Theoretical identification of pyrimidine on Si(100) by means of X-ray photoelectron spectroscopy and near-edge X-ray absorption fine structure spectra

The C 1s and N 1s X-ray photoelectron spectroscopy (XPS) and near-edge X-ray absorption fine structure structure (NEXAFS) spectroscopy of 11 possible configurations of pyrimidine molecules adsorbed on the Si(100) surface have been studied by the first principle method. These structures for pyrimidine adsorbed on the silicon surface have also been predicted and theoretically characterised. Our results show that the XPS and NEXAFS spectra of these adsorption configurations are structurally dependent. In contrast to the XPS spectra, it is found that the NEXAFS spectra are significantly dependent on the structures of pyrimidine absorbed on the Si(100) surface, which can effectively identify different molecular structures. In addition, since spectra vary with the local environment of carbon atoms, the NEXAFS spectra are useful to investigate the relationship between spectra and the structures of carbon atoms. GRAPHICAL ABSTRACT


Introduction
Silicon is the most significant semiconductor material, including silicon polycrystal, silicon single crystal, silicon wafer, silicon epitaxial sheet, amorphous silicon film, etc., which can be directly or indirectly used to fabricate semiconductor devices.Also, the function of silicon surface has been applied in the field of conventional electronics in recent years such as CMOS [1] (complementary metal oxide semiconductor) devices, mixed molecular silicon memory capacitor structures [2,3] and molecular sensors [4].For the rapid development of microelectronics industry, silicon material industry as the basic material of semiconductor industry makes a great contribution.And the rapid development of optoelectronic technology also accelerates the pace of semiconductor optoelectronic materials research, so it is imperative to study the development status and future trend of semiconductor silicon.In addition to the requirements of processing technology and equipment, modern microelectronics industry also puts forward newer and higher requirements for silicon materials.At present, the Si(100) surface is the main surface used in the microelectronics industry due to its complex reactivity [5].The direct covalent bonding of organic molecules to silicon has long been considered a way to combine molecular functions with conventional semiconductors [6][7][8].Aromatic hydrocarbons play an important role in the production of organic materials, but their interaction with the silicon surface is complex [8].Aromatic molecules adsorbed on silicon surface can produce a variety of structures.In the aromatic groups, pyrimidine is a heterocyclic compound formed by replacing two carbons at the intermolecular site of benzene with two nitrogen atoms.It retains aromaticity and is an important heterobenzene substance, mainly used as raw materials for pharmaceutical intermediates and photosensitisers.Prior to this, there have been some studies on pyrimidine adsorption [9][10][11][12].The pyrimidine molecules adsorbed on the silicon surface can produce different structures, which will lead to different physical and chemical properties.It is of great importance to identify the structures of pyrimidine adsorbed on the Si(100) surface.
For many different molecular structures, various spectral techniques can be used to study them [13][14][15][16].Regarding the X-ray spectrum technology, it is an effective method to study the interface structure of organic molecules and silicon surface, which is of great significance to identify the structures.It includes X-ray photoelectron spectroscopy (XPS) and near-edge Xray absorption fine structure spectrum (NEXAFS).XPS spectra are connected with the system of nuclear orbital electron ionisation process and NEXAFS spectra correspond to the system of the nuclear level electrons excited to unoccupied track process, which are both effective to distinguish diverse molecular structures.Previous studies have shown that these spectra can be used as an effective tool to identify the structures of pyrimidine adsorbed on silicon surface [17,18].In this work, we theoretically characterised the structures of pyrimidine adsorbed on the Si(100) surfaces by XPS and NEXAFS spectra to effectively identify them.Studying the structures of pyrimidine adsorbed on the silicon surface can stimulate the identification and characterisation of relevant molecular structures in the future, and promote the further development of the semiconductor materials devices.It is also a relatively basic and important work in the research of organic functional semiconductors, which is conducive to the development of microelectronics industry.

Computational details
In this project, the GaussView [19] software was first used to map the different structures that might appear on the Si(100) surface of the pyrimidine adsorption.Then, the DFT method at B3LYP/6-31G(d,p) level was used to optimise 11 different structures by the Gaus-sian16 package [20] and got their optimised structures.After this, the C 1s XPS and NEXAFS spectra of these adsorption structures were calculated.Among them, the calculation of spectra was completed by using the StoBe package [21] which adopted the gradientcorrected Becke (BE88) [22] exchange functional and the gradient-corrected Perdew86 (PD86) [23] exchange functional.Previous studies have confirmed that this set of commutation-related function can be able to give vibrator strength which was in great agreement with experimental results [24,25].For describing different atoms, the DZVP (double-ζ for valence electrons plus polarisation) was used to all silicon atoms.The tripleζ quality individual gauge for localised orbital (IGLO-III) [26] basis was selected to describe the core-exited carbon atom, and the triple-ζ plus valence polarisation (TZVP) basis set was chosen for the remaining ones.For facilitating self-consistent field (SCF) convergence of the core-hole state, various auxiliary basis sets were used for all atoms and the main potential function of the model was used to describe unexcited atoms.For excited carbon atoms, in addition to the general orbital basis group, the amplified diffuse basis set (19s, 19p, 19d) [27] was used to calculate the transition dipole moment and transition energy of nuclear-excited state.
We used Kohn-Sham ( KS) [28,29] method to calculate the ionisation potential (IP) of C 1s, IP is considered as the energy difference between the ground state (GS) and the completely optimised nuclear ionisation state.After calculating the ionisation energy, the C 1s XPS spectra were obtained by Gaussian broadening at the centre of the ionisation energy.The X-ray absorption process involves two states, the initial state (GS) and the final state (the core-exited state).According to the final state rule [30][31][32], only the wave function of the final state can obtain the accurate absorption spectra of a finite molecular system in the X-ray absorption process.In addition, a reference electronic configuration representing the characteristics of all transitions needs to be found, so that we can calculate all transitions using a single electronic structure.Then the full-core hole (FCH) approximation method was used to calculate NEXAFS by taking the nuclear ionisation state as the reference state [33].Considering the random orientation of molecules in reality, the intensity of the absorption oscillator was calculated through the average value of x,y and z components to give: where ψ i,f expresses two molecular orbitals (i denotes initial state, and f denotes final state) in the process of X-ray absorption, and ε fi denotes the transition energy from i to f.In order to gain absolute energy positions of peaks, the transition energy from C 1s to the lowest unoccupied molecular orbital (LUMO) was calculated with KS to standardise the absorption spectra, The conversion energy from 1s to LUMO was called the energy difference between the initial state and the fully optimised final state.The first spectral feature describing the conversion to LUMO after energy calibration was consistent with the precisely calculated excitation energy from 1s to LUMO.In addition, the relativistic effect of the introduced nuclear pore was considered by adding 0.2 eV to the calculated ionisation energy or transition energy.
The XPS spectrum adopted a half-height full width at half maximum (FWHM) of 0.2 eV Gaussian expansion.
For the NEXAFS spectrum, we adopted FWHM of 0.3 eV in the region below the ionisation energy and used the Stieltjes imaging method [34] to expand in the continuous state region above the ionisation energy.

Structures
Based on previous experience [11], we predicted 11 adsorption configurations of pyrimidine adsorbed on Si(100) surface, as shown in Figure 1.After replacing two non-adjacent carbon atoms in the benzene ring with nitrogen atoms, the remaining carbon atoms are marked as C The molecular structures in Figure 1 are divided into two types according to the number of binding bonds.To compare the stability of these adsorption configurations, we calculated their binding energies, where E C 4 H 4 N 2 is the total energy of a single thiophene molecule, E si is the energy of the Si surface, and is the total energy of the entire adsorption structure.The binding energies of the 11 structures are shown in Table 1.
The structure of the model with high binding energy is relatively stable.Compared to the 'Tight-Bridge' structures, the 'Tilted' and 'Butterfly' structures have higher binding energies, so their structures are more stable.The 'Tilted" structures of M1 and M3 have the higher binding energies, and the binding energy of M3 is 0.06 kcal/mol higher than that of M1, so the structure of M3 is more stable.

XPS
The XPS spectra of structures of pyrimidine adsorbed on Si(100) surface are shown in Figure 2.There are two characteristic peaks in the spectra of M1 and M8, and the spectral profiles are similar, which are not easy to distinguish.For M5, M6 and M7, there are three spectral features, and feature c, feature a and feature b are the strongest peak in the spectra of M5, M6 and M7 respectively, which can be taken as evidence to identify them.As for the remaining six adsorption structures, the spectral profiles of M2, M3, M4 and M9 are similar, so it is not easy to distinguish.The obvious shoulder features appear in the spectra of M10 and M11 and appear twice in M11, so M10 and M11 can be distinguished from other structures.It can be seen that we cannot fully identify these adsorption configurations by XPS spectra.To better reflect the relationship between structures and spectra, we also plot the spectrum of the original pyrimidine molecule in Figure 2. In the XPS spectra, these 11 adsorption structures show a significant red shift compared with the original molecule, that is, the carbon atoms of the original pyrimidine have higher k-edge IPs than those of the adsorption structures, which reflects the charge transfer from Si100) to the carbon atom of pyrimidine during the adsorption process caused by the non-metallic interaction between carbon and silicon.
In order to further study the spectral sources of these 11 structures of pyrimidine adsorbed on Si(100) surface, we calculated the spectra of non-equivalent carbon atoms of each adsorption structure.Figures 3 and 4 show the total C 1s XPS spectra for all adsorption structures and the spectral components corresponding to nonequivalent carbon atoms of each structure.In the spectrum of M1, peak a comes from C   The N 1s excited XPS spectra of structures for pyrimidine molecules adsorbed on Si(100) are also demonstrated in Figure 5.It can be seen from the figure that M1-M5, M8 and M11 all have only one spectral feature, which is not easy to identify, while the remaining M6, M7, M9 and M10 have two obvious features.Feature a and feature b of M6 show shoulder features, which can be taken as evidence to distinguish it, while the spectral profiles of M7, M9 and M10 are relatively similar, so it is difficult to accurately identify.In conclusion, XPS spectra cannot fully distinguish these adsorption structures, so we need to use NEXAFS spectra to provide more specific information for further identification.

NEXAFS spectra
NEXAFS spectra can reflect more information about the molecular structure because it describes the process of the excitation of electrons from the core layer to the unoccupied orbital.Thus, it can be better utilised to identify and characterise different molecular structures.Figure 6 shows the absorption spectra of structures (M1-M5), where (a) is the total spectra of M1-M5, and (b)-(f) shows the spectral components corresponding to carbon atoms.For (a) of Figure 6, we first focus on the spectral features between 284 and 286.6 eV.Combined with Table 2, we can clearly know the energy positions of the individual structure and mark them out in this range, 4. Calculated C 1s XPS spectra of the structures (M1-M5) for pyrimidine adsorbed on Si(100) and the spectral components corresponding to non-equivalent carbon atoms.
including a-e.In line with the number of major features, we can identify M1 with three features of a, c and e and M5 with two features of c and e from M1-M5.Moreover, the unique feature a of M3 sets it apart from other structures.And the identification of M2 and M4 depends on different intensity of c and d.In M2, the strength of feature c is stronger than that of feature d, while in M4, the opposite is true.Therefore, M1-M5 can be distinguished well on the basis of Figure 6 Finally, the feature c (1s → LUMO) in the spectrum of M5 is mainly generated by C 2 and C 4 , while e is mainly composed of C 3 .And C 1 of M5 and M1 contributes little to the main features of the total spectra.
Figure 7 shows the NEXAFS spectra of the 'Tight-Bridge' structures (M6-M11).Similarly, according to the energy position of the main features a-d given in Table 3, the main features are marked on each spectral line.M6 and M9 have only the main feature c, M7 and M10 have feature a and feature b, M8 and M11 have feature b and feature d.In M7 and M10, M10 contains feature f that M7 does not possess.In M8 and M11, the weak peak appears about 0.2 eV on the right side of feature d of M8.And the energy range of feature d of M11 is relatively wide.Moreover, we also mapped the spectra corresponding to the different carbon atoms of M6-M11, as shown  in Figure 8. From (a) and (d) of Figure 8, it can be seen that feature c of the total spectrum in M6 is mainly generated by C 1 and C 3 , while it is generated by C 1 in M9.The N 1s excited NEXAFS spectra of structures for pyrimidine molecules adsorbed on Si(100) are shown in Figure 9. M2 and M8 have only one obvious characteristic peak, and the spectral intensity for peak a of M2 is obviously greater than that of M8.Only M5 has two obvious characteristic peaks, and peak a is a weak shoulder peak on the left of peak b.As for M1, M3, M4, M6 and M10, there are three obvious characteristics.Among them, the spectral line profiles of M1 and M3, M4 and M6 are relatively similar, which is difficult to distinguish them, while the intensity of the three characteristic peaks is relatively small.Regarding M7, M9 and M11, the intensity of peak c in M7 is stronger than that of peak b, while in M9, the opposite is true.In the spectrum of M11, the intensity of the four distinct characteristic peaks decreases sequentially.
In conclusion, through the spectra generated by nonequivalent carbon atoms, it is found that carbon atoms in different environments contribute differently to the main features of the total spectrum, and carbon atoms nonbinding to silicon surface contribute more to the main characteristic peaks of the absorption spectrum, which helps us to understand the dependence of the NEXAFS spectra on different local structures.Through the NEX-AFS spectra of C 1s and N 1s, these structures of pyrimidine adsorbed on the Si(100) surface can be accurately identified, which provides a new idea for the development of organic functional semiconductor.

Summary
To sum up, we theoretically simulated the C 1s and N 1s X-ray photoelectron spectroscopy (XPS) and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy  of different molecular structures of pyrimidine adsorbed on the surface of Si(100).Compared with XPS, NEXAFS can distinguish 11 structures of pyrimidine adsorbed on the surface of Si(100) in more detail.The NEXAFS spectral properties of different carbon atoms in various molecular structures have also been investigated.It has been shown that carbon atoms in different environments contribute differently to the main features of the total spectrum, and carbon atoms non-binding to silicon surface contribute more to the main characteristic peaks of the absorption spectrum.It will provide support for related molecular structures in the future and promote the further development of silicon surface technology.

Figure 1 .
Figure 1.All possible structures of pyrimidine absorbed on Si(100) surface for the theoretical research.
1 − C 4 according to the clockwise direction.These 11 adsorption configurations are labelled as Model1-Model11 (M1-M11 are replaced in the following sections, respectively).According to the number of bonds formed between pyrimidine and Si(100) surface, we divided these 11 adsorption systems into two categories.M1-M5 combine the pyrimidine molecule with Si(100) by two bonds.In these structures, M1-M3 is constructed by connecting two adjacent atoms of the organic ring to the silicon surface, called the 'Tilted' structure.The other, M4 and M5, organic molecules and silicon surfaces are linked together by two equivalent Si-C bonds or Si-N bonds on the same dimer, called the 'Butterfly' structure.M6-M11 are established by adsorbing the pyrimidine molecule to two silicon dimers along the same row, which are called the 'Tight-Bridge' structure.In different adsorption structures, carbon atoms at different positions in the pyrimidine molecule bind to the silicon surface.Carbon atoms in one configuration are in a different environment in another configuration, that is, carbon atoms bound to the silicon surface may not bind to silicon in another structure.
1 and peak b is generated by C 2 , C 3 and C 4 .Peaks a, b, c and d in the spectra of M2 and M10 originate from C 3 , C 4 , C 2 and C 1 , respectively.The four peaks of M3 and M4 are from C 2 , C 1 , C 3 , C 4 and C 2 , C 3 , C 4 , C 1 , respectively.In the spectrum of

Figure 2 .
Figure 2. Calculated C 1s XPS spectra of the structures for pyrimidine adsorbed on Si(100) and pyrimidine, and the major features are labelled.

Figure 3 .
Figure 3. Calculated C 1s XPS spectra of the structures (M1-M5) for pyrimidine adsorbed on Si(100) and the spectral components corresponding to non-equivalent carbon atoms.
(a).In (b)-(f) of Figure 6, it can be observed that different carbon atoms contribute differently to the total spectrum in different structures.The feature c (1s → LUMO) in the total spectrum of M1 is mainly generated by C 2 and C 3 .Besides, C 4 contributes significantly to the feature a and e in the total spectrum of M1.With regard to the total spectrum of M2, b (1s → LUMO), c (1s → LUMO), d and e are respectively generated by C 4 , C 1 , C 3 and C 2 .The features a (1s → LUMO), b (1s → LUMO), d and e in the total spectrum of M3 are generated by C 3 , C 4 , C 1 and C 2 , respectively.About the characteristics of b (1s → LUMO + 1), c (1s → LUMO), d and e of M4 total spectrum are, respectively, contributed by C 3 , C 4 , C 1 and C 2 .

Figure 5 .
Figure 5. Calculated N 1s XPS spectra of the structures for pyrimidine adsorbed on Si(100), and the major features are labelled.

Figure 6 .
Figure 6.Calculated C 1s NEXAFS spectra and the individual component corresponding to the classification of carbon atoms for the structures (M1-M5), and the major features are labelled.
Therefore, feature c (1s → LUMO) in M6 is stronger than that (1s → LUMO) in M9, which can be used to distinguish molecular structures of M6 and M9.In (b) and (e) of Figure 8, feature a (1s → LUMO) and feature b (1s → LUMO) of the total spectra in M7 and M10 are respectively generated by C 3 and C 4 .But feature f of the total spectrum in M10 generated by C 1 and C 4 is more obvious, which makes M10 different from M7.In (c) and (f) of Figure 8, both feature b (1s → LUMO) in M8 and feature b (1s → LUMO + 1) in M11 is generated by C 4 , while feature f is generated by C 1 , C 2 and C 4 .The differences between them are that the weak peak appears about 0.2 eV on the right of C 1 in M8 and the energy position of feature f produced by C 2 in M11 is skewed to the right.

Figure 7 .
Figure 7. NEXAFS spectra at the C-edge of the structures (M6-M11) for pyrimidine adsorbed on Si(100), and the major features are labelled.

Figure 8 .
Figure 8. Calculated C 1s NEXAFS spectra and the individual component corresponding to the classification of carbon atoms for the structures (M6-M11), and the major features are labelled.

Figure 9 .
Figure 9. Calculated N 1s NEXAFS spectra of the structures for pyrimidine adsorbed on Si(100), and the major features are labelled.

Table 1 .
The binding energies of the 11 adsorption configurations at DFT level.

Table 2 .
Energies (eV)for main features between 284 and 286.6 eV in the calculated NEXAFS spectra of the structures(M1-M5).

Table 3 .
Energies (eV)for main features between 284.4 and 289 eV in the calculated NEXAFS spectra of the structures(M6-M11).