Thermal Fatigue Crack Propagation Process and Mechanism of Multicomponent Al-7Si-0.3Mg Alloy

Abstract

The thermal fatigue behavior of multicomponent Al-7Si-0.3Mg alloys in four different treatment states at typical temperature amplitudes 20 °C→350 °C was studied. The morphology of the second phase particles and crack propagation, and distribution characteristics of dislocations in the thermal fatigue specimens of multicomponent Al-7Si-0.3Mg alloys, were analyzed by optical microscopy (OM), scanning electron microscopy (SEM), energy dispersive X-ray spectroscopy (EDAX spectrum), and transmission electron microscopy (TEM). The influencing factors, the process, and the mechanism of thermal fatigue crack propagation were mainly studied. The results show that under the same temperature amplitude, the thermal fatigue properties and dislocation densities of the new aluminum alloy and the new aluminum alloy under T6 heat treatment are significantly higher than that of the multicomponent Al-7Si-0.3Mg alloy in cast and refined and modified treatment. The crack growth of thermal fatigue specimen depends on three factors: the temperature amplitude, oxidation, and residual stress. The process of thermal fatigue crack propagation mainly experiences crack initiation and the formation of microcracks, but only a few microcracks continue to expand rapidly or preferentially expand into main cracks. The mechanism of thermal fatigue crack propagation is mainly under the action of thermal stress, the crack tip undergoes a cycle of repeated alternation of sharpening → passivation → sharpening, and the crack continues to move forward from its tip intermittently in the way of propagation → stopping → propagation until fracture failure.

1. Introduction

In recent years, with the requirement of energy conservation, emission reduction, and “carbon neutrality”, increasingly stringent regulations on greenhouse gas emissions from transportation have prompted efforts to re-examine the materials used in vehicles [1]. Aluminum alloys are widely used in preparing automobiles and other structural parts due to their high specific strength, excellent casting property, corrosion resistance, and machinability [2,3]. Increasing numbers of automobile engine blocks and cylinder heads are made of all-aluminum alloy. High strength lightweight Al-Si-Mg alloys have become the preferred lightweight materials for heat-resistant automobile components below 400 °C [4,5,6,7,8]. The automotive engines are subjected to mechanical and thermal oscillation loads during service. Thermal fatigue is the most critical cause of failure in such parts [9,10,11]. When the parts are subjected to high temperature thermal cycles, the thermal fatigue process leads to microstructural damage and eventual component failure [12,13]. Therefore, it is very important to study the process and mechanism of thermal fatigue propagation. Thermal fatigue, a common type of damage in engineering structures, can be generated in the form of periodic temperature changes, which often occur as a consequence of the complete or partial limitation of thermal deformation [14]. However, the mismatch in the coefficient of thermal expansion (CTE) of the matrix and the second-phase particles may cause fluctuations at ambient temperature. It can bring about significant thermal stresses, which may lead to localized plastic deformation in matrix close to second phase particles and instability in dimensions and possible failure in mechanical properties [15,16,17]. Nowadays, car manufacturers are developing new technologies to replace steel materials with aluminum materials. There have been numerous studies about the thermal properties of steels [18,19]. The thermal cycling tests mainly focused on specific temperature ranges to evaluate their thermal fatigue behavior. The thermal cycling tests were performed in different ways. Three essential parameters determined the operating conditions of the cycle, which included the maximum temperature, the amount of strain, and the time per cycle. Generally, the difference between these methods was in the way of cooling and heating [20,21]. Their results showed that thermal strain changed slightly throughout the thermal cycle after aging. Moreover, heat treatment led to an increase in matrix yield stress. The distribution of second phase particles was a contributing factor in preventing thermal fatigue crack propagation. Hardness values of alloys decreased with an increase in a number of cycles [22]. Nevertheless, compared with steel materials, the thermal fatigue behavior of aluminum alloy materials is less studied, even if the amount of research and production used is very large to cast Al-Si-Mg aluminum alloys. The process and mechanism of thermal fatigue crack propagation in aluminum alloys have not been studied yet, and there are minimal reports [23,24,25]. At present, the research on casting Al-Si-Mg alloys at home and abroad mainly focuses on three aspects: melting process, casting process optimization, and heat treatment process. A great deal of research has been carried out on the correlation and mechanism of action between the composition, microstructure, and properties of cast Al-Si-Mg alloys [26,27,28,29]. In view of this situation, this article focuses on the study of a multicomponent Al-7Si-0.3Mg alloys used for preparing cylinder bodies, and the thermal fatigue behavior of Al-7Si-0.3Mg alloys reinforced with elements such as Cu, Mn, Ti, as well as the thermal fatigue propagation process and mechanism under heating and cooling cycles from room temperature 20 °C to 350 °C, were studied to fully tap the potential of multicomponent Al-7Si-0.3Mg alloys and lay the foundation for equipping high-performance aluminum alloy engines. It also provides a theoretical basis for improving the thermal fatigue properties of multicomponent Al-7Si-0.3Mg cast alloys and expanding the application of aluminum alloy in automobiles.

2. Materials and Methods

2.1. Materials

The experiments were carried out on cast Al-7Si-0.3Mg, directly refined and modified Al-7Si-0.3Mg, alloyed (with a design composition of 1.80 wt% Cu, 0.30 wt% Mn, and 0.25 wt% Ti by the addition of Al50Cu, Al10Mn, and Al5Ti master alloys to the cast Al-7Si-0.3Mg alloy), refined, and modified (hereinafter referred to as the new aluminum alloy), and the T6 heat treatment of the new aluminum alloy (in the following study, the aluminum alloys with the above four different process states are recorded as A, B, C, and D, respectively, for the convenience of writing).

2.2. Methods

2.2.1. The Preparation of Each of Samples (A, B, C, and D)

The graphite crucibles containing Al-7Si-0.3Mg alloy of equal quality were heated in three KSL-12-JY well resistance furnaces with identical conditions, and the temperature was raised to 750 °C. When the alloys became soft and concave, aluminum and aluminum alloy special covering agents were sprinkled on their surface, and then they were fully melted, degassed, refined, stirred, and slagged. The melt in the first furnace was cooled to 730 °C, slagged, and poured into a metal mold (in the following test, the alloy melts were poured using a metal mold with the same insulation temperature) to obtain A Alloy. In the second furnace, Al-5Ti-1B-1RE master alloy refiner with 0.80% melt mass fraction and Al-10Sr master alloy modifier with 0.30% melt mass fraction were pressed into the graphite bell jar in advance, and then added to the melt in turn, vigorously stirred for 1 min, kept warm for 5 min, refining, underwent slag removal, cooled to 730 °C, underwent slag removal, and poured to obtain B alloy. At the same time, the melt in the third graphite crucible was heated to 780 °C, and a certain quality of Al-10Mn master alloy, Al-50Cu master alloy, and Al-5Ti master alloy were added sequentially under certain process conditions. The melt was cooled to 750 °C, the slag was slagged, and then the refinement and modification process was carried out by adding refiner and modifier with the same mass fraction as the second furnace. It was cooled to 730 °C, slagging, poured and cooled. It was sawn transversely along the same middle position of the sample, half of which was C alloy. The other half of the sample was subjected to T6 heat treatment by a 12 kW model SX2-12-10 (Shanghai Shenguang Instrument Co., Ltd., Shanghai, China) box resistance furnace, followed by a three-stage solid solution treatment: (490 ± 2) °C insulation for 3 h, (510 ± 2) °C insulation for 5 h, (525 ± 2) °C insulation for 6 h, 80 °C water quenching, and aging treatment: (175 ± 2) °C insulation for 6 h aging, air cooling to room temperature to obtain D alloy. This was detected by IRIS ER/S full-spectrum direct-reading inductively coupled plasma emission spectrometer, and its chemical composition is shown in

Table 1. Chemical compositions of unrefined and unmodified alloys (wt. %).

Element	Si	Mg	Cu	Mn	Ti	B	RE	Sr	Fe	Various Trace Elements	Al
A	7.03	0.32	-	-	-	-	-	-	-	≤0.20	Bal.
B	7.03	0.32	-	-	0.042	0.0074	0.0086	0.032	-	≤0.20	Bal.
C	6.92	0.29	1.85	0.33	0.29	0.0076	0.0082	0.034	0.08	≤0.20	Bal.
D	6.92	0.29	1.85	0.33	0.29	0.0076	0.0082	0.034	0.08	≤0.20	Bal.

.

2.2.2. The Tensile Test

The tensile test was carried out on the microcomputer-controlled electronic universal testing machine (model WDW-10, Jinan Hensgrand Instrument Co., Ltd., Jinan, China), and the tensile specimen, as shown in Figure 1, had a total length of 50 mm, wherein the specimen gauge distance before tensile at room temperature was 16 mm, the cross-sectional width was 4 mm, the thickness was 2 mm, the clamping length at both ends of the specimen was 12 mm, and the tensile speed was set to 0.5 mm/min. The two parts of the specimen after the tension are closely docked together, ensuring that the axis of the two parts is located in the same straight line, and we measured the gauge distance after the specimen was broken. In all samples, five tensile specimens were measured per sample and the mean of the tensile was calculated.

2.2.3. Hardness

The Vickers microhardness of the specimens was tested on microhardness tester (model HV-1000, Laizhou Weiyi Experimental Machine Manufacturing Co., Ltd., Laizhou, China), and the test conditions were pressure 4.90 N and load holding time 15 s. Each specimen measured five hardness and we calculated the average hardness.

2.2.4. Impact Toughness

The impact toughness was tested on high and low temperature impact testing machine (model JBGD-300, Yima Optoelec Co., Ltd., Shanxi, China), the impact energy was 300 J, the impact speed was 5.2 m/s, the pendulum pre-elevation angle was 150°, the distance from the pendulum axis to the impact point was 750 mm, the angle between the impact knife was 30°, and the span of the support was 40 mm. In all samples, five impact toughness specimens were measured per sample and the mean of the impact toughness was calculated.

2.2.5. Thermal Fatigue Test

The above four different process treatment aluminum alloys were cut into thermal fatigue specimens with EDM wire, as shown in Figure 2, and thermal fatigue tests were carried out in a hot and cold fatigue testing machine (model LRP1200, Foshan Soontom Ventilation Equipment Co., Ltd., Foshan, China) at a temperature range of 20 °C to 350 °C. The temperature accuracy was ±1 °C. The four aluminum alloy thermal fatigue specimens with different processes were mounted on a movable lever. The lever can transfer the specimens from the furnace to the cooling tank and vice versa. We moved the thermal fatigue specimens up to the middle position inside the heating furnace, closed the furnace door, and adopted timing control, and the jogging speed of the specimens was 10 mm/s. In the first test, in order to evenly heat the thermal fatigue specimen, after reaching the set test temperature of 350 °C and holding for 10 min we pressed the start test button. After the system started counting for 120 s, the automatic control device drove the specimens to enter the 20 °C circulating cooling water tank, with a depth of (15 ± 1) mm of quenching water, a cooling time of 10 s, and the number of runs was 1 cycle. The automatic control device drove the specimens back to the heating furnace for heating, and after each heating of 120 s, the specimen entered the cooling water tank to cool for 10 s, and the number of runs increased by 1 cycle. After that, it was repeatedly heated for 120 s and cooled for 10 s to complete a cooling cycle, and the number of runs increased by 1 cycle. Until the specimens had completed the set number of cycles (1000 cycles) we turned off the corresponding switch button, took out the specimen, weighed it, and recorded the number of cycles. Following that, thermal fatigue specimens were mechanically polished, corroded, and observed. The morphology of the second phase particles, crack morphology, and distribution characteristics of dislocations in the thermal fatigue specimens of multi-component aluminum silicon alloys were analyzed by transmission light microscope (OM: model LEICA DM 2500M, Leica Mikrosystems Vertrieb GmbH, Wetlzar, Germany), scanning electron microscopy (SEM: model S-3400N, Hitachi, Tokyo, Japan), high-resolution transmission electron microscope (TEM: model JEM-2100, JEOL Ltd., Tokyo, Japan) and Energy Dispersive Spectrometer (model GENESIS200XM30T, EDAX, Mahwah, NJ, USA). The specimen without defects, such as cracks and pores in the notched area, was selected for thermal fatigue testing and was removed from the experimental device after each thermal cycle of a certain number of times. After mechanical polishing, the length of the surface crack at the V-shaped notch was measured. We selected the crack connected to the specimen notch. Both surface cracks should be measured. If there are multiple cracks, the longest one should be taken as the crack length. According to the actual shape, the measured crack is approximated as one or more line segments, and its length is the sum of all line segments. When the specified surface crack length a = 0.10 mm was adopted, the required number of cycles N was the crack initiation life [30].

3. Results and Discussion

3.1. Analysis of Microstructure of Thermal Fatigue Specimens

The microstructure of the thermal fatigue specimens of aluminum alloy treated by four different processes before thermal cycling is shown in Figure 3. The V-notch is relatively flat and almost free of defects such as pores and cracks. In as-cast A sample alloy, the α- Al phase presents coarse columnar crystal, and the eutectic Si has a large difference in morphology and presents as an elongated plate-like or long-needle-like shape, distributed in aluminum matrix without directionality and irregularity. The edge has sharp corners.

The B specimen alloy was subjected to composite refinement and modification of Al-5Ti-1B-1RE and Al-l0Sr master alloys. The shape and size of the α-Al phase of the matrix become uniform and fine, closely arranged, the shape is relatively regular, the grain boundaries are clearly visible, and the coarse columnar crystals are greatly reduced, mainly fine equiaxed crystal. The morphology of eutectic Si has undergone significant changes, and the lamellar eutectic silicon has almost completely disappeared, almost all of which have transformed into uniformly diffused and spheroidized particles. They were mainly uniformly concentrated at the α-Al grain boundary of the matrix. In addition to the microstructure and morphology of specimen B, the new aluminum alloy of specimen C also has a relatively uniform and fine second phase distributed in the grain and on the grain boundary, which becomes the main role of the alloy strengthening phase and hinders dislocation movement. After T6 heat treatment of the specimen D, the eutectic silicon in the alloy structure is obviously spheroidized, which was punctuated with good spheroidization effect and uniform distribution of the second phase.

From the microstructure of the aluminum alloy treated by four different processes, it can be seen that the morphologies and distributions of the α-Al phase and eutectic silicon phase of the alloy of specimens B, C, and D are significantly better than that of cast specimen A.

3.2. Analysis of Mechanical Properties of Thermal Fatigue Specimens

The mechanical properties of thermal fatigue specimens of multicomponent Al-Si-Mg aluminum alloys treated by four different processes are shown in

Table 2. Mechanical properties of Al-Si-Mg alloys.

Sample	Tensile Strength (MPa)	Yield Strength (MPa)	Elongation (%)	Vickers Hardness (HV)	Impact Toughness (J/cm2)
A	177.56	128.21	1.17	72.50	8.36
B	258.74	196.35	7.73	83.41	13.19
C	282.36	223.89	6.96	101.80	20.31
D	352.45	267.58	5.75	120.30	26.25

.

The difference in the mechanical properties of multicomponent Al-7Si-0.3Mg alloys treated in these four different states is mainly due to the existence of different dislocation densities in the alloy, that is, the total length of the dislocations contained in the unit [31]:

$$\rho =\frac{L}{V}$$

(1)

where: ρ—dislocation density, m⁻²; V—alloy volume, m³; L—the total length of the dislocation lines in alloy, m.

The existence of dislocations has an important impact on the mechanical properties of alloys, especially the tensile strength. From the relationship between the strength of the alloy and the dislocation density in Figure 4, it can be seen that the dislocation density ρ_min is the smallest and the tensile strength of the alloy is the smallest. When the alloy is alloyed or heat treated, the dislocation density will be greatly increased, and the strength of the alloy will be greatly increased due to the interaction and restriction between dislocations.

According to the TEM image analysis of the multicomponent Al-7Si-0.3Mg alloy in four different treatment states, the dislocation densities contained in these four alloys are significantly different. The sampling area of TEM micrographs is the area where each specimen is equidistant from the notch. Figure 5a is a TEM image of cast Al-7Si-0.3Mg alloy, from which it can be seen that the alloy has low dislocation density, low dislocation line entanglement, weak hindrance to alloy deformation, and low alloy performance. Figure 5b shows a TEM image of the refined and modified Al-7Si-0.3Mg alloy. It can be seen from the figure that there are obvious dislocations, and the dislocation density has been greatly improved. The interaction between fine eutectic silicon and dislocations increases the resistance to alloy deformation and improves its mechanical properties. Figure 5c is a TEM image of the new aluminum alloy, from which it can be seen that the configuration distribution and number of dislocations have changed significantly, and its dislocation density is greatly improved compared with the cast and refined and modified Al-7Si-0.3Mg alloy, and a large number of dislocations and the presence of some fine second phase particles can be clearly observed. A large number of dislocations gradually formed dislocation cells inside the alloy through slipping and climbing, dislocation walls formed in some areas, and their mechanical properties were further improved. Figure 5d shows a TEM photo of a new aluminum alloy after T6 heat treatment. From the figure, it can be seen that a large number of dislocations are entangled together, and most of the areas have formed obvious dislocation walls, greatly increasing the dislocation density of the alloy. This is mainly due to the occurrence of a large amount of dislocation increment and expansion inside the alloy during T6 heat treatment process. In the tangle of high-density dislocations, there are some small second-phase particles that are not easy to find.

Through the above analysis, it can be concluded that the dislocations in the four different treatment states of the multicomponent Al-7Si-0.3Mg alloys undergo a finite dislocation line in the cast aluminum alloy, followed by the obvious dislocation line in the refined and modified aluminum alloy, the new alloy dislocation cell, and a small portion of the dislocation wall. After T6 heat treatment, the new alloy has a large number of dislocation walls. The mechanical properties of the new alloy after T6 heat treatment, especially the tensile strength and hardness, are the largest among the four different treatment states of the aluminum alloy, which is consistent with the mechanical properties of the four different treatment states of the multicomponent Al-7Si-0.3Mg alloys in Table 2.

3.3. Analysis of Thermal Fatigue Crack Growth Behavior

Since the multicomponent Al-7Si-0.3Mg alloy treated by different processes is mainly used to manufacture automobile cylinder blocks and cylinder heads, its conventional working temperature ranges from room temperature to about 300 °C, so the temperature range of 20 °C↔350 °C can better meet the practical application conditions, which is representative. Therefore, the temperature range of 20 °C↔350 °C was selected to systematically study the thermal fatigue crack growth behavior of aluminum alloy treated by these four different processes. The relationship between the length of thermal fatigue crack and the number of thermal cycles of four different alloys is shown in Figure 6. When the number of thermal cycles N = 8000 cycles, specimen A had produced obvious cracks. The crack length reached 0.085 mm, which was close to the surface crack length of the specimen at the specified crack initiation a = 0.10 mm. The crack length produced by specimen B was relatively short, about 0.03 mm, while specimens C and D had not yet produced cracks. With the increase of the number of thermal cycles, the crack length showed an increasing trend. When the number of thermal cycles N = 11,600 cycles, the crack of specimen A propagated rapidly, the crack length reached nearly 0.65 mm, and the crack length of specimen B was about 0.53 mm. At this time, the crack propagation of specimen C and D was relatively short, and the crack length was about 0.22 mm and 0.10 mm, respectively, which had exceeded or reached the specified crack length of the surface of the specimen crack initiation a = 0.10 mm. The crack length data showed that the thermal fatigue resistance of specimens C and D was much higher than that of A and B at a temperature range of 20 °C↔350 °C. From the perspective of crack length growth rate, it first increased and then decreased with the increase of the number of cycles. This is mainly due to the fact that after removing the crack initiation stage with a = 0.10 mm, the crack length a had an approximate linear relationship with the cycle number N in the crack growth region [32]. The crack propagation rate $\frac{\mathrm{d}a}{\mathrm{d}N}$ is controlled by the intensity factor amplitude $\Delta K=Y\sqrt{a}\cdot \Delta \sigma$ of the thermal stress field at the crack tip, and the relationship between them meets the Paris formula [33,34]:

$$\frac{\mathrm{d}a}{\mathrm{d}N}=\mathrm{c}(\Delta K{)}^{\mathrm{n}}=\mathrm{c}(K_{max}-K_{min}{)}^{\mathrm{n}}=\mathrm{c}{(Y\sqrt{a}{\sigma }_{max}-Y\sqrt{a}{\sigma }_{min})}^{\mathrm{n}}=\mathrm{c}(Y\sqrt{a}\cdot \Delta \sigma {)}^{\mathrm{n}}$$

(2)

where ΔK is the stress intensity factor range (MPa$\sqrt{\mathrm{m}}$), Y is the crack shape and load coefficient, which takes values from 0 to 2 (many times in various tests it is about 0.9), σ_max and σ_min represent maximum stress and minimum stress (MPa), respectively; a—crack length (m); c, n—the material test constant, $\lg \frac{\mathrm{d}a}{\mathrm{d}N}-\lg \Delta K$ determined by the intercept and slope of the test curve, and the n value of most materials varies between 2–4.

At the initial stage of crack propagation, it can be seen from the Paris formula that, on the one hand, with the increase of crack length a, the crack propagation rate increases. On the other hand, with the increase of crack length a, the local constraint ratio decreases, which leads to thermal stress relaxation and crack growth rate reduction. On the other hand, the increase of crack length a leads to thermal stress relaxation and crack propagation rate $\frac{\mathrm{d}a}{\mathrm{d}N}$ reduction. Within a certain range, the effects of these two effects cancel each other out, making $\frac{\mathrm{d}a}{\mathrm{d}N}$ approximately constant. With the increase of crack size, it gradually plays a leading role in the thermal stress relaxation, so $\frac{\mathrm{d}a}{\mathrm{d}N}$ starts to become smaller. When the thermal stress is relaxed to a certain value, the crack propagation stops [35].

3.4. Morphology Analysis of Thermal Fatigue Crack Propagation

After the thermal fatigue specimens of four different processes of aluminum alloys undergo multiple hot and cold cycle thermal fatigue tests, the crack propagation of the specimen was revealed, as shown in Figure 6. When the number of thermal cycles of the thermal fatigue specimen N = 27,500 cycles, specimen A, as shown in Figure 7a, was due to the presence of needle-like eutectic silicon in the structure at the grain boundary under the action of alternating cyclic thermal stress. When the stress concentration at the grain boundary cannot be relaxed, the stress peak was higher and higher, and when the grain boundary strength was exceeded, cracks would occur at the grain boundary. As the binding force of aluminum and silicon decreased obviously under the action of high temperature, the grain boundary strength further decreased. Under high temperature atmosphere, aluminum and oxygen in the air continuously combined to form an extremely thin Al₂O₃ film. Under the action of thermal stress, the specimen cracks continue to propagate, resulting in the rupture of the oxide film, and then a new oxide film was rapidly generated so that the crack propagation and oxidation cycle continued. The oxidation further weakened the bonding strength between the grain boundary and the matrix and made the as-cast Al-7Si-0.3Mg alloy thermal fatigue specimen show high brittleness at high temperature, resulting in a significant reduction in the driving force required for the forward propagation of the crack tip. Due to the bending of the grain boundary, the crack propagation of the specimen followed the principle of minimum energy consumption and propagation along the grain boundary was preferred, resulting in the bending of crack propagation [30,36].

The grain of specimen B was refined, the morphology of eutectic silicon was significantly improved, showing roundness and granularization, and the splitting effect on the matrix was greatly reduced. At high temperatures, the bonding strength of aluminum and silicon phases was much higher than that of as-cast Al-7Si-0.3Mg alloy. During the thermal crack propagation, due to the thermal expansion and contraction of the specimen and the large temperature gradient, the crack tip had a sufficiently high thermal stress as the driving force to pass through the grain interior and the grain boundary to propagate to the adjacent grains. With the propagation of the crack, plastic deformation should occur near the crack tip, and thermal stress can be released and reduced. The plastic relaxation and the local constraint became small, and the alternating thermal stress experienced by the crack tip also decreased and was not enough to pass through the grain. Therefore, following the principle of minimum energy consumption, it will mainly propagate along the crystal, as shown in Figure 7b. The crack propagation path was less bent than that of specimen A [30,36]. It is shown that the grain size strongly affects the process of crack growth [37].

Specimen C, the thermal resistance of refining and modification after alloying, was greatly improved. Under the condition that the solid solution saturation was equal, the atomic bond binding energy of the multi-element alloy was higher than that of the binary alloy. The more the strengthening phase elements were, the finer the second phase particles were, the higher the bonding strength was, and the greater the hindrance to the solid solution grain deformation was. Therefore, the yield strength and toughness of the new aluminum alloy are significantly higher than those of specimens A and B. Because the sublimation energy of the added element Cu is 81.2 kJ · mol⁻¹, that of Mn was 74 kJ · mol⁻¹, which was much higher than that of Al, which was 55 kJ · mol⁻¹, and the alloy element with high sublimation energy had high melting point [38]. It formed complex heat-resistant phases with Al and other elements, which greatly improved the heat resistance and thermal fatigue resistance of the alloy. The Cu-containing phase and Mn-rich phase in the second phases had a strong blocking effect on the thermal fatigue crack, which greatly improved the thermal strength of the alloy, made up for the weakness of the grain boundary strength of the multi-component Al-7Si-0.3Mg alloy at high temperature, effectively prevented the grain boundary sliding, and improved the high-temperature strength and thermal fatigue performance of the alloy. For multiphase alloys, each grain had the most easily cracked face-crack surface. When the crack propagated to the grain boundary, the angle between the crack faces of adjacent grains was small, and the crack propagated in a transgranular mode. When the angle was large, the crack propagated in an intergranular mode. The crack propagation of specimen C follows the principle of minimum energy consumption, and the thermal fatigue crack propagation was mainly propagated in a transgranular–intergranular hybrid manner, as shown in Figure 7c [30,36]. During the solution treatment of the specimen D, the non-equilibrium copper-containing second phase was dissolved into the α solid solution, and the supersaturated Mn-rich phase was distributed and precipitated in a diffuse manner. In the aging process, the copper-containing phase was precipitated in a diffuse strengthening phase. Below 350 °C, the solubility of the Mn-rich phase in the α solid solution changed very little. It was relatively stable at high temperature, it was not easy to condense and grow, and it had strong thermal hardness. The addition of a small amount of Ti to the alloy could further refine the alloy structure and improved the strengthening effect. Therefore, there were a dense and high strength second phases distributed on the grain boundaries, and when the crack propagation direction and grain boundaries orientation deviation was large, the thermal fatigue crack preferentially propagated through the grain boundaries and interconnects between different microcracks in the grains to form cracks. When the deviation between the crack propagation direction and the grain boundaries orientation was small, the crack propagated along the grain.

The grain boundaries of specimen D were significantly strengthened, the thermal fatigue crack propagation was mainly transgranular–intergranular mixed mode, and the crack length was the shortest, as shown in Figure 7d. After the crack initiation, the new aluminum alloy and T6 heat treatment were not easy to propagate. Because of the high strength and toughness of its grain boundaries, the crack propagation path needed to be changed many times, which would consume more energy, and its thermal fatigue resistance was greatly improved. A suitable selection of heat treatment parameters allowed us to obtain high strength properties under a relatively low density of these materials [39].

It could be seen from Figure 7 that when the number of thermal cycles of thermal fatigue sample N = 27,500 cycles, the bending degree of crack propagation of specimen A was the highest, the length was the largest, and the width was thick and uneven. The cracks of specimens B, C, and D were relatively straight and fine, especially specimen D, which was the smallest in length and the narrowest in width. This showed that good microstructure and suitable heat treatment processes can significantly improve the thermal fatigue resistance and reduced crack propagation of alloys.

3.5. Influencing Factors of Thermal Fatigue Crack Propagation

3.5.1. Effect of Temperature Amplitude

In the test, the thermal fatigue specimens were not subjected to any external mechanical load, and the fatigue failure of the specimens was completely the result of repeated action of the thermal stress cycles. When the lower limit temperature was the same, increasing the upper limit temperature would reduce the thermal fatigue life of the alloy. The main reason was that the thermal stress increases with the increase of the temperature difference. The thermal stress difference was proportional to the temperature difference [40]:

Δσ = K · E · α · ΔT

(3)

where Δσ—thermal stress difference, K—constraint coefficient, E—modulus of elasticity, α—coefficient of thermal expansion, and ΔT—temperature difference.

It can be seen from Equation (3) that when the lower limit temperature is constant and the upper limit temperature is increased, the temperature difference ΔT becomes larger and the thermal stress also increases. The thermal stress generated by the thermal fatigue specimen mainly consists of two parts, one of which came from the thermal stress caused by the temperature difference between the surface and the interior of the specimen. When the specimen was cooled, it could be reduced from the upper limit temperature to the lower limit temperature in a very short time, as there was a large temperature difference between the surface and the interior of the specimen in a relatively short time. The surface of the specimen shrunk sharply due to the sudden decrease in temperature, while the interior of the specimen was still in a state of high-temperature expansion and the contraction of the specimen surface was constrained by the internal high-temperature expansion, which instantly produced great thermal stress. Within the temperature difference ΔT, the resulting expansion deformation was α·ΔT, and if the deformation was completely constrained, the thermal stress generated was Δσ = K·E·α·ΔT. When heated, the opposite was true. Additionally, due to the temperature difference, the thermal expansion of the surface of the specimen was constrained by the internal, which also produced thermal stress. After a certain number of cycles, when the thermal stress exceeded the elastic limit of the alloy at high temperature, local plastic deformation would occur, which could cause fatigue cracks, as shown in Figure 8. Under certain cyclic stress conditions, the length a of thermal fatigue cracks was increasing when they propagate, and the propagation rate $\frac{\mathrm{d}a}{\mathrm{d}N}$ was also increasing. When the cyclic thermal stress was changed, from Δσ₁ to Δσ₂, Δσ₃, and Δσ₃ > Δσ₂ > Δσ₁, when the lower limit temperature was the same 20 °C, the upper limit temperature was increased. When T₃ > T₂ > T₁, from Δσ = K·E·α·ΔT, it can be seen that the larger the temperature difference ΔT, the greater the thermal stress Δσ. According to the Paris formula: $\frac{\mathrm{d}a}{\mathrm{d}N}=\mathrm{c}(Y\sqrt{a}\cdot \Delta \sigma {)}^{\mathrm{n}}$, the greater the thermal fatigue crack propagation rate $\frac{\mathrm{d}a}{\mathrm{d}N}$, the greater the slope of the a-N curve, namely ${(\frac{\mathrm{d}a}{\mathrm{d}N})}_3>{(\frac{\mathrm{d}a}{\mathrm{d}N})}_2>{(\frac{\mathrm{d}a}{\mathrm{d}N})}_1$. At this point, the a-N curve shifts to the left, and a_c2 and N_p2, and a_c3 and N_p3 decreased accordingly, that was, in a shorter cycle, the critical crack size a_c can be reached, resulting in thermal fatigue crack instability propagation.

The other part of the thermal stress generated by the thermal fatigue specimen was that when the specimen was heated, it was in the uniform heating temperature zone, and when cooled, only the notched end and above parts of the specimen were quenched into the water, while the rest were not quenched into the water. During the cooling of the thermal fatigue specimen, the quenched part shrunk sharply due to the sudden decreased in temperature, while the unquenched part remained in a state of high-temperature expansion. The unquenched part of the specimen had a restraining effect on the quenched part, thus producing tensile stress on the quenched part. The increase of the upper limit temperature increased the two kinds of thermal stresses at the same time, and leaded to decrease in the yield strength of the alloy itself. Thus, the thermal fatigue resistance of the alloy was reduced, and the crack propagation was accelerated. Therefore, improving the thermal stability of alloy was the most effective measure to improve its thermal fatigue life [41].

3.5.2. Effect of Oxidation

During the cold and hot cycles, the plastic deformation of the thermal fatigue specimen increased continuously, the oxidation of the specimen surface and the V-notch continued to occur, and the oxide continued to fall off. On the one hand, at high temperatures, O atom activity increased, the surface of the alloy was easily oxidized, and the plasticity of the surface of the specimen was reduced. The plasticity of the surface of the specimen was different from that of the inside of the specimen. Due to the continuity of displacement at the surface of the specimen and the interior interface, the thermal stress would be generated inside. When the yield limit of the alloy was reached, plastic deformation occurred in local areas, resulting in the accumulation of plastic deformation. Under the induction of thermal stress, O atoms diffused into the specimen, resulting in local surface structure porosity. On the other hand, in the process of cold and hot cycling, due to the temperature gradient on the surface and inside of the alloy specimens, the temperature distribution was uneven. Although the specimen was not subject to external constraints, because only part of the length of the specimen was quenched into the water, when the specimen was quenched and cooled and heated away from the water, due to the different temperatures of the specimen, each part cannot expand freely due to the influence of adjacent parts at different temperatures, thus generating thermal stress in the specimen. At 350 °C, the expansion coefficient α_Al = 24.90 × 10⁻⁶ K⁻¹ of pure aluminum was much larger than that of silicon expansion coefficient α_Si = 6.95 × 10⁻⁶ K⁻¹, and the difference between the two was more than three times. In the process of the temperature rising and falling, the silicon phase and the aluminum matrix would undergo significantly different thermal expansion and contraction, so mutual constraints occurred. In addition, micro-stress was generated at the interface of aluminum-silicon phase. As a result, the bond strength of the silicon phase and the aluminum matrix was greatly weakened. The alloy was easy to crack along the grain boundaries. Some oxides fell off from the alloy under the action of thermal stress and its own gravity, forming oxidation pits, as shown in Figure 9a. Some of these oxides can connect with each other to form a “bridge”, as shown in Figure 9b. This was mainly because the thermal fatigue specimen met the cooling water in the circulating system at high temperature and cools down, and when the specimen returned to the heating furnace for heating and heat preservation, the water on the surface of the specimen was quickly heated into water vapor. Because of the high chemical activity of aluminum, the following reactions would occur quickly in contact with water vapor [38]:

$${\mathrm{Al}\left(\mathrm{s}\right)+3\mathrm{H}}_2\mathrm{O}\left(\mathrm{g}\right)\to {\mathrm{Al}\left(\mathrm{OH}\right)}_3\left(\mathrm{s}\right)+\frac{3}{2}{\mathrm{H}}_2\left(\mathrm{g}\right)$$

(4)

Al(OH)₃ remained on the surface of the specimen, the structure was loose, and there was no protective effect on the aluminum alloy specimen. When the temperature rose to about 350 °C, the following decomposition was as follows:

$${2\mathrm{Al}\left(\mathrm{OH}\right)}_3\left(\mathrm{s}\right)\to {\mathrm{Al}}_2{\mathrm{O}}_3{\left(\mathrm{s}\right)+3\mathrm{H}}_2\mathrm{O}\left(\mathrm{g}\right)$$

(5)

The decomposition product Al₂O₃ was loose and powdery, which could absorb water vapor and oxygen, increase the content of gas and oxide inclusions near the surface of the thermal fatigue specimen, so that the oxidation corrosion of the thermal fatigue specimen increased with the increase of thermal cycles, and the oxide would fall off under the effect of thermal stress on the surface of the specimen, forming a “bridge”. At the late stage of crack propagation, O was accumulated again on the new surface of the specimen, and the “bridge” rate decreased slightly. The propagation direction of the crack was from right to left, and the main thermal fatigue crack oxidation and propagation were connected to each other, as shown in Figure 9c. Mutual promotion led to the formation of microcracks or micropores, as shown in Figure 9d, which grew and connected with the main crack at the same time so that the main crack grew in leaps, which was also in line with the principle of minimum energy consumption of crack propagation, and the optimal propagation was selected [30,36].

3.5.3. Effect of Residual Stress

Thermal fatigue cracks propagated forward from their tip, as shown in Figure 10a, when thermal fatigue crack cores were formed and initiated at a point on the slip line, and then propagated under the thermal stress at that point. However, whether the thermal fatigue microcrack at this point can propagate depends on the strength of the alloy. At the same average stress, cracks would propagate along weaker alloy strength, which was consistent with the stress principle and strength principle of the crack propagation path. Under the same conditions, thermal fatigue cracks at grain boundaries with lower strength were prone to propagation. On the other hand, the thermal stress that the crack bears during the crack propagation was constantly changing. When the sample was heated, the thermal stress borne by the crack was the residual tensile stress, which made the crack open and had the driving force of forward propagation. When the specimen was cooled by quenching water, the thermal stress borne by the crack was residual compressive stress, which made the crack closed, so the thermal crack could not propagate under the action of compressive stress. Therefore, during the cold and hot cycles, the thermal fatigue cracks propagate in the direction perpendicular to the tensile stress under the cyclic action of residual tensile stress→residual compressive stress→residual tensile stress. With the increase of the number of thermal cycles, the plastic deformation of the thermal fatigue specimen accumulated continuously. The sliding occurred along the crystal plane with the highest atomic density (closely arranged plane) and the crystal direction with the highest density above it (closely arranged direction), resulting in great stress concentration and damage to the crack core, crack growth, causing some silicon phases, second phases, or inclusion particles (mainly including Al₂Cu, AlSiTiCu, etc.) in the thermal fatigue specimen to break under the action of thermal stress, generating microcracks inside and absorbing the driving force required for the propagation of some main cracks, as shown in Figure 10b. The second phase of the torn block through energy spectrum scanning analysis was AlSiTiCu phase, as shown in Figure 10c.

3.6. Thermal Fatigue Crack Propagation Process and Mechanism

The thermal fatigue crack propagation process is relatively complex, and mainly consists of two stages. The first stage is on the basis of crack initiation. Under the action of thermal stress, microcracks are first formed, and propagate inward in the direction of maximum shear stress. Most microcracks will not become the main cracks and only a few microcracks can continue to propagate rapidly or preferentially. These few microcracks, such as crack ① in Figure 11, greatly alleviate the stress concentration at the tip of other microcracks (such as cracks ② and ③ in Figure 11) during the propagation process, so that the propagation process of other microcracks slows down or stops propagating.

Under the action of alternating cyclic thermal stress, a small number of microcracks can quickly continue to propagate or preferentially propagate into the main crack, and the plastic opening passivation and closure sharpening of the main crack tip will make the crack extend and propagate forward, as shown in the schematic diagram of the thermal crack propagation process in Figure 11. Under the action of alternating cyclic thermal stress, when the specimen was cooled by quenching water, the crack was closed under pressure within a half cycle of compressive stress, and the crack presented a tip shape at M position. This was mainly due to the sharpening of the front end of the crack to reduce the driving force required for the self-crack to continue to propagate forward, as shown in Figure 12a. When the thermal fatigue specimen was heated in the hot and cold fatigue testing machine, the crack was tensile and opened within half a cycle of tensile stress. Due to the stress concentration, the crack tip slipped in the direction of 45°, and when the tensile stress reached the maximum, the slip zone expanded, the crack tip became semi-circular, plastic passivation occurred, and the crack front end moved to N. At this time, due to the plastic deformation, the stress concentration at the front end of the crack became smaller, and the driving force of the crack propagation was not enough to cause the front end of the crack to slip forward. As a consequence, the slip stopped and the crack stopped propagating, as shown in Figure 12b. When the thermal fatigue sample was quenched and cooled again, the crack was closed again under pressure in the next half cycle of compressive stress. The sliding was carried out in the opposite direction when the crack was opened under tension, the original crack and the newly propagated crack surface were pressed close, the front end of the crack was pressed, and the compressive stress concentration slide along the 45° direction. When the compressive stress reached the maximum, the crack surface was pressed, and the front end of the crack changed from passivation to sharpening, forming a sharp corner. However, the crack tip had moved to N, propagating the ΔL distance forward, as shown in Figure 12c. It can be seen that under the action of alternating cyclic thermal stress, the thermal fatigue specimen propagated forward due to the plastic closure sharpening and opening passivation effect of the crack tip.

At the end of the first stage of crack propagation, due to the continuous obstruction of the grain boundaries, the crack propagation gradually turned to the direction perpendicular to the tensile stress and entered the second stage until the final fracture failure [39]. The crack propagation rate $\frac{\mathrm{d}a}{\mathrm{d}N}$ of the second stage was relatively small compared with that of the first stage in most of the cycle cycles, as shown in Figure 6, which was the main part of thermal fatigue crack propagation and took a long time. The second stage of thermal fatigue crack propagation was under the action of thermal stress Δσ = K·E·α·ΔT, the crack tip alternately changed the cycle by sharpening→passivation→sharpening, and the crack was intermittently propagated forward from its tip in a way that promoted expansion→stopping→expansion until the specimen failed, and the thermal crack propagation of these two stages was also consistent with the thermal fatigue test results, as shown in Figure 13. The crack propagation of the new aluminum alloy thermal fatigue specimen was that when the specimen was cooled by quenching water, the crack was closed under pressure, and the crack tip was sharpened at position I, as indicated by the arrow in the circle in Figure 13a, and the crack tip produced a plastic closure effect. When the thermal fatigue specimen was heated in the heating furnace, it was within the half cycle of tensile stress. The crack was tensile and opened, and the crack tip was subject to plastic passivation. The hot crack tip was passivated at position II, as indicated by the arrow in the circle in Figure 13b, and the crack tip produced a plastic opening effect. When the thermal fatigue specimen was quenched and cooled again by water, the crack was closed again under pressure, and the crack tip changed from passivation to sharpening, forming a sharp corner in the next half cycle of compressive stress. However, at this time, the crack tip had propagated forward, and the crack tip was sharpened at the III position, as indicated by the arrow in the circle of Figure 13c, and the crack tip again produced a plastic closing effect. The thermal fatigue crack continued to propagate forward at a certain rate until the specimen broke and failed.

4. Conclusions

(1)

When the temperature amplitude was 20 °C↔350 °C and the number of thermal cycles was N = 11,600 cycles, the crack of specimen A increases rapidly, the crack length reaches nearly 0.65 mm, and the crack length of specimen B is about 0.53 mm. At this time, the crack propagation of specimen C and specimen D was relatively short, and the crack length was about 0.22 mm and 0.10 mm, respectively, which had exceeded or reached the specified crack length of the surface of the specimen crack initiation a = 0.10 mm. The thermal fatigue resistance of C and D specimens was much higher than that of specimens in both A and B.

(2)

When the number of thermal cycles N = 27,500 cycles, the crack propagation of specimen A had the highest curvature, the longest length, and the largest width and unevenness. The cracks of B, C, and D specimens were relatively straight and fine, especially specimen D, which had the smallest length and narrowest width. Good microstructure and suitable heat treatment process can significantly improve the thermal fatigue resistance of the alloy and reduce crack propagation.

(3)

Under the same lower limit temperature, increasing the upper limit temperature can make the thermal fatigue specimens reach the critical crack propagation size in a shorter cycle, resulting in thermal fatigue crack instability propagation.

(4)

With the increase of the number of heating cycles, the oxidation corrosion of the thermal fatigue specimen also increases. The oxide will fall off under the thermal stress on the surface of the specimen, forming a “bridging” phenomenon.

(5)

With the increase of the number of thermal cycles, the plastic deformation of the thermal fatigue specimen accumulated continuously. The sliding occurred along the crystal plane with the highest atomic density (closely arranged plane) and the crystal direction with the highest density above it (closely arranged direction), resulting in great stress concentration and damage to the crack core, growth. Some silicon phases, second phases, or inclusion particles (mainly including Al₂Cu, AlSiTiCu, etc.) in the thermal fatigue specimen broke under the action of thermal stress, generating microcracks inside and absorbing the driving force required for the propagation of some main cracks.

(6)

Under the action of alternating cyclic thermal stress, only a few microcracks can continue to quickly propagate or preferentially propagate into the main crack. The crack tip changes the cycle in a way of sharpening→passivation→ sharpening repeatedly alternating, and the crack intermittently propagates forward from its tip in a way that expansion→stopping→expansion until the specimen breaks and fails.