Low-energy positron diffraction (LEPD) is used in conjunction with low-energy electron diffraction (LEED) to determine the relaxed atomic geometries of the CdSe cleavage surfaces. The LEPD analyses yield optimal fits at smaller top-layer perpendicular relaxations than LEED for both cleavage faces, and significantly better agreement between theoretical and experimental intensity profiles.During the past decade, low-energy electron diffraction (LEED) has become a powerful tool for the structural determination of ordered surfaces. ! " 3 In spite of the success of LEED in determining the surface structure of compound semiconductors, the agreement between observed diffracted-beam profiles I ex (V) and calculated profiles Ith(V) has not been as good as for metal and adsorbate-on-metal surfaces. Although the reason for this is not yet fully understood, the higher degree of complexity due to relaxations extending below the outermost one or two atomic layers is commonly regarded as a major factor. Because of the large differences between diffracted-beam intensity profiles observed via LEED and low-energy positron diffraction (LEPD), 4 " 6 comparison of LEED and LEPD structural determinations of a compound semiconductor surface would provide additional insight into this issue and could yield a more accurate structural determination than one using LEED alone. Accordingly, we have carried out LEPD and LEED structural determinations of the relaxed atomic geometries of the recently determined 7 (lOTO) and previously undetermined (1120) cleavage faces of CdSe. We find that both the LEPD and LEED results are consistent with the relaxation predicted by Wang and Duke. 8 In addition, the agreement between I ex (V) and Ith (V) for LEPD is, surprisingly, observed to be significantly better than for LEED. For both cleavage faces, we also find smaller top-layer perpendicular relaxations for the LEPD structure analyses than for the LEED analyses, although the differences may be viewed as lying barely within the uncertainty of the analyses. We have established that the LEPD-LEED differences are not experimental artifacts by obtaining both LEPD and LEED data from the same sample surfaces in the same apparatus at Brandeis University. The validity of the Brandeis LEED results was then checked by comparison with LEED data and structural analyses carried out at Princeton University.The LEPD intensities were obtained with the brightness-enhanced, slow positron beam at Brandeis. 9 A 230-mCi 58 Co positron source produced a final positron beam of 5000 s _1 , monochromatic within 0.2 eV and having a phase space of 1 mmdeg over the energy range 20-200 eV. The beam optics also produced a 1-mm-deg electron beam, allowing the collection of LEED intensities from the same sample surfaces as the LEPD. The normal-incidence diffractometer incorporated a channel electron multiplier array and resistive anode encoder. Details of the diffractometer and beam design are given elsewhere. 1011 The CdSe(lOlO) and CdSe(1120) samples were high-purity, Cd-rich, lo...