高周疲劳一般定义为5×10⁵至10⁷周次的疲劳阶段，而超高周疲劳是指材料经受10⁷周次以上的循环载荷而发生疲劳损伤断裂的过程。在高周疲劳阶段，裂纹往往从材料表面起源，至超高周疲劳阶段，裂纹主要从试样内部萌生。裂纹内部萌生区作为超高周疲劳的特征区域，消耗了总疲劳寿命的绝大部分。在超高周疲劳阶段，研究材料内部萌生区的裂纹扩展行为和微结构特征有助于理解裂纹萌生的形成机制。本文通过变幅加载（VA，Variable amplitude）方法研究了高强钢内部萌生区裂纹扩展行为，利用聚焦离子束（FIB，Focused ion beam）方法、扫描电镜（SEM，Scanning electron microscope）观测、透射电镜（TEM，Transmission electron microscope）观测和选区电子衍射（SAD，Selected area electron diffraction）检测等手段表征了高强钢在VA加载下的超高周疲劳裂纹萌生区微结构特征。此外，由于超声疲劳测试技术可以大大缩短实验时间，其在高周与超高周疲劳领域的应用广泛。但加载频率的提高会对材料的疲劳性能产生影响，即加载频率效应。目前，关于加载频率效应的产生机理还未有统一的解释。因此，本文还从应变率和温度的角度研究了加载频率对高强钢高周和超高周疲劳行为的影响。

在超高周疲劳萌生区裂纹扩展行为方面，本文开展了GCr15材料在旋转弯曲疲劳试验机（RB，Rotating bending）和电磁共振疲劳试验机（ER，Electro-magnetic resonance）下的恒幅（CA，Constant amplitude）和变幅加载实验。通过SEM仔细观测了试样的裂纹萌生区。结果表明，VA加载下材料的总损伤主要由高应力主导。通过精细控制加载参数，在RB方法VA加载下试样疲劳断口的裂纹萌生区（FGA区，Fine-granular-area）内首次捕获到了裂纹扩展痕迹，进而计算出了FGA区域内的裂纹扩展速率，这些速率和文献中萌生区外的鱼眼区获得的数据有很好的线性汇合趋势，同时这些数据和整个FGA区的等效裂纹扩展速率相吻合。另外，通过对FGA区内不同区域的应力强度因子和裂尖塑性区尺寸的计算以及裂纹面的往复挤压作用分析了裂纹扩展痕迹的形成可能性。

对VA加载下RB和ER方法试样FGA区域的微结构进行了TEM观测，剖面样品的观测结果显示裂纹萌生区表层为纳米晶层，且萌生区有裂纹扩展痕迹和没有裂纹扩展痕迹出现位置的微结构特征无明显差别，其FGA区域皆为纳米晶层，进一步支持了我们组提出的大数往复挤压模型。实验结果表明VA加载和CA加载的情况相同，只要满足原始裂纹面挤压的压应力和裂纹面的相互接触有足够多的周次两个条件即可形成纳米晶层。

在加载频率对高强钢的疲劳强度和寿命的影响方面，本文对三种热处理状态的GCr15试样（T.T. 150°C：150°C回火，T.T. 200°C：200°C回火，T.T. 400°C：400°C回火）分别在RB（f = 52.5 Hz）、ER（f = 120 Hz）和超声疲劳机（UL，Ultrasonic loading with cooling：超声冷却，f = 20 kHz；UL-NC，Ultrasonic loading without cooling：超声不冷却，f = 20 kHz）下进行了CA加载疲劳实验。实验结果表明，加载频率对材料的疲劳强度和寿命有明显影响。对于T.T. 150°C和T.T. 200°C试样，RB下测得的疲劳强度明显比其余三组（ER, UL, UL-NC）测得的疲劳强度高。应力较低时UL测得的疲劳强度略大于ER，随加载应力提高，在高周疲劳区域，UL测得的的疲劳强度略低于或相当于ER。UL-NC测得的的疲劳强度明显低于UL和ER。对于T.T. 400°C试样，RB、ER和UL-NC三种加载下测得的疲劳强度大致相同，而UL下测得的疲劳强度要明显高于这三种加载情况的疲劳强度。通过一种引入控制体积的统计分析方法运用RB下的数据评估了大控制体积下ER的数据，使RB和ER的结果相对应，即协同了不同加载方式导致的影响。采用SEM对断口观测表明，加载频率并不改变材料的失效模式。

本文从应变率、温度的角度解释了加载频率的形成机理。实验结果显示，加载应变率的提高会增加材料的强度，而加载频率导致的温升会降低材料的强度。对于UL-NC试样，三种材料状态的温升都相对较高，对材料强度有明显影响。对于UL试样，在疲劳加载应力相对较低时，三种材料状态在超声频率即高应变率下的温升都不明显，温度对材料的影响较小。由温升数据外推结果可知，在疲劳加载应力相对较高时，T.T. 150°C和T.T. 200°C试样在超声频率即高应变率下的温升明显，这显著降低材料的强度；而T.T. 400°C试样的温升依然较低，对材料强度影响较小。应变率强化效应和热软化效应的共同作用造成了不同形式的加载频率效应。进而提出了频率效应的判据参数η：当η>1时，出现I型频率效应；当η<1时，出现II型频率效应；当η=1时，出现III型频率效应。计算结果显示三种材料状态在低应力（700 MPa或600 MPa）下的η值皆大于1，说明在相应加载条件下UL的材料强度要高于ER，进而出现I型频率效应，这和S-N结果相一致。在相同应力下UL和UL-NC的η值之比大于1，表明在相应加载条件下UL的材料强度高于UL-NC，也和实验结果吻合，进一步说明了η判据进行频率效应分析的可行性。

        High cycle fatigue (HCF) is known as the fatigue regime with 5×10⁵ to 10⁷ cycles of loading, and very-high-cycle fatigue (VHCF) is the phenomenon of fatigue damage and failure subjected to over 10⁷ loading cycles. In HCF regime, cracks always initiate from the surface of specimen for metallic materials, and in VHCF regime, cracks initiate commonly from the interior of specimen. As the characteristic zone in VHCF, the interior crack initiation region consumes almost all the fatigue life. The investigation of crack growth behavior in the initiation region and the related microstructure feature is helpful for understanding the formation mechanism of crack initiation in VHCF. In this dissertation, the crack growth behavior in initiation region was investigated by means of variable amplitude (VA) cycling for a high-strength steel, and the microstructure of crack initiation region under VA cycling in VHCF regime was characterized with the methods of focused ion beam (FIB) technique, scanning electron microscopy (SEM) and transmission electron microscopy (TEM) with selected area electron diffraction (SAD). In addition, the ultrasonic frequency technique is routinely used in HCF and VHCF testing due to its high testing efficiency. But the increasing of loading frequency may influence the fatigue property of materials, namely loading frequency effect. So far, no unified explanation is proposed about the mechanism of loading frequency effect. Hence, the effect of loading frequency on HCF and VHCF behavior of a high-strength steel was also investigated from the view of strain rate and temperature.

In this dissertation, the S-N data up to VHCF regime under constant amplitude (CA) and VA for the GCr15 steel were obtained by fatigue tests via rotating bending (RB) and electro-magnetic resonance (ER) cycling to investigate the crack growth behavior in the initiation region. Crack initiation for VHCF was from the interior of specimen and the initiation region was carefully examined by SEM. Crack growth traces in the initiation region of fine-granular-area (FGA) were the first time captured for the specimens under VA cycling by RB method. The obtained crack growth rates in FGA were upwards to connect well with those in fish-eye region outside FGA available in the literature, and were associated well with the calculated equivalent rates of FGA. In addition, the appearance possibility of crack growth traces in FGA was discussed.

The microstructure in FGA region under VA cycling by RB and ER method was systematically examined by TEM. The observations of profile samples revealed that FGA is a nanograin layer, and the microstructure in the initiation region with crack growth traces and without traces did not show a significant difference both being a nanograin layer, which is a new evidence to support the previously proposed numerous cyclic pressing model. The experimental result indicated that the formation of nanograins under VA cycling was the same with that under CA cycling if specimen satisfied the two conditions: (1) the compression loading that results in the contacting between the surfaces of originated crack, and (2) the sufficient number of loading cycles that ensures the enough contacts between the crack surfaces.

HCF and VHCF tests under CA via RB ( f = 52.5 Hz), ER ( f = 120 Hz) and ultrasonic loading (UL: with cooling, f = 20 kHz; UL-NC: without cooling, f = 20 kHz) testing machines were carried out for a high-strength steel GCr15 with three heat treatment conditions (T.T. 150°C, T.T. 200°C, T.T. 400°C: tempered temperature at 150°C, 200°C and 400°C, respectively). It is observed that the fatigue strength and fatigue life of the material were evidently influenced by loading frequency. For specimens T.T. 150°C and T.T. 200°C, the measured fatigue strength under RB was significantly higher than that under ER, UL and UL-NC. A slightly higher measured fatigue strength was presented under UL than that under ER at low stress, whereas the measured fatigue strength under UL was slightly lower than or equal to that under ER at high stress, i.e. in HCF regime. Under UL-NC, it is seen that the measured fatigue strength was lower than that under UL and ER. For specimens T.T. 400°C, it was almost the same for the measured fatigue strength under RB, ER and UL-NC, which were all lower than that under UL. A statistical analysis method in the light of control volume was used to reconcile the effect of loading type, and the predicted data were consistent with the experimental results. Fracture surface observation with SEM indicated that loading frequency does not change the failure mode.

    Furthermore, the present investigation explained the effect of loading frequency from the view of strain rate and its induced temperature. Experimental data indicated that the increase of strain rate raised the strength of material, and the induced temperature rise reduced the strength of material. For UL-NC specimens, the temperature rise of three material conditions was high enough to affect the strength of material. For UL specimens, when the loading stress was relatively low, the temperature rise under ultrasonic frequency, i.e. high strain rate for the three groups of specimens was not obvious, i.e. the effect of temperature was negligible. According to the extrapolated data, when the loading stress was relatively high, the temperature rise under ultrasonic frequency, i.e. high strain rate for specimens T.T. 150°C and T.T. 200°C was high enough to affect the strength of material; while for specimens T.T. 400°C, the high strain rate just induced very little temperature rise and has little effect on the strength of material. The combined response caused by strain rate and the induced temperature rise resulted in the different forms of loading frequency effect. A parameter η was proposed to judge the form of loading frequency effect: η>1 representing type I loading frequency effect; η<1 representing type II loading frequency effect; η=1 representing type III loading frequency effect. The results by calculating the values of parameter η under low stress (700 MPa or 600 MPa) showed that η>1 for three material conditions, indicating that the material strength of UL was higher than that of ER under corresponding loading conditions, which agreed with the experimental data. Under the same stress, the ratio of η value of UL to UL-NC specimens was higher than 1, representing the material strength of UL was higher than that of UL-NC under corresponding loading conditions, which also agreed with the experimental data. All the calculated results showed that the η critetion is reasonable.