*4.3. VHCF Behavior*

VHCF behavior was investigated at AT using an ultrasonic fatigue system (Figure 3d) at stress amplitudes 180 MPa ≤ σ<sup>a</sup> ≤ 283 MPa, and a load frequency f = 20 kHz, while at T = 300 ◦C a servohydraulic test system was used and stress amplitudes 120 MPa ≤ σ<sup>a</sup> ≤ 190 MPa were applied at a load frequency f = 980 Hz. All VHCF tests were performed on specimens from batch A.

#### 4.3.1. Ambient Temperature

As σ-ε hysteresis measurements are (to date) not realizable during ultrasonic fatigue testing, and due to the macroscopically elastic behavior in the lower HCF/VHCF regime, the cyclic deformation behavior cannot be characterized using conventional data like εa,p, N curves. However, in situ measurement of further physical data, such as changes in the specimen temperature before and after a pulse sequence, dissipated energy and generator power, as well as "quasi in situ" measurement of magnetic properties, can be used to describe the cyclic deformation and transformation behavior of metastable austenites during fatigue testing in the VHCF regime. Figure 10 shows the aforementioned data and, in addition, the displacement amplitude as well as load frequency during an ultrasonic fatigue test with a stress amplitude of 250 MPa. After an approximately stationary phase up to N = 106, cyclic hardening occurred which could be detected by a strong increase in α'-martensite content. As mentioned before, and illustrated in Figure 4, α'-martensite formation promotes transient material behavior. To keep the displacement amplitude close to a constant value of 9.5 μm, corresponding to a stress amplitude of 250 MPa, the PID parameters of the ultrasonic testing system were stepwise readjusted. The cyclic hardening of the material resulted in lower self-heating illustrated by a decreasing change in temperature. This aspect enabled longer pulse and shorter pause times which resulted in a higher effective frequency. As described in Section 3, a stepwise increase of the pulse pause ratio up to 0.72 s/0.8 s, which represents an effective frequency feff = 1650 Hz, led to achieving the ultimate number

of cycles within 25 days. After N = 1 <sup>×</sup> 108 load cycles, a lower <sup>α</sup>'-martensite formation rate dξ/dN occurred and until the limiting number of cycles a saturation state with a stabilized α'-martensite content of ξ = 2.2 FE% was reached. At the same time, temperature as well as displacement amplitude remained constant. In this phase, further adjustments of the displacement control were not necessary. The S-N curve resulting from single step tests is given in Figure 12c. Cyclic plastic deformation of metastable austenite at ambient temperature led to significant changes in phase distribution from single-phase austenitic to two-phase austenite/α´-martensite microstructures. At ambient temperature, and at all load amplitudes σ<sup>a</sup> > 240 MPa, formation of α'-martensite in the range 0.3 FE% ≤ ξ ≤ 2.3 FE% took place. However, at smaller stress amplitudes σ<sup>a</sup> < 240 MPa no deformation induced α'-martensite was measured. Fatigue failure only occurred in the HCF regime and no specimen failed in the VHCF regime beyond N = 107 load cycles. Accordingly, a true fatigue limit exists for metastable austenite [80,81].

**Figure 10.** Displacement amplitude sa, frequency f, temperature change ΔT, power P, dissipated energy Edis and α'-martensite content ξ of batch A during VHCF test at σ<sup>a</sup> = 250 MPa.
