Surface Charging on Insulating Films with Different Thicknesses in UPS

Abstract

The conventional view holds that ultraviolet photoelectron spectroscopy (UPS) measurements are not applicable to insulating materials due to interference from charging effects. To avoid surface charging, researchers typically restrict valence band structure investigations to ultra-thin films. However, the UPS spectral performance of ultra-thin films tends to correlate with the substrate characteristics and film thickness, while charging effects, which still unavoidably occur, can also affect the realism of the results. This study systematically investigates the charging effects and valence band structural evolution in SiO₂ insulating films with controlled thickness variations through XPS and UPS depth profiling. By analyzing spectral shifts, surface potential dynamics, and work function variations, three continuous regimes are identified. The results demonstrate that the surface potential undergoes abrupt intensification when exceeding critical thickness thresholds (about 8 nm), a phenomenon governed by substrate resistivity and charge compensation pathways. Conventional work function determination methods remain valid only when the actual effect of the applied bias exceeds the surface potential values. For thicker films, the limited efficacy of negative bias fails to compensate for the spectral shifts caused by surface charging, consequently rendering work function measurements unreliable. These findings provide critical guidance for optimizing UPS measurements and spectral interpretation in insulating films.

1. Introduction

Ultraviolet photoelectron spectroscopy (UPS) is one of the most commonly used techniques for measuring the valence band electron structure of materials. This is particularly important for understanding the interactions between chemical adsorbents and surfaces, as well as for measuring the band structure of inorganic or organic semiconductors [1]. UPS also aids in determining the work function (Φ) of metal or semiconductor surfaces [2,3]. The work function is a crucial physical parameter of materials, directly related to their photoelectric, tribological, oxidation, corrosion, and catalytic properties. The accurate measurement of the work function is also essential in experiments.

However, insulating materials are often deemed unsuitable for UPS measurements due to the impact of the charging effect [1,4]. Based on the photoelectric effect, electrons (i.e., photoelectrons) are continuously excited from the surface area irradiated by ultraviolet light during the measurement process. To maintain the electrical neutrality of the surface, the rate of electron compensation on the surface needs to be sufficiently rapid. However, for samples with poor electrical conductivity, such as insulating materials, the lost electrons cannot be promptly replenished, leading to the accumulation of positive charges on the surface, a phenomenon commonly referred to as the charging effect [5,6,7].

The impacts of charging effects have been studied deeply with an X-ray photoelectron spectrometer (XPS). The presence of the charging effect reduces the initial kinetic energy of the emitted photoelectrons, causing the peak in the spectrum to shift towards a higher binding energy [8,9]. Moreover, the spatial non-uniformity of charge distribution results in phenomena such as peak shape distortion, broadening, and intensity attenuation, thereby significantly reducing the reliability of the information obtained from the spectrum [10,11,12,13]. In XPS, a suitable charge compensation method in combination with binding energy (BE) referencing is typically used to avoid the misleading of charging effects on spectral analysis [6,8,14,15,16,17]. However, this method is not applicable to UPS measurements due to the limitations of the measurement range of BE [1,18].

Therefore, researchers investigating valence band structures typically circumvent direct measurements on bulk insulating materials, opting instead for ultra-thin films [1,19,20]. However, the film fabrication requires precise thickness control to balance competing requirements: sufficient thickness to suppress substrate photoelectron interference while maintaining thinness to prevent charging effects [1]. However, significant discrepancies persist in valence band configurations across different film thicknesses, with the critical thickness selection criteria remaining inadequately characterized [18]. Furthermore, while reduced resistivity in thin films mitigates charging artifacts [21], the conductivity of the films actually depends on the film composition and on the low thickness limit on the composition of the substrate. Structural/energetic defects inevitably introduced during deposition (e.g., point defects, dislocations, and interfacial strain) further distort electronic structures [22,23]. These combined factors raise fundamental questions regarding the representativeness of thin-film electronic structures compared to bulk insulators. Consequently, establishing reliable UPS measurement protocols for insulators necessitates comprehensive thickness-dependent studies correlating charge dynamics with spectral evolution.

This study investigates the evolution of valence band spectra and charging-induced surface potential in SiO₂ thin films with varying thicknesses obtained by the sputtering method. The observed UPS spectra exhibit a stage-dependent evolution with increasing film thickness. Furthermore, the research systematically explores the combined influence of applied bias and surface potential on the validity of work function determination through UPS measurements. The findings demonstrate that understanding surface charging effects and selecting appropriate film thicknesses are critical for accurately interpretating UPS spectra in insulating materials. This work establishes a thickness-dependent framework for surface potential calibration and electronic structure analysis in dielectric thin-film systems.

2. Materials and Methods

The experiment employed two thermally grown SiO₂ thin-film samples (designated as Sample 1 and Sample 2 in this study), commercially acquired from Zhejiang Lijing Technology Co., Ltd., China. Both samples utilized p-type single-crystal silicon (100) substrates with boron doping. The substrate resistivity was measured as 0.001~0.005 Ω·cm for Sample 1 and 1~10 Ω·cm for Sample 2, ensuring distinct electrical properties for comparative analysis. The actual thicknesses of the SiO₂ layers were determined through spectroscopic ellipsometry with nine random measurements across each sample surface. The measured thicknesses with standard deviations were 93.0 ± 0.2 nm for Sample 1 and 58.3 ± 0.2 nm for Sample 2, respectively. All subsequent experimental procedures and data analyses were conducted using these precisely measured values. For practical handling during measurements, the original wafers were precisely diced into 10 × 10 mm square specimens using a laser ablation cutting technique.

The photoelectron spectrometer used in the experiment was the 5000 VersaProbe Ⅲ, which was from ULVAC-PHI, Japan. This spectrometer included a focused monochromatic X-ray beam and an ultraviolet light source, a spherical capacitor energy analyzer for multi-channel detection, a high-performance argon ion sputtering gun, and a dual-beam charge compensation system. For UPS and XPS measurements, a He I ultraviolet light source (hν = 21.2 eV) and a monochromatic Al Kα X-ray source (hν = 1486.6 eV) were used, respectively. The energy of the photoelectrons was analyzed by a spherical capacitor analyzer. All the binding energies of the obtained spectra were referenced to the Fermi level (FL) of the spectrometer, which was determined by the Fermi edge of sputter-cleaned silver particles. The work function of the spectrometer (Φ_SP) was 4.453 eV. The dual-beam neutralization system used in the instrument consisted of a low-energy electron source and a low-energy ion beam. The low-energy electron source was low-energy thermally emitted electrons (about 1 eV) from a hot filament, and the emission current was set to 20 μA by default. The high-energy argon ion (2 kV) sputtering method was used for the depth analysis of the material. The dual-beam neutralization system was used during sputtering to improve the sputtering removal efficiency. After each period of sputtering, XPS and UPS measurements were carried out on the surface, respectively.

3. Results and Discussion

3.1. Time-Dependent Residual Thickness

Stratified material information can be obtained by the analytical capability of depth profiling. This approach enables the sequential spectral characterization of varied film thicknesses on a single specimen through controlled material removal. For a thin film with a known thickness, a prolonged sputtering duration until the detection of the substrate’s characteristic spectral signatures allows for the retrospective determination of the residual film thickness at intermediate sputtering stages.

Given the identical XPS spectral evolution during sputtering removal observed in both samples, the 93 nm thick SiO₂ film (Sample 1) was selected as a representative case to illustrate the time-dependent residual thickness of the SiO₂ film. As shown in Figure 1a,b, in the early stage of sputtering (0–120 min), both the O 1s (531.2 ± 0.05 eV) and Si 2p (102.1 ± 0.06 eV) spectra maintained stable intensity profiles and binding energy positions, confirming unaltered SiO₂ stoichiometry. Subsequently (120–145 min), a two-stage interfacial transition occurred: (1) 120–130 min marked a +1.5 eV shift for O 1s and a 1.1 eV shift for Si 2p, respectively, indicative of SiOx (0 < x < 2) interfacial species formation [24]; (2) 130–145 min saw the progressive attenuation of Si-O bonding signals alongside the emergence of Si-Si bonding signals. Finally (≥145 min), complete SiO₂ removal was evidenced by oxide signal extinction. Combined with the evolution of atomic concentration (Figure 1c), the sputter rate was calculated as 0.64 nm/min (93 nm/145 min). Therefore, the remaining thickness of the film after different sputtering times could be obtained (as shown in Figure 1d). Consistent application of the sputter rate calculation method yielded a sputter rate of 0.56 nm/min (58.3 nm/105 min) for the 58.3 nm thickness SiO₂ film (Sample 2), enabling the determination of time-dependent residual thickness throughout the measurement process.

Furthermore, despite the different sputtering rates of O and Si atoms, the system maintains stable stoichiometric ratios during the first 120 min of sputtering (as seen in Figure 1c), as evidenced by the correspondingly stable O 1s and Si 2p spectral features (Figure 1a,b). This temporal stability confirms that changes in chemical state do not dominate the structural tuning of the valence band spectra in this thickness range. However, during the sputtering period of 120–145 min, the observed stoichiometric deviation transforms into a key factor influencing the evolution of the XPS/UPS valence band spectra.

3.2. Observations of Photoelectron Spectra

Through the progressive sputtering removal of SiO₂ films with known thicknesses, samples 1 and 2 underwent valence band characterization using XPS under dual-beam charge neutralization conditions (Figure 2a,c) and UPS without charge neutralization (Figure 2b,d). For XPS analysis, the C 1s peak was deemed unsuitable for charge correction due to the sputter removal of adventitious carbon layers. The O 2s peak was selected for charge correction until significant oxygen depletion occurred [25], after which the literature-reported reference peak for Si-Si bonding was employed during the final stages of sputtering when the Si substrate became exposed. UPS spectra were acquired without charge neutralization and remained uncorrected.

The emergence of the O 2s peak in XPS valence band spectra with an increasing film thickness corroborates our thickness determination. XPS results demonstrate stabilized valence band structures beyond critical thicknesses of 16 nm and 8.3 nm for respective samples, indicating uniform SiO₂ film distribution consistent with Si 2p and O 1s narrow spectra analyses. The difference in detection depth between XPS (1–5 nm) and UPS (0.5–2 nm) [1,9,26] leads to distinct manifestations of band structure features in their respective spectra, particularly pronounced in ultra-thin films. As shown in Figure 2c,d, when analyzing progressively thicker films, the characteristic band structure features (0–5 eV range) originating from the substrate persist longer in XPS valence band spectra compared to UPS measurements. At a critical thickness of 2.8 nm, these substrate-induced features remain clearly observable in XPS spectra while having completely vanished in UPS spectra. This phenomenon arises from the deeper information depth of XPS, which enables the simultaneous detection of signals from both the film and substrate, whereas the shallower probing depth of UPS restricts its detection to the film layer exclusively.

Notably, as film thickness continues to increase beyond this critical value, both techniques eventually detect signals solely from the film, resulting in diminished spectral discrepancies. Furthermore, the implementation of effective charge neutralization in XPS significantly minimizes spectral distortions during valence band measurements. In contrast, UPS spectra exhibit more pronounced spectral changes due to charging effects arising from the absence of effective charge compensation (Figure 2b,d).

The UPS spectral measurements across varying film thicknesses provide detailed insights into the evolution of electronic band structures and charge accumulation effects. Taking Sample 1 as a representative case (Figure 2b), the UPS spectra evolution can be systematically categorized into three continuous regimes:

Regime I (0–16 nm, Figure 3a): Initial spectral intensity enhancement with thickness increment contrasts and with the absence of a discernible Fermi edge and the progressive attenuation of band features within 0–5 eV binding energy. This transitional behavior stems from two concurrent mechanisms: (1) electronic structure modification during the substrate transition from p-type Si to a SiO₂-dominated surface, and (2) deteriorating electrical conductivity exacerbating charge accumulation effects. The characteristic intensity suppression in the 0–5 eV range becomes progressively pronounced with increasing thickness. Regime II (16–32.1 nm, Figure 3b): Spectral profiles exhibit remarkable consistency across this thickness range, indicating the stabilization of both interfacial electronic states and charge compensation dynamics. Regime III (>32.1 nm, Figure 3c): Though a further increase in thickness induces measurable spectral intensity reduction, normalized comparative analysis reveals an only 0.64 eV valence band maximum (VBM) shift across the 16–77 nm range (Figure 3d), indicating that the enhancement of surface charging is not significant.

A comparable evolution pattern emerges in Sample 2, as constrained by its reduced dimension. UPS spectra demonstrate the truncated progression through two characteristic regimes (Figure 3e,f). In the first regime (0–16.7 nm), the valence band maximum (VBM) undergoes an apparent 6.63 eV displacement. This pronounced displacement correlates with the process of establishing continuous charge dissipation pathways through the SiO₂ layer. In the second regime (16.7–50 nm), spectral stabilization occurs with a merely 0.72 eV VBM shift. The spectral shape preservation in this regime indicates the establishment of thermodynamic equilibrium between surface charging and bulk-mediated charge neutralization regimes.

Cross-validation with XPS measurements (Figure 2a,c) conclusively demonstrates that this substantial VBM shift primarily originates from charge-induced artifacts rather than genuine band structure modifications. The differential charge accumulation behavior between samples correlates with their distinct substrate resistivities (0.001–0.005 vs. 1–10 Ω·cm), where the higher substrate conductivity in Sample 1 enables more effective charge neutralization through the Si/SiO₂ interface.

3.3. Charging-Induced Surface Potential

The investigation reveals that charging effects persistently manifest and interfere with spectral measurements. The direct measurement of surface potential induced by charging effects during photoelectron spectroscopy remains technically challenging. However, a practical methodology has been implemented by analyzing the discrepancy between VBM positions derived from XPS and UPS spectra. This approach, though constrained by the inherent energy resolution limit of XPS, provides a viable solution for quantifying charging-induced surface potential on insulating surfaces. As demonstrated in Figure 4 for Sample 2, the surface potential values corresponding to different film thicknesses are labeled separately. Parallel analysis is also performed on Sample 1, and both results are shown in Figure 5.

Both samples exhibit analogous trends in surface potential evolution, with direct correlations observed between surface potential variations and UPS spectral shifts. For Sample 2, the surface potential shows gradual growth (0–1 V) within the 0~8 nm thickness range, where sufficient tunneling-assisted conduction maintains effective charge compensation [27], resulting in minimal spectral distortion. Beyond the critical thickness threshold (about 8 nm), the value of surface potential undergoes abrupt intensification (3–5 V) due to the weakening of the tunneling of electrons with the growth of the film resistivity. Meanwhile, internal electric fields between the charged surface and the conductive substrate are formed, and therefore the competing migration of UV-generated holes and compensation electrons under the field leads to the fact that the surface potential does not increase further.

This stabilization mechanism persists through subsequent thickness increments. Similar behavior is also observed in Sample 1, though with modified dynamics: Enhanced substrate conductivity (0.001~0.005 Ω·cm vs. 1~10 Ω·cm for Sample 2) facilitates more efficient charge dissipation through the Si/SiO₂ interface. However, at thicknesses (>50 nm), weakened electron compensation via the substrate–film–surface pathway leads to progressive charge accumulation, and further results in the emergence of lateral potential inhomogeneity and consequent spectral attenuation and broadening.

3.4. Work Function Analysis

The determination of the material work function (WF) through valence band analysis constitutes a principal application of UPS. Current methodologies for thin-film characterization predominantly adopt the conductor-based WF calculation framework: WF = hν − (E_Cutoff − E_Fermi), where E_Cutoff and E_Fermi correspond to the secondary electron cutoff edge and Fermi edge in the spectra, respectively. However, the validity of this conventional approach for progressively thickened films requires systematic verification due to charging effects.

To unambiguously distinguish the spectrometer cutoff from the sample cutoff, a negative bias voltage (−5 V) was applied during subsequent measurements. While this methodology demonstrates proven efficacy for conductive specimens, its applicability to thickness-dependent insulating films remains insufficiently explored. UPS analyses with bias voltage were performed on both samples (Figure 6a,b) throughout the sputtering process. Cross-referencing with unbiased measurements (Figure 2b,d) revealed comparable spectral evolution trends: the three-regime behavior of Sample 1 and the two-regime behavior of Sample 2 remained consistent under biased conditions.

The spectral shifts induced by the negative bias were quantitatively analyzed through Fermi edge displacement measurements (Figure 6c). Notably, the biasing efficiency exhibited pronounced substrate resistivity dependence: Sample 1 maintained an approximate −5 V shift across the film thickness range (<50 nm), whereas Sample 2 preserved this magnitude of shift only within a reduced thickness threshold (<16 nm). This thickness-dependent divergence highlights the influence of substrate resistivity in the biasing effectiveness for insulating films.

The derived WF values calculated using the standard formula are presented in Figure 6d. For Sample 1 (low-resistivity substrate), the bias voltage effectively counteracts charging across the 10–50 nm thickness range, yielding reliable WF measurements. Beyond 50 nm, however, the weakening of the bias efficacy makes the cutoff edges of the spectrometer and sample overlap, artificially inflating WF values. In contrast, Sample 2 (high-resistivity substrate) demonstrates limited biasing effectiveness, with WF values exhibiting an upward deviation beyond 20 nm thickness. The competition between biasing effectiveness (Figure 6c) and charging-induced surface potentials (Figure 5) directly affects the credibility of the WF measurement results. While increasing bias voltage may temporarily alleviate this issue, its utility diminishes as film thickness grows and insulating properties dominate. Beyond a critical thickness threshold, where bias influence becomes negligible, conventional WF determination methods based on the framework lose validity.

4. Conclusions

This study demonstrates that the charging dynamics in SiO₂ insulating films during UPS measurements exhibit a three-stage evolution strongly correlated with film thickness. In the transitional phase, the electronic state recombination dominated by interfacial remodeling together with conductivity degradation triggers the characteristic spectral attenuation. As film thickness increases, stabilization occurs through established charge transport pathways, followed by absolute intensity reduction due to the decrease in the quantum yield, while the valence band top (VBM) undergoes a weak shift.

A direct correlation exists between surface potential evolution and UPS spectral shifts. Benefiting from tunneling-assisted charge compensation, minimal surface potential forms on the film with the thickness range of 0–8 nm. Beyond the critical thickness threshold, there is a sudden enhancement of the surface potential value (3–5 V) due to the growth of film resistivity. Concurrently, the formation of an internal electric field between the charged surface and conductive substrate establishes competing migration pathways for UV-generated holes and compensating electrons under field-driven conditions, thereby suppressing further surface potential elevation through a dynamic charge compensation equilibrium.

The validity of conventional work function determination methods is constrained by film thickness and substrate properties. While bias voltage application partially mitigates charging effects within a certain thickness, its efficacy diminishes progressively in thicker films and the reliability of the results of the analysis of work function is lost due to the overlap of the spectrometer with the sample cutoff edges. This work further highlights the necessity of substrate resistivity and strategies in UPS measurements of insulating films to bridge the gap between ultra-thin-film measurements and bulk insulator characterization in valence band spectroscopy.