Noncolinear spin polarization from frustrated antiferromagnetism: A possible scenario for molecular oxygen at high pressure R. Gebauer CECAMCentre Europe ´en de Calcul Atomique et Mole ´culaire, ENSL, 46 Alle ´e d’Italie, F-69364 Lyon Cedex 07, France S. Serra, G. L. Chiarotti, and S. Scandolo INFMIstituto Nazionale di Fisica della Materia and SISSAScuola Internazionale Superiore di Studi Avanzati, via Beirut 2/4, I-34014 Trieste, Italy S. Baroni CECAMCentre Europe ´en de Calcul Atomique et Mole ´culaire, ENSL, 46 Alle ´e d’Italie, F-69364 Lyon Cedex 07, France and INFMIstituto Nazionale di Fisica della Materia and SISSAScuola Internazionale Superiore di Studi Avanzati, via Beirut 2/4, I-34014 Trieste, Italy E. Tosatti INFMIstituto Nazionale di Fisica della Materia and SISSAScuola Internazionale Superiore di Studi Avanzati, via Beirut 2/4, I-34014 Trieste, Italy and ICTPThe Abdus Salam International Centre for Theoretical Physics, I-34014 Trieste, Italy Received 11 June 1999; revised manuscript received 23 August 1999 We perform density-functional calculations of the magnetic properties of a simplified structure aimed at capturing some of the features of the elusive phase of molecular oxygen at high pressure. Starting with the phase—which is a quasi-two-dimensional distorted triangular arrangement of antiferromagnetically ordered molecules—pressure could decrease the b / a ratio in the basal planes pushing it toward the ideal triangular value of 1/3, thus increasing magnetic frustration. We conjecture that when frustration takes over, the magnetic order may turn into a 120° planar spin-spiral structure inside the phase, until at higher pressures band-overlap metallization suppresses magnetization in the phase. This conjecture is substantiated by calcu- lations that also represent the attempt to apply state-of-the-art pseudopotential techniques to the magnetic properties of a frustrated antiferromagnet. I. INTRODUCTION Most magnetic materials are characterized by atomic mo- ments or electronic spins, in an itinerant pictureall aligned, parallel or antiparallel, to the same direction everywhere in space. A number of notable exceptions to this rule exist, in which the direction of the magnetization varies from point to point in space. Such exceptions include, e.g., spin spirals in the lanthanides and the complex structures occurring in to- pologically frustrated antiferromagnets. Density-functional theory DFTcalculations of noncolin- ear magnetic structures have been available for more than a decade. 1 Most of these studies, however, rely on some kind of atomic-sphere approximation ASAin which different spin quantization axes are chosen within different spheres. The stable magnetic structure is then determined a posteriori as the one that minimizes the total energy with respect to the directions of the quantization axes chosen as inputs of the calculation. Although spin colinearity may be broken even within individual atoms by, e.g., spin-orbit effects, 2 the con- cept that the same direction of magnetization is associated with each atom is physically well motivated, and it has been recently confirmed by calculations on iron clusters per- formed without requiring ASA. 3 Nevertheless, going beyond the ASA treatment of magnetic noncolinearity is important, because only releasing all prior constraints on the magnetic structure such as, notably, that on the relative orientation of different quantization axes within different atomic spheres can DFT calculations display their full predictive power. In this paper we present a fully unconstrained calculation of the magnetic structure of a topologically frustrated anti- ferromagnet, following an approach whose bases are concep- tually similar to that of Ref. 3. The system we choose to study is a layered triangular arrangement of oxygen mol- ecules aimed at capturing some of the features of the hitherto ill-characterized phase of molecular oxygen at high pres- sure. II. OXYGEN AT HIGH PRESSURE In the gas phase, the ground state of the O 2 molecule is a triplet, as required by Hund’s rule. In solid, and Mott- Hubbard insulating, O 2 , at low temperature and moderate pressure, weak electron hopping gives rise to antiferromag- netic intermolecular superexchange, whose magnitude rises considerably with pressure, 4,5 from the zero-pressure value of 5 meV Ref. 6to hundreds of meV at tens of GPa. 7 The relevant pre-1990 work on high-pressure phases and magne- tism of oxygen is reviewed by Freiman. 8 Antiferromagnetic order is realized in the insulating low-temperature crystalline phase -O 2 , 9 which is stable up to 1 GPa. With tempera- ture this converts to -O 2 , which is magnetically disordered. PHYSICAL REVIEW B 1 MARCH 2000-I VOLUME 61, NUMBER 9 PRB 61 0163-1829/2000/619/61455/$15.00 6145 ©2000 The American Physical Society