Nondestructive quantification of single crystal elasticity in additively manufactured polycrystals

Resonant ultrasound spectroscopy (RUS) is capable of determining the single crystal elastic constants from polycrystalline specimens with known crystallographic texture. However, the calculated single crystal elastic constants vary with the measured texture, resulting in inconsistent estimates for additively manufactured (AM) specimens with heterogenous texture regions. In this work, the accuracy of the determined single crystal elastic constants is improved by incorporating the uncertainty of the texture in the determination of single crystal elasticity and requiring only small quantities of electron backscatter diffraction data (EBSD) to do so. The single crystal elastic constants are determined by Bayesian inference with parallelized sequential Monte Carlo (SMC), enabling an order of magnitude reduction in computational cost. AM specimens of a cobalt-nickel-base superalloy (SB-CoNi-10C) demonstrate that the incorporation of texture variability enables the single crystal elastic constants to converge to the reported literature values within one standard deviation, avoiding any dependence on the initial texture values. The single crystal elastic constants of nickel-base-superalloy Inconel 625 (IN625) and Ti-6Al-4V (Ti64) are determined from AM specimens, using only RUS and EBSD data. The determined single crystal elastic constants of IN625 agree between two different texture conditions (induced by AM raster strategy), as well as with the literature values, within one standard deviation. The single crystal elastic constants determined from three AM Ti64 specimens, printed with different beam powers, agree with the range of literature values within two standard deviations but demonstrate variability between AM specimens, indicating that the frequencies may be susceptible to the effects of secondary phases.

共振超声光谱法（Resonant ultrasound spectroscopy, RUS）可通过已知晶体学织构的多晶试样测定单晶弹性常数。然而，计算得到的单晶弹性常数会随实测织构发生变化，导致针对存在异质织构区域的增材制造（additively manufactured, AM）试样的估算结果不一致。本研究通过在单晶弹性常数测定过程中引入织构的不确定性，且仅需少量电子背散射衍射（electron backscatter diffraction, EBSD）数据，提升了单晶弹性常数测定的准确性。本次研究采用并行化序列蒙特卡洛（parallelized sequential Monte Carlo, SMC）结合贝叶斯推断来确定单晶弹性常数，将计算成本降低了一个数量级。针对钴镍基高温合金（SB-CoNi-10C）的增材制造试样的验证结果表明，引入织构变异性可使单晶弹性常数收敛至文献报道值的一个标准偏差范围内，且无需依赖初始织构数值。本研究仅通过RUS与EBSD数据，从增材制造试样中测定了镍基高温合金因科内尔625（Inconel 625, IN625）与Ti-6Al-4V（Ti64）的单晶弹性常数。所测定的IN625单晶弹性常数在两种不同织构条件（由增材制造扫描策略诱导产生）下，以及与文献报道值之间，均在一个标准偏差范围内保持一致。针对三种采用不同束流功率打印的增材制造Ti64试样所测定的单晶弹性常数，与文献值范围的偏差在两个标准偏差以内，但不同增材制造试样间存在一定差异，这表明共振频率可能易受第二相的影响。