Analyst, December, 1983, Vol. 108, $9. 1477-1480 147 7 Analytical Potential of Valence State and Ligand Atom Effects in Titanium K X-ray Spectra John B. Jones and David S. Urch Admiralty Marine Technology Establishment, Holton Heath, Poole, Dorset, BH16 6 J U Chemistry Department, Queen Mary College, Mile End Road, London, E l 4NS Changes in peak profiles and in peak energies of titanium Kcc, KPl,3 + K/3’ and K& + KP” X-rays with both valency and ligand environment changes have bekn observed. Both factors affect Kcc but a distinct KP’ feature identifies Ti(I1) whilst ligand atoms can be identified from the KP2,+ KjY energy difference. When the ligands are known, the small shifts of the sharp and intense Ti Ka, peak provide the best indicator of valency. It is suggested that such shifts, together with changes in KP peak profiles, could be used analytically to determine the valence state of titanium and the chemical nature of the atoms that surround it.Keywords : Titanium X-ray spectra ; valence state effeects ; ligand atom effects The wavelengths of characteristic X-ray emissions can be slightly but perceptibly altered by changes in the formal valency or ligand environment of an atom, even if the X-ray results from electronic transitions between core orbitals. More dramatic effects such as larger wavelength shifts and the formation of new peaks are observed as a result of valence shell to core transi- tions. The analytical potential of these effects has, however, been exploited only in a few iso- lated instances, e.g., to determine the co-ordination number of aluminium using Ka,,, shifts‘ or to determine the valency of manganese using the KP’: KP1,3 intensity ratio2 or to determine the chemical nature of a ligand atom attached to a second-row atom using the KP’-KP1,3 energy differen~e.~ Even so there is sufficient variety in the parameters that can be measured and the correlations that can be established to suggest that a detailed study of peak shifts, etc., should yield data that could be useful in analysis by providing structural and valence state information.In this work an examination had been made of the X-ray emission spectra from a wide range of titanium compounds in which both the valence state and the ligand environ- ment of the titanium atom were varied. Experimental The titanium X-ray emission spectra of the following substances were measured : titanium metal, titanium carbide, titanium oxides (TiO, Ti203 and TiO,), titanium sulphide (Ti,S,), titanium trifluoride, potassium hexafluorotitanate and titanates of barium and neodymium.Either commercially available samples of AnalaR purity were used or compounds were pre- pared by standard techniques. The stoicheiometries of the titanates, which were supplied by S.T.L., Harlow, were checked by X-ray diffraction. Discs of all the samples (except titanium metal) were made by pressing the powdered compound with terephthalic acid in a ring press to 5 tons. These discs were irradiated with X-rays from a sealed chromium anode X-ray tube (50 mA, 50 kV) in a Philips PW1410 X-ray fluorescence spectrometer.To obtain the best dispersion of the characteristic radiation, a rubidium acid phthalate crystal (2d = 2612 pm) was used for Ti Karl,, (ninth order) and ammonium dihydrogen orthophosphate (2d = 1064 pm) was used for Ti KP1?3 and Ti KP2,5 (fourth order). The exact positions of the Ka,,, and KP,,, peaks were determined by counting for pre-set periods of time every O.Ol”26’ in the region of the peak maximum. The more complex KP,,, region was measured by scanning the wave- length automatically and recording the fluctuations in count rate (from the ratemeter) graphically. Results and Discussion The energies of Ka,, KP1,3 and KP,,, (with K/3” if seen) are listed in Table I and some typical results for the Kp”-Kp2,5 and KP,,,--KP’ regions are shown in Figs.1 and 2, respectively.1478 JONES AND URCH: ANALYTICAL POTENTIAL OF VALENCE Analyst, VoZ. 108 TABLE I ENERGIES OF TITANIUM X-RAYS X-ray energy/eV b r Titanium metal.. . . 4510.0 4931.0 4960.3 Ti0 . . .. . . 4509.9 4930.4 4955.9 Ti,O, . . .. . . 4509.6 4930.7 4956.6 Ti,S, . . .. . . 4509.8 4930.9 4958.0 TiF, . . .. . . 4509.4 4930.9 4959.6 Tic . . .. . . 4510.0 4931.0 4959.9 TiO, . . .. . . 4509.5 4930.9 4960.8 BaTiO, . . .. . . 4509.2 4930.7 4961.5 BaTi,O, . . . . . . 4509.3 4930.7 4960.8 Nd,Ti,O, . . .. . . 4509.3 4930.7 4960.8 K,TiF, . . .. , . 4509.3 4930.9 4962.4 Compound Kal* Kfi1,3 KfiZ,5 * Kcc, = Kor, -5.6 eV. 7 K r 4953.5 4 945.1 4945.8 4945.3 4945.3 4941.0 Kw,2 The most intense emission feature in the X-ray spectrum of titanium is Kocl,,, which is generated by an electronic transition from the 2p orbitals to a 1s vacancy.The final 2p-1 “hole” configuration is, however, split by spin - orbit exchange coupling into two possible states, which differ in energy by 5.6 eV, 2P, and 2P+ This causes the Ka peak to be split into Kcc, [ls-1(2S+) -+ 2p-1(2P;)] and Kcc, [1s-1(2S,) -+ 2~-l(~P,)], the former having a slightly higher energy and being twice as intense as the latter. Because of its intensity and because it can be resolved from Koc,, chemical effects in Ti Kcx spectra will be discussed in terms of the Koc, peak alone. The shifts observed for Kcxl are very small, the whole range of 0.8 eV corresponding to only 0.06’28, but the sharpness of the Kcx peaks enables their angular position to be determined precisely (&0.01”28).The shifts that are observed, relative to the value of Ti Karl for titanium metal, are a function of both ligand and valence state. These two effects can be disentangled as shown in Fig. 3. From this diagram it can be seen that any effect that would increase the 4 :i”\. 2 n- 4940 4960 4980 E ne rg yleV Fig. 1. Titanium K/3”- Kfi,,& spectra. A 4920 4930 4940 E ne rgyleV Fig. 2. Titanium Kfi’ -ICfll,s spectra from (A) TiO, and (B) TiO. For Ti0 note KP’ at 4916 eV.December, 1983 STATE AND LIGAND ATOM EFFECTS IN Ti K X-RAY SPECTRA 1479 effective charge on the titanium atom, be it either an increase in the electronegativity of the ligands or an increase in the formal valence state, causes a reduction in the Ka, energy. This is because valence shell orbitals, which would, from either cause, experience a reduction in electron density, overlap more with 2p orbitals than with 1s.Thus, as charge is withdrawn from the valence shell both 2p and 1s orbitals became more tightly bound owing to a reduction in electron - electron repulsion, but because of orbital overlap this effect will be greater for 2p than for Is. The energy difference between 2p and Is ionisation.energies, the Ka, and Kaz X-ray energy, will therefore diminish as the titanium atom becomes more positive. Unfortunately, the shifts brought about by changes in ligands and in valence state are com- parable, so that it is only possible to determine the valence state from a knowledge of the Ka shift if the ligand atoms are known or vice versa; the chemical nature of the ligands can only be indicated if the valency is known.It should be noted, however, that the required information about ligand atoms can sometimes be found from KP” and KP,,, spectra. Carbon Sulphur X I Fluorine I x o 0 -0.2 -0.4 -0.6 -0.8 ShiWeV Fig. 3. Shifts of the titanium Ka, peak from titanium metal for different ligands and different titanium valences: e, (11); X , (111); and C, (IV). KPl3 The Ti K/3,,, peak is caused by tfie transition 1s-l -+ 3p-l, but as 3p orbitals are more diffuse than 2p orbitals their overlap with a 1s core hole is less so that KP1,3 peaks are less intense than Ka. For titanium the 3p spin - orbit exchange splitting is smaller than for 2p so that distinct KP, and KP3 peaks are not seen; however, overall KP1,3 is broader than either Ka, or Ka2.This greater width makes it more difficult to locate the Kp1,3 peak maximum than Ka,. Also it is found that the KP1,3 shifts are in fact smaller than Kor, shifts and behave in a different way, the shift for Ti(I1) being greater than those of Ti(II1) or Ti(1V). This shows that other factors apart from the effective charge at titanium help to determine the KP1,3 energy. With an ionisation energy of only about 38 eV it seems most likely that Ti 3p orbitals will be directly influenced by ligand atoms, i.e., they will be subject to chemical bond effects4 The breadth of the Kp1,3 peak may be due to many possible final states being accessed by the “3p” -+ 1s relaxation ; states in which other electronic excitation (“shake-up”) has taken place involving vacant 3d orbitals.The anticipated simple decrease in KP1,3 energy with increasing positive charge at the titanium would then be masked by these other configurational excitation effects. The simplest of these is exchange coupling between unpaired 3d electrons and the final state vacancy in the 3p shell.5 The hole state spin can be parallel or anti-parallel to the 3d electron spin so that there would be two possible energy states associated with 3p5, for Ti(I1) and Ti(II1). These two states give rise to two peaks in the Kp spectrum, the main KP1.3 peak and a low-energy satellite KP’ (the relative intensity of which is, for light transition elements, pro- portional to the number of unpaired d electrons5). The presence of such a peak in the Ti KP spectrum of Ti0 can be clearly seen in Fig.2, with a relative intensity of about 10% of KP1,3 and about 12 eV less energy. By analogy with manganese2 the presence of such a peak can be used diagnostically to indicate the presence of divalent titanium. The single d electron in Ti(II1) does not, however, generate an intense enough feature to be of use analytically.1480 JONES AND URCH KP2.5- KP” These very weak peaks result from electronic transitions from valence band orbitals to the titanium 1s core orbital; the transition will therefore be due to Ti 4p character in these valence band orbitals. However, these orbitals are predominantly ligand in character and so their energies will be determined by the nature of the ligand atom and the effective charge at the ligand atom.If for all titanium oxides and for titanates the oxygen 2p ionisation energy is taken, as a first approximation, to be constant, then it follows that the KP,,, transition energy should decrease in the valency sequence Ti(1V) > Ti(II1) > Ti(I1) because the Ti 1s ionisa- tion energy will be largest for the titanium state with the greatest positive charge. This anticipated trend is in fact observed and can be seen from the results in Table I : TiO, (4960.8 eV), Ti203 (4956.6 eV) and Ti0 (4955.9 eV). A similar trend is found for titanium in different valence states with fluorine ligands, TiF,” (4962.4 eV) and TiF, (4959.6 eV). For a particular valence state increasing the electronegativity of the ligands should also increase the Ti 1s ionisation energy but this trend runs parallel to the increase in ligand valence orbital ionisation energy so that only slight changes may be observed.Actually a small increase is found: for TIC, TiO, and K,TiF, the energies are 4959.9, 4960.8 and 4962.4 eV, respectively. Although the shifts found for KP2,5 are much larger than for the other K peaks of titanium their lack of intensity limits their use for analysis. If, however, both KP,,, and KP” can be observed then the location of Kp” is of use in determining the chemical nature of the ligand atoms. This is because the origin of KP2,5-KP” for first-row transition elements is strictly analogous to that for KP, 3-K/3’ for second-row main group elements,6 i.e., KP2,5 reflects the 4p character in orbitals principally ligand 2p in character and KP” the 4p character in orbitals principally ligand 2s in character. The KP2,5-KP” region of the titanium X-ray spectrum is shown in Fig.1, where it can be seen that KP” is only clearly observed in Ti(1V) compounds. This is because the intensity ratio Kp” : KP2,5. decreases with metal valency’ and also because the KP” -+ KP,,, feature shifts to lower energies with lower valencies, KP” then tending to be lost* on the side of the steeply rising KP,,, peak (TiO, Fig. 1). Fig. 1 shows that for tetra- valent compounds KP” can be found for K,TiF, at about 20eV less than the KP2,5 peak energy, for TiO, and the titanates at about 16 eV less than KP,,,, and for TIC at about 6 eV less than KP2,5. These energy separations can, therefore, be used to determine the nature of the ligand atoms that surround the titanium.Conclusions Shifts in Ti Kcc,, whilst small, can be used to determine the valency of titanium provided that the ligand environment is known. This can be found, at least for Ti(IV), by an examination of the energy separation between KP” and Kp2,5. The presence of low-valency titanium [i.e., Ti(II)] can be confirmed by the presence of a distinct KP’ peak on the low-energy side of K/31,3. Thus if all the K X-ray emission spectra are examined it should be possible to deter- mine both the valency of the titanium and the chemical nature of atoms to which it is bound. The authors thank The Royal Society arid the Central Research Fund of London University One of them (J.B. J.) is grateful to for financial assistance in the purchase of equipment. both the SERC and to S.T.L. (Harlow) for a CASE award and research grant. References 1. 2. 3. 4. 5. 6. 7. 8. Day, D. E., Nature, London, 1963, 200, 649. Wood, P. R., and Urch, D. S., X-ray Spectrom., 1978, 7, 9. Esmail, E. I., Nicholls, C. J., and Urch, I>. S., Analyst, 1973, 98, 725. Ichikawa, K., Nakamori, H., Tsutsumi, K., and Watanabe, T., J. Phys. SOC. Jpn., 1977, 43, 1255. Slater, R. A., and Urch, D. S., J. Chem. SOC. Chem. Commun., 1972, 564. Urch, D. S., J. Phys. C , 1970, 3, 1275. Asada, E., Takiguchi, T., and Suzuki, Y., X-ray Spectrom., 1975, 4, 186. Nemnonov, S. A., and Kurmayev, E. S., Fiz. Met. Metalloved., 1969, 27, 816. Received May 3rd, 1983 Accepted July 4th, 1983