The segregation of solute atoms at austenite grain boundaries may be classified into equilibrium and non-equilibrium segregation. The theories on the thermodynamics and isothermal kinetics of equilibrium segregation have been investigated in detail [1-3]. Equilibrium segregation is a thermodynamic process and occurs mainly during isothermal holding of material at a certain temperature and decreases with increasing holding temperature. Up to now investigations of non-equilibrium grain boundary segregation have been sparse. Non-equilibrium segregation is a process of kinetics and occurs during cooling of material from a higher temperature and increases with increasing cooling starting temperature at the same coolng rate and decreases with increasing cooling rate at the same cooling starting temperature [4, 5]. Nowadays most investigations on boron grain boundary segregation are those of accumulating effects of the two types of boron segregation [6], whereas the investigations of purely equilibrium or purely non-equilibrium segregation are sparse. In the present work the purely non-equilibrium segregation of boron at grain boundaries was studied in detail. A B-doped Fe -30 wt % Ni austenitic alloy was chosen as the experimental alloy to investigate the behaviour of boron segregation at austenite grain boundaries. The alloy was prepared by vacuum induction melting and then forged into a cylindrical bar 12 mm in diameter. After the region of B-depletion of the bar was removed by lathing, it was cut into cylindrical discs 8 mm in diameter and 2 mm thick. The alloy composition is given in Table I. All of the sample discs were pre-heat-treated in argon at 1220 °C for 0.5 h to gain the same grain size of approximately 290/xm in diameter in order to eliminate the effect on the level of boron segregation caused by the differences in grain size in the samples. The samples were then cooled at a rate of 600 °C s -1 to room temperature. The samples that were pre-heat-treated were, respectively, heated to 1050, 1150 and 1220°C, again for a duration of 0.5 h, and then cooled to room temperature at rates of 1200, 600 and 50 °C s-~, respectively. In the present work the distribution of boron in the samples was detected by par tMe tracking autoradiography (PTA). Three acetic acid fibre foils were used as detecting foils; their detecting sensitivity to boron atoms is about 2/xm [7]. The integral flux of hot neutron irradiation of 1.3 x 10 ~5 cm -2 was chosen for the work. The detecting foils were etched for 20 min by an aqueous soution of 7.5 M NaOH. The surface of each detecting foil was then coated with chromium and observed by optical microscopy. The distributions of boron in the samples cooled at a rate of 1200 °C s -z from different temperatures are illustrated in Fig. 1. Ueno and Inoui [8] reported that, in the PTA, the cotninuous etched pit belt on the etched foils represents boron segregation, whereas the etched pit clustering represents boride precipitation. According to their study, Fig. 1 shows that no apparent boron segregation and boride precipitation appears in the samples cooled at a rate of 1200 °Cs -z from 1050 and 1150 °C. The distributions of boron for the samples cooled from different temperatures (1050, 1150 and 1220 °C) and at cooling rates of 600 and 50 °Cs -1, respectively, are shown in Figs 2 and 3, respectively