Low-Energy Inverse Photoelectron Spectroscopy

• Low-Energy Inverse Photoelectron Spectroscopy (LEIPS) is a technique that probes unoccupied electronic states by injecting low-energy electrons and detecting emitted photons.
• LEIPS enables low-damage material analysis with enhanced signal-to-noise ratio while facing challenges from instrumental broadening and overlapping spectral features.
• Innovations in electron source design and analytical protocols in LEIPS have led to improved energy resolution and quantitative peak separation for precise band mapping.

Searching arXiv for recent and foundational LEIPS-related papers to ground the article. Low-Energy Inverse Photoelectron Spectroscopy (LEIPS) is a low-kinetic-energy form of inverse photoelectron spectroscopy (IPES) for probing the unoccupied electronic states of materials. IPES may be regarded as the inversion process of photoelectron spectroscopy (PES), which examines occupied states, whereas LEIPS injects low-energy electrons into a target and detects the emitted photons associated with population of unoccupied states. Recent LEIPS developments are characterized by improved signal-to-noise ratio and minimal sample damage compared to conventional IPES, but the instrumental resolution remains approximately $0.2\text{ eV}$ to $0.25\text{ eV}$, one order of magnitude lower than that of PES; as a result, spectral broadening and peak overlap remain central analytical constraints (Nakazawa et al., 25 Sep 2025). Advances in electron-source design and analysis protocols have consequently been central to the technique’s maturation, including high-current low-energy sources for IPES and quantitative frameworks for peak separation and angle-resolved band mapping (Ibach et al., 2022, Yang et al., 2020).

1. Physical basis and measurement concept

LEIPS belongs to the IPES family of photon-detection spectroscopies for unoccupied electronic structure. In the formulation used in recent angle-resolved work, the method probes unoccupied conduction states by injecting low-energy electrons and measuring photons emitted as electrons occupy conduction-band states; in that description, the electron goes from the vacuum into unoccupied conduction states, forming an anion state as the final state of the process (Yang et al., 2020). This distinguishes LEIPS from PES by the direction of the elementary process and from angle-resolved two-photon photoemission (AR-2PPE) by the nature of the accessed state manifold.

The practical appeal of LEIPS follows from the combination of low kinetic energy and improved signal-to-noise ratio. In the LEIPS-focused peak-separation study, these characteristics are explicitly linked to minimal sample damage relative to conventional IPES, which is especially relevant for delicate molecular and organic systems (Nakazawa et al., 25 Sep 2025). A plausible implication is that LEIPS is particularly well suited to materials classes in which conventional electron-beam conditions would compromise surface integrity or induce charging.

The principal limitation is spectral resolution. Even in recently developed LEIPS, instrumental broadening remains large enough that closely spaced unoccupied states are often measured as overlapping structures rather than isolated peaks. In practice, this places LEIPS in a regime where extraction of physically meaningful peak positions and intensities depends not only on the instrument but also on post-acquisition spectral analysis (Nakazawa et al., 25 Sep 2025).

2. Instrumentation and low-kinetic-energy electron sources

A recent high-current electron source for IPES comprises a thermal cathode electron emission system, an electrostatic deflector-monochromator, and a lens system that provides variable kinetic energy from $1.6$ to $20\text{ eV}$ at the target (Ibach et al., 2022). The emission system uses a thermal LaB$_6$ cathode with a $6\,\mu\text{m}$ tip and $60^\circ$ cone. Electrons are accelerated and directed toward the monochromator through three copper lens elements, denoted A1, A2, and A3, together with a repeller. The design is described as more compact, approximately half the previous size, with the stated consequence of boosting space-charge-limited current and improving beam matching to the monochromator entrance slit.

The monochromator is an Ibach-type electrostatic deflector, extended from $146^\circ$ to $162^\circ$. In the reported implementation, convex deflector plates and shaped cover electrodes control focusing, while precisely defined entrance and exit slits reduce aberrations. Electrons with the appropriate longitudinal velocity are focused onto the exit slit; other trajectories are filtered out, yielding a monochromatic beam. Downstream, a lens system consisting principally of B1 and B2 brings the beam onto the sample while preserving tunability of the target kinetic energy (Ibach et al., 2022).

Performance metrics reported for this source are unusually favorable for low-energy IPES. When scaled to energy resolution, the output current is an order of magnitude higher than that of previously described electron sources developed in the context of electron energy loss spectroscopy. The measured full width at half maximum (FWHM) reaches as low as $78\text{ meV}$ with output current up to approximately $0.25\text{ eV}$0, and transmission to the target reaches up to $0.25\text{ eV}$1 of the monochromatized current (Ibach et al., 2022). The same study reports that output is well described by Gaussian energy distributions without pronounced low-energy tails.

A central interpretive point in that work is the reported “inverse Boersch effect.” Standard continuum space-charge calculations would predict current-dependent energy broadening, and the empirical current–resolution relation is given as $0.25\text{ eV}$2, compared with $0.25\text{ eV}$3 for a simpler space-charge-limited theory. Experimentally, however, the energy spread decreases at high currents rather than increasing. The proposed mechanism is velocity selection in the forward direction by binary electron-electron collisions, which preferentially remove electrons with off-pass longitudinal velocity from the transmitted beam (Ibach et al., 2022). This does not eliminate space-charge effects as a concern in general LEIPS practice, but it demonstrates that under specific monochromator conditions high current and high resolution need not be mutually exclusive.

3. Resolution limits, spectral broadening, and the LEIPS analysis problem

The central analytical problem in LEIPS is instrumental broadening. In the methodological comparison of peak-separation approaches, the instrumental resolution is given as approximately $0.25\text{ eV}$4 to $0.25\text{ eV}$5, and the broadening is modeled as convolution with an instrumental response function, typically Gaussian (Nakazawa et al., 25 Sep 2025). Overlapping peaks are therefore not exceptional cases but common outcomes whenever the intrinsic separation of electronic features is comparable to the instrumental width.

This issue is especially acute for systems with split or closely spaced unoccupied bands. The study on pentacene uses the lowest unoccupied molecular orbital-derived band as a representative LEIPS case: the band consists of two splitting peaks due to the two inequivalent molecules in the unit cell, and these become difficult to distinguish once instrumental broadening is applied (Nakazawa et al., 25 Sep 2025). The modeled test case uses two Gaussians with $0.25\text{ eV}$6 and $0.25\text{ eV}$7, convolved with the instrumental function and contaminated with noise to simulate realistic LEIPS data.

The underlying inverse problem is to recover peak positions and intensities from broadened and noisy spectra. In the deconvolution formalism used in that work, the observed spectrum is represented as

$0.25\text{ eV}$8

followed by

$0.25\text{ eV}$9

where $1.6$0 is the undistorted spectrum, $1.6$1 the instrumental function, $1.6$2 the background, and $1.6$3 a Poisson noise process (Nakazawa et al., 25 Sep 2025). This formulation makes explicit that LEIPS analysis is constrained simultaneously by the instrument response, background treatment, and counting noise.

A further practical limitation is that the value of deconvolution diminishes when intrinsic peak width exceeds instrumental width. The same comparison explicitly notes that if the intrinsic peak width is larger than the instrumental width, the benefits of deconvolution decrease (Nakazawa et al., 25 Sep 2025). This point is important because it limits the scope of claims about resolution recovery: deconvolution can compensate instrumental broadening, but it does not create information absent from intrinsically broad spectral features.

4. Peak-separation methodologies

Three peak-separation strategies have been compared systematically for LEIPS: second derivative analysis, curve fitting, and deconvolution (Nakazawa et al., 25 Sep 2025). Each method addresses the overlap problem differently and imposes distinct assumptions on the data.

The second derivative method identifies peak positions through maxima in the negative second derivative of a smoothed spectrum. Because numerical differentiation amplifies noise, smoothing is essential; the reported implementation uses a Savitzky–Golay filter based on local polynomial fits. The smoothing window size is optimized empirically, since windows that are too small enhance noise and false peaks, whereas windows that are too large oversmooth and suppress genuine structure. The method has the practical advantage of not requiring background subtraction, but it is highly sensitive to noise and to user-defined smoothing parameters, and no strict objective criterion for the smoothing parameter is available (Nakazawa et al., 25 Sep 2025).

Curve fitting represents the measured spectrum as a sum of analytical peak functions together with a residual,

$1.6$4

where $1.6$5 contains the peak model with parameters $1.6$6 for positions, widths, and amplitudes. In the reported implementation, the parameters are determined by nonlinear least squares using the Levenberg–Marquardt algorithm, minimizing

$1.6$7

This strategy is more robust to noise than the second derivative, but it requires a priori selection of the number and type of line shapes, typically Gaussian, Lorentzian, or Voigt, and it requires prior background subtraction. The comparison emphasizes that model dependency can introduce bias when peak shapes are asymmetric or otherwise uncertain (Nakazawa et al., 25 Sep 2025).

Deconvolution aims to recover the undistorted spectrum by removing the known instrumental broadening. The method selected in the LEIPS comparison is Jansson’s iterative deconvolution,

$1.6$8

with

$1.6$9

Iteration number is controlled by monitoring the root mean square error between the measured spectrum and the reconvolved deconvolved spectrum,

$20\text{ eV}$0

In this framework, background subtraction and spectral edge smoothing are not optional refinements but critical preprocessing steps; improper treatment produces artifacts, including false peaks and spectral distortions (Nakazawa et al., 25 Sep 2025).

5. Quantitative performance in LEIPS peak separation

The comparative study provides explicit quantitative thresholds for noise robustness and resolvable peak separation. Under simulated noise conditions quantified by RMSE, the second derivative method remains reliable only for $20\text{ eV}$1, curve fitting remains reliable for $20\text{ eV}$2, and deconvolution remains reliable for $20\text{ eV}$3 (Nakazawa et al., 25 Sep 2025). On this metric, curve fitting and deconvolution outperform second derivative analysis, and deconvolution is the most noise-robust of the three.

Peak-separation capability was evaluated as the minimum resolvable energy separation $20\text{ eV}$4 between two peaks. The second derivative requires $20\text{ eV}$5 to resolve two peaks. Curve fitting resolves $20\text{ eV}$6. Deconvolution achieves $20\text{ eV}$7, particularly when the deconvoluted spectrum is subsequently smoothed and analyzed by the second derivative (Nakazawa et al., 25 Sep 2025). These thresholds provide a practical basis for method selection in systems with narrow band splittings near the instrumental resolution limit.

The same analysis also clarifies the conditions under which each method fails. Second derivative analysis is vulnerable to false structure induced by noise or by unsuitable smoothing windows. Curve fitting can return stable but biased solutions when the chosen line-shape model is not physically appropriate. Deconvolution is especially sensitive to background treatment: raw background or sharp spectral edges generate artifacts, while linear background subtraction or smooth connection to the baseline is essential for reliable recovery (Nakazawa et al., 25 Sep 2025).

The resulting practical framework is explicit. Second derivative analysis is a rapid, model-free check, but should be used with care. Curve fitting is appropriate when physics-based assumptions about the number and type of peaks are defensible. Deconvolution, preferably with Jansson’s method and RMSE-monitored termination, is the most robust approach when background handling and edge preprocessing are done properly (Nakazawa et al., 25 Sep 2025). The study further notes that the framework extends beyond LEIPS to PES and a wide range of spectroscopies.

6. Angle-resolved LEIPS and conduction-band dispersion

LEIPS is not restricted to angle-integrated lineshape analysis. In CH$20\text{ eV}$8NH$20\text{ eV}$9PbI$_6$0 single crystals, angle-resolved low-energy inverse photoelectron spectroscopy (AR-LEIPS) was used together with AR-2PPE to study conduction-band dispersion (Yang et al., 2020). The AR-LEIPS implementation employed incident electrons with kinetic energies $_6$1 to $_6$2, an electron source current density of $_6$3 to $_6$4, sample current of approximately $_6$5, a photon detector based on an elliptical mirror, bandpass filter, and photomultiplier tube, and a detected photon energy of $_6$6. The reported energy resolution was $_6$7, determined from the Ag Fermi edge, and no bias voltage was applied during AR-LEIPS measurements.

Spectra were recorded as a function of emission angle $_6$8 along the high-symmetry IM direction in $_6$9-space at different temperatures. The lowest conduction-band feature, denoted CB$6\,\mu\text{m}$0, showed clear angular dispersion. Representative values given in the study include shifts at $6\,\mu\text{m}$1 from $6\,\mu\text{m}$2 at $6\,\mu\text{m}$3 to $6\,\mu\text{m}$4 at $6\,\mu\text{m}$5 and $6\,\mu\text{m}$6 at $6\,\mu\text{m}$7, and at $6\,\mu\text{m}$8 from $6\,\mu\text{m}$9 at $60^\circ$0 to $60^\circ$1 at $60^\circ$2 and $60^\circ$3 at $60^\circ$4 (Yang et al., 2020). The resulting dispersion is concave around $60^\circ$5, consistent with calculated conduction-band structure in the cubic phase.

Peak positions in those AR-LEIPS data were obtained by Gaussian fitting, and angular coordinates were converted to parallel momentum for $60^\circ$6-$60^\circ$7 mapping. Near the I-point, the effective mass was extracted from a parabolic fit using

$60^\circ$8

Reported values were $60^\circ$9 at $146^\circ$0 and $146^\circ$1 at $146^\circ$2 in AR-LEIPS. These are larger than the AR-2PPE values of $146^\circ$3 at $146^\circ$4 and $146^\circ$5 at $146^\circ$6, a difference that the study states can be ascribed to final-state effects, possibly including electron correlation and upper Hubbard band formation (Yang et al., 2020).

The same comparison reveals that energy positions obtained by AR-LEIPS and AR-2PPE are not interchangeable. The conduction-band minimum CB$146^\circ$7 at the I-point appears at approximately $146^\circ$8 above $146^\circ$9 in AR-LEIPS and approximately $162^\circ$0 above $162^\circ$1 in AR-2PPE, with a disparity of $162^\circ$2. In the paper this is interpreted as approximately half of the on-site Coulomb energy $162^\circ$3, following Hubbard-model reasoning (Yang et al., 2020). Combined with AR-UPS, the corresponding band-gap estimates are $162^\circ$4 and $162^\circ$5. These results show that AR-LEIPS can resolve conduction-band curvature while also remaining sensitive to correlation and final-state effects.

7. Methodological significance, limitations, and interpretive issues

A recurring misconception is to treat LEIPS as a purely low-resolution density-of-states probe. The angle-resolved work on CH$162^\circ$6NH$162^\circ$7PbI$162^\circ$8 shows that clear conduction-band dispersion can be measured, and that the observed dispersion agrees closely with band calculations under the cubic phase (Yang et al., 2020). This suggests that, under suitably controlled low-energy and low-current conditions, LEIPS can function as a momentum-resolved probe of unoccupied bands rather than only as an angle-integrated spectroscopic complement to PES.

An opposite misconception is that improved source design alone resolves LEIPS’s interpretive problems. The high-current low-kinetic-energy source demonstrates that current, transmission, and energy resolution can be substantially improved, including FWHM values as low as $162^\circ$9 in the source characterization (Ibach et al., 2022). Yet the LEIPS peak-separation study shows that instrumental broadening near $78\text{ meV}$0 still produces peak overlap in realistic measurements and that data interpretation remains strongly dependent on background treatment, noise level, and analysis strategy (Nakazawa et al., 25 Sep 2025). Instrumental progress and analytical rigor are therefore complementary rather than substitutive.

The most consequential methodological conclusion in current LEIPS literature is that deconvolution and curve fitting generally outperform second derivative analysis for overlapping LEIPS peaks, but that deconvolution only remains reliable when background subtraction, spectral-edge smoothing, and iteration control are handled carefully (Nakazawa et al., 25 Sep 2025). Conversely, AR-LEIPS studies indicate that even when dispersion is measured reliably, absolute energy positions may differ from those obtained by AR-2PPE because the techniques access different final states and reflect electron-correlation effects differently (Yang et al., 2020). The technique therefore combines direct access to unoccupied states with a nontrivial final-state problem.

Taken together, these studies define LEIPS as a technically specialized but increasingly versatile spectroscopy. Its present profile is that of a low-energy, comparatively low-damage probe of unoccupied states whose utility depends on three intertwined factors: electron-source performance, analysis of broadened spectra, and careful interpretation of final-state and correlation effects. The same studies further indicate that the practical framework developed for LEIPS peak separation is extensible to PES and to a wider class of broadened spectroscopies (Nakazawa et al., 25 Sep 2025).