Monochromatic x-ray diffraction data collected in-situ within the diamond anvil cell show perovskite structured Neighborite (NaMgF3) transforms to the CaIrO3-type post-perovskite structure between 28 and 30 GPa. Upon laser heating, the CaIrO3-type structure transforms further to an unknown structure (Pnnm, designated N-phase). Upon pressure release, N-phase NaMgF3 becomes x-ray amorphous. A structure transformation in post-perovskite MgSiO3 and MgGeO3 to N-phase may account for previous observations of extra x-ray reflections during high pressure experiments and tomographic observations of an additional boundary in the lower mantle below the D″ discontinuity.
 Recently, a post-perovskite structure of MgSiO3 has been observed at pressures in excess of 120 GPa and 2500 K, conditions thought to coincide with the onset of the D″ layer [Murakami et al., 2004; Oganov and Ono, 2004]. The post-perovskite structure is orthorhombic (Cmcm) and rarely adopted among oxides and sulfides (Figure 1). Having significant axial anisotropy, this layered structure may provide a suitable explanation for both the sharp horizontal discontinuity marking the onset of D″ and the zones of strong anisotropy observed within the layer [Garnero, 2004; Murakami et al., 2004; Oganov and Ono, 2004; Tsuchiya et al., 2004].
 Neighborite (NaMgF3) [Chao et al., 1961] is a well studied analog for MgSiO3 perovskite, helpful in demonstrating the effects of pressure and temperature as well as cation substitution on the perovskite structure [Liu et al., 2005; Martin et al., 2005; Zhao et al., 1993]. NaMgF3 perovskite has a bulk modulus less than one third that of MgSiO3 and nearly half that of MgGeO3 perovskite. Thus NaMgF3 may require a lower pressure to stabilize post-perovskite, and thereby facilitate Rietveld structural modeling.
 Previous work [Liu et al., 2005] finds evidence for a post-perovskite phase transition in NaMgF3 at just 19 GPa, however the data quality is not sufficient for further analysis. Later, enthalpy calculations [Parise et al., 2004] support the work of Liu et al. , finding NaMgF3 should transform to the CaIrO3 structure around 35 GPa, before dissociating to constituent fluorides at even higher pressures and temperatures.
 In the current study, we report results of x-ray diffraction as we observe high pressure structural phase transformation of NaMgF3 perovskite to post-perovskite. Utilizing laser heating within the diamond anvil cell we overcome kinetic transition barriers and observe the formation of an unknown post-CaIrO3-type phase of NaMgF3 (designated N-phase).
 Polycrystalline samples of perovskite analog NaMgF3, synthesized using conventional solid state techniques [Zhao et al., 1993], were ground and loaded in diamond anvil cells (DACs) fitted with 350 μm culets. Anvil gaskets (Tungsten or stainless steel) were pre-indented to 80 μm thickness. The transparent sample was mixed with a dark material (platinum or graphite) to absorb the laser radiation before loading into the DAC. Samples were sandwiched between layers of insulating material (NaCl or MgO). By conducting multiple experiments with several different combinations of gasket, laser absorber, and insulator material, we distinguish possible chemical reactions from structural transitions. The equation of state of NaCl [Decker, 1971], or MgO [Speziale et al., 2001] with Pt [Holmes et al., 1989] was used as an internal pressure marker.
 Data were collected using monochromatic radiation at several synchrotron beamlines; at GeoSoilEnviroCARS (GSECARS) 13-ID-D with λ = 0.3344(2) Å, and at the High Pressure Collaborative Access Team (HPCAT) 16-ID-B with λ = 0.4018(2) Å [Shen et al., 2005], both at the Advanced Photon Source (APS). At the European Synchrotron Radiation Facility (ESRF) data were collected at ID-27 [Mezouar et al., 2005] with λ = 0.3738(2) Å. Data at HPCAT and ESRF were collected by a MAR345 imaging plate, while that at GSECARS was collected by a Bruker CCD. Raw 2-D data were integrated to 1-D powder patterns suitable for Rietveld structure refinement using program FIT2D [Hammersley et al., 1996].
 Double-sided infrared laser heating was performed in-situ during each experiment [Schultz et al., 2005]. X-ray diffraction data were collected before during and after heating to temperatures of 2000 K ± 200 K [Benedetti and Loubeyre, 2004].
 Our pressure-volume data for NaMgF3 perovskite are consistent with previous estimates of zero pressure bulk modulus [Liu et al., 2005; Zhao et al., 1994], where KT = 76.5 GPa. Fitting pressure-volume data of the post-perovskite phase to a 2nd order Birch-Murnaghan equation of state yields a zero pressure bulk modulus of 137(18) GPa. The significant error in this value stems from uncertainty in the volume of the unit cell at 1 bar, 200(3) Å3.
 Contributing to error in unit cell axis measurements of post-perovskite is the presence of an unknown phase (N-phase) which appears upon laser heating in the post-perovskite region (Figure 3). This phase transition is reproducible and observed in 3 different sample loadings with different pressure markers, laser absorber, and gasket from data collected at both GSECARS and ESRF. The new peaks are not those expected of both MgF2 (cotunnite) [Haines et al., 2001] and NaF (B2) as dissociation of NaMgF3 would imply [Parise et al., 2004]. Solutions derived from indexing the new reflections favor an orthorhombic unit cell with 6 formula units (Z). The Inorganic Crystal Structure Database (ICSD) contains no orthorhombic ABX3 structures with Z = 6, and few with Z = 12. Considering monoclinic and triclinic structure entries with these criteria are scarce in the ICSD, in addition to the quality of our Le Bail refinement (Pnnm, Figure 3), we believe the structure of N-phase NaMgF3 (a = 8.353(3) Å, b = 5.265(2), c = 5.857(3), V = 257.58(3) Å3 at 37(1) GPa) may be new and need not belong space group Pnnm necessarily. The density difference between N-phase and CaIrO3-type NaMgF3 is smaller than that between perovskite and post-perovskite, about 1%, yet a value is difficult to derive since relaxing stress in post-perovskite requires heating, driving the phase transition to N-phase. Upon decompression to the perovskite field, diffraction peaks broaden suggesting the onset of amorphization of N-phase at low pressure.
 It is possible the topology of N-phase NaMgF3 is also layered, containing cations in coordination schemes other than that existing in the CaIrO3-type structure. While more work is necessary to identify a structure for N-phase NaMgF3, we can gain insight to possible high pressure structures through comparative analysis of other high pressure fluoride phases. While sodium in post-perovskite NaMgF3 is contained within a tri-capped trigonal prism, at 38 GPa MgF2 (Sellaite) is reported to form a structure based on mineral cotunnite [Haines et al., 2001], where magnesium also resides in a tri-capped trigonal prism. Thus, N-phase NaMgF3, or further high pressure forms of NaMgF3, could contain both sodium and magnesium in these 9-fold coordination units, and perhaps adopt structures [Caracas and Cohen, 2005] expected for A2X3 compounds as opposed to dissociation [Parise et al., 2004; Umemoto et al., 2006].
 A transformation in CaIrO3-type NaMgF3 to N-phase has possible implications for MgSiO3 and the lower mantle. A post-post-perovskite phase such as N-phase may account for previous observations of extra x-ray reflections during studies of MgSiO3 and MgGeO3 post-perovskite [Guignot et al., 2005] and several recent studies of lower mantle tomography [Flores and Lay, 2005; Hernlund et al., 2005; Thomas, 2005] suggest a seismic discontinuity below the D″. Recent rationale invokes a double-crossing of the post-perovskite phase boundary by the geotherm at two different depths. NaMgF3 has proven to be a satisfactory analogue material for MgSiO3 thus, solid-solid transformation in silicate post-perovskite via N-phase may be possible within the lower mantle considering geotherms [Boehler, 2000; Zerr et al., 1998] above the core-mantle boundary as well as within rocky interiors of extraterrestrial bodies.
 This work was supported by grant NSF-EAR-0510501 to JBP and we acknowledge the ESRF for provision of beamtime to proposal number HS-2780 at ID-27. Portions of this work were performed at GeoSoilEnviroCARS (Sector 13) as well as HPCAT (Sector 16), Advanced Photon Source (APS), Argonne National Laboratory. GeoSoilEnviroCARS is supported by the National Science Foundation - Earth Sciences (EAR-0217473), Department of Energy - Geosciences (DE-FG02-94ER14466) and the State of Illinois. Use of the HPCAT facility was supported by DOE-BES, DOE-NNSA (CDAC), NSF, DOD –TACOM, and the W.M. Keck Foundation. Use of the APS was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract W-31-109-ENG-38. The thoughtful comments and suggestions of two anonymous reviewers strengthened this work.