Synchrotron-based ambient pressure X-ray photoelectron spectroscopy of hydrogen and helium

From all the elements in the periodic table, hydrogen is arguably among the most important one. It is a prime candidate to fuel a sustainable society; it is key in organic chemistry and essential in biology.^1,2 The importance of hydrogen makes it imperative to have analytical tools to detect and quantify it. Due to its single electron, hydrogen diffraction intensities for X-rays and electrons are weak. Furthermore, hydrogen does not have an Auger transition and, hence, is invisible in Auger electron spectroscopy. Finally, it is widely claimed to be undetectable by X-ray photoelectron spectroscopy (XPS), one of the most frequently used techniques in surface science, materials science, and catalytic research. For example, an excellent systematic review on the effect of the presence of hydrogen in the XPS shifts of a variety of compounds states that “While XPS is recognized as a preeminent tool for surface chemical analysis, a major shortcoming is that it cannot see hydrogen directly.”³ Another excellent article by Stojilovic entitled “Why Can't We See Hydrogen in X-ray Photoelectron Spectroscopy?” nicely explains why most students misunderstand the reason for the apparent impossibility of doing XPS on elements lighter than Li in relation to the confusing information in the literature.⁴ Even textbooks explaining the principles of the technique state that neither hydrogen nor helium are detectable by XPS.⁵ 

In XPS, photons with energies higher than 100 eV excite electrons from orbitals with a very specific binding energy, and the kinetic energy of the ejected electrons is detected using an electron energy analyzer. The binding energy is element specific, and it often contains valuable information regarding the chemical environment of the atom. In fact, the technique was originally named electron spectroscopy for chemical analysis (ESCA).⁶ The highest photoionization cross-section is obtained for photon energies slightly above the required energy for a particular transition.⁷ With increasing energy, the cross-section diminishes greatly. To measure hydrogen, the electrons in the H 1s orbitals must be excited. In atomic hydrogen, these electrons have a binding energy of 13.6 eV,⁸ and to maintain a reasonable cross-section, photon sources in the ultraviolet (UV) range are employed. For example, the cross-section has a value of 1.9 Mbarn for a He1α source,⁷ which emits photons at 21.2 eV. With UV sources, the spectra of H₂, D₂, and He have been recorded.^9,10

In XPS, the higher photon energy is disastrous for the H 1s photoionization cross-section. To illustrate, the cross-sections are 5.5 and 2.8 barn for Mg Kα and Al Kα radiation sources,^7,11 respectively. This is almost 6 orders of magnitude lower than for a He 1α source. Additionally, the photon flux of typical X-ray sources is 4–5 orders of magnitude lower than that originating from a UV source, with the latter being typically 10^12–14 photons/s.^12,13 Both these facts result in an exceedingly small number of H 1s or He 1s photoelectrons, making it understandable that it was claimed, in both textbooks⁵ and research papers,³ that it is impossible to detect hydrogen and helium using XPS. However, this is merely a technical limitation and not a fundamental one as sometimes portrayed, which often generates confusion when students try to sort through the literature on the reason behind the limitation.⁴ 

This letter presents experimental XPS of H 1s as well as He 1s, definitely proving that both elements can be detected in XPS. This is of tremendous academic value, as it is at odds with previous misconception.⁴ The use of synchrotron radiation is crucial to observing H and He in XPS.¹⁴ These sources commonly have photon fluxes in the order of ∼10¹³ photon/s, about 3 orders of magnitude more intense compared to lab sources. It is equally important that synchrotrons provide tunable photon energies, e.g., the Coherent Soft X-ray Scattering and Spectroscopy beamline (CSX-2) at the National Synchrotron Light Source II (NSLS-2) has an energy range of 250–2000 eV,^15,16 putting it within the so-called “soft X-ray” energy range. At the lowest obtainable energy, the H 1s cross-section is ∼1 kBarn.⁷ Together, the H 1s photoelectron intensity will be approximately 6 orders of magnitude higher with synchrotron-based XPS than with lab-based XPS. The ability to detect hydrogen is particularly relevant for ambient pressure (AP)-XPS. Using this approach, which was pioneered by Siegbahn¹⁷ and significantly improved in the last two decades,^18–20 solid or liquid samples are being exposed to gas environments with pressures up to a few Torr.^17,21 In this work, we took advantage of the use of both synchrotron light and AP-XPS to obtain photoelectron spectra of H₂ and He in the gas phase and H on a Pt(111) surface.

XPS spectra of H 1s and He 1s were taken at a pressure of 0.5 Torr, and they are shown in Fig. 1. These gas-phase spectra were taken at the AP-XPS end station at the CSX-2 beamline of the NSLS-II.^15,16 For diatomic molecules (e.g., H₂), photoionization normally consists of non-dissociative and dissociative ionization processes²² 

However, the H⁺/H₂⁺ ratio is relatively small, especially in the high photon energy regions.²² The neutral hydrogen molecule has filled bonding 1σ_g and unoccupied antibonding 1σ_u^* molecular orbitals. In the case of non-dissociative photoionization, the XPS peak corresponds to the emission of a 1σ_g electron. Several different photoionization energies can be defined related to the various possible vibrational states of the H₂⁺ cation

As shown in Fig. 1(a), the broad H 1s peak is asymmetric with fine structures assigned to different vibrational transitions (see the inset).¹⁰ Note that while at first sight, the fine structure appears relatively small and close to the order of magnitude of the background noise, this measurement was repeated several times and the location and separation between the features in the fine structure are reproducible. According to theoretical calculations,²³ there are in total 19 different vibrational states of the hydrogen molecular ion, and these are fitted in the inset in Fig. 1(a). The full width at half maximum (FWHM) of each of these components is 0.33 eV. The first ionization energy (or the adiabatic ionization energy, ν' = 0) is located at 15.41 eV, which is defined as the negative of the orbital energy of the highest occupied molecular orbital (HOMO), i.e., the minimum energy required to remove an electron from the molecule in its ground state, while the vertical ionization energy corresponds to the ionization energy associated with the transitions from the neutral ground state to different vibrational levels (ν' = 1, 2, 3…). It should be noted that the energy position in this AP-XPS spectra was calibrated according to the literature.²³ The actual measured energy is artificially higher (by 0.71–0.73 eV) than previous experimental and theoretical results,^9,10,23 and this is due to the fact that the gas phase molecules are subjected to an electrostatic potential at the location where the ionization takes place, which is very difficult to determine and depends on various factors including the pressure of the gas and the population of ionized molecules. This has been discussed before in the AP-XPS literature.²⁴ In contrast, the XPS spectra of He (i.e., a monatomic species) in Fig. 1(b) shows a sharp and symmetric peak, which has been referenced to the literature ionization energy of 24.59 eV. Note that the actual measured binding energy is 0.91 eV higher, for the same reason described above for hydrogen. The FWHM of the He 1s peak is 0.21 eV, as shown in the inset in Fig. 1(b). We want to emphasize that a lab-based AP-XPS system would not allow these measurements given its much lower flux and lack of tunability of the photon energy, as described in more detail in the first three paragraphs. Mass spectra were taken in order to verify the purity of the gases and to be absolutely certain that the spectra we observed corresponded indeed to hydrogen and helium.

To further examine the potential of the study of hydrogen, AP-XPS measurements were performed on a Pt(111) surface. Pt(111) is an important model system for hydrogen dissociative adsorption studies.^25,26 It is widely accepted that the dissociative adsorption of hydrogen on the Pt surface is a structure sensitive process and the binding energy strongly decreases with increasing hydrogen coverage.²⁷ There are only a few experimental techniques which are able to directly detect hydrogen on Pt(111), such as low-energy electron diffraction (LEED) and electron energy loss spectroscopy (EELS).²⁸ Here, we examine the detailed XPS of hydrogen adsorption on Pt(111) as shown in Fig. 2. Figure 2(a) shows in black the XPS spectrum in the valence band region obtained in ultra-high vacuum (UHV) conditions on a clean Pt(111) surface, featuring a rich d-band electronic structure near the Fermi level.²⁹ The Pt(111) surface was subsequently exposed to 0.5 Torr of H₂. A new feature located at a binding energy of 11.90 eV can be distinguished as shown in the red spectrum, which is assigned tentatively to hydrogen from the H-Pt bonds. This peak remains upon evacuating the hydrogen gas (green spectrum). It should be noted that no gas phase hydrogen signal can be clearly distinguished in this AP-XPS measurement, which is due to the low counts from the gas phase hydrogen when compared to the background signal. For comparison, the H 1s peak in Fig. 1(a) is 4 × 10² cps, while the background intensity around this energy region in Fig. 2(a) is 5 × 10⁴ cps (both cases were obtained with the exact same analyzer settings).

Figure 2(b) shows the evolution of the Pt 4f spectra. Interestingly, there is a significant satellite peak located at a binding energy of 83.67 eV under 0.5 Torr H₂, which is 12.70 eV higher in the binding energy scale than the Pt 4f peak. These peaks are probably caused by the inelastic collisions between the Pt 4f photoelectrons and the gas phase H₂ molecules (i.e., the kinetic energy of the emitted Pt 4f photoelectron from Pt(111) is reduced by the excitation of a 1σ_g electron from H₂, which has a binding energy of ∼12.70 eV when referenced to the Fermi level of Pt).²⁴ A careful inspection of Pt 4f_7/2 further reveals indications of the adsorption of hydrogen on the Pt(111) surface as shown in the inset of Fig. 2(b). For a clean Pt(111) surface, the lower binding energy component at 70.57 eV originates from the surface atoms, while the bulk component is located 0.4 eV above.³⁰ Under 0.5 Torr H₂, the surface component is significantly reduced and a new small shoulder appears at higher binding energies (71.29 eV, as shown in the deconvoluted Pt 4f_7/2 region in the right panel of Fig. 2(b), which we tentatively assign to Pt bound to H. It should be noted that the intensity of this shoulder peak increases slightly after evacuating the H₂ gas, mostly attributed to the lower screening by H₂ gas molecules on the H-Pt surface dipoles.³¹ 

The peak assignments were further supported by density functional theory (DFT) calculations. DFT calculations were performed using the Vienna Ab initio simulation package (VASP).^32,33 The exchange and correlation energies were described by the PBE functional.³⁴ A kinetic energy cutoff of 520 eV was used, together with a 5 × 5 × 1 k-point grid for the Brillouin zone sampling. The Pt(111) substrate was modeled by five layers of Pt atoms. The top two layers of Pt atoms and H atoms were allowed to relax until forces were smaller than 0.01 eV/Å, while the bottom three layers were kept fixed. The core-level binding energies (E_BE) were calculated using the transition state model.³⁵ E_BE values of bulk Pt atoms were averaged over all Pt atoms in the middle three layers. As shown in Fig. 3, three systems with hydrogen coverages (Θ) of 0.25 ML, 0.50 ML, and 0.75 ML were studied, corresponding to 1, 2, and 3 H atoms per unit cell (a 2 × 2 unit cell was defined). H atoms are adsorbed at the fcc hollow sites with the most negative adsorption energy (⁠$Eads$⁠)³⁶ defined as $Eads=EPt−H−EPt−12EH2$⁠, with $EPt−H$ being the energy of the Pt (111) surface with a H atom adsorbed per unit cell, $EPt$ the energy of the clean Pt (111) surface, and $EH2$ the energy of a free H gas molecule. At Θ = 0.25, $Eads$ is −0.48 eV, which is consistent with the published results.³⁶ $Eads$ of the second H atom is −0.41 eV, which is calculated by $EPt−2H−EPt−H−12EH2$⁠, where $EPt−2H$ is the energy of the system with a hydrogen coverage Θ = 0.50. $Eads$ of the third H atom is −0.36 eV for Θ = 0.75. The adsorption energy remains negative as the H coverage increases, indicating that the coverages of 0.25 ML, 0.50 ML, and 0.75 ML give stable structures during H adsorption. Experimentally, hydrogen coverages up to 0.75 ML were achieved by dosing 0.1–500 L of H₂ (1 L = 1.33 × 10⁻⁶ mbar × s) at 85 K.³⁷ 

At Θ = 0.25, E_BE for clean surface Pt atoms (Pt_surf, red dashed circle in Fig. 3(a)) is 0.34 eV lower than that for Pt atoms in the bulk (Pt_bulk), which is consistent with the 0.40 eV red shift from the XPS spectra. At Θ = 0.25, E_BE of Pt atoms connected with H atoms (Pt_1H) is 0.13 eV lower than Pt_bulk. At Θ = 0.50, 50% of the surface Pt atoms are bonded to 2H atoms (Pt_2H). E_BE of Pt_2H is 0.02 eV higher than that of Pt_bulk. As Θ increases to 0.75, 25% of the surface Pt atoms are connected to 3 H atoms (Pt_3H). E_BE of Pt_3H is 0.18 eV higher than that of Pt_bulk. The blue shift of Pt_3H from Pt_1H is similar to that observed in a study of hydrogen adsorption on Rh(111) where E_BE shifted 0.19 eV from Rh_1H to Rh_3H as measured by XPS.³⁸ In the XPS spectra in Fig. 2(b), the Pt_H atoms are 0.32 eV higher in core level binding energies than Pt atoms in the bulk, indicating a high H coverage under the experimental conditions used in this work.

In summary, we have clearly demonstrated, by means of synchrotron-based AP-XPS measurements, that gas phase XPS spectra of hydrogen and helium can be obtained if a bright enough X-ray source is used. The H 1σ_g peak is asymmetric, which is related to the different possible vibrational modes of the final state. The He 1s peak is symmetric, as expected. In addition, we report what appears to be hydrogen adsorbed on (or dissolved in) Pt(111), as suggested by a peak at 11.9 eV during (and more clearly after) exposure to elevated pressure of hydrogen. Two other features related to hydrogen are also evident in the experiment, namely, electron energy loss features in Pt 4f and a shoulder on the higher binding energy side of Pt 4f_7/2. These results put to rest the common misconception that X-ray photoelectron spectra of elements lighter than lithium are impossible to obtain and also help in clarifying matters on the usual misunderstanding of the reason for this apparent impossibility. In addition, this work shows that the most abundant chemical element in the universe, namely, hydrogen, can be studied using one of the most useful analytical techniques in materials science.

The research carried out in part at the Center for Functional Nanomaterials and the CSX-2 beamline of the National Synchrotron Light Source II, Brookhaven National Laboratory, was supported by the U.S. Department of Energy, Office of Basic Energy Sciences, under Contract No. DE-SC0012704. J.Q. Zhong and M. Wang were supported by BNL LDRD Project No. 15-010, and W. H. Hoffmann was supported by DOE SULI program. This work was supported as part of the Integrated Mesoscale Architectures for Sustainable Catalysis (IMASC), an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under award #DE-SC0012573. This research used resources of the National Energy Research Scientific Computing Center, a DOE Office of Science User Facility supported by the Office of Science of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.

The authors declare no competing financial interests.