The definition of oxidation level was introduced for single crystal surface oxidation evaluation. The oxidation level is a parameter describing the total share of the A and B atoms bound with oxygen in an III–V crystal in one of the configurations illustrated in Fig. 1 for the example of GaAs.

Figure 1.

Schematic of four Ga and As bonds with oxygen in single crystal GaAs: (a) Ga[O, 3As]; (b) Ga[2O, 2As]; (c) Ga[3O, As]; (d) Ga[3O]. Dashed line shows bond between unshared pair of Ga atom valence electrons

The Ga[O, 3As], Ga[2O, 2As] and Ga[3O, As] configurations indicate partial oxidation of trivalent Ga (III), and the Ga[3O] configuration corresponds to Ga₂O₃. For the As and Sb elements forming III- and V-valence oxides, the As[5O] and Sb[5O] configurations indicate As₂O₅ and Sb₂O₅ oxides. The oxidation level ω_ox takes into account the bond configurations and A and B atomic concentrations and is calculated as follows:

$\begin{array}{rl}& {\omega }_{\mathrm{ox}}=\frac{C(A)}{C(A)+C(B)}\sum _i\left(k_{{\mathrm{ox}}_i}I{\left(A_{\mathrm{ox}}\right)}_i\right)+\frac{C(B)}{C(A)+C(B)}\sum _j\left(k_{{\mathrm{ox}}_j}I{\left(B_{\mathrm{ox}}\right)}_j\right)\end{array}$ (3)

where C (A) and C (B) are the A and B element concentrations (at.%); i and j are the configuration indices as determined from the high-resolution A and B element spectra; I (A_ox) and I (B_ox) are the relative intensities (%) of spectral features pertaining to specific configurations; k_ox is the coefficient taking into account bond configuration and valence.

The k_ox coefficient was determined from the quantity of oxygen atoms bound with A or B elements and accepted to be 1 for the A[3O] and B[3O] configurations and 5/3 for the B[5O] configuration. For partial oxidation configurations k_ox was 1/3 (one oxygen atom) or 2/3 (two oxygen atoms).

It should be noted that the oxidation level has limitations imposed by its evaluation method. This parameter takes into account the number of oxygen bonds in the surface layer the thickness of which depends on the photoelectron free path. The oxidation level is 100% if all the bonds in the surface layer are occupied. This means that the oxide layer thickness is more than 5 nm since it is three times the inelastic mean free path (IMFP) of electrons emitted from the Ga 3d (1.7 nm), Sb 3d3 (1.3 nm), In 3d5 (1.3 nm) and As 3d (1.9 nm) levels (calculated for Ga₂O₃, Sb₂O₃, In₂O₃ and As₂O₃ oxides for a θ = 45 deg angle) [20]. The method is not applicable beyond this depth.

The ionicity of chemical bonds was evaluated using the Sanderson and Phillips models on the basis of the calculation formulas and earlier data [15–17].

The Sanderson model is based on electronegativity interpretation as a chemical potential that is to some extent equalized upon molecule formation. This approach allows evaluating the ionicity degree of a chemical bond from the partial anion charge generated as a result of electronegativity equalization. In turn, the partial charge is calculated as

$\delta =\frac{\mathrm{\Delta }S}{\mathrm{\Delta }{S}_c}$ (2)

where ΔS is the difference between the electronegativities of the element S and the intermediate electronegativity S_b; ΔS_c is the change in electronegativity required for the element to acquire unit positive or negative charge.

The S and ΔS parameters of all the elements are available in Sanderson’s tables [16]. The intermediate electronegativity S_b is given by the geometric mean of the individual electronegativities of the component atoms. For the III–V compounds in question the intermediate electronegativity was evaluated using the formula

$\mathrm{\Delta }{S}_b=\sqrt{S_{\text{III}}{S}_{\mathrm{V}}},$ (3)

For the Phillips approach the ionicity degree of a chemical bond is evaluated from the optical absorption spectra.

The ionicity degrees of the GaAs, InAs, GaS band InSb bonds were calculated as per the Sanderson model using Eqs. (2) and (3) and the chemical bond ionicity degrees calculated using the Phillips model were borrowed directly from available tabulated data [17]. The ionicity degrees of the A–B bonds calculated using both models are summarized in Table 1.

The ionicity degree estimates made using the above two models differ significantly. The Phillips ionicity degrees of the GaAs, InAs, GaSb and InSb compounds are close and range from 26 to 36%. The Sanderson ionicity degrees differ tangibly: it is 1% for GaSb and 14% for InAs. Moreover, the Sanderson ionicity degrees are far lower than the Phillips ones. The difference originates from a discrepancy between the different ionicity degree evaluation methods and the choice of ionicity degree scale limits. Sanderson takes into account the chemical bond ionicity degrees of all possible compounds [16] whereas the range of compounds considered by Phillips is limited [17].

Direct effect of ionicity degree on the surface chemical reactivity of the crystals was assessed for natural air oxidation under similar conditions (temperature, humidity and atmospheric exposure time). The wafers were cleaved along their cleavage planes (110), and the cleaved specimens were clamped in the spectrometer holder which was then loaded in the spectrometer preparation chamber evacuated to a less than 2∙10^-6 Pa residual gas pressure. The atmospheric exposure time was 2 min for this procedure. The cleavage surface survey spectra taken in the spectrometer analytical chamber exhibited carbon and oxygen photoelectron bands indicating adsorption of those elements by the atomically pure cleavage surface during the short air exposure time. The element atomic concentrations are summarized in Table 2.

The carbon concentrations on the cleavage surfaces are within 22 at.%, the lowest one being that on the GaAs cleavage. As regards oxygen, its concentration on the gallium and indium antimonide cleavages is far higher than that of carbon, and on the arsenide cleavages it is far lower than that on the antimonide cleavages. There are no clear regularities in the carbon and oxygen atomic concentration ratios. However, direct comparison between the oxygen concentrations calculated for the surfaces of the test materials from the survey spectra does not allow one to draw any definitive conclusion as to the chemical reactivity of the surfaces. The XPS-measured oxygen concentration is an integral parameter since it includes both chemically bound oxygen and adsorbed species. Selective analysis of chemically bound oxygen can be done from the high-resolution spectra of A and B as shown in Fig. 2. The presence of single peaks or doublets shifted to the left-hand side of the E_b energy scale (i.e., to higher energies) relative to the main state peaks in the Ga 3d, In 3d5, As 3d and Sb 3d5 spectra indicates oxygen bonds. The greater the shift (the chemical shift of ∆E_b in XPS terms), the greater the number of oxygen bonds, i.e., different configurations shown in Fig. 1 produce different shifts.

Figure 2.

High-resolution photoelectron spectra of (a) A and (b) B elements for (110) cleavages of III–V compounds (GaAs, GaSb, InAs and InSb). Peaks of oxidized element states are green-filled

Correct interpretation of the chemical shifts was achieved using literary available reference data [6, 21–26] and the Authors’ own experimental results for III–V polished commercial wafer specimens with thick stoichiometric oxide layers on the surface. The chemical shifts ∆E_b of the oxide peaks from the main element state peaks (hereinafter the latter is the chemical state of the elements A and B in the III–V lattice) were as follows:

– +1.4 eV for Ga 3d (Ga₂O₃ – GaAs);

– +3.2 eV for As 3d (As₂O₃ – GaAs);

– +2.8 eV for Sb 3d (Sb₂O₃ – InSb);

– +0.9 eV for In 3d (In₂O₃ – InSb).

Partial oxidation configurations were identified by introducing the definition of unit chemical shift ∆₊₁ which is the magnitude of chemical shift caused by attachment of one oxygen atom to the element A or B. The unit shift was accepted to be one third of the chemical shift for a trivalent oxide in the assumption that the chemical shift is linearly dependent on the number of oxygen bonds.

Figure 2 illustrates the differences (between compounds) in the intensities of spectral features pertaining to A and B atoms bound with oxygen atoms (green-filled areas). It can be seen that the area under those spectral features is the greatest for the GaSb and InSb cleavages. The quantitative data (including standard deviation of repeatability) obtained by approximation of high-resolution cleavage spectra are summarized in Table 3. The oxidation level ω_ox was calculated using Eq. (1).

The data in Table 3 suggest a general regularity for all the four test III–V compositions, i.e., preferential oxidation of the A elements: the shares of oxide peaks are 31 and 41% in the Ga 3d and In 3d5 spectra, respectively, and only 3 and 14% in the As 3d and Sb 3d3 spectra. However, the overall oxidation level ω_ox suggests a tangible difference in the reactivities of the above compounds. The oxidation level of cleavages increases in the InAs–GaAs–InSb–GaSb sequence, from arsenides to antimonides. The Sanderson A–B chemical bond ionicity degree (see Table 1) decreases in the same sequence. The InAs spectra of the test compositions with the greatest ionicity degree did not contain oxidation-related peaks, hence ω_ox = 0, with the highest ω_ox being that for the GaSb surface where the ionicity contribution is the lowest.

Figure 3 shows oxidation level of (110) cleavage as a function of chemical bond ionicity degree. The curves demonstrate a clear decrease in the oxidation level of GaAs, GaSb, InAs and InSb single crystal surfaces with an increase in the chemical bond ionicity degree. Interestingly, the pattern of the curves is similar regardless of whether Sanderson or Phillips ionicity degree calculation method was used (different order values).

Thus, the effect of chemical bond ioncity degree on natural chemical reactivity was revealed. One more question to be solved is whether this regularity is valid for oxidation resulting from III–V crystal wafer surface treatment during industrial processes.

Figure 3.

GaAs, GaSb, InAs and InSb single crystal (110) cleavage surface oxidation level as a function of A–B chemical bond ionicity degree

Surface oxidation of treated GaAs, GaSb, InAs and InSb wafers was studied for chemically-mechanically polished specimens. In general, wafer fabrication route includes the following operations:

– slicing of Cz-grown single crystals;

– wafer grinding and chemical-mechanical polishing in chemical reactant solutions;

– multistage rinsing.

The as-treated wafers were stored in open cassettes for 4 months except for GaAs which were stored for 6 months.

The atomic concentrations of elements on the wafer surfaces after storage are summarized in Table 4. The adsorbed carbon and oxygen concentrations prove to be far higher than those for cleavages. Furthermore, there was a disproportional change in the A and B element concentrations, most expressed for GaSb. The C_A/C_B concentration ratio was > 1 (except for InAs) indicating preferential oxidation of the A elements.

The high-resolution spectra for the wafers (Fig. 4) exhibit significantly larger areas of the oxide peaks. The Ga 3d and Sb 3d3 spectra of the GaSb specimen do not show peaks relating to the main states of gallium and antimony. Quantitative data obtained by high-resolution peak approximation are presented in Table 5.

One can see higher oxygen bond peak intensities and greater chemical shift ∆E_b. For example, ∆E_b of the Ga 3d band was 0.5 eV for the GaAs cleavage and 1.3 eV for the GaAs wafer indicating partial gallium oxidation on the cleavage surface and Ga₂O₃ oxide formation on the wafer surface. The As 3d and Sb 3d spectra contain peaks indicating the presence of pentavalent arsenic and antimony oxides As₂O₅ and Sb₂O₅. As a result, the oxidation levels of the wafers were several times those of the cleavages for the same specimens (see Tables 4 and 5). Despite the limitations of the oxidation level assessment method used, the oxidation patterns for all the four test compositions were the same for the wafers and for the cleavages. Figure 5 shows chemically-mechanically polished wafer surface oxidation level as a function of chemical bond ionicity degree. It can be seen that the oxidation level decreases with an increase in the chemical bond ionicity degree, both for natural oxidation and chemical-mechanical polishing stimulated oxidation.

Figure 4.

High-resolution photoelectron spectra of (a) A and (b) B elements for III–V (100) wafers (GaAs, GaSb, InAs, InSb). Peaks of trivalent oxides are green-filled and peaks of pentavalent oxides are yellow-filled

Figure 5.

GaAs, GaSb, InAs and InSb single crystal (100) wafer surface oxidation level as a function of A–B chemical bond ionicity degree