a single lens that is as large and close to the tip-sample junction as possible.^131,132 Perhaps due to the fact that the STM tip becomes the excitation source and no external source is required, recent advances in STML measurements have focused on the development of new detection methods, moving beyond a spectrometer to attain unprecedented time resolution of dynamic processes that will be mentioned later. STML has become an extremely powerful method to probe a material at the atomic scale since the light generated within the STM junction carries information that describes the structure under investigation.

Initial demonstrations of STML attributed the light emission to one of two possible mechanisms resulting in a technique capable of studying optical and electronic properties of molecules and materials. Light emission from the tip-sample junction was proposed to arise from inelastic electron tunneling processes or the injection of hot electrons.¹³³ This provides the ability to use STML to measure multiple types of phenomena such as overbias emission,¹³⁴ photon pair emission,¹³⁵ and electron-hole recombination that leads to excitation emission.¹³⁶ Importantly, the observation of light emission from recombination processes requires that the emitter be decoupled from a metallic substrate. In the case of molecular systems, this has been accomplished by the growth of multilayers of molecules or thin insulating layers, such as Al₂O₃ or NaCl,^15,136–138 with studies even reporting the ability to resolve vibronic features that correspond to a Raman vibrational fingerprint.^139,140 Some early STML studies of materials used noble metal substrates^141,142 demonstrating atomic resolution in STM-induced photon emission images of the Au(110) surface.¹⁴³ More recently, STML has been used to study engineered materials, such as several-atom long silver chains on the NiAl(110) surface¹⁴⁴ and thin films of oxides on gold substrates.^145–147 Here, we highlight a few recent studies that used unique methods to probe materials with STML.

Aside from insulating layers, the STM tip itself can also be used to decouple a material from the substrate enabling the observation of its optoelectronic properties. As shown in Fig. 8(a), Chong et al. used the STM to lift one end of an individual GNR from an Au(111) surface,¹⁴⁸ forming a GNR junction in a manner that has previously been used to study the conductance of a single GNR.¹⁴⁹ By placing the STM tip over the terminus of a GNR and ramping the bias voltage, they were able to dehydrogenate a terminal carbon, where H-terminated [Fig. 8(b)] and C-terminated [Fig. 8(c)] species have distinct appearances in STM imaging. Furthermore, the decay of the conductance while the tip was retracted was used to verify the formation of a GNR junction based on the decreased decay rate for retraction compared to the approach as shown in Fig. 8(d). They found that STML spectra of H-terminated GNRs yielded weak featureless emission spectra, while spectra of C-terminated GNRs exhibited sharp features suggestive of a radiative transition intrinsic to the GNR [Fig. 8(e)]. Furthermore, the spectra were found to exhibit both applied voltage [Fig. 8(f)] and tip-sample gap distance [Fig. 8(g)] dependence. In this manner, they were able to reveal new insight into the optical transitions of GNRs by considering a single GNR bridging the gap between the gold-coated tungsten tip and Au(111) substrate.

The atomic resolution of STML makes it an ideal method to study photon emission that arises from individual atomic defects. Schuler et al. grew WS₂ islands on epitaxial graphene supported on an SiC substrate and considered atomic defects within the top layer of WS₂ with a combination of SPM methods and STML.¹⁵⁰ They found that inelastic electron tunneling from the gold-coated tip into single sulfur top vacancies led to photon emission. Emission was found to be highly localized and dependent upon the location of charge injection by the STM tip. Furthermore, the defect also exhibited photon emission at a tunneling bias significantly lower than the surrounding pristine WS₂. They supplemented these spectral measurements with high-resolution SPM imaging. Ultimately, with a focus on the high spatial resolution, they were able to characterize individual atomic defects within the 2D semiconductors, therefore realizing and imaging atomic-scale single-photon sources.

Fluctuations induced within the well-controlled atomic-scale environment observed in a low-temperature UHV-STM can be used to probe the effects of atomic-scale processes on light emission from a picocavity. Rosławska et al. used STML to develop an understanding of how the plasmonic properties of the tip-sample junction change at the single-atom limit.¹⁵¹ As shown in Fig. 9(a), while recording an STM image of the Au(111) surface with a gold STM tip at T = 4 K, they interrupted scanning and deliberately deposited a single atom onto the surface. By simultaneously recording light emission, they observed that the deposition of an atom from the gold tip significantly changed the observed photon yield [Fig. 9(b)]. The overall electroluminescence was found to reduce due to this change in the atomic apex of the tip, with a notably reduced intensity observed when the tip was positioned on top of the deposited atom and a surface defect. Significantly, even though the atomic apex of the tip was modified, the shape of the spectral features found with STML was unchanged aside from an overall reduction in intensity [Fig. 9(c)]. Beyond capturing the effects of the deposition of a single atom on STML spectra, the authors also considered dynamic changes when the tip was in atomic point contact with the Au(111) surface as shown in Fig. 9(d). They observed that the optical spectra and integrated light intensity both undergo changes as a function of time [Figs. 9(e) and 9(f)], with the feedback loop of the STM causing the emission to return to its relative original intensity. They attributed these spectral fluctuations to current changes, as well as mechanical stress and thermal effects that cause changes and modifications to the atomic structure of the tip. However, while the intensity was found to change significantly over the course of their measurements, they found that the shape of STML spectra remained relatively unchanged. This suggests that in addition to spectral features, the overall intensity of an STML spectrum can be used to define the tip-sample junction. Based on their observations, they stressed the importance of atomic-scale features in the STM tip-sample junction in terms of near-field optical microscopy measurements that rely on the formation of a picocavity.^152,153 They specifically note that in their case, even at slightly elevated temperatures, i.e., ∼50 K compared to 4 K, spontaneous atomic rearrangements become probable, preventing the desired observation of atomic-scale light emission.

More recently, STML has been used to capture dynamic processes, with reviews of recent progress available elsewhere.^23,154 Time-resolved detection of single photons has been used to probe excitation and relaxation dynamics within the STM tip-sample junction.^155,156 Time-resolved measurements have permitted the differentiation of emission mechanisms due to their different dynamics. Specifically, the excitonic and plasmonic components of emission were characterized, separated, and tuned in thin films of C₆₀ supported on noble metal substrates.¹⁵⁷ Furthermore, the exciton¹⁵⁸ or plasmonic¹⁵⁹ dynamics of single-photon emitters within these molecular films have been probed by the combination of STM with a Hanbury Brown and Twiss interferometer. The same detection method has also recently been used to observe photon superbunching from an STM junction consisting of an Ag tip and an Ag(111) substrate through a process that was found to be electrically driven.¹³⁵ Since STML relies on the STM tip to serve as the excitation source, some instrumental efforts have been focused on the development of new methods to detect the emitted photons. The ability to detect single photons with exceptional time resolution offers the potential to study dynamic processes in materials at the atomic scale.

From its initial demonstrations in its various forms, scanning near-field optical microscopy has offered unique insight into materials by probing light-matter interactions below the diffraction limit.¹⁶⁰ The combination of optical spectroscopy with a scanning probe microscope provides a highly useful tool to investigate the properties of a material within the probe-sample interaction volume. Recently, atomic-scale features at the apex of the probe have become considered essential to achieving the highest degree of spatial resolution. As a result, the technique has benefited significantly from the stability found with implementations in UHV and further with low-temperature measurements. In this Review, we discussed the development and recent studies of materials with two methods that rely on the detection of visible light that describes the material in the atomic-scale tip-sample junction of a cryogenic UHV-STM. The insight from methods such as these is expected to inform design principles for engineering materials with a focus on atomic-scale phenomena.¹⁶¹ 

TERS relies on the highly enhanced and localized electric field generated at a plasmonic tip apex under light illumination to measure Raman scattering in the near-field. This provides a platform to measure the vibrational modes or phonons of materials near the atomic scale. With an initial focus on molecular systems, various low-temperature UHV-TERS studies have shown intramolecular (angstrom-scale) resolution of individual vibrational modes within a single molecule and, in some cases, even shown sensitivity to the formation or dissociation of a single chemical bond.^41,162–166 Comparatively, UHV-TERS studies of materials have focused on slightly larger scales, detecting and identifying distinct phases, but a few early studies have demonstrated the ability to detect and identify atomic defects as well as the effects of adsorption or chemisorption of individual chemical species. Furthermore, TERS in ambient conditions has recently been used to study low-angle twisted bilayer graphene¹⁶⁷ and mixed heterostructures.¹⁰⁴ This avenue remains currently unexplored with UHV-TERS, perhaps due to a degree of technical incompatibility with fabrication methods for these materials. However, the ability to probe highly sensitive phonon modes at the angstrom scale in these and other newly designed materials offers tremendous potential once these obstacles are surmounted. Finally, thus far most UHV-TERS studies have been focused on steady-state measurements. However, the use of pulsed laser sources in the future offers the ability to capture dynamic processes^168,169 as has been demonstrated with ultrafast surface-enhanced Raman spectroscopy.¹⁷⁰ 

STML uses an inelastic electron tunneling process or the injection of hot electrons to excite light emission at the STM junction. As shown in the discussed studies, this allows the precise spectral identification of emission down to a single atom or defect. Time correlated measurements provide a method to differentiate between the different mechanisms as well as track energy flow. In this way, the development of new detection schemes improves the flexibility of STML to consider dynamic processes with exceptional temporal resolution. Although STML measurements require that a molecule or material be decoupled from the underlying substrate to prevent quenching effects, the study of heterostructures with this technique remains limited, offering a possible future direction for this powerful technique.

While the studies mentioned here focused on the detection of visible light, infrared,^45,171–173 THz,^174–180 and even longer wavelength radiation¹⁸¹ has been effectively coupled into UHV scanning probe microscope junctions with various excitation and detection schemes.^23,182–185 In this regard, these methods have led to studies of light-matter interactions with exceptional spatial as well as temporal resolution. The ongoing development of these techniques continues to offer exciting new opportunities to examine both established and new materials.

J.F.S. acknowledges a National Research Council Postdoctoral Fellowship. N.J. acknowledges support from the National Science Foundation (No. CHE-1944796).

The authors have no conflicts to disclose.

All data shown within this article has been previously published. Suitable references can be found in the corresponding figure captions.

Jeremy F. Schultz received his Ph.D. in chemistry in 2021 from the University of Illinois Chicago (UIC) where he worked with Professor Nan Jiang to develop and apply a cryogenic ultrahigh vacuum (UHV) scanning tunneling microscope coupled with tip-enhanced Raman spectroscopy. This tandem technique enables simultaneous nanoimaging and nanospectroscopy to study molecules, nanostructures, and low-dimensional materials on surfaces at the angstrom scale. His work has been recognized by various awards, including the Wayne B. Nottingham Prize at the 81st Physical Electronics Conference 2021, the Barbara Stull Graduate Student Award from the Society for Applied Spectroscopy, and the Graduate Research Award from the American Vacuum Society. Jeremy is currently a National Research Council (NRC) Research Associate at the National Institute for Standards and Technology (NIST) where he is developing a cryogenic UHV scanning probe microscope capable of new nanospectroscopic methods to characterize quantum and 2D materials.