Influence of lattice dynamics on electron momentum density of lithium C. Sternemann a, * , G. Do ¨ring a , C. Wittkop a , W. Schu ¨lke a , A. Shukla b , T. Buslaps b , P. Suortti b a Institute of Physics, University of Dortmund, D-44221 Dortmund, Germany b European Synchrotron Radiation Facility, F-38043 Grenoble, France Abstract High resolution Compton profile measurements of single crystalline Li were performed for momentum transfer ~ q k110 with 0.12 a.u. momentum space resolution at room temperature and at 95 K using the Compton spectrometer of ID 15 B at ESRF. The valence electron Compton profile at room temperature is higher around p z  0a:u: and lower for |p z |  0:4a:u: when compared with the one at 95 K. The experimental results are in good agreement with calculations using an empirical pseudopotential scheme, where the variation of lattice constant with temperature is taken into account when calculating the Fermi energy and Fourier coefficients of the empirical pseudopotential are multiplied by the corresponding Debye–Waller factor to simulate the thermal disorder. Thus the measured temperature effect can be traced back not only to the variation of the lattice constant with temperature but also to the decreasing contributions of higher momentum components to the total momentum density with increasing temperature.  2000 Elsevier Science Ltd. All rights reserved. Keywords: Single crystalline lithium; D. Lattice dynamics Recent high resolution Compton scattering measurements on Li[1,2] have shown significant discrepancies between experiment and LDA calculations. The experimental valence Compton profiles are below the calculated ones around p z  0 and above in the high momentum regime for |p z |  p F : When electron–electron correlation is intro- duced into the LDA based calculations via the Lam–Platz- mann correction[3], a better agreement between theory and experiment is achieved but the discrepancies are still much larger than the experimental error. Most recently calcula- tions of Kubo[4] using a GW scheme exhibit a much better agreement with experiment. Another approach to overcome the discrepancies between experiment and LDA calculation has been tried by Dugdale and Jarlborg [5]. These authors simulated thermal disorder in Li by introducing a static disorder into a supercell approach within a self consistent LMTO scheme resulting in a better agreement between theory and experiment. They claim that thermal disorder leads to increasing delocalization of the Compton profiles. Indeed, the influence of temperature on the momentum space density of the valence electrons in LDA theory is usually neglected, even though thermal disorder in the case of Li should be significant considering the low melting point of 456 K and the low Debye temperature of 344 K. Therefore it seems to be worthwhile to perform an experi- mental test of the theoretical statement of Ref. [5] by measuring the Compton profile of Li at different tempera- tures. We performed high resolution Compton profile measure- ments of single crystalline Li for momentum transfer ~ q k 110 with 0.12 a.u. momentum space resolution, an inci- dent energy of 30 keV and a scattering angle of 173° at the Compton spectrometer of ID 15 B at the ESRF. The sample was measured at room temperature and at 95 K, which is above the martensitic transition of Li [6]. For both profiles 2:8 × 10 5 counts were collected at the Compton peak. The measurement at room temperature was repeated after the 95 K profile was taken to prove the reversibility of the temperature effect, where 1:3 × 10 5 counts were collected Journal of Physics and Chemistry of Solids 61 (2000) 379–382 0022-3697/00/$ - see front matter  2000 Elsevier Science Ltd. All rights reserved. PII: S0022-3697(99)00321-2 www.elsevier.nl/locate/jpcs * Corresponding author. Tel.: +49-231-755-3523; fax: +49-231- 755-3657. E-mail address: sternemann@physik.uni-dortmund.de (C. Sternemann).