**1. Introduction**

Polymer bonded explosive (PBX) is a class of heterogeneous composite material that mainly consists of explosive crystal particles and polymer binder matrix. PBX is widely used in weapon systems because of its excellent performance. In the course of the service of the weapon systems, because of various environment temperature and mechanical actions, such as compression and tension, there could be a series of damage occurring, structure evolution, and even fracture behaviors in PBX. These damages, structure evolution, and fractures directly affect the mechanical properties, safety performance, and detonation performance of PBX [1,2]. Therefore, it is important to study the response of PBX under thermo-mechanical loading.

Previous researches mainly studied the single-factor effect on PBX. When considering the mechanical loading, the mechanical properties of PBX were widely studied [3,4]. The deformation [5], creep [6], cohesive [7], and fracture [8] behavior of PBX under mechanical loading were investigated

in the digital image correlation (DIC) method. The micro-mechanical evolution was investigated through scanning electron microscopy (SEM). It was found that there was a variety of types of damages in PBX, such as intragranular voids, crystal fractures, interfacial debonding, and deformation twinning [9–11]. The above detection techniques could only obtain the surface morphology, however X-ray micro-computed tomography (μCT) allows for the observation of the internal three-dimensional (3D) structures. The internal deformation of PBX simulant in compression was analyzed in detail by digital volume correlation (DVC) of in-situ μCT [12].

While in the terms of thermal loading effects, state and phase change of both explosives [13] and binders [14] at a wide range of temperatures were studied. The mesoscale structure evolution of PBX during heating was analyzed by ultra-small angle X-ray scattering (USAXS) and μCT [15]. However, mechanical properties and structure evolution of PBX under thermal-mechanical coupling loading still need to be investigated deeply. Because of the technical difficulties of experiments, most of the researches on properties of PBX in thermal-mechanical loading were simulation calculations [16,17]. In respect of experiments, Willamson and others comprehensively studied temperature-time response of an cyclotetramethylenetetranitramine (HMX)-based PBX at a wide temperature range [18]. It was found that the failure strain of PBX is non-sensitive to temperature, so the modules of PBX at different temperatures have a liner relation with failure stress. Other researchers studied the mechanical [19–21] and fracture [22] behavior of PBX under different temperature and loading conditions from different perspectives. The previous researches were mainly focus on the mechanical properties of PBX at high temperatures, while the structure evolution of PBX at both low and high temperatures still needs more investigation. Especially, most of the researches were about HMX-based PBX, while few researches studied 2,4,6-triamino-1,3,5-trinitrobenzene (TATB)-based PBX. Furthermore, quantitative characterization techniques should be built up to describe and analyze the observation result of fracture and structure evolution in mathematical language.

In this paper, in-situ Brazilian test with an improved arc loading head was conducted on aμCT apparatus in order to investigate the interior structure evolution of a TATB-PBX. The test was under quasi-static loading and five different temperatures, ranged from −20 ◦C to 70 ◦C. Three-dimensional morphology of cracks was investigated by digital image process. Fracture degree and complexity were defined and used to quantitatively characterize the crack properties. Fractal dimension was used to characterize the roughness of the crack surface. The test samples were also investigated by SEM, and the results of different kinds of detection and analysis were compared. Slice images of μCT were also analyzed by the displacement of particles, and the displacement field of the interior structure of PBX was analyzed.

#### **2. Materials and Methods**

The experimental specimen is a kind of TATB-based PBX. The main components are TATB as explosive and F2314 fluororubber as binder. TATB and F2314 are firstly made into molding powders, which are particles with some TATB powder crystals that are wrapped in an F2314 binder. The mass percentage of TATB in this PBX is higher than 90%. The specimen was made into a disk with diameter of 10 mm, and a thickness of 3 mm. The experiment took the way of Brazilian disc quasi-static displacement loading, the loading speed was 0.1 mm/min. As a method of indirect tension, Brazilian disc test is widely used to study the damage behavior of brittle materials. Awaji and Sato [23] improved the loading head in diameter disc compression to reduce the stress concentration at the head and analyzed the stress distribution under this type of loading. Pang [24,25] and others found that when the radius of the arc loading head is 1.35 times of the radius of the specimen, the test result is the closest to that of the direct tensile test. In this experiment, the radius of the arc loading head was designed in this method. Because of the small size of the specimen, and in order to ensure the stability of the specimen, a group of fenders were used in experiment. The test loading diagram was shown in Figure 1.

**Figure 1.** The loading method diagram.

The experiment was carried out with the same scan parameters in every group of tests in order to ensure that the scan results are consistent and comparable. The scan voltage was 60 kV, scan current was 150 mA, exposure time was 0.4 s, and with an image merging number of 5. In this condition, the spatial resolution of this experiment was 21.182 μm/voxel. The loading device was an in-situ CT material testing machine, Deben Microtest CT5000-TEC (Deben UK Ltd, London, UK). In every group of test, the specimen was firstly scanned at room temperature without any loading for structure information of original state. Then, the test temperature, −20 ◦C, 0 ◦C, room temperature (22 ◦C), 55 ◦C (the glass transition temperature of F2314 is around 50 ◦C), or 70 ◦C was set, and we waited for 30 min in order to make the specimen temperature stable. Then, the specimen was loaded to fracture, and was scanned under the fracture state. Replicated tests were taken in order to ensure the repeatability and accuracy of the experiment. The specimen was also loaded to 75% fracture extension at different temperatures, which was obtained in the previous tests, and was scanned to ge<sup>t</sup> the structure information of intermediate state of loading at different temperatures. After in-situ CT tests, some of the specimens were also detected by SEM (CamScan Apollo300, CamScan, Cambridgeshire, UK) for more information of more micro-scale structure to compare with the results of CT detection.
