This perspective explores the complexity of nitrogen chemistry and its interpretation from high-resolution x-ray photoelectron spectra by examining metal-nitrogen-doped carbon systems used in many technological applications. Current understanding of factors contributing to nitrogen 1s photoelectron spectroscopic signature is reviewed, and limitations on assigning precise chemistries to individual spectral ranges are discussed. The importance of proper curve fitting of XPS spectra based on appropriate peak widths and shapes for correct data interpretation is highlighted. Appropriate peak fitting and chemical identification are critical to developing structure-to-property correlations for functional materials in which nitrogen chemistry plays a vital role.

Oversimplifications and erroneous interpretations of nitrogen chemistry are especially prevalent in applied scientific reports where analysts use a large set of multiple analytical techniques in which XPS is one of the analytical methods used. Misinterpretations are caused by common mistakes made during curve fitting of XPS spectra and the fact that multiple nitrogen bonding configurations can have the same binding energy (BE). Propagation of erroneous conclusions on the contributions of different types of N-containing structures into XPS spectra is a significant problem in the literature that needs to be acknowledged and addressed. This opinion paper summarizes the challenges in analyzing XPS nitrogen spectra and discusses uncertainties in peak identification, which should be taken into account when structure-to-property correlations are being identified based on XPS spectra interpretations and emphasizes the importance of integration with other analytical methods.

In the development of platinum group metal (PGM) free electrocatalysts for oxygen reduction reaction (ORR) based on transition metals-nitrogen-carbon (MNC), the role of metal and nitrogen chemistry in catalysts’ selectivity and activity is central. The major focus of analytical characterization used for building structure-to-property relationships for MNC electrocatalysts has been the chemistry of the transition metal. In addition to surface analytical techniques, such as x-ray photoelectron spectroscopy (XPS), x-ray absorption spectroscopy, and Mössbauer spectroscopy have advanced the understanding of the coordination of transition metal in the nitrogen-doped carbon network.1–6 

However, in addition to transition metal coordinated to nitrogen, these materials also contain multiple types of nitrogen embedded in the carbon network in a variety of coordinations. The possible participation of various types of nitrogen in parallel oxygen reduction reactions cannot be overlooked in designing efficient and selective electrocatalysts. XPS has been a workhorse for analyzing bonding states of nitrogen in different fields of science and technology—from fuel cokes to N-doped nanotubes and graphenes.4,7–18 Even though reference compounds with single unique nitrogen chemistry are readily available, the complexity of interpreting high-resolution photoelectron nitrogen spectra comes from the fact that N-containing materials synthesized by chemical, physical, and thermal routes, such as MNC electrocatalysts discussed herein, have a mixture of possible nitrogen bonding configurations that are prone to hydrogenation and protonation.15,19 A recent review discussed the issue of interpreting nitrogen chemistry from high-resolution N 1s spectra for carbon nitride films.13 

This perspective reviews the current understanding of nitrogen chemistry with a focus on MNC electrocatalysts, discusses the uncertainty that exists in the identification of nitrogen species, and indicates the care needed to be taken to avoid incorrectly interpreting XPS data.

The primary contribution to the exact value of BE for a particular type of chemical bonding of that element depends on the total number of bonds between atoms of different electronegativity values. Generally, BE increases if electron-withdrawing groups are attached to the probed atom.

For nitrogen, the coordination number (N bonded to two or three neighbors) has a larger effect than the geometry (five- or six-membered rings). The three-coordinated nitrogen atoms have higher binding energy (∼1.5–1.7eV) than two-coordinated nitrogen atoms.20,21 Within carbon-rich networks that have a structure similar to multilayer graphene with defects, the coordination of the same moiety will be different depending on if it is located within the plane and at the edge of a graphene sheet. As a reference for this, in carbon nitride C3N4, nitrogen in sp2 coordination contributes to larger binding energy than nitrogen in sp3 coordination.22 

The local chemical and physical environments have a significant secondary effect on binding energy. The local environment of nitrogen moiety in an MNC electrocatalyst can vary in several ways. Saturation of the neighboring carbon atoms enhances the degree of localization of the nitrogen lone pair and, therefore, increases the screening effect on the 1s state. Nitrogen moieties with a similar coordination number can be located at the edge of the graphene sheet or within the plane such as shown in Fig. 1 for hydrogenated five-ring nitrogen (aka pyrrolic). The chemical environment in proximity of an in-plane defect or termination around the edge, as shown for hydrogenated pyridinic N in Fig. 1, can also have a significant effect.

FIG. 1.

CLS calculated for pyrrolic N in-plane (a) and on-edge (b); and hydrogenated edge N terminated by C—Ox (c) and by C—CH (d). Colors used for elements: light blue (light grey)–carbon, dark blue (dark grey)–nitrogen, white–hydrogen, and red (black)–oxygen. The same defect located at the edge of the carbon sheet can have up to ∼2 eV higher binding energy than the one in-plane. Adapted with permission from Matanovic et al., Curr. Opin. Electrochem. 9, 137 (2018). Copyright 2018, Elsevier (Ref. 23).

FIG. 1.

CLS calculated for pyrrolic N in-plane (a) and on-edge (b); and hydrogenated edge N terminated by C—Ox (c) and by C—CH (d). Colors used for elements: light blue (light grey)–carbon, dark blue (dark grey)–nitrogen, white–hydrogen, and red (black)–oxygen. The same defect located at the edge of the carbon sheet can have up to ∼2 eV higher binding energy than the one in-plane. Adapted with permission from Matanovic et al., Curr. Opin. Electrochem. 9, 137 (2018). Copyright 2018, Elsevier (Ref. 23).

Close modal

In recent years, density functional theory (DFT) calculations have become crucial in the accurate interpretation of the spectroscopic features obtained by improved spectroscopic techniques. It was recently shown that the DFT calculations of core-level binding energy shifts (CLS) for well-defined defects of the PGM-free catalysts can facilitate the interpretation of XPS measurements.24–26 Based on DFT and experimental studies, it was shown that the difference in N 1s binding energies of the same defect at the edge and in-plane defects is more than 2eV with it being higher for an edge defect.23 As shown in Fig. 1 for a hydrogenated pyridinic edge defect, the binding energy of nitrogen surrounded by aliphatic carbons (—C—CH) is much larger than for the same defect surrounded by carbon oxide groups (C—C—OH/C—C=O).23 

For model reference materials with a single type of chemistry, the primary contribution from the coordination number will dominate the exact value of binding energy. In more complex materials synthesized by chemical, physical, and thermal routes, other types of sites with different chemistry may surround the defect of interest; hence, in addition to immediate coordination, local physical and chemical environments will have a significant contribution onto binding energy of a defect, explaining ambiguity in identifications of nitrogen existing in the literature.

The direct comparison between spectroscopic identifications of nitrogen chemistry existing in the literature is also complicated by different acquisition and data processing parameters used. The most important effect on binding energy determination is the way instrument calibration and charge referencing are performed. The linearity and accuracy of binding energy scale is assumed to be properly calibrated (Ref. 27 and references therein). This means that for conductive materials, such as MNC electrocatalysts, no charge neutralization and, as a consequence, no binding energy referencing are needed. However, for reference and precursor materials, such as polymers or small organic molecules, or for MNC materials with inherent poorer conductivity, the use of charge neutralization during spectral acquisition requires proper referencing of spectra to a specific known binding energy.

The use of the maximum of C 1s to internally reference the BE scale became routine. Although using C 1s signal to reference spectra has important uses, recent work has highlighted multiple challenges with the accuracy of this approach.28,29 The use of the maximum of the C 1s signal for referencing a spectra can be particularly problematic for polymer materials, precursors, and electrocatalysts themselves, where carbon can have primary and secondary effects of various amounts of aliphatic, aromatic, C—N and surface oxide species, the use of carbon for referencing can cause errors in absolute peak positions unless details of the known C components of the material are taken into consideration.

Figures 2(a) and 2(b) show high-resolution C 1s spectra for two different NC materials in which carbon shape reflects different proportions of lower binding energy components—aromatic and aliphatic carbons—and higher binding energy components—CxOy/CxNz species. The first material is the M-N-C precursor mixture prior to pyrolysis in which high amounts of higher binding energy components are observed and the second material is a pyrolyzed M-N-C catalyst in which graphitization of carbon and removal of carbon oxides occurred during pyrolysis causing a dramatically different carbon chemistry.30,31 The spectra used for this demonstration were chosen deliberately to represent one of the extreme situations where carbon has a much higher contribution from a higher binding energy component than from aliphatic carbon.

FIG. 2.

High-resolution C 1s and N 1s for two different materials: an unpyrolyzed mixture of N and C precursor and Fe salt [(a), (c), and (e)]; pyrolyzed MNC catalyst [(b), (d), and (f)]. Material synthesis and data acquisition details including charge neutralization are described in Refs. 30 and 31. (a) and (b) Curve fit of C 1s spectra; (c) and (d) overlay of normalized high-resolution C 1s spectra calibrated by C and by the internal Au reference sample; (c) shows experimental setup in which a small amount of 99.9% pure Au powder is put on a sample and multiple areas are analyzed including one area on the internal Au reference powder; (e) and (f) overlay of normalized high-resolution N 1s spectra calibrated by C and by the internal Au reference sample. The abundance of high binding energy components due to C—O/C—N species in C 1s for unpyrolyzed sample causes incorrect calibration when a maximum of the C 1s spectrum is used for calibration.

FIG. 2.

High-resolution C 1s and N 1s for two different materials: an unpyrolyzed mixture of N and C precursor and Fe salt [(a), (c), and (e)]; pyrolyzed MNC catalyst [(b), (d), and (f)]. Material synthesis and data acquisition details including charge neutralization are described in Refs. 30 and 31. (a) and (b) Curve fit of C 1s spectra; (c) and (d) overlay of normalized high-resolution C 1s spectra calibrated by C and by the internal Au reference sample; (c) shows experimental setup in which a small amount of 99.9% pure Au powder is put on a sample and multiple areas are analyzed including one area on the internal Au reference powder; (e) and (f) overlay of normalized high-resolution N 1s spectra calibrated by C and by the internal Au reference sample. The abundance of high binding energy components due to C—O/C—N species in C 1s for unpyrolyzed sample causes incorrect calibration when a maximum of the C 1s spectrum is used for calibration.

Close modal

Calibrating by adjusting the position of C maximum to 284.8 eV will not reflect the chemistry distribution of carbon correctly. One of the solutions to this problem is curve fitting the C 1s spectra first and then adjusting the position of the lower binding energy peak to 284.8 eV. An alternative solution is using a pure gold powder deposited on the sample and calibrating all the spectra acquired from the sample itself by adjusting the maximum of Au peak to a reference value of 84 eV. A very small amount of 99.9% pure Au powder is deposited in one small area on the sample as shown in the inset in Fig. 2(c).

Spectra in Figs. 2(c) and 2(d) show a comparison between referencing by a maximum of C and by Au. For the pyrolyzed MNC material, the difference in referencing by C and Au is only 0.2 eV, which is within the accuracy of BE referencing and identification, as this sample is mainly rich in aliphatic-graphitic carbon as shown in fitted spectrum in Fig. 2(b), so referencing by adjusting the position of C maximum to 284.8 eV results in an accurate charge correction of BE and accurate identification of N chemistry in Fig. 2(f).

For the unpyrolyzed mixture [Figs. 2(a)2(e)], the abundance of higher binding energy species in C 1s photoelectron peak causes an error when the maximum of this peak is adjusted to the position of C=C/C—C at 284.8 eV. When all spectra are referenced by adjusting by Au position and the C 1s spectrum is curve fitted [Fig. 2(a)], the peak due to C=C at 284.8 eV is smaller than the peak at 286 eV due to an abundance of C—N/C—O species. The difference in referencing by Au and C is as large as 0.9 eV, which would result in the erroneous identification of chemistries in nitrogen as shown in Fig. 2(e).

All the spectra from materials included in this report were referenced by using an internal 99.9% pure gold powder deposited on each sample, when charge neutralization was used.

The electrocatalysts in the MNC family are derived by pyrolytic routes so the knowledge of the evolution of chemistry in fuel coals serves as a reliable basis for identifying chemistries resulted during synthesis. The occurrence of peaks at 398.8 and 400.2–400.4eV was depicted for coals and chars during the past few decades.7–9 The peak at 398.8eV was attributed to pyridinic nitrogen, i.e., N-6, while that at 400.2eV was to N-5 or pyrrolic N. Pyridinic nitrogen invariably refers to N bonded to two sp2-coordinated carbons, either at an edge or next to a carbon vacancy inside the structure while pyrrolic refers to N bonded to two carbons and one hydrogen.

The identification of these species is straightforward as reference materials are readily accessible. Figure 3 shows N 1s high-resolution spectra for reference pyridinic and pyrrolic compounds.

FIG. 3.

High-resolution N 1s spectra for three reference samples: polypyridine, polypyrrole, and pyridine HCl. Structures contributing to each peak are shown. Both pyrrolic N and hydrogenated pyridine contribute to the peak at 400.2eV.

FIG. 3.

High-resolution N 1s spectra for three reference samples: polypyridine, polypyrrole, and pyridine HCl. Structures contributing to each peak are shown. Both pyrrolic N and hydrogenated pyridine contribute to the peak at 400.2eV.

Close modal

Coordination to a transition metal. There is good evidence in the literature based on theoretical calculations, reference materials, and correlation of XPS with other techniques sensitive to the chemistry of transition metal that the nitrogen coordinated to transition metal contributes to the range between 399.5 and 400.5eV, depending on coordination number x in MNx and on the type of transition metal M.23 In addition, based on analysis of materials that do not contain any transition metal, a small peak at the position expected for amino nitrogen at 399.4eV is sometimes present.9 Hence, the binding energy region between 399.4 and 400eV is where M-Nx species plus amines are contributing.

Hydrogenation. Hydrogenation changes the coordination of nitrogen from two to three, therefore, causing the expected shift to the higher binding energy. In a previous paper in reference to carbon nitride, it was suggested that N bonded to two sp2-coordinated C and one hydrogen independent of the number of members in the ring, such as pyrrolic or hydrogenated pyridinic, should be referred to as pyrrolic. This suggestion was not successful as all scientific dictionaries refer to pyrrolic as a five-member ring N.

Figure 3 includes the high-resolution N 1s spectrum from the hydrogenated pyridine reference sample overlayed with reference from pyrrolic N. The position of this peak is in perfect alignment with the position of pyrrolic nitrogen. In previous reports combining experimental and theoretical work, it was demonstrated that the peak at 400.2–400.5 eV, which is usually identified as pyrrolic also has contribution from hydrogenated pyridine, hence the naming of this peak has been suggested to be hydrogenated N.23,25 The shift to higher binding energy by 1.3 eV upon hydrogenation of not only pyridinic but also of graphitic nitrogen was reported.17,32

Protonation. In coals and tars, the third peak at ∼402eV was introduced to the fit of the N 1s spectra and has been usually assigned to quaternary nitrogen. Quaternary nitrogen, usually known as an R4N+ group, is not the only type of cationic nitrogen species that can be present at the surface. Figure 3 shows that some of the nitrogen in pyridine HCl are protonated and contribute to 1.8eV higher than hydrogenated nitrogen. Quaternary nitrogen can also be described as “graphitic nitrogen,” in which nitrogen is within a graphite plane and bonded to three carbon atoms causing a positive charge on the N atom. Different cationic species, such as protonated pyridine, quaternary nitrogen side groups, and any other protonated nitrogen within the plane can contribute to the range of binding energies between 401.5 and 403.5eV.

Oxidation. The presence of N-Ox species in nitrogen-carbon materials has been debated as well. At binding energies higher than 403eV, some of the N—OH and N=O species can be contributing with the relative contribution being small in comparison to other chemistries present. It is important to note that the comparison between in situ and air-exposed samples suggests that the peak components at ∼402–403eV may be due only to protonated nitrogen and not oxidized nitrogen as discussed above.13 

Graphitic nitrogen. In terms of its contribution to the binding energy of XPS N 1s photoelectron graphitic nitrogen has been the most debated type of nitrogen in the literature. The range of binding energies where graphitic nitrogen has been reported span a wide range from 400.2 to 401.8eV.12,33 The main factor in such a large range is the effect of immediate coordination of the nitrogen atom and long-range neighboring atoms.21  Figure 4 shows the binding energy scale with contributing graphitic carbon structures discussed herein.

FIG. 4.

Binding energy scale and different types of graphitic N substitution defects in carbon. 1—isolated Ngr, 2—triazene, 3—melamine, 4—two Ngr within a single aromatic ring, 5—hydrogenated isolated Ngr, and 6—protonated isolated Ngr. Colors separate the types of substitution defects (from right to left): blue (2,3)—multiple, green (1,4)—isolated, and yellow (5,6)—protonated/hydrogenated.

FIG. 4.

Binding energy scale and different types of graphitic N substitution defects in carbon. 1—isolated Ngr, 2—triazene, 3—melamine, 4—two Ngr within a single aromatic ring, 5—hydrogenated isolated Ngr, and 6—protonated isolated Ngr. Colors separate the types of substitution defects (from right to left): blue (2,3)—multiple, green (1,4)—isolated, and yellow (5,6)—protonated/hydrogenated.

Close modal

An important starting point for interpreting the N 1s core level of graphitic types of defects are azafullerenes—fullerenes containing a single nitrogen substitution. These are one of the only systems where there is no ambiguity about the underlying atomic structure. The N 1s core level of C59N measured at 400.7 eV provides the first estimate for the energy of the isolated graphitic N type 1, as shown in Fig. 4.12 

At the same time, highly nitrided carbon structures represented by the triazene fragment (type 2 in Fig. 4), which is the basic tectonic units for g-C3N4 such as seen in S-Triazine and S-Heprazine, have been reported to have a binding energy of 398.0 eV.14 XPS reference for identification of the C3N4 phase using melamine, which has a triazene fragment, was reported to a have binding energy of C=N—C bond (type 3 in Fig. 4) of 399.1 eV.34 Amorphous carbon nitride (a-C:N) has reported having a binding energy of 398.2 eV (Ref. 13) and 398.8 eV.10 To complicate things even more, in C3N4, nitrogen substitution in sp2 coordination contributes to the binding energy of 400 eV while that in sp3 coordination to the smaller binding energy of 398.9 eV.22 

In pyrolyzed MNC electrocatalysts, the nitrogen content is of the order of 3–5 at. %. For an isolated N defect substituting carbon in a triazene segment, there is 8 at. % of N present. With smaller weight loadings of nitrogen in MNC materials and the fact that only a fraction of all nitrogen can be present as substituted carbon defects, one should not expect to see fully nitride structures such as melamine and triazene (2 and 3). However, neighboring nitrogen atoms in proximity to the single N defect, as shown in structure 4, can be present and these can cause a change in the binding energy of graphitic nitrogen.

Moreover, as discussed above, graphitic nitrogen can also undergo hydrogenation (structure 5) and protonation (structure 6) shifting the binding energy of the pristine Ngr atom by 1.8 and 3 eV to higher binding energy, respectively. In experimental and theoretical studies of graphitic N-rich structures, it was shown that depending on the density of nitrogen per graphene sheet, surrounding chemical and physical environments, and possible protonation, the binding energy of localized graphitic N can contribute to as high binding energy as 402.1 eV.26 

Challenges in the interpretation of nitrogen chemistry affect multiple fields of material chemistry and engineering where nitrogen-carbon-rich compounds are utilized. To demonstrate these challenges, MNC electrocatalysts and their rich nitrogen chemistry will be used as an example.

Based on individual nitrogen chemistries discussed above, Fig. 5 shows the typical N 1s spectrum from Fe-NC electrocatalysts, reported elsewhere.19,31,35 The likely contributions of various nitrogen species into different regions of the binding energy spectrum are labeled.

FIG. 5.

Typical high-resolution N 1s spectrum from Fe-NC catalyst and contribution from nitrogen structures. The range of binding values is reflected by the width of the blocks. Moieties in gray boxes have a smaller surface concentration [as evaluated by previous multiple studies (Refs. 23, 25, 26, 30, 31, 35–38, and 45)] and hence smaller contribution to the total nitrogen than other moieties.

FIG. 5.

Typical high-resolution N 1s spectrum from Fe-NC catalyst and contribution from nitrogen structures. The range of binding values is reflected by the width of the blocks. Moieties in gray boxes have a smaller surface concentration [as evaluated by previous multiple studies (Refs. 23, 25, 26, 30, 31, 35–38, and 45)] and hence smaller contribution to the total nitrogen than other moieties.

Close modal

sp2 nitrogen moieties, such as pyridinic and transition metal coordinated to metal, contribute to the lower binding energy range (398–400eV), which is free from overlap with other nitrogen chemistries discussed. Numerous previous structure-to-property correlations for Fe-NC materials show the positive contribution of these nitrogen species to the selective and efficient ORR.1–3,6,19,23,31,35–44

A higher binding energy range (400–403eV) has a contribution from multiple types of species, most of which are present in sp3 coordination—such as hydrogenated and protonated nitrogen. Isolated graphitic N defects can also contribute here. Due to the high overlap between contributions from multiple nitrogen species, it is not possible to exclusively assign binding energy in this range to one type of species. The presence of these hydrogenated and protonated species has been shown to inhibit the selectivity of ORR by promoting the first step of reducing oxygen to hydrogen peroxide.19,31,45

Even though there is a large spectral overlap between different nitrogen-containing chemical structures, it does not suggest that a single wider peak should be used to curve fit this higher binding energy region as was reported recently.46 The authors made a suggestion that because of the high overlap of graphitic nitrogen with pyrrolic nitrogen (which is only true for isolated graphitic nitrogen defect), two peaks—one for pyridinic and one for the mixture of graphitic/pyrrolic—should be used in large spectral envelopes encompassing all chemistries of N-doped graphene.46 One should not confuse the physical meaning of curve fitting XPS spectra, which is based on physical processes of photoelectron emission and analyzer settings that contribute to the shape and width of XPS spectra with the interpretation of curve fits.

Curve fitting using an appropriate number of peaks having a proper width and shape guided by XPS physics should still be done, but interpretation should be made with an understanding of the overlap between binding energy ranges where multiple types of nitrogen structures can contribute.

Indeed, the assignment of peaks above 400eV to a specific N type should be done with care and represent the understanding captured by Fig. 5. Spectral separation between hydrogenated, protonated, and graphitic N is not possible for materials such as N-doped carbons or analogous N-C-rich materials discussed here. Such a significant overlap between peaks due to different chemistries in the higher binding energy range of 400–403eV may require the analyst to add the peaks contributing to that region together and interpret them as a mixture of possible chemistries discussed.

In the case of MNC electrocatalysts, the ambiguity of N chemistry identification does not prevent from using XPS for effective structure-to-property relationships. As mentioned above, the nitrogen species contributing below 400eV are positively contributing to the selectivity of oxygen reduction reaction while species contributing above 400eV are diminishing the selectivity of ORR. Hence, very simple metrics beneficial for ORR nitrogen chemistry can be the ratio of area under the peak below 400eV to that above 400eV. In the previous report on more than 50 electrocatalytic materials for ORR, similarly, the ratio of pyridinic to hydrogenated nitrogen as a measure of a structure resulting in a better electrocatalytic activity was used.19 The data have been revisited here and plotted both as original metrics and also as a ratio of the sum of all the peaks below 400eV to that above 400eV. Figure 6 shows a very strong positive correlation of both these metrics with electrocatalytic activity confirming that XPS is very efficient in reflecting on the chemistry of nitrogen that is beneficial for ORR.

FIG. 6.

Correlation between the ratio of all peaks lower than 400 eV to that above 400 eV (left y-axis); the ratio of the peak due to pyridinic N to that of hydrogenated N (right y-axis) vs half-way potential—data presented in Ref. 19 and reanalyzed. Both correlations reflecting structures beneficial for ORR should have less hydrogenated and protonated N (contribute to BE above 400 eV) and more N coordinated to metal and pyridinic (contribute to BE below 400 eV).

FIG. 6.

Correlation between the ratio of all peaks lower than 400 eV to that above 400 eV (left y-axis); the ratio of the peak due to pyridinic N to that of hydrogenated N (right y-axis) vs half-way potential—data presented in Ref. 19 and reanalyzed. Both correlations reflecting structures beneficial for ORR should have less hydrogenated and protonated N (contribute to BE above 400 eV) and more N coordinated to metal and pyridinic (contribute to BE below 400 eV).

Close modal

The oversimplification example of XPS spectra presented in this perspective is distinctive and specific to this type of material system. Such a simplification is only possible when care is taken in following steps of converting XPS data to information:

  1. Correct charge correction of spectra.47 

  2. Proper curve fitting of N 1s XPS spectra using adequate background subtraction, proper shape, and width of peaks driven by physics of photoelectron spectroscopy and chemistry of the material.

  3. Assignment of species to peaks based on (1) knowledge of the sample history, i.e., materials origin and synthesis steps, (2) microscopic and structural information available from other analytical techniques, (3) cross-confirmation of interpretation by all the elements present, and (4) understanding the possible spectroscopic overlap between peaks originating from different chemical moieties.

  4. Reporting experimental protocols of data acquisition, data processing, including full details of curve fitting spectra based on international standard.48 Complete curve fit spectra should always be presented.

  5. A thorough analysis of structural relationships between the surface concentration of N species and metrics reflecting the function of the material. Univariate or multivariate methods of data analysis can be used to assist in the interpretation of possible correlations.19 

1.
T.
Mineva
,
I.
Matanovic
,
P.
Atanassov
,
M.-T.
Sougrati
,
L.
Stievano
,
M.
Clémancey
,
A.
Kochem
,
J.-M.
Latour
, and
F.
Jaouen
,
ACS Catal.
9
,
9359
(
2019
).
2.
W.
Wang
,
Q.
Jia
,
S.
Mukerjee
, and
S.
Chen
,
ACS Catal.
9
,
10126
(
2019
).
3.
U. I.
Kramm
,
M.
Lefèvre
,
N.
Larouche
,
D.
Schmeisser
, and
J.-P.
Dodelet
,
J. Am. Chem. Soc.
136
,
978
(
2014
).
4.
K.
Mamtani
,
D.
Singh
,
J.
Tian
,
J.-M. M.
Millet
,
J. T.
Miller
,
A. C.
Co
, and
U. S.
Ozkan
,
Catal. Lett.
146
,
1749
(
2016
).
5.
J.
Li
et al,
J. Am. Chem. Soc.
142
, 1417 (
2019
).
6.
Q.
Jia
,
E.
Liu
,
L.
Jiao
,
S.
Pann
, and
S.
Mukerjee
,
Adv. Mater.
31
,
1805157
(
2019
).
7.
M. A.
Wójtowicz
,
J. R.
Pels
, and
J. A.
Moulijn
,
Fuel
74
,
507
(
1995
).
8.
P.
Chambrion
,
T.
Suzuki
,
Z.-G.
Zhang
,
T.
Kyotani
, and
A.
Tomita
,
Energy Fuels
11
,
681
(
1997
).
9.
R.
Kelemen
,
H.
Freund
,
M. L.
Gorbaty
, and
P. J.
Kwiatek
,
Energy Fuels
13
,
529
(
1999
).
10.
E.
Ech-chamikh
,
A.
Essafti
,
Y.
Ijdiyaou
, and
M.
Azizan
,
Sol. Energy Mater. Sol. Cells
90
,
1420
(
2006
).
11.
C.
Morant
,
J.
Andrey
,
P.
Prieto
,
D.
Mendiola
,
J. M.
Sanz
, and
E.
Elizalde
,
Phys. Status Solidi A
203
,
1069
(
2006
).
12.
T.
Susi
,
T.
Pichler
, and
P.
Ayala
,
Beilstein J. Nanotechnol.
6
,
177
(
2015
).
13.
N.
Hellgren
,
R. T.
Haasch
,
S.
Schmidt
,
L.
Hultman
, and
I.
Petrov
,
Carbon
108
,
242
(
2016
).
14.
K. N.
Wood
,
S. T.
Christensen
,
D.
Nordlund
,
A. A.
Dameron
,
C.
Ngo
,
H.
Dinh
,
T.
Gennett
,
R.
O'Hayre
, and
S.
Pylypenko
,
Surf. Interface Anal.
48
,
283
(
2016
).
15.
S.
Kabir
,
K.
Artyushkova
,
A.
Serov
, and
P.
Atanassov
,
ACS Appl. Mater. Interfaces
10
,
11623
(
2018
).
16.
E.
Contreras
,
D.
Dominguez
,
H.
Tiznado
,
J.
Guerrero-Sanchez
,
N.
Takeuchi
,
G.
Alonso-Nunez
,
O. E.
Contreras
,
M. T.
Oropeza,-Guzmán
, and
J. M.
Romo-Herrera
,
Nanoscale
11
,
2829
(
2019
).
17.
P.
Lazar
,
R.
Mach
, and
M.
Otyepka
,
J. Phys. Chem. C
123
,
10695
(
2019
).
18.
S. J.
Mao
,
C. P.
Wang
, and
Y.
Wang
,
J. Catal.
375
,
456
(
2019
).
19.
K.
Artyushkova
,
A.
Serov
,
S.
Rojas-Carbonell
, and
P.
Atanassov
,
J. Phys. Chem. C
119
,
25917
(
2015
).
20.
J.
Fang
,
H.
Fan
,
M.
Li
, and
C.
Long
,
J. Mater. Chem. A
3
,
13819
(
2015
).
21.
J.
Fang
,
H.
Fan
,
M.
Li
, and
C.
Long
,
Diamond Relat. Mater.
16
,
581
(
2007
).
22.
H.
Dai
,
X.
Gao
,
E.
Liu
,
Y.
Yang
,
W.
Hou
,
L.
Kang
,
J.
Fan
, and
X.
Hu
,
Diamond Relat. Mater.
38
,
109
(
2013
).
23.
I.
Matanovic
,
K.
Artyushkova
, and
P.
Atanassov
,
Curr. Opin. Electrochem.
9
,
137
(
2018
).
24.
K.
Artyushkova
,
B.
Kiefer
,
B.
Halevi
,
A.
Knop-Gericke
,
R.
Schlogl
, and
P.
Atanassov
,
Chem. Commun.
49
,
2539
(
2013
).
25.
I.
Matanovic
,
K.
Artyushkova
,
M. B.
Strand
,
M. J.
Dzara
,
S.
Pylypenko
, and
P.
Atanassov
, “
Core level shifts of hydrogenated pyridinic and pyrrolic nitrogen in the nitrogen-containing graphene-based electrocatalysts: In-plane vs edge defects
,”
J. Phys. Chem. C
120
(
51
),
29225
(
2016
).
26.
S.
Kabir
,
K.
Artyushkova
,
A.
Serov
,
B.
Kiefer
, and
P.
Atanassov
,
Surf. Interface Anal.
48
,
293
(
2016
).
27.
W. E. S.
Unger
,
J. Vac. Sci. Technol. A
38
,
021201
(
2020
).
28.
G.
Greczynski
and
L.
Hultman
,
Chem. Phys. Chem.
18
,
1507
(
2017
).
29.
G.
Greczynski
and
L.
Hultman
,
Prog. Mater. Sci.
107
,
100591
(
2020
).
31.
K.
Artyushkova
,
S.
Rojas-Carbonell
,
C.
Santoro
,
E.
Weiler
,
A.
Serov
,
R.
Awais
,
R. R.
Gokhale
, and
P.
Atanassov
,
ACS Appl. Energy Mater.
2
,
5406
(
2019
).
32.
F.
Späth
,
W.
Zhao
,
C.
Gleichweit
,
K.
Gotterbarm
,
U.
Bauer
,
O.
Höfert
,
H. P.
Steinrück
, and
C.
Papp
,
Surf. Sci.
634
,
89
(
2015
).
33.
P. H.
Matter
,
L.
Zhang
, and
U. S.
Ozkan
,
J. Catal.
239
,
83
(
2006
).
34.
A. P.
Dementjev
,
A.
de Graaf
,
M. C. M.
van de Sanden
,
K. I.
Maslakov
,
A. V.
Naumkin
, and
A. A.
Serov
,
Diamond Relat. Mater.
9
,
1904
(
2000
).
35.
U. I.
Kramm
,
I.
Herrmann-Geppert
,
J.
Behrends
,
K.
Lips
,
S.
Fiechter
, and
P.
Bogdanoff
,
J. Am. Chem. Soc.
138
,
635
(
2016
).
36.
Y. C.
Chen
,
I.
Matanovic
,
E.
Weiler
,
P.
Atanassov
, and
K.
Artyushkova
,
ACS Appl. Energy Mater.
1
,
5948
(
2018
).
37.
U. I.
Kramm
,
L. M.
Ni
, and
S.
Wagner
,
Adv. Mater.
31
,
e1805623
(
2019
).
38.
U.
Martinez
,
E. F.
Holby
,
S. K.
Babu
,
K.
Artyushkova
,
L.
Lin
,
S.
Choudhury
,
G. M.
Purdy
, and
P.
Zelenay
,
J. Electrochem. Soc.
166
,
F3136
(
2019
).
39.
N. R.
Sahraie
,
U. I.
Kramm
,
J.
Steinberg
,
Y. J.
Zhang
,
A.
Thomas
,
T.
Reier
,
J. P.
Paraknowitsch
, and
P.
Strasser
,
Nat. Commun.
6
,
8618
(
2015
).
40.
P. G.
Santori
,
F. D.
Speck
,
J.
Li
,
A.
Zitolo
,
Q. Y.
Jia
,
S.
Mukerjee
,
S.
Cherevko
, and
F.
Jaouen
,
J. Electrochem. Soc.
166
,
7
(
2019
).
41.
M. T.
Sougrati
,
V.
Goellner
,
A. K.
Schuppert
,
L.
Stievano
, and
F.
Jaouen
,
Catal. Today
262
,
110
(
2016
).
42.
S.
Wagner
et al,
Angew. Chem. Int. Ed.
58
,
10486
(
2019
).
43.
H. G.
Zhang
,
H. T.
Chung
,
D. A.
Cullen
,
S.
Wagner
,
U. I.
Kramm
,
K. L.
More
,
P.
Zelenay
, and
G.
Wu
,
Energy Environ. Sci.
12
,
2548
(
2019
).
45.
S.
Rojas-Carbonell
,
K.
Artyushkova
,
A.
Serov
,
C.
Santoro
,
I.
Matanovic
, and
P.
Atanassov
,
ACS Catal.
8
,
3041
(
2018
).
46.
M.
Figueras
,
I. J.
Villar-Garcia
,
F.
Viñes
,
C.
Sousa
,
V. A.
de la Peña O’Shea
, and
F.
Illas
,
J. Phys. Chem. C
123
,
11319
(
2019
).
47.
ISO
19318:2004, Surface chemical analysis—X-ray photoelectron spectroscopy—Reporting of methods used for charge control and charge correction (
2004
).
48.
ISO
19830:2015, Surface Chemical Analysis. Minimum reporting requirements for peak fitting in XPS defined how peak fitting and the results of peak fitting in X-ray photoelectron spectroscopy (
2015
).

Kateryna Artyushkova is a Senior Staff Scientist at Physical Electronics. Her research focuses on developing applications and methodologies for surface analysis focusing on x-ray photoelectron spectroscopy. She has received an M.A. Degree with Honors in chemistry from the National Technical University of Ukraine, Kyiv, in 1995. She obtained her Ph.D. in chemistry from the Kent State University in December 2001 under the supervision of Julia E. Fulghum. She was a Research Associate Professor at the Chemical Engineering Department of the University of New Mexico and Associate Director for Center for Microengineered Materials until 2018. She has more than 20 years of experience with all aspects of x-ray photoelectron spectroscopy, including instrumentation, experimental design for optimizing time and information content, and data analysis. She has coauthored 200+ peer-reviewed publications and four book chapters. She is a short course instructor of AVS, Past chair of Applied Surface Science Division of AVS, Past Chair of NMAVS chapter, and executive member of Minnesota Chapter AVS. She is a North American Editor of the Surface and Interface Analysis Journal.