Wang et al. previously annealed MgO in the TEM to ~1500 K, observed a phase transformation at the surface, and identified the phase as MgO 2 [1]. They primarily studied bulk, cleaved {100} surfaces in the reflection geometry (REM, RHEED, and REELS), but also studied thin cleavage fragments by transmission electron microscopy (TEM), selected area diffraction (SAD), and transmission electron energy-loss spectroscopy (EELS). In light of the extensive use of annealed single crystal MgO as a substrate for epitactic growth, numerous previous studies of MgO surfaces, and especially the seemingly unlikely formation of MgO 2 (stable only at high oxygen partial pressures) in the vacuum of the TEM, additional in-situ annealing studies have been performed. Single-crystal {001} MgO TEM specimens, prepared following an involved and proven protocol [2], were annealed in situ with the use of a Gatan model 628 single-tilt heating holder. Most anneals employed a Philips CM30; additional anneals used an oil-free, dry-pumped Philips CM200-FEG. Specimens were heated in situ to ~1500 K at ~1 x 10 -7 Torr, held at that temperature for a few minutes, and cooled. Other than preliminary experiments to determine appropriate conditions, in-situ anneals were performed with specimens kept out of the electron beam. Comparison specimens were annealed in a vacuum furnace at 10 -5 to 10 -6 Torr, but no second phase formed.In figure 1, the weak-beam image shows a contamination-free, step-terrace structure, the SAD pattern shows only expected MgO reflections, and the EELS shows only O and Mg edges, quantified with Gatan EL/P software to yield Mg/O = 1.0 ± 0.1. Following in-situ annealing (figure 2) the bright-field image shows ~100-nm grains on the specimen surfaces, the SAD pattern has the same features as in Wang et al. [1] (extra reflections at 3/2{100} (type 1) and {110} (type2) plus double diffraction), and EELS shows Si, possibly Al, and lower Mg relative to O. Quantification yields Mg/O = 0.50 ± 0.05 and Si/Mg = 0.49 ± 0.05, consistent with forsterite Mg 2 SiO 4 . Upon additional annealing in air at 1623 K for 15 min the polycrystalline film dewets the surface and forms small islands with moiré fringes visible on some particles [3]. Large-area SAD patterns exhibit the same features as after in-situ annealing, except 4-fold spots around MgO reflections (possibly <111> relrods from defects on {111}) are no longer present. Tilting experiments (on carefully selected individual particles) around 13 -0 f (type 1)//200 M reflections to [312] (10°) and [311] (25°) forsterite zones and around 002 f (type 2)//220 M reflections to [210] (5°), [310] (16.5°) and [410] (24°) forsterite zones, determined the orientation relationship as (100) f //(1 -1 -1) M , (010) f //(2 -11 -) M and (001) f //(011) M (12 variants), where subscripts f and M refer to forsterite and MgO, respectively [3]. Notable d-spacing matches are d 400 forsterite ~ d 222 MgO and d 004 forsterite ~ d 220 MgO.Possible sources of Si contamination were considered. The SHaRE CM30 always use...