1. Introduction
Metal 3D printing technology, as an additive manufacturing technology, has the advantages of high deposition efficiency, low manufacturing cost, and a high degree of design freedom, lightweight design and functional integration over the conventional formative or subtractive manufacturing [
1,
2,
3,
4]. It shows broad prospects for technological industrialization in the fields of medical, automotive, nuclear power, and aerospace [
4,
5]. Laser Melting Deposition (LMD) technology, as one of the metal 3D printing technologies, has attracted researchers’ attention. Based on different forms of raw materials, there are two LMD forming processes: powder feed and wire feed, which have different advantages and applications. The LMD with powder feed gives better resolution and surface finish and it has the advantages of high matrix bonding strength, and uniform and dense microstructure. While the LMD with wire feed is considered to be a promising method for forming large components owing to its high deposition rate, high material utilization rate, and relatively low costs of manufacturing and equipment [
6,
7,
8,
9,
10]. During the LMD forming process, the sample will experience transient heating and melting at a point heat source followed by rapid cooling, resulting in a large temperature gradient between the melt pool and the substrate. For near-α titanium alloys, which are widely used in aircraft engine, the large temperature gradient can lead to the appearance of large microtextured clusters (MTC) running through several cladding layers, which are composed of fine α-phase crystallites with preferred crystallographic orientations [
11,
12]. Studies have shown that the presence of large MTCs can result in poor performance and significant anisotropy of the components, and even lead to early failure due to dwell fatigue [
13,
14,
15]. Since the large MTCs have remarkable effects on the mechanical and fatigue properties, it is crucial to evaluate the mean size of MTCs for optimizing the mechanical and fatigue properties of laser melting deposited components [
16,
17]. LMD is considered as a one-step direct forming technology, the destructive methods for grain size evaluation, such as optical metallography and electron backscatter diffraction (EBSD), can no longer meet the detection requirement, and thus the non-destructive evaluation of microstructural properties is particularly important. Among many non-destructive testing technologies, the ultrasonic testing technology has shown great potentials due to its high sensitivity, strong penetrability, and wide applicability for materials. In particular, laser ultrasonic technology has the capability of non-contact and on-line real-time inspection at high temperatures or on moving surfaces. It shows great potentials in both off-line and on-line characterization of material microstructures.
When an ultrasonic wave propagates through polycrystalline materials, the elastic properties and density inhomogeneity of individual crystallites result in the variation of velocities of ultrasonic waves in each crystallite, causing scattering of ultrasonic waves at grain boundaries and resulting in the variation of ultrasonic propagation properties, such as the propagation velocity, the amplitude attenuation and the scattered noises. It is believed that information on microstructure features, such as grain size, crystallographic and morphological textures, can be obtained by inversion of ultrasonic wave properties. Correlation of ultrasonic wave properties to microstructural features is a fundamental research for the non-destructive evaluation of microstructural evolution.
Ultrasonic velocity is often used to evaluate the mean grain size since its measurement is relatively simple and straightforward. Hirsekorn et al. [
18,
19] proposed a theoretical model for planar ultrasonic longitudinal and transverse wave velocities as a function of grain size in polycrystalline materials. Ünal et al. [
20] measured ultrasonic wave velocity in boron carbide (B
4C)-aluminium (Al) and B
4C-nickel (Ni) composites using a piezoelectric ultrasonic transducer and found an approximately linear relationship between the ultrasonic velocity and the grain size. However, studies have shown that this method has less sensitivity to some metal materials, causing a relatively large error of grain size evaluation [
21].
Recently, more and more researchers focus on the spectral analysis, and use the ultrasonic features, such as the attenuation and the backscattered noises, to evaluate grain size of polycrystalline materials. Özkan et al. [
22] compared ultrasonic velocity, ultrasonic attenuation, the rate of peak heights for grain size characterization of B
4C-Al composite materials. Results showed a higher correlation between ultrasonic attenuation and grain size variation. Furthermore, theoretical studies [
23,
24] have showed that ultrasonic attenuation has an exponential relationship with grain size, and a large number of experiments [
25,
26,
27,
28,
29,
30] have been performed for grain size characterization by inversion of ultrasonic attenuation in both weakly [
25] and strongly [
26,
27,
28,
29,
30] scattering materials. Nevertheless, the research on ultrasonic attenuation is usually restricted by signal-to-noise ratio (SNR) for strongly scattering materials. Thus, research efforts have also been dedicated to backscattered noise signals. Theoretical models for ultrasonic backscattering coefficient in both single-phase [
31] and duplex [
32] microstructures have been proposed based on the isolated scatterer model, i.e., the single-scattering assumption, and effects of grain size and grain shape on ultrasonic backscattering have been investigated [
15,
33]. However, the aforementioned methods are still amplitude-based approaches. Since the amplitude measurement accuracy of ultrasonic signals is closely related to the stability of the ultrasonic inspection system, it can largely affect the accuracy of grain size evaluation [
34]. Besides, the accuracy of the ultrasonic attenuation method is also influenced by the attenuation sources from absorption, geometric spreading and dislocation in materials [
35]. The backscattering coefficient tends to be stable with the increase of grain size at a given frequency, so the range of grain sizes evaluated by the backscattering method is relatively narrow [
33,
36].
In view of the limitations of the aforementioned methods, emphasis has been placed on research about grain size effects on ultrasonic centroid frequency shift. Studies have showed that high-frequency components of ultrasonic signals experience greater attenuation than low-frequency components, which results in a downshift in the centroid frequency of the ultrasonic pulse echo [
37]. Moreover, the centroid frequency downshift is more significant as the grain size increases. The centroid frequency shift method is insensitive to the geometric spreading, wave reflection and transmission effects, and can effectively avoid the error caused by the instability of laser-ultrasonic signal amplitude measurement. Thus this method can evaluate the grain size more accurately [
38,
39].
There have been some advances in ultrasonic characterization for grain size of 3D printed metals or composites by use of the above-mentioned methods. Traditional ultrasonic techniques are first used since the detection signals have a good SNR. Yang et al. [
40] found that the longitudinal wave velocity along the deposition direction was about 2% lower than the velocity perpendicular to the deposition direction using immersion ultrasonic testing method, thus verifying the anisotropy of additively manufactured TC18. Laser ultrasonics, as a new technology capable of online real-time non-contact detection at high temperatures, has been demonstrated to be promising for inspection of internal defects of additively manufactured components [
41,
42]. Nevertheless, research on its application to the microstructure characterization is still insufficient. Ma et al. [
43] evaluated the grain size of 3D printed titanium alloy using laser ultrasonic technology, and results showed that the ultrasonic attenuation exhibited a linear relationship with the mean grain size, but the applicable range of material and frequency of this linear relationship model still need to be confirmed by theoretical derivation and extensive experimental studies. In this paper, the laser ultrasonic technology is used for detecting Ti6Al4V/B
4C composite specimens manufactured by the LMD of powder feed. Theoretical models for scattering-induced attenuation and centroid frequency downshift of ultrasonic waves propagating in the studied composite materials are presented. Effects of microstructural properties on laser-generated ultrasonic wave characteristics, such as the ultrasonic velocity and the spectral centroid frequency shift of ultrasonic echoes are analyzed. This would lay a research foundation for the non-destructive evaluation of grain size in 3D printed metal composite materials.
This paper proceeds as follows:
Section 2 is devoted to methods for sample preparation and ultrasonic experiment. In
Section 3, theoretical models for scattering-induced attenuation and centroid frequency downshift of ultrasonic waves are presented. Then, the correlation of laser-ultrasound to microstructural properties of laser melting deposited Ti6Al4V/B
4C composites is described in
Section 4. Finally, main conclusions are summarized in
Section 5.