In recent years, the development of thulium-doped fiber lasers (TDFL) gradually followed the footsteps of ytterbium-doped and erbium-doped fiber lasers. The tunable range of TDFL is 1400-2200 nm, covering the atmospheric transmission window. In this window, the allowed power transmission of light in free space can be several orders of magnitude higher than that of the other wavelength bands. In particular, optical power transmission exceeds 80% in the 2050 nm wavelength band, making it possible to use TDFL in this band for free-space optical communications and atmospheric Doppler lidar. The TDFL operating wavelength is in the 2 μm band, which is a safe operating band for human eyes, and in which there is a high transmittance atmospheric window and strong absorption peaks of multiple gas molecules and OH^- ions. Therefore, lasers in this band are favored by various application industries, especially in the free-space optical communication field, where human eye safety is a requirement. Single longitudinal mode (SLM) fiber lasers have excellent characteristics, such as high beam quality, good coherence, and narrow linewidth, and are widely used as the preferred light source in multiple important fields. For example, fiber lasers with a narrow linewidth output have been used in ultra-long-range coherent optical communication, fiber optic sensing, optical metrology, high-resolution spectroscopy, and lidar, and have potential applications in optical atomic clocks, fundamental constant measurements, and physics. Therefore, the realization of a stable single longitudinal-mode narrow-linewidth laser source in the 2050 nm band is indispensable.

The proposed structure predominantly consists of a ring main cavity and a compound sub-cavity (Fig.1). The 793 nm pumped source output is input into the ring cavity through the fiber combiner, a 4 m long double-clad thulium-doped fiber is used as the gain medium, and the circulator ensures unidirectional transmission of light inside the ring cavity. A 0.5 m length of unpumped thulium-doped fiber is added to port 2 of the ring as the saturable absorber (SA), making it equivalent to a dynamic self-tracking narrow-band filter, which effectively suppresses the multi-longitudinal mode oscillation, realizing single longitudinal mode operation and compressing the narrow linewidth. A fiber Bragg grating (FBG) is used as a wavelength-selective device. The optical signal reflected by the FBG is injected into a composite double-loop cavity composed of three couplers and a composite double-loop cavity structure (Fig.3), which consists of one 3×3 coupler OC1 and two 2×2 couplers OC2 and OC3 which are connected in sequence. The 3×3 coupler OC1 has a 33∶33∶33 output and divides the input optical signal into three optical signals. The coupling ratio of both 2×2 couplers is 20∶80. The output laser is generated from the 10% port of the 90∶10 coupler.

The laser was developed and tested on an ultrastable optical stage at room temperature. A stable laser output was obtained when the pumping power reached 4 W. The central wavelength was 2048.76 nm, and the optical signal-to-noise ratio was 68 dB. The output spectrum was measured every 6 min for 60 min, and the spectrum obtained after ten consecutive scans [Fig. 8(a)]. To further quantify the stability of the laser, the power jitter and wavelength drift results over 60 min were analyzed with power fluctuation less than 0.15 dB and wavelength drift less than 0.02 nm [Fig.8(b)], indicating that the laser had good output stability at room temperature. The results of the single longitudinal mode of the output laser using the self-homodyne method show no obvious mode-hopping phenomenon over the three measurement ranges (Fig.9). To demonstrate the stability of the laser’s single longitudinal mode, ten sets of measurements were performed within 60 min, and no beat frequency signal generated by the longitudinal mode was captured [ Fig.9(a) inset]. When the SA was removed, a nonzero frequency rate peak was observed in the 0-500 MHz measurement range [Fig.9(d)]. The results show that the composite double-ring cavity effectively suppresses most longitudinal modes in the cavity; however, the remaining longitudinal modes must be further suppressed using saturable absorbers. To further characterize the linewidth characteristics of the TDFL, the frequency noise of the laser was measured using an unbalanced Michelson interferometer based on a 3×3 fiber coupler, and the linewidth of the laser was calculated using the β-separation line method. The laser linewidth was 9.17 kHz at 0.001 s and the relative intensity noise is below -129.69 dB/Hz at frequencies above 1 MHz.

A single longitudinal mode narrow linewidth TDFL based on a compound double-ring cavity with a saturable absorber operating in the 2050 nm wavelength band is reported, with its output stability and linewidth characteristics characterized in detail. The performance of the proposed filter was analyzed in detail, and it was confirmed that the structure suppresses dense multilongitudinal modes well and has the advantages of simple fabrication and high tolerance. In combination with the excellent single longitudinal mode selection capability of the unpumped thulium-doped fiber, the laser was guaranteed to be in a stable single longitudinal mode state. Experimental results demonstrate that the proposed laser has the advantages of a high optical signal-to-noise ratio (OSNR), high stability, and narrow linewidth, and can be more widely used in lidar and space optical communication systems by reducing fusion loss, good vibration isolation, and temperature compensation to achieve a superior laser output.

The miniaturization and customization of corrugated sheets have become development trends to meet the demand of the high-efficiency heat and mass transfer in methanol reforming microreactors. To avoid problems such as springback, cracking, and wrinkling caused by conventional metal-forming processes, laser-forming technology becomes a potential solution to manufacture such structural components. However, when laser forming is adopted to fabricate specialty-shaped microcorrugated sheets, the sheets are prone to plane distortion defects owing to the uneven distribution of the energy input and free-end constraint in the curved scanning process. The plane distortion may lead to a gap between the corrugated sheets in the microreactor assembly, resulting in fluid leakage and a significant reduction in the heat and mass transfer performance of the microreactor. Because studies focusing on the laser forming of specialty-shaped microcorrugated sheets remain limited, the heat accumulation in laser scanning and the relationship between the temperature field and bending angle with various scanning strategies are analyzed in this study based on numerical simulations and experimental studies. A segmented variable-velocity strategy is explored to suppress the plane distortion. The forming quality of the corrugated sheets is further evaluated. This study provides a reference for the analysis and suppression of the plane distortion in the laser forming of specialty-shaped components.

The heat accumulation and temperature field distribution during the laser scanning along a double-arc curve are studied via numerical simulation, and the temperature difference between the top and bottom surface on the scanning path is analyzed using a segmented variable-velocity strategy. An experimental setup for the laser forming of a specialty-shaped corrugated sheet is developed (Fig. 6). The material of the specimen used in the experiments is 304 stainless steel with the size of 65.0 mm×65.0 mm×0.2 mm. A 1.5 kW oscillator continuous fiber laser with a laser power of 250 W and velocity range of 70-140 mm/s is used. Twelve cycles of scans are performed with the interval time of 10 s. During the experiments, one end of the sheet is fixed with a clamping width of 10 mm and can therefore be turned 180° using a three-jaw chuck. The laser scanning path can be drawn during the experiments and modeling process using the definition of a double-arc specialty-shaped line (Fig. 1). The x-directional and normal-plane bending angles are defined (Fig. 7), and the angles after laser forming are measured using a confocal microscope.

The segmented variable-velocity strategy effectively reduces the heat accumulation in the scanning process along the double-arc curve by discretely controlling the energy distribution along the scanning line. Using the segmented variable-velocity strategy, the heat accumulation at the exit is reduced, and the overall peak temperature decreases (Fig. 3). In addition, the segmented variable-velocity strategy improves the uniformity of the temperature distribution owing to the varying laser irradiation time and heat input (Fig. 5). After the local thermal stress generated during the forming process is balanced by the constraint force of the free end to the laser heating location, a relatively uniform thermal deformation is generated along the scanning line, thereby suppressing the distortion of the forming plane during laser forming. Without considering the heat accumulation, the thermal stress must maintain a downward trend while balancing the binding force at each part of the scanning line (Fig. 4). Combined with the double-layer model, the variation trend of the bending angle of the normal plane depends on the change of the temperature difference (Fig. 5). The accuracy of the numerical model is verified by comparing the numerical simulation and experiment results of the sheet surface temperature during laser bending forming (Fig. 8). The results demonstrate that a suitable number of velocity-varying segments can balance the energy distribution and reduce the unevenness of the bending angle in the normal plane, whereas the velocity should be set within a reasonable range. The angular difference is reduced to less than 1.0°. The forming efficiency of the corrugated sheet is also improved by using the segmented variable-velocity strategy (Figs. 9 and 10). By using the segmented variable-velocity strategy, the parallelism between the top and bottom surfaces of the corrugated sheet increases by 87.5% and the waviness values of the bottom and wall surfaces decrease by a maximum of 51.2% (Figs. 11 and 12).

A numerical model for the laser forming of a 304 stainless-steel sheet is developed, and the temperature distribution of the sheet at different numbers of velocity-varying segments is investigated. The results demonstrate that an increase in the segmental velocity can reduce the peak temperature of the top surface and heat accumulation at the exit. Meanwhile, the temperature gradient decreases as the velocity increases. Compared to constant-velocity scanning, the peak temperature at the terminal is significantly reduced by using the segmented variable-velocity strategy. For example, the peak temperature at the exit with eight velocity-varying segments is reduced by 13.4%. The experimental studies based on various scanning strategies and the detection of the normal-plane bending angle demonstrate that the distribution of the normal-plane bending angle is more uniform with an increase in the segmental number and velocity. Owing to the use of the segmented variable-velocity strategy, the forming plane distortion is significantly suppressed and the angle difference is reduced from 5.6° to 1.0° compared to constant-velocity scanning. Furthermore, the parallelism between the top and bottom surfaces of the corrugated sheet increases by 87.5% and the waviness decreases by a maximum of 51.2% by using the suppression strategy. The segmented variable-velocity strategy can be concluded to significantly improve the quality of specialty-shaped corrugated sheet forming.

With the rapid development of modern industrial technology, silicon carbide has broad application prospects owing to its excellent physical and chemical properties. Compared to conventional cutting methods, laser stealth dicing has the benefits of less debris with higher cutting accuracy. The research on the effect of laser parameters on surface ablation, edge chipping and cross-section roughness, and the development of new laser cutting techniques are of great practical significance to the development of silicon carbide cutting technology.

In this study a high power ultrafast laser processing system is used to cut out 300 μm thick SiC wafers with the diameter of 4 inch (1 inch=2.54 cm). Firstly, the effects of laser single pulse energy, feed distance, pulse repetition frequency, pulse width and scanning speed on cutting results are investigated using the control variables method. Based on the results of the multi-pulse mode (Fig.5(a)), the burst mode (Figs.5(b) and (c)) is used to reduce the edge chipping size and cross-section roughness. In burst mode, sub-pulses with the same pulse repetition rate as the output pulse sequence are selected from seed source pulses by adjusting the transistor-transistor logic (TTL)signal in the acousto-optic modulator (Fig.4). The surface ablation, edge chipping size and cross-section roughness are analyzed using laser confocal microscopy.

The effects of laser single pulse energy, feed distance, pulse repetition frequency, pulse width and scanning speed on cutting results under the multi-pulse mode are investigated. If the pulse energy is lower than 4 μJ, modified layers cannot be formed inside the SiC wafers, resulting in failure to separate the wafer (Fig.6). The feed distance has little effect on kerf width, however, the significant effects on edge chipping size and cross-section roughness are observed (Fig.7). A too low or too high pulse repetition frequency results in large kerf width, large edge chipping size and high cross-section roughness (Fig.8). Appropriately increasing pulse width can improve the quality of surface, edge and cross-section (Fig.9). Utilizing the appropriate scanning speed can reduce kerf width, edge chipping size and cross-sectional roughness (Fig.10). Based on these results, the burst mode is used to cut the wafers. It is confirmed that the cutting accuracy significantly improves under the burst mode (Fig.11). Because interaction time between the laser and material is too short in multi-pulse mode, the density of free electrons is too low and the internal material modification is insufficient, which affects the quality of the edge and cross-section. The burst mode extends the interaction time between the laser and material which induces a high density of free electrons and good internal crack continuity. Therefore, the edge chipping size and cross-sectional roughness are reduced.

This study investigates the effects of laser pulse energy, feed distance, pulse repetition frequency, pulse width and scanning speed on the top and bottom surfaces, edge chipping and cross-section of SiC wafers in multi-pulse mode using the control variable method. It is identified that the best cutting results are produced under the single pulse energy of 6 μJ, +5 μm feed distance, 100 kHz pulse repetition frequency, 10 ps pulse width, and 100 mm/s scanning speed. The kerf width on the top and bottom surfaces is 15.9 μm and 5.1 μm, respectively, the top and bottom surface edge chipping size is 7.8 μm and 2.1 μm, respectively, and the cross-section roughness is 3.1 μm. To further improve the edge size and cross-section morphology, the burst mode effect on the cutting results is investigated. It is confirmed that burst mode improves the continuity of modified cracks and reduces edge chipping size. When the number of sub-pulses is five, the kerf width on the top and bottom surfaces is 21.4 μm and 7.6 μm, respectively, and the minimum edge chipping size on the top and bottom surfaces is 1.2 μm and 1.0 μm, respectively, which is 85 and 52% less than those under the same cutting parameters in multi-pulse mode. Also at five sub-pulses, the lowest cross-section roughness is 2.3 μm, which is 26% less than that under the same cutting parameters in multi-pulse mode. This is because of the high density of free electrons generated in multi-pulse mode, which results in full and homogeneous material modification, thus reducing cross-section roughness. The burst mode increases dimensionalities of the laser stealth dicing parameters compared with multi-pulse laser stealth cutting and facilitates better cutting results.

To improve the relatively low hardness and unsatisfactory wear resistance of 304 stainless steel, corresponding surface coatings with high hardness and wear resistance are prepared in this study using the laser cladding surface modification technology. Accordingly, the service life of 304 stainless steel is extended. Medium-entropy alloys (MEAs) are derived from high-entropy alloys (HEAs) and primarily comprise three to four major elements. Compared with traditional alloys, MEAs have unique phase structures and excellent mechanical properties. Based on the design concept of HEAs with unequal atomic and eutectic structures, Ta is incorporated into FeCrNi MEAs and its content is adjusted to realize FeCrNiTa_x eutectic medium-entropy alloy coatings (EMEACs). Using this in-situ nanoscale laminate structure which combines both soft and hard phases, the overall strength and wear resistance of the alloy are greatly improved, while considerable plasticity is maintained, achieving a balance between strength and plasticity. In addition, the rapid laser cladding solidification technology is beneficial for refining the characteristic structure and improving the mechanical properties of the coating.

In this study, five FeCrNiTa_x (atomic fraction x=0, 0.2, 0.4, 0.6, and 0.8) alloy coatings with varying Ta contents are designed as research samples. The 304 stainless steel is chosen as the substrate for laser cladding. Elemental powders with different components are uniformly mixed and preplaced on a polished substrate using a planetary ball mill. The coatings are prepared using a laser cladding device with the laser power of 1400 W, scanning speed of 8 mm/s, spot diameter of 3.6 mm, and overlap rate of 35%. The crystal structures of the coatings with different Ta contents are analyzed using X-ray diffractometer (XRD), and their microstructures and chemical compositions are characterized via scanning electron microscope (SEM). Subsequently, the Vickers hardness values of the coatings from the cladding layer to the substrate are measured using a Vickers hardness tester and the change curves are recorded. Finally, a friction and wear tester is employed to measure the friction coefficients of the coatings, and an electronic balance is used to measure the mass loss. The wear morphologies and elemental compositions of the coatings are then characterized using SEM and XRD .

From the XRD patterns, only the face center cubic (FCC) phase diffraction peak is detected for the FeCrNi MEAs. After the addition of Ta, the Laves phase corresponding to the hexagonal structure appears when x=0.2, and the respective diffraction peak becomes clearer with increasing Ta content. Meanwhile, the FCC phase diffraction peak continuously weakens, indicating that the volume fraction of the Laves phase increases from x=0.2 to x=0.8 (Fig. 2). A clear boundary line between the coating and substrate is observed in the cross section of the laser-melted alloy coating (Fig. 3), indicating good metallurgical bonding of the alloy. Moreover, the microstructure of the coating cross section (Fig. 4) shows that the primary phase is the FCC phase when x=0.2, accompanied by a small amount of the Laves phase, which is a typical hypoeutectic structure. With increasing Ta content (x=0.4), the Laves phase increases significantly and a completely layered eutectic structure forms. When the Ta content further increases, the primary phase also changes from the FCC phase to the Laves phase, and large Laves phase blocks precipitate in the structure, forming a peritectic structure. From the hardness measurement results of the coatings (Fig. 7), the corresponding curves generally show a downward stepwise distribution, and the higher the Ta content, the higher the hardness value. The alloy with x=0.2 exhibits the highest hardness value of 705.3 HV, approximately 3.7 times that (190 HV) of the 304 stainless steel substrate. In addition, the improvement in hardness significantly influences the wear resistance (Fig. 9). The wear surface of the substrate exhibits obvious plastic deformation and severe adhesive wear. After the addition of Ta, the wear of the coating decreases. The coating with x=0.8 is generally smoother and flatter with the best wear performance, and the wear mechanism is slight adhesive and oxidative wear.

In this study, FeCrNiTa_x EMEACs with different Ta contents are successfully prepared on the surface of 304 stainless steel using the laser-cladding technology. The coating exhibits good metallurgical bonding with the substrate and no obvious defects appear on the surface. The microstructural analysis shows that an increase in the Ta content results in the transition of the coating structure from the sub-eutectic to fully eutectic, and finally to super-eutectic structure. In addition, the coating hardness gradually increases with increasing Ta content, and the hardness value of 705.3 HV is the highest when x=0.8, which is approximately 3.7 times that of the 304 stainless steel substrate. The solid-solution strengthening and precipitation strengthening caused by the addition of Ta are important factors in improving the coating hardness. Regarding the wear resistance of the coating, the Laves phase generated by the addition of Ta has an inhibitory effect on the coating wear, and the friction coefficient is the lowest and the wear resistance is the best when x=0.8. The wear mechanisms mainly include adhesive and oxidative wear.

Silicon is an important material utilized in various fields, such as biology and energy, particularly in the realm of integrated circuits. With advancements in semiconductor manufacturing toward large formats and thin substrates, the wafer-edge chipping and damage caused by traditional blade dicing have become increasingly significant in the field of semiconductor packaging. The introduction of low-dielectric-constant materials in the 90 nm integrated circuit technology node has presented a significant challenge to wafer dicing processes. The industry has responded by implementing a combination of laser surface grooving and mechanical dicing to address the low-dielectric-constant material separation issue. Laser processing equipment manufacturers have redirected their focus from nanosecond laser processing equipment toward ultrafast laser processing equipment that offers reduced thermal affected zones. However, existing research mainly focuses on optimizing the laser power, frequency, defocus, and feed speed, and comparatively analyzes multi-beam laser scribing spot distributions. A comprehensive theoretical analysis remains lacking. To address this gap, the current study utilizes a 517 nm femtosecond laser as a light source and shapes the energy of the focal spot into a flat-top distribution via a diffractive optical element for the surface grooving of silicon materials. The mechanism and quality of the groove process are investigated and discussed.

The experimental setup comprises a diode-pumped femtosecond fiber laser as the light source, while a galvanometer and an F-theta lens served as the beam movement and focusing tools. A diffractive optical element (DOE) is used to shape the near-focus spot into a top-hat square distribution on the surface of silicon for grooving. First, a one-dimensional simulation model is established using finite element analysis software, and the dual-temperature model is coupled with the excess carrier balance equation to analyze the interdependence among the material electrons, lattice, and carrier density in the femtosecond laser ablation process. This results in the establishment of a theoretical model for the femtosecond laser ablation process for silicon materials. Subsequently, a two-dimensional simulation model is established to simulate the morphology evolution during the silicon laser ablation process, and the impact of multipulse lasers with different spot intervals and energies on the groove quality is calculated. Finally, an experimental optical system is constructed for the grooving experiments, and the actual morphology is tested using a laser confocal microscope and the results are compared with those of the two-dimensional simulation.

The one-dimensional model (Fig. 1) demonstrates that the temperature of the free electrons in the silicon material increases rapidly during the 517 nm femtosecond laser irradiation, reaching a maximum temperature of approximately 28000 K, while the lattice temperature remains constant. The energy of the electrons is transferred to the lattice system after the laser pulse, causing the temperature of the electrons to decrease and that of the lattice to increase (Fig. 2). This suggests that the femtosecond laser primarily damages the silicon material through electron excitation, thereby achieving "cold" processing and reducing thermal damage to the material. The two-dimensional model (Fig. 7) demonstrates that ablation occurs when the laser flux exceeds 0.3 J·cm^-2 and the ablation depth increases significantly. The simulation results indicate that the thermal effect of the femtosecond laser ablation is small when the luminous flux is below 1.5 J·cm^-2, and the ablation depth remains relatively stable (Fig. 8). The experimental results confirm that the top-hat square spot produces a flat-bottomed inverted trapezoidal groove with a width of approximately 35 μm and a depth of approximately 16 μm (Fig. 11). The sidewalls of the grooves are vertical and their shapes are consistent. However, the actual groove depth is slightly smaller than the depth calculated from the two-dimensional simulation model owing to the non-ideal fluence distribution of the light spot and the presence of plasma clusters generated during processing. These results indicate that the model produces better results with fewer processing pulses.

This study proposes the use of a DOE element to shape a femtosecond Gaussian spot with a wavelength of 517 nm into a top-hat square spot to achieve laser grooving on silicon surfaces. First, a theoretical model for the femtosecond laser ablation of silicon materials is established. The simulation results demonstrate that the femtosecond laser can excite a significant number of free electrons within the pulse duration, causing the electronic system temperature to surpass the damage threshold, while the crystal lattice temperature remains constant. A flat-top Gaussian distribution function is utilized to establish a two-dimensional flat-top spot ablation model, and the laser single-pulse ablation depth is calculated. The results indicate that the ablation groove shape of the flat-top square spot closely resembles an inverted trapezoid, with a high energy utilization rate and a linear relationship between the ablation depth and the number of pulses when the number of pulses is small. A self-constructed silicon wafer femtosecond grooving system is used to adjust the grooving process by modifying the processing speed of the galvanometer and laser power at a constant laser frequency. Under the conditions of an incident light power of 22 W and a stage speed of 2000 mm/s, silicon wafer grooving is obtained with a high groove bottom level, good side wall verticality, and a depth of 16 μm. The experiments demonstrate that the use of flat-top laser grooving significantly improves the processing efficiency and spot utilization, with a flat groove bottom and vertical side wall, which are advantageous for subsequent processing technologies.

In electronics and precision devices, achieving reliable connections for composite joints of thin metal sheets such as Al/Cu is critical, as this helps reduce Cu usage costs. Conventional welding methods such as brazing and stir friction welding are limited by the sample size and tend to introduce excessive heat input into the joint. Thus, they are not suitable for joining thin metals. Suitable methods for joining thin metal sheets include ultrasonic and laser welding. Laser welding, particularly short-pulse laser welding, has precise heat input, high energy density, and good controllability, making it suitable for connecting thin sheets. Accordingly, in this study, a short-pulse laser is used to weld Cu/Al/Cu laminar joints. The effects of scanning speed on the joint microstructures and properties are investigated to provide guidance in joining laminar metal sheets.

A nanosecond-pulsed laser is used to join Cu/Al/Cu laminates. The laser scanning speed is used as a major parameter that affects the joint quality. First, the effect of scanning speed on weld formation is studied. The microstructural evolution of typical joints is then analyzed. Finally, the effects of welding speed on the mechanical properties are investigated, and the fracture mechanism of the joints is elucidated.

In terms of weld formation, the heat-affected zone at the joint edge and the penetration depth both decrease as the welding speed increases (Fig.3). The pores in the weld migrate upward as the melting depth decreases, and the joint quality improves when the welding speed is 21 mm/s (Fig.4). In terms of microstructures, the interface produces numerous bulk CuAl₂ phases at a low welding speed (15 mm/s) (Fig.6). As the welding speed increases, the number of interfacial compounds decreases. In addition, the compound Cu₉Al₄ phase appears near the copper base material, whereas the part near the aluminum is mainly subeutectic, eutectic, and perieutectic structures composed of α-Al and CuAl₂ phases. When the welding speed reaches 27 mm/s, the Cu₄Al₃ phase appears in the Cu clusters (Fig.6). In tensile tests, the maximum shear force of 95.1 N is obtained in the joint at a welding speed of 21 mm/s (Fig.9). Following fracture analysis, three fracture modes are observed. When the welding speed is low, the joint experiences the tearing failure, which is caused by deep longitudinal cracks in the outer ring of the welded joint due to considerable heat input, and brittle compounds such as CuAl₂ are produced at the weld edges (Figs.10 and 11). As the welding speed increases (21 mm/s), the cracks and porosity defects in the weld decrease, and the joint has a certain melting depth, resulting in a combined failure mode (weld shear and nugget pullout failure) in the joint (Fig.11). As the welding speed becomes excessively high, the molten nucleus has difficulty resisting the tensile shear force because of the smaller penetration depth, resulting in a button pullout fracture mode.

A nanosecond-pulsed laser is used to weld three-layer Cu/Al/Cu lamellar structures, where defects such as porosity and cracking at the joint interface can be reduced by adjusting the welding speed. The joint formation quality is found to be excellent at a welding speed of 21 mm/s. The diffusion of Al into the upper and lower Cu base materials to produce nail-shaped areas improves joint strength. The Al/Cu binary compound type is found to be related to the welding speed. The CuAl₂ phase is prevalent in the Cu welds, whereas Cu₉Al₄ and Cu₄Al₃ are present only near the Cu base material or Cu clusters. The maximum joint shear force of 95.1 N is obtained at a welding speed of 21 mm/s. Three types of fractures occur in the Cu/Al/Cu joints at different welding speeds, namely, joint tearing, mixed-mode fracture (weld shear and nugget pullout failure), and button pullout failure. The nail-shaped structure has a hindering effect on joint fracture.