This article is to capture some of the important developments in the rapidly growing areas of laser-based manufacturing and materials processing and also to describe important technological issues pertaining to various laser-based manufacturing processes. The topics to be covered in this paper include more popularly used processes in industry such as laser additive manufacturing, laser-assisted machining, laser micromachining, laser forming, laser surface texturing, laser welding, and laser shock peening, although there are several additional areas of laser applications. In each section, a brief overview of the process is provided, followed by critical issues in implementing the process, such as properties, predictive modeling, and process monitoring, and finally some remarks on future issues that can guide researchers and practitioners.
Recent years have witnessed a rapid growth of the laser market. Figure 1 shows the recent revenues of laser sales with about 50% increase in revenue over the recent four years , which far exceeds the growth of other industrial sectors. More than one-third of this revenue belongs to the category of laser materials processing arena. Undoubtedly, lasers have been playing increasingly important roles in various industrial manufacturing sectors. This paper is intended to highlight some of the important and growing areas of laser applications in manufacturing and materials processing.
1 Laser Additive Manufacturing
1.1 Introduction and Types of Additive Manufacturing.
Additive manufacturing has been gaining rapidly growing attention over the last few decades, and recent years have witnessed a rapid increase of the market and a vast amount of research. Lasers have been the increasingly important core of additive manufacturing, with laser-based additive manufacturing systems accounting for more than half of metal additive manufacturing market revenues in recent years as shown in Fig. 2 . The global market of the additive manufacturing market is estimated to be $774M in 2019 and projected to reach $3.2B by 2024 . Undoubtedly, laser-based additive manufacturing has become a very important application area of lasers.
In this section, the relevant activities on laser-based additive manufacturing are described in terms of different additive manufacturing processes, properties, predictive modeling, and future issues. There have been a number of additive manufacturing processes that have been introduced to the market, which are based on material extrusion, powderbed fusion, vat photopolymerization, material jetting, binder jetting and sheet lamination, and directed energy deposition, according to ASTM International [4,5]. Among these, only the representative laser-based additive manufacturing (AM) processes, i.e., stereolithography (SLA) or vat photopolymerization, selective laser sintering (SLS), selective laser melting (SLM), and directed energy depositions (DED), are discussed in this paper. Additive manufacturing can produce parts with complex shapes from design drawings. Sample parts produced by AM are illustrated in Fig. 3.
Stereolithography (SLA) is one of the oldest additive manufacturing processes, first patented in 1984 . In SLA, a ultraviolet (UV) laser light is focused onto a certain spot and scanned by a galvanometer to photo polymerize the liquid polymers with photoinitiators voxel by voxel. Since the photoinitiators exhibit high absorption in the UV wavelength range, the UV laser beams are mostly used for this process. Once the planar shape is created by polymerization, then the liquid resin bath is lowered and another layer is formed, thus forming 3D structures layer by layer. For most resins, the critical UV laser energy density between 4.3 and 7.6 mJ/cm2 is needed , thus making a relatively low-power laser sufficient for this process. The process is limited to creation of 3D structures of polymers. Two-photon polymerization (TPP) can create nano-sized features using a similar principle . Femtosecond (fs) lasers with a high numerical aperture objective lens are mostly used for two-photon polymerization because it requires high photon energy and short pulse duration to get good resolution. An example of a part created by TPP is shown in Fig. 4 .
Selective laser sintering (SLS) and selective laser melting (SLM) share similarity in the process as they both use a powderbed to lay a layer of powder and use subsequent scanning of the powderbed surface with a laser according to the programmed scanning pattern of the slice of the 3D part to be built. The difference between SLS and SLM is whether the powders are partially melted or fully melted. Both processes can be used for a wide range of materials, including polymers, ceramics, and metals. Selective laser sintering was introduced and patented by Beaman and Deckard in the mid-1980s , while SLM was invented by the Fraunhofer ILT in 1995  and is now one of the fastest-growing AM processes .
Directed energy deposition (DED) on the other hand uses either blown powder or wire fed into the laser heating spot. The powder focal spot can coincide with the laser focal spot or can be offset depending on the nozzle design and applications. DED can also be used for a wide range of materials and especially can provide advantages in creating functionally gradient structures by varying the mixing ratios of multiple powders or selectively adding materials to existing parts such as in cladding and repair. While various materials have been used for AM, this review mainly focuses on metal additive manufacturing, although metal AM shares many similar attributes to AM of nonmetallic materials.
1.2 Properties of Laser Additive Manufacturing–Built Parts.
Due to the local heating and cooling of laser additive manufacturing processes, the parts built by AM inherently exhibit heterogeneous microstructures, often with defects such as porosity and cracks, varying phases and grain structures and high residual stresses. Therefore, AM-built parts exhibit quite different mechanical properties than the counterparts produced by conventional processes such as casting, forming, and machining. Ensuring the quality of the AM parts is still a challenge for functional parts and real structures. This is because most AM processes are still in the development phase, and as a result, there exist no standard process parameters for each material undergoing AM, particularly when they are built on different platforms. Accordingly, any changes in the parameters may affect the thermal history during the process, and consequently, it is possible that some defects may be induced in the microstructure of the manufactured parts. Such defects can degrade the mechanical performance of the parts and consequently can prevent them from being used in their intended applications . Examples of defects can include porosity, inhomogeneity in the microstructure, and surface roughness. Porosity is the undesirable microscopic voids in the AM-built parts, which can result from insufficient input energy, incomplete melting of the powder particles, and unbalanced melt pool [14–17]. On the other hand, the microstructure implies the phases of the material and the morphology of the grains inside these phases at room temperature, which controls the deformation behavior when subjected to external loading. A large number of efforts have been made to quantify the resultant mechanical properties based on the microstructural information. Many studies showed that the rapid localized heating and cooling can result in the formation of undesired microstructure as well as non-uniform grain size, which can degrade the mechanical behavior [18–20]. Furthermore, the build orientation of the AM has a big influence on the morphology of the microstructure and porosity due to the variation in the cooling rates, which can affect the material response [14,18]. In addition, the parts manufactured by AM and especially SLM can have higher surface roughness compared to those of conventional manufacturing process due to the incompletely melted particles on the surface in addition to the successive molten pools forming the parts, which results in overlapping peaks and valleys [20,21].
The important mechanical properties of concern include anisotropic elastic modulus, ultimate strengths in tension and compression, ductility, and fatigue life, which are dependent on the build parameters and build directions. Due to the variation of properties, local variations of temperature, and cooling rates, the homogeneity of mechanical properties cannot be guaranteed either. Especially, the defects and porosity as well as surface roughness are known to greatly affect the resultant mechanical properties. Therefore, achieving desired mechanical properties has been the main challenge in the industrial adoption of metal additive manufacturing. To alleviate these problems, additive manufactured parts often go through post-processing treatments such as furnace sintering, hot isostatic pressing, and machining. Depending on the AM process and platforms, build direction, build parameters, and post-processing conditions, even the same material shows a wide range of properties in ultimate tensile strength, yield strength, and ductility. Such issues were illustrated by Liu and Shin  with an extensive review of mechanical properties of Ti6Al4V parts built by different AM processes, as reported by various researchers, which showed a wide range of property variations. At this point, there are no generally agreed mechanical properties of AM-built parts for specific material, which creates a great challenge for the qualification of AM-produced parts. Various efforts using predictive modeling, experimental testing, and data-driven modeling have been made to address these issues in recent years.
1.3 Predictive Modeling of Laser Additive Manufacturing.
One of the critical requirements in using additive manufacturing, especially for metals, is to achieve the desired mechanical properties. Since additive manufacturing involves many variables that affect the process conditions, an attempt to determine the resultant mechanical properties only by experiments could be time-consuming and expensive. To alleviate this problem and also gain insights into the process, many attempts to develop predictive process models have been made. The predictive models for AM can be largely divided into three categories: AM process thermal models, microstructure prediction models, and mechanical property prediction models. Important recent progresses in each of these areas are described in the following sections.
1.3.1 Modeling of Laser Powderbed Fusion Processes.
Predictive modeling of laser powderbed fusion processes (LPBF) involves two critical physical processes: powder placement in the powderbed and laser selective melting. Powder placement has been mostly modeled using the discrete element method (DEM), which could capture the interactions of powders in a large-scale system [22–25]. The next issue is dealing with the interaction between incident laser beam and the placed powder. This problem has been handled using either the so-called ray-tracing method [26,27] that keeps track of incident laser rays interacting with the powders throughout the powderbed or the volumetric absorption method based on the Beer–Lambert law [28,29]. While the ray-tracing method provides a better physical representation, it is much more computationally intensive. Besides, the ray-tracing method requires the configuration of particles to be known to determine the trajectories of incident rays. The volumetric absorption method requires some empirical data to calibrate the model depending on the powder material, powder size distribution, laser wavelength, etc., but is computationally efficient. The third portion is handling of the molten pool dynamics and associated heat transfer to the surrounding regions while accounting for laser-beam interaction with the powder and the substrate (or previously formed tracks and layers), including denudation. This has been carried out by using analytical models, conduction models, or solving fluid motions with heat transfer by finite element methods or computational fluid dynamics (CFD) modeling solving full hydrodynamic equations using the finite volume method. Analytical models using a moving heat source [30,31], while simple, ignore many physical phenomena and properties such as fluid motion and temperature-dependent properties. Finite element modeling approaches have been used by a number of researchers [32–36], which however neglected melting and fluid motions as wells interaction with the powders. While they provide quick solutions, the prediction accuracy is less than desired. Finite volume methods have been used to couple the fluid motions in the molten pool coupled with heat and mass transfer equations, using commercial software such as flow3d [22,37,38] or fluent . The use of such commercial codes, however, does not allow for the utilization of the optimal algorithm for the interaction of the laser with the powder, powder melting, and consolidation. Another prominent model describing the LPBF is the work of Lawrence Livermore National Lab, which uses the in-house developed software, ale3d, to solve the powder melting and the resultant densification, and is coupled to the effective medium model which is based on the finite element method [40,41]. An example of their predictions is shown in Fig. 5 . However, the work was limited to a single-track case and the very high computational cost makes it difficult to apply to a large domain practical problem at present.
1.3.2 Modeling of Directed Energy Deposition.
Modeling of directed energy deposition requires handling the feeding of material in blown powder or wire stock into the molten pool. Some researchers attempted to predict temperature distribution and track heights without considering fluid dynamics [42–44] and considering fluid motions in the molten pool [45,46] using commercial code COMSOL. Finite element analysis suffers from one major drawback: only global conservation of solved parameters is guaranteed. In addition, it uses element activation and deactivation to predict track geometry instead of predicting free surface based on fluid motion and interaction of fed material with the molten pool. In finite volume solutions, local conservation can be guaranteed. Furthermore, in finite element methods, flux limiting is not feasible, leading to unstable solutions depending on the discretization scheme utilized, and thus, finite volume methods have been more popularly used for the directed energy deposition modeling.
In directed energy deposition, it requires modeling of material addition during the process. When blown powders are used, then powder flow emerging from the nozzle and its interaction with the incident laser and the molten pool must be simulated. Blown-powder heating and attenuation were investigated analytically by Diniz Neto et al. , who presented analytical treatment of blown-powder heating and attenuation, but they assumed their powder distribution to follow two prescribed powder mathematical representations. Pinkerton  also presented an anlytical model for the powder flow and temperature distribution of particles ejected from a coaxial nozzle considering particles of a single diameter. Wen et al.  performed a computational fluid dynamics simulation of the powder flow from a coaxial nozzle and included a particle heating model. In their analysis, a Rosin–Rammler particle distribution was considered, which more closely resembles the particle size distribution observed in manufactured powders and could be applied to practical AM processes. Lin  used fluent, a commercially available numerical computational fluid dynamics software package, to simulate the powder flow emerging from the nozzle using a gas–particle mixture known as a discrete phase model. Modeling of the wire fed directed energy deposition process was also presented by Wei et al. .
The molten pool dynamics of the directed energy deposition is solved similarly to the case of the laser powder bed fusion process. One major difference is that the modeling of deposited track geometry needs to be carried out, considering the addition of feedstock material to the molten pool. Perhaps, the most sound approach would be CFD-based finite volume methods, considering energy, momentum, and continuity equations simultaneously with free surface tracking. He et al.  developed a numerical CFD model to study the temperature, fluid flow, and species transport for two-track laser cladding of H13 powder onto an H13 substrate. However, some critical assumption and the lack of experimental data supporting the track geometry and molten pool boundary limited its general applicability, such as constant thermal and physical properties, assuming small (∼10 µm diameter) and uniformly sized spherical powder which melted upon contact at the liquid surface, etc. The most generically applicable work came with Wen and Shin  who developed the more general frame of the DED process model with the capability to handle temperature-dependent thermal and physical properties, an arbitrary laser irradiation source, and arbitrary powder size and distribution. By coupling powder concentration predictions with the CFD molten pool dynamics model using the level-set function to track the free surface, they predicted free surface, molten pool, and heat-affected zone boundaries, which agreed well with experimental results for varying conditions. Wen and Shin  subsequently extended their single-track model to a multi-track simulation, which dealt with more complex blown-powder depositions. Manvatkar et al.  built a numerical model that predicts the temperature and approximate geometry during the process and used it to analyze the deposition of a thin-walled structure of 316SS by including a powder heating model with temperature-dependent thermal properties. In their analysis, particles with large diameters (175 µm) were used, which is known to increase surface roughness .
1.3.3 Modeling of Microstructure.
Due to localized rapid heating and cooling, additive manufacturing yields very heterogeneous microstructures, which will govern the resultant mechanical properties. Prediction of resultant microstructure such as grains and phases has been much of the focus in the AM community in recent years. Currently, there are three popularly used methods for the prediction of microstructure: cellular automata (CA), phase-field (PF) modeling and the Monte Carlo (MC) method. Each of these methods provides advantages and disadvantages.
The CA models are computationally efficient and hence can be used to predict the microstructure evolution of large domains . Meso-scale CA models have been used to predict the resultant microstructure. Coupling CA models with the finite element model, several researchers attempted to predict the dendrite growth in 2D [57–60] and also in 3D diffusion [61,62]. All these studies based on finite element models however did not capture the AM process accurately because they assumed a fixed heat source without melting instead of a moving heat source. More realistic cases with the use of a moving heat source with the molten pool dynamics were presented by Shi et al.  and Tan and Shin . However, they have been mostly used for binary alloys and require growth kinetics data, which are often acquired from the experimental data.
The Monte Carlo (MC) method has also been used to predict grain distributions with the user-defined molten pool [65,66]. However, the MC model does not provide the means of directly coupling thermal history during solidification and does not account for texture and anisotropy . A kinetic Monte Carlo method has been implemented in the open-source code SSPARKS.1 SSPARKS based on the Potts kinetic Monte Carlo technique has been used to predict microstructure evolution during additive manufacturing .
Phase-field (PF) modeling, which was designed to solve interface dynamics, has been used to simulate solidification processes. The PF method is based on a set of partial differential equations, and thermodynamic principles use CALPHAD for diffusion coefficients and free energies. Most PF simulations in the literature have been carried out under ideal or simple conditions, i.e., pure substance or simple binary alloys, an isothermal temperature field or imposed temperature gradient and cooling rates, although it is applicable to more complex materials with transient temperature fields. Suzuki et al.  developed one of the first multicomponent PF models based on their earlier work for binary alloys . Karma  addressed some inaccuracies associated with determining concentration fields due to solute trapping, particularly during fast cooling cases such as in additive manufacturing, and introduced “anti-trapping current term” to alleviate excessive solute trapping. Some notable PF codes include Tusas [70,71] and AMPE [72,73]. Some recent cases of 2D PF modeling include those by Kundin et al.  for the selective laser melting, Karma and Tourret , Nie et al. for AM of nickel-based superalloy , Sahoo and Chou  for prediction of microstructure evolution of Ti6Al4V, Liu and Wang  for the simulation of β-grain solidification during the SLM process for Ti6Al4V and Liu and Shin  for Ti–TiC composites. The major shortcoming of PF modeling is the high computational cost, and hence, PF modeling has been carried out on very small domains of a few tens of microns, often in 2D configuration although the actual microstructure should be 3D in nature. Recently, some 3D PF modeling efforts have been carried out by some researchers [80–83], who demonstrated that 3D modeling should be used to correctly represent the actual 3D microstructure. The challenge associated with the prediction of the 3D microstructure is that even the fusion zone in AM is too large for the PF simulation and large-scale parts produced by AM processes are a way beyond the computational capability at present. Therefore, PF simulations at present can show only the snapshots of small local regions.
A novel scheme to overcome the prohibitive computation requirements of PF modeling has been proposed by Tan et al. , who combined the respective advantages of cellular automata and phase-field modeling, the so-called cellular automata-phase-field (CAPF) modeling. The CAPF utilizes the fast computation capabilities of CA with the PF model calculating the local growth kinetics. Therefore, without losing the fidelity of physics, the computational speed was improved by five orders of magnitude, thereby offering promise for prediction of microstructure in a large domain covering the entire molten pool or even the entire part. They used the CAPF model to simulate the dendritic growth of multicomponent alloys in the entire fusion zone during the laser cladding process .
1.3.4 Prediction of Mechanical Properties.
There is a large variation among the parts produced by AM due to a large number of process parameters and the processing environment as well as initial powder characteristics . Even within a single part, the microstructure heterogeneity and porosity vary due to the geometrically sensitive thermal conditions the part is experiencing during building as surrounding temperatures and boundary conditions vary with changing part geometry and laser scanning patterns. It is also well known that additively manufactured metal parts exhibit directional properties, viz., horizontal versus vertical . Furthermore, additive manufactured metal parts are known to have some porosity. Most of the efforts thus far have been empirical, involving a large number of experiments [86–88]. Determining the effects of porosity and microstructure variations solely based on experiments will not only cost a significant amount of money but also a very long lead time. Therefore, there is an urgent need to develop predictive modeling capabilities of mechanical properties considering the heterogeneous microstructure and defects as well as surface roughness.
Some of the efforts considered the topological aspect of AM-built parts and tried to predict the associated mechanical properties. Munford et al.  presented prediction of anisotropic mechanical properties such as apparent modulus and strength for lattice structures based on structure density and fabric, using well-known power–law relationships. Others tried to study the effects of porosity on mechanical properties. For example, Ebrahimi and Mohammadi  used a combined FEM-analytical approach to predict the fatigue life of DMLS-MS1. Mechanical properties of porous materials have been modeled by Ahmadi et al.  who adopted a vector technique to predict mechanical properties of porous biomaterial and Campoli et al.  for open-cell biomaterials. Despite some of these efforts, truly predictive models accounting for the actual morphology of microstructural details and porosity are still lacking.
1.4 Future Directions.
Metal additive manufacturing faces great challenges that need to be overcome to be accepted as an economically viable manufacturing process in the industry, but at the same time offers unprecedented opportunities for manufacturing new breeds of products that were not feasible to build by traditional manufacturing processes. Obviously one of the grand challenges is to build process–microstructure–property relations either by physics-based modeling or data-driven approaches to facilitate the qualification process of additively manufactured parts. In addition, reliable in-process monitoring methods must be developed at the same time. The new opportunities that laser metal AM provides include manufacturing of mass-customized parts such as medical implants, functionally gradient parts with desired local properties, topological design for fabricating smart or metamaterial structures, geometrically complex parts such as heat exchangers and synthesis of novel materials. In order to exploit these opportunities, people across multiple disciplines need to work together to generate new designs and materials that can accelerate the adoption of AM in the manufacturing industry.
2 Laser-Assisted Machining
Laser-assisted machining (LAM) uses a laser as a heating source to elevate the workpiece temperature at the cutting zone to facilitate the material removal process by a cutting tool in an otherwise conventional machining process, as illustrated in Fig. 6 . The purpose of laser heating in LAM is to reduce the shear strength and enhance the ductility of the workpiece material and thus to assist the cutting tool in the material removal process. Ideally, only the material at the shear zone should be rapidly heated to a sufficiently high temperature and most of the heat input will be taken away by the chip. The expected gains from LAM include reduced cutting force and power consumption, improved surface finish, longer tool life, improved subsurface integrity, and higher productivity. World-wide attention in recent years has propelled LAM to a new level. This section aims to report the advances in LAM over the past few decades.
For at least 50 years, it has been recognized that, as the temperature of a material increases, its shear strength decreases, thereby reducing strain hardening effects and the energy expended in the cutting zone [94–96]. Various hot machining techniques have been attempted, including the use of furnaces, torches, and electric resistance or inductance heating . However, such techniques have not been put into practice due to the major deficiencies stemming from insufficient localization and intensity of the heating agent and hence the inability to adequately control the temperature in a small volume of the workpiece. In the late 1970s, laser-assisted machining with the advantage of precisely controllable heat input was considered for mainly difficult-to-machine metals such as Inconel 718 and Ti6Al4V [98,99]. However, these earlier studies did not lead to positive conclusions due to the high cost of lasers and insufficient economic benefits and the interest waned until the 1990s. The pioneering concurrent studies in the 1990s and 2000s by the Fraunhofer Institute of Production Technology in Germany  and Purdue University [93,101–104] in the US clearly demonstrated the benefits of LAM in machining engineering ceramics and revitalized the global interest in LAM. The reduction in hardness with laser heating significantly improved the machinability of ceramics and made it possible for single-point cutting operations. Enhanced material plasticity eliminated tool sudden failure, promoted plastic flow, and produced a smooth surface finish without many micro-fractured zones compared to room temperature grinding (Fig. 7). Since then extensive studies have documented the beneficial effects that can be gained when applying LAM to a wide range of difficult-to-machine materials, such as high-temperature alloys, ceramics, metal matrix composites, and high hardness materials.
In recent years, the material base for LAM has expanded to a wide range including ceramics [93,104–109], high hardness steel [110,111], stainless steel , titanium alloys [113,114], nickel-based alloys , metal matrix composites [116–119], glass , semiconductors , and ceramic matrix composites . The need for components with diverse geometric shapes has pushed the application of LAM from the dominant turning operation to milling [114,123–125] and micromilling [126–129], drilling , orthogonal cutting , and grinding [132,133]. Over the years, the LAM field has attracted various research groups in the world and their engagement has advanced the field in various fronts as evidenced by the large number of studies reported every year. Table 1 provides an overview of the types of machining methods, lasers, cutting tools, and workpiece materials that have been used in the LAM studies, while Table 2 shows some of the benefits identified as results of implementing LAM.
The economic feasibility of LAM has been a concern from the beginning in the 1980s, mainly due to the high cost and low efficiency of high-power CO2 lasers at that time. Since then, improvement in the laser technology has resulted in a significant reduction in the equipment cost, improved efficiency, and a greater variety of wavelengths to choose from. For example, metals absorb a greater amount of incident laser energy at 0.8–0.9 µm wavelength (diode laser) or 1.06 µm wavelength (fiber and Nd:YAG lasers) than at 10.6-µm wavelength emitted by CO2 lasers. Studies have shown that cost reductions in LAM over conventional processes can be achieved for both metallic components and ceramic materials .
Experimental investigations remain a dominant portion of the LAM literature. The pioneering systematic work at Purdue University [93,102,104] provided the means of precisely controlling temperature distributions in the workpiece, particularly near the shear zone. The three-dimensional transient numerical models allowed for the prediction of temperature fields in terms of laser and machining operating conditions, thereby moving the design of LAM conditions from empirical approaches to the science-based process. At the same time, in-process temperature measurement by using a pyrometer was introduced to maintain the desired temperature during LAM . Subsequently, the thermal model was extended to LAM turning processes with complex workpiece shapes  and milling processes .
New thrusts in the experimental studies have emerged over the past years in the areas of new LAM processes, new laser heating strategies, energy efficiency, and process optimization. For example, a microscale laser-assisted machining (µ-LAM) process was developed by Mohammandi et al.  for high precision machining of SiC. As illustrated in Fig. 8, the laser beam is delivered through a single-point diamond tool toward the cutting zone, which is a novel way for beam delivery. The surface finish of optical quality was achieved using µ-LAM. A different configuration of a laser-assisted micromilling was proposed by Shelton and Shin [127,128]. A new heating strategy in LAM was reported by using distributed lasers for laser-assisted turning  and using a combination of laser and induction heating by Woo and Lee . The dual heat source provided a more uniform temperature at the cutting zone and was shown to be most effective in tool wear reduction. Another new heating method was reported by Shang et al.  based on spatially and temporally controlled laser scanning. The benefit includes a large heating area with a small laser spot size and is free from workpiece overheating below the depth of cut.
Modeling and simulation have been an active topic in the LAM literature since the 1990s. Although analytical modeling continues to be attempted for LAM, numerical approaches are usually used to treat the highly nonlinear nature of the process. A transient, three-dimensional thermal model was first developed by Rozzi et al.  for laser-assisted turning of silicon nitride ceramics. A comprehensive treatment was made to account for the incident laser flux distribution, conduction within the workpiece, convection and radiation on all surfaces, heat and mass loss resulting from material removal, temperature-dependent thermophysical properties of the workpiece, and heat generation at the machining zone due to plastic deformation and friction. This model was extended later by Pfefferkorn et al.  to LAM of semi-transparent, homogeneous materials. Tian and Shin  extended the thermal model to account for complex geometry and material removal using a Lagrangian method in laser-assisted turning and Tian et al.  developed another such a model for prismatic parts in laser-assisted milling.
To handle both the heating and chip formation in LAM, coupled thermo-mechanical modeling and simulation studies have been reported. Dandekar and Shin  carried out a simulation of laser-assisted machining of Ti6Al4V in terms of laser heating conditions and showed the reduction of cutting forces with the increasing shear zone temperature. Ding et al.  carried out thermal and mechanical modeling analysis of laser-assisted micromilling of Ti6Al4V, Inconel 718, and AISI 422, and predicted the temperature increase in the shear zone due to laser heating, chip formation, and cutting tool temperature using the thermal model developed by Tian et al.  and strain gradient constitutive material models. Shen et al.  used a hybrid approach for LAM of silicon nitride ceramics where finite element analysis (FEA) was used for thermal modeling to determine the temperature field in the workpiece while a distinct element method was used to treat the initiation and propagation of a large number of microcracks in the material removal process. Xi et al.  also used a hybrid approach by combining the smoothed particle hydrodynamics method for large deformation analysis and FEA for thermal modeling to simulate laser-assisted machining of a beta titanium alloy. The finite element method remains the most commonly used means for LAM simulations. Tian and Shin  were among the first to treat the granular microstructure of silicon nitride using continuum elements embedded in thin interfacial cohesive elements in a multiscale FEA model. The continuum elements simulated the deformation of the bulk workpiece while the interfacial cohesive elements accounted for the initiation and propagation of intergranular cracks. The model revealed that discontinuous chips form by the propagation of cracks in the shear zone while the machined surface is generated by plastic deformation of the workpiece material under confined high pressure. The multiscale FEA model was later applied to LAM of a fiber-reinforced metal matrix composite  and LAM of a particulate reinforced metal matrix composite to investigate the subsurface damage as illustrated in Fig. 9 . To better model the interfacial layer, molecular dynamics simulations were performed to obtain the atomistic structures and parameterize traction-separation laws for the aluminum–silicon carbide interface. The interfacial model was then coupled into the FEA model to simulate damage formation in the LAM process.
2.4 Future Directions.
Laser-assisted machining is now a rather mature processing technology since its capabilities and benefits have been fully demonstrated. However, the industrial implementation requires robust and flexible operation of laser-assisted machining. Therefore, flexible and portable methods of delivering a laser beam or multiple laser beams to the target with respect to the cutting tool position and movement direction need to be developed. This is particularly important for processes such as milling and profile turning of internal surfaces. In addition, in-process monitoring capabilities to ensure the optimal operating conditions when machining complex parts, particularly with unknown surface profiles or properties, must be developed to avoid unexpected failure. Multiscale modeling capabilities that can predict subsurface damages and surface integrity, although not limited to laser-assisted machining, need to be continuously developed to avoid expensive post-inspection and trial and errors.
3 Laser Surface Texturing
Laser surface texturing is a surface engineering process that uses a laser to create periodic microstructures on the surface of a material to induce desired surface properties targeting various applications. Early studies on laser surface texturing in the 1990s used lasers to create patterned micro dimples and investigated the effects of a textured surface on tribological performances of mechanical components including mechanical seals, piston rings, and thrust bearings [147,148]. Since then, this field has grown extensively together with the rapid development of laser technologies, and a wide range of applications have emerged well beyond the tribological domain. This section aims to report the advances in laser surface texturing over the past two decades.
3.2.1 Laser Surface Texturing Methods and Diverse Patterned Microstructures.
Laser surface texturing is commonly performed through direct laser irradiation of a thin layer of surface material. Usually, one laser beam is employed and the laser scans the workpiece surface at a specified speed. For high-speed surface texturing, optical laser scanning can be used. Laser interference-based surface texturing methods are often used to create well-defined long-range micro- and submicron structures. As illustrated in Fig. 10, two laser beams propagating at an angle overlap on the sample surface and form an interference pattern . The modulated intensity profile is used to produce a patterned structure with the same geometry on the surface. More recently, a new texturing method called shifted laser surface texturing has been developed to enable large-area high-speed surface structuring with high precision and without heat accumulation by Martan et al. . As shown in Fig. 11, the entire array of objects is produced with only linear laser raster plus sequential shifting. This method is suitable for patterning a large surface with small complex features which render conventional methods time-consuming.
Pulsed lasers (especially ultrashort pulse lasers) are normally used for surface texturing work. Among the diverse surface textures created, the most commonly observed structure is laser-induced periodic surface structure (LIPSS), also known as surface ripples. LIPSS was discovered more than 50 years ago by Birnbaum  with a ruby laser. Since then, this area has continued to evolve with the development of laser technology and remains an active research area with numerous new developments being reported in recent years. LIPSS can be created on metals [152–154], semiconductors [155–157], polymers , and dielectrics . Figure 12 shows two examples of LIPSS structures created using fs laser. The LIPSS structures are generally oriented perpendicular to the laser-beam polarization direction. By changing the laser beam to elliptical polarization, the ripples can be rotated; for circularly polarized beam, uniformly distributed bumps can be produced. LIPSS can fall into one of the two types: low spatial frequency LIPSS (LSFL) and high spatial frequency LIPSS (HSFL). The spatial period of LSFL is close to the laser wavelength, but it can be smaller or larger depending on the laser wavelength, incident angle, laser fluence, and material type [153,159]. Unlike LSFLs that occur for all major material types, HSFLs appear to be a phenomenon for dielectrics and semiconductors only. HSFLs have spatial periods much smaller than incidence laser wavelengths and are usually produced using ultrafast lasers below a material's ablation threshold . On the theoretical front, Bonse et al.  and He et al.  further advanced an early, widely accepted theory developed by Sipe et al.  to explain the formation of LIPSS. A notable contribution was made recently by Gnilitskyi et al.  with the creation of long, highly uniform LIPSS (Fig. 12(b)) and the discovery of its origin. They found that LIPSS regularity is governed by the decay length of the excited surface electromagnetic waves, and the shorter the decay length, the more uniform the LIPSS is.
Some other examples of laser-textured surfaces are shown in Fig. 13, among which the conical structure on titanium (Fig. 13(a) ) represents another widely known laser-induced ordered surface structure. Generally speaking, the necessary conditions for micro spike/cone formation involve relatively high laser fluence and a large number of laser shots (e.g., hundreds), and normally, LIPSS is created first after a small number of shots before the transition to conical structures occurs [166,167]. According to Carey et al. , micro spikes on silicon are formed after tens or hundreds of laser shots at about 100 fs pulse duration and medium fluence levels (e.g., 1–3 J/cm2). The first few laser pulses induce micro craters/ripples scattered over the surface based on the initial surface morphology. The subsequent laser pulses strike this roughened surface and cause preferentially more absorption in the valleys, resulting in conical microstructures after a certain number of shots. To produce such textures economically over a large surface, single scan with overlapping pulses can be used. Interference-based methods have been used to gain more control and precision of the texture geometry. Using two-beam interference, uniform long-range microscale island-like structures are produced by overlapping laser scans at a 90-deg angle, as shown in Fig. 13(b) . With a single shot of four interfering fs laser beams at each spot, circular bump arrays on a gold film are formed . In addition, ellipsoidal bumps form with three beams and linear bumps form with two beams. Other beam shaping techniques for surface structuring include the use of blazed gratings and spatial light modulator (SLM). High-quality and more uniform LSFLs on a few metallic materials are produced using spatiotemporally modified fs pulses with a grating pair .
It should be noted that although surface texturing is generally processed with ultrashort laser pulses nowadays, long pulse durations such as nanosecond (ns) and microsecond (µs) lasers have also been frequently used in creating surface textures. The difference is that because of the thermal effects, long pulses are more suitable for creating microstructures of tens to a few hundreds of microns in size, which is evident from the numerous recent studies of using ns and µs lasers to create microdomes , microdimples , micropillars , and microgrooves  for various applications.
3.2.2 Surface Property and Potential Applications.
Laser surface structuring not only alters a material's surface morphology but often imparts some new functionality and properties to the surface, notably optical, mechanical, wettability, and chemical properties. Femtosecond laser-textured silicon covered with micro spikes, often known as black silicon, is among the early findings to demonstrate nearly 100% absorptance in the visible range, which can be extended to 2.5 µm when processed in an ambient SF6 gas environment . Parmar and Shin  found that fs laser creates globular micro/nano structures in air and elongated columnar structure in water, and both produce black silicon with wideband antireflection effect. Tunable reflectivity (Fig. 14(a)) over the wavelength range from UV to mid-infrared (MIR) is achieved with ps laser-induced micro/nano structures on Cu surface . Yang et al.  reported a significant enhancement of thermal emission to about 100% for a NiTi alloy due to a microscale coral-like surface structure produced by an fs laser. Surface textured silicon and metals have a wide range of potential applications including solar cells, detectors, sensors, field emission devices, plasmonics, broadband thermal sources, and radiative heat transfer devices.
Laser surface texturing has also been used to modify and control a material's wetting property. Superwetting surfaces were created using fs laser for silicon and human enamel and dentin tissues by Vorobyev and Guo [180,181]. The structure consists of parallel microgrooves with a period of 100 µm with naturally formed micro/nano scale morphology over the microgrooves. When water is dropped on the processed area, it spreads rapidly along the microgroove orientation and can even climb up the surface vertically. The wetting property of fs laser-patterned surfaces is determined by the combined effect of surface morphology and surface chemistry. Skoulas et al.  demonstrated the effect of surface morphology on wettability for a Ni film with different complex micro/nano structures as shown in Fig. 14(b). Yang et al.  used a ps laser to create a hierarchical texture on a titanium alloy consisting of nano ripples superimposed on microdimple arrays that display superhydrophobic behavior. Materials with tailored wetting properties have potential applications in microfluidics, biomaterials, anti-icing, self-cleaning, etc. For example, Shashank and Shin  developed a fast and inexpensive method to create microfluidic devices with textured superhydrophobic inner channel walls for controlled fluid flow. A textured copper mold transferred the structure to a polydimethylsiloxane (PDMS) device that increased the fluid flow rate by 186%.
Laser surface texturing for tribological applications has been an area of interest for more than 20 years. LIPSS is most effective to reduce friction under regular sliding contact conditions . In contrast, fs laser-textured surfaces with microholes or microgrooves of tens or hundreds of microns in size are effective in reducing friction and wear under both dry and lubricated conditions in machining applications [186,187], where high temperature, high pressure, and high speed occur at the tool–chip interface. This area has been active for about 10 years, with current publications targeting new textured tool development and different machining operations [188,189]. Recent years have seen an increase in studies on laser surface texturing for biomedical applications. There seem to be two thrust areas: (1) biocompatibility enhancement between cell/tissue and textured implant surface and (2) improvement of tribological behavior at the bone–implant interface. Several studies have shown that laser-textured surface can promote cell adhesion, proliferation, and spreading on various biocompatible metal alloys [190–192]. The effects of surface texture on the tribological behavior of orthopedic implants have been investigated, and the main benefits include improved wettability, friction, and wear performance when compared to smooth surfaces [193–195].
3.4 Future Directions.
Laser surface texturing has gone through significant development over the past two decades and remains an active research area at present. This field is driven by the diverse and ever-increasing applications spanning from mechanical to biological systems, and it is a more leveled field because all types of pulsed lasers from µs to fs can play a role in this area. Despite the vast work from more than a half-century, challenges remain and future research could help in the following aspects. First, more theoretical work and numerical modeling and simulation should be developed to determine the mechanisms of experimentally observed phenomena such as the formation of HSFL. Second, scalable micro/nano surface texturing methods should be developed to enable large-area texturing at high efficiency for industrial applications. Finally, more hierarchical and nanoscale texturing techniques are needed for applications in microscale systems such as micro-electro-mechanical systems.
4 Advances in Laser Forming of Metal Foam
Metal foam is a relatively new material that has stimulated interest due to its high strength-to-weight ratio and its excellent shock and noise absorption properties . In many engineering applications such as car bumpers or spacecraft components, metal foam needs to be in specific shapes. Since near-net-shape manufacturing is difficult and expensive, it becomes necessary to bend metal foam to the desired shape. Bending metal foam is not trivial because the cell walls can only sustain low stresses and crack easily. As a consequence, conventional mechanical bending methods cause fracture and cell collapse [197,198]. Within the last decade, several research groups have attempted the laser forming of metal foam and reported positive results. None of the studies have addressed the underlying bending mechanism in sufficient detail. So far, it has been assumed that the temperature gradient mechanism (TGM), identified by Vollertsen  for sheet metal laser forming, is always the governing bending mechanism in metal foam laser forming. In many cases, it is desirable to encapsulate metal foam within a shell of solid metal. Of particular interest are the so-called sandwich panels, where plates of metal foam are “sandwiched” in between solid metal sheets. Potential applications exist for sandwich panels, especially in the aerospace industry. They could in some applications replace honeycomb sandwich panels that are frequently used in airplanes nowadays. Unlike honeycomb structures, which are anisotropic and cannot be readily curved, metal foam is isotropic and can be bent to doubly-curved shapes . The metal facesheets, however, render such sandwiches exceptionally stiff and hard to bend.
Among the four mechanisms through which heat can be transferred in foams: solid conduction, gas conduction through the cavities, natural convection inside the cavities, and radiation heat transfer, the solid conduction is most significant and others can be neglected. Metal foam shows fundamentally different deformation behaviors in tension and compression. In tension, the material can undergo only a small amount of plastic deformation and has a low strength. In compression, the stress–strain curve can be divided into three distinct stages, as shown in Fig. 15 , the linear stage, a large “plateau” where cells collapse, followed by an exponential increase after the foam is fully compressed. Due to the wide plateau, the area under the stress–strain curve is large, explaining why metal foam is an excellent energy absorber. Metal foam bending involves a combination of tensile and compressive deformation. Compared to solid material, metal foam has a very large bending stiffness. The reason is that the moment of area of the foam is more than an order of magnitude greater than the moment of area of a solid with the same net cross-sectional area.
4.2 Numerical Simulation and Experiments.
Three different geometries were explored and are shown in Fig. 16 . The equivalent model [203–205] simplifies the porous structure using a solid but the model uses material properties “equivalent” to these of the foam. In the second model, the foam geometry was explicitly approximated using a Kelvin-cell geometry, whose cell size and pitch can be readily adjusted. The third model, a voxel model, aimed to replicate the exact foam geometry by using a FEM model that was based on an X-ray computed tomography (CT) scan. The second and third models use solid material properties. To model the interfaces of sandwiches, an exceedingly thin layer at the interface consisting of cohesive elements is inserted, with the additional benefit that delamination can be monitored and visualized , even though delamination normally does not occur in laser forming. Closed-cell AlSi7 foam was used. Its volume fraction was 11%, and its density was 279 kg/m3. Temperature-dependent material properties of AlSi7 were extracted from Ref. . Test specimens were of the size of 100 by 35 and 10 mm. Laser-beam diameter ranges from 4 to 12 mm, laser power is up to 120 W, and the laser scanning velocity ranges from 10 to 30 mm/s. Sandwich panels consist of AW 5005 facesheets and AlSi10 metal foam cores and have the same size as the foam specimens with each facesheet of about 1 mm thick and an average density of 700 kg/m3.
4.3 Results and Discussions.
Figure 17 compares the top and bottom temperature history plots of the experiment and the three numerical models during a 2-s laser pulse with a - mm radius at 30 W. As expected, there is a steep temperature gradient throughout the thickness. Like in solid materials, the temperature gradient mechanism (TGM) dominates in foams from the heat transfer point of view. From a mechanical point of view, it is questionable if the metal foam is able to develop the plastic compressive strains as in the solid material, due to its crushability and low compressive strength. Laser forming experiments revealed that no matter how low within the processing window the laser power was chosen, localized melting of thin cell walls was unavoidable. As shown in Fig. 18 [208,209], thin cell walls started melting from the top surface, forming u-shaped trenches that deepened until either the entire cell wall was melted away or the cell wall thickness increased. If metal foam bending indeed occurred due to plastic compressive strains, melting would drastically impede bending, because it reduces the amount of compressible material. Experiments have shown the contrary, however, implying that plastic compressive strains cannot be the major cause of bending. It turns out that cell wall bending occurred close to the bending axis, clearly indicating that the cell walls are unable to withstand high compressive stresses. These findings indicate cell wall bending and cell crushing in metal foam laser forming are equivalent to plastic compressive strains in steel sheet laser forming and we call it a modified temperature gradient mechanism (MTGM).
For sandwich panels, the bending mechanism is yet more involved. Two cases were examined. As seen from Fig. 19, at D = 4 mm and v = 30 mm/s, a steep temperature gradient exists in the top facesheet, indicating that the TGM is always the governing mechanism. At D = 12 mm and v = 10 mm/s, there is hardly any gradient over the top facesheet in both scenarios, indicating that the unsetting mechanism (UM) governs . As a result, the maximum achievable bending angle is very sensitive to the laser spot size. At D = 4 mm, the maximum bending angle is merely ∼15 deg. Large bending angles can only be achieved by performing multiple parallel laser scans. In Fig. 20(a), for instance, two scans were performed per scan line, and each scan line was offset by 1 mm. At D = 12 mm, on the other hand, bending angles up to 65 deg and beyond can be achieved over a single scan line as shown in Fig. 20(b).
4.4 Future Research.
A better understanding should be developed of how laser forming affects the material properties and structural attributes. Cylindrical coupons can be extracted from untreated and laser formed specimens, and compression tests can be performed to quantify losses in the crushability. Impact tests should be conducted on laser formed specimens to quantify its impact-absorbing capability. Additionally, cyclic tensile strength tests can be performed to investigate whether thermally induced changes to the material properties affect the fatigue life of the material.
The effect of the material composition and the facesheet adhesion method on the formability and bending mechanism should be investigated in more detail. A previous study has shown that the outcome of laser forming processes is strongly dependent on the material or alloy that comprises the foam . Adding silicon carbide (SiC) as an alloy element to aluminum, for example, rendered laser forming nearly impossible due to the significant drop in thermal expansion caused by SiC.
A subject requiring further study is the process synthesis of metal foam sandwich panels. For sheet metal, FEM-based methods were formulated based on thin-plate assumptions  to determine the required plain strain field, which in turn allows the synthesis of laser scan trajectory, power, and speed. Those assumptions no longer hold for metal foam, due to the large sheet thickness, and the algorithm needs to be revisited. A possible route is to extend the FEM-based methods to determine the required plain strain field in the top facesheet, which is relatively thin, while additional algorithms are to be developed to determine the bending strain throughout the panel thickness. The synthesis then proceeds to determine process parameters to realize both strain fields.
The equivalent, Kelvin-cell, and voxel models all predicted steep temperature gradients during the laser forming of Al-foam, and this result was both validated by experiments and theory. Metal foam was found to undergo compressive shortening via cell wall bending and cell crushing near the irradiated surface, as opposed to compressive plastic strains that occur in laser forming of solid sheet metal. Based on this deviation from the traditional TGM, a modified temperature gradient mechanism (MTGM) was proposed. For sandwich panels with metal foam core, the top facesheet bending mechanism varies from TGM for smaller laser spots to UM for larger spot sizes. The metal foam core bends via the MTGM for all spot sizes. It was shown that much greater bending angles can be achieved at large spot sizes, due to the reduced impact of facesheet thickening on the bending mechanism efficiency.
5 Laser Micromachining
In this section, laser micromachining is defined as the process of creating or modifying microscale features through laser irradiation of the target material. Laser micromachining often has several advantages, such as high-resolution (due to small achievable laser spot sizes), non-contact and no mechanical tool wear, good flexibility (laser spot trajectory can be easily programmed and varied), and wide applicability (most materials can be potentially micromachined by a sufficiently intense laser beam, either conductive or non-conductive). This section gives a short review of laser micromachining. Due to the space limit, the review is not intended to be comprehensive, but only to present some key points in several related aspects. The cited references may or may not be the first, the latest or the most significant relevant studies in the literature. This review mainly emphasizes on laser micromachining using short laser pulses, particularly (although not necessarily limited to) nanosecond-scale or shorter pulses.
5.1.1 Material Removal Mechanisms in Laser Micromachining.
How material is removed in laser micromachining is often a complicated issue, and one or more mechanisms may be involved, surface vaporization, melt ejection, phase explosion, spallation, Coulomb explosion, critical point phase separation, fragmentation and hydrodynamic expansion, etc. [212–225]. The exact mechanism(s) may depend on workpiece material types and laser parameters.
Obvious surface vaporization can occur when a target surface is melted and driven to a sufficiently high temperature (for vaporization flux to be significant [222,223]), but also sufficiently low (for a sharp liquid–vapor interface to exist). When the target near-surface region is driven to a temperature near the thermodynamic critical point, homogeneous bubble nucleation and growth may occur in the superheated melt and can lead to the ejection of a mixture of bubbles and liquid droplets from the target. This mechanism is the so-called “phase explosion” mechanism [213,220]. However, it should be noted that “phase explosion” is not necessarily the only mechanism that can cause melt ejection from the target. Pressure spatial gradients (e.g., those induced due to the gas-dynamic process right above the target surface) may also drive target melt ejection without phase explosion occurring . When laser intensity is sufficiently high, the target surface temperature can be driven above the thermodynamic critical point, where the material is in a “supercritical state.” The target surface no longer has a sharp liquid–vapor interface; instead, hydrodynamic expansion may occur, where the material may move from a high-density condensed phase to a low-density phase across a transition zone with a finite width (instead of a sharp liquid–vapor interface as for surface vaporization) [217,224].
If a sufficiently high electrical field is developed in a target near-surface region due to laser-induced electron emission (through photoemission and/or thermal emission mechanisms), then some positive ions of the target material may be expelled out, which is the so-called Coulomb explosion mechanism .
5.1.2 Research Approaches in the Study of Laser Micromachining.
Besides characterizations of workpiece geometries, surface morphologies, microsctructures, and properties after laser micromachining, other important research approaches include in situ time-resolved observations and physics-based modeling to improve the understanding of laser micromachining processes [212–221,225–239].
In situ time-resolved observations include direct imaging and shadowgraph imaging [213,216,226–228]; the former often involves the use of an intensified charge-coupled device (ICCD) camera with short gate widths , while the latter typically uses a probe beam which is also often short-pulsed [213,226–228]. These observations have revealed important physical processes that may occur during laser ablation. For example, the shadowgraph imaging in Ref.  shows liquid ejection in ns laser ablation which is believed to be due to phase explosion, while the direct imaging in Ref.  shows liquid ejection that is analyzed to be due to surface pressure gradient. Besides imaging, optical emission spectroscopy is another important research approach to study the evolution of plasma induced by laser ablation [230,231], by which plasma temperature and electron densities can be obtained.
5.1.3 Hybrid or Modified Laser Micromachining Processes.
If “conventional laser micromachining” is defined as the process where a workpiece surface in air (or other gas environments) is irradiated by one or multiple individual laser pulses at a certain repetition rate, then unconventional hybrid or modified laser micromachining processes are reported in the literature, some of which will be briefly reviewed in this sub-section.
Laser micromachining in water is a process where the workpiece surface to be machined is placed under water [240–243], and quality improvement due to the application of water (e.g., debris and recast reduction) is often reported in the literature. Some studies report an increase in the machining rate, while some report decreased machining rates due to the application of water . Water jet-guided laser micromachining is reported in the literature [244,245], where water jet can help guide the laser beam, increase the working distance, reduce contamination, and produce a cooling effect.
Ultrasound-assisted water-confined laser micromachining (UWLM) is a novel machining technology proposed by Wu  and studied in recent years [247–249], where besides water confinement of the front surface region to be machined, in situ ultrasonic waves are applied to induce ultrasonic cavitation and energize water to produce beneficial effect(s), such as in situ ultrasonic cleaning, to improve the machining process. The ultrasonic waves can be applied through different approaches such as an ultrasonic horn [248,249] or a high-intensity focused ultrasound (HIFU) transducer that can produce focused and intense ultrasonic pulses with good timing control . Under the conditions studied, it is found that with similar incoming ns laser pulses, UWLM can produce much less debris deposition than laser micromachining in air, and much higher (up to a few times) ablation depths per pulse than those in water without ultrasound [247–249]. An ultrasonic-assisted machining process, where ultrasonic waves are applied by vibrating the water tank, is also studied in the literature .
Ultrasonic vibration-assisted laser machining is a process where the workpiece surface to be machined is ultrasonically vibrated, but not water immersed [251–254]. It is fundamentally different from the aforementioned UWLM process because it does not involve the UWLM's critical component of ultrasonic in-water cleaning (and/or other) effect(s) around a water-immersed workpiece surface machining region. Under the conditions studied, it has been found that the applied ultrasonic vibration can improve machining efficiency and/or drilled hole geometry. Besides ultrasounds, the combinations of laser machining with electrochemical machining , electric field , or magnetic field [257,258] are also reported in the literature. Laser micromachining typically involves direct laser irradiation onto a workpiece surface to remove material. A modified process, called “laser-induced plasma micromachining” (LIPMM), is reported , where laser-induced plasma in a medium is utilized as the energy source for material removal. Some related advantages and drawbacks were discussed in Ref. .
Another way to improve laser micromachining quality and/or efficiency is the use of laser pulse trains (where each train contains two or more pulses with proper pulse energies and/or relative timing, etc.) [260–264]. For example, Ref.  shows that double-pulse laser ablation, where each pulse train contains two pulses with about an equal energy that are ∼30 to 150 ns away, can significantly enhance material removal rates for drilling high-aspect-ratio micro holes. Reference  shows that in double-pulse laser ablation using pulse trains of two pulses of different energies, different sequences of the high- and low-energy pulses can lead to significantly different average ablation rates, particularly for drilling high-aspect-ratio holes.
For continuum-scale modeling, if the dominant material removal mechanism is surface vaporization, the material removal is affected by both the target surface temperature and the vapor gas-dynamic process in the ambient environment right above the target surface [216,222]. The process can be modeled by solving the heat transfer equation in the target condensed phases (solid and melt) to obtain the target surface temperature and solving compressible gas-dynamic equations in the ambient environment to obtain the vapor status right above the target surface [216,222]. The two sets of equations can be coupled via the Knudsen layer relations at the condensed–gaseous phase interface [216,222,223]. If the laser intensity is high enough to drive the condensed phase surface temperature above the thermodynamic critical temperature, then the aforementioned sharp interface may disappear and change into a continuous transition zone [217,224]. In such a situation, one set of continuum-scale mass, momentum, and energy conservation equations can be solved for the entire domain including both the condensed and the gaseous phases [217,224], which need to be supplemented by the equation of state(s) (EOS) that can cover the entire temperature and density ranges involved. Besides, as introduced earlier, material removal in laser micromachining may proceed through other mechanism(s). For example, Ref.  reports a model for ns laser ablation of aluminum by solving hydrodynamic equations, where possible explosive boiling has been considered.
Continuum-scale models have the advantage of being able to handle relatively large temporal and spatial scales, but often lack the capability of directly revealing atomic-scale phenomena, which can be potentially described by a molecular dynamics (MD) model for laser ablation. In an MD model, motions of atoms or molecules are simulated by solving their equations of motion, where the forces on the atoms or molecules are calculated based on relevant molecular potential(s) . An MD model can help clarify complicated thermodynamic evolutions of material during laser ablation, particularly those by an ultrashort laser pulse . Hybrid models, such as those combining two-temperature model (TTM), molecular dynamics and/or the Monte Carlo method, etc., are also reported in the literature (e.g., Refs. [220,227,239]), which can potentially utilize the advantages of all the modeling approaches involved synergistically. For example, Ref.  reports a hybrid TTM–MD model for short laser pulse ablation of metals. The electron temperature increase due to laser energy absorption by free electrons is calculated by solving the electron heat transfer equation. MD replaces the lattice heat transfer equation in a thin surface layer to simulate atomic dynamics due to energy transfer from electrons to the lattice. Reference  reports a hybrid MD–MC ultrashort laser ablation model, where the MC method is used to simulate the electron dynamics and particle scattering events. Reference  reports the simulation of double-pulse laser ablation of aluminum by a TTM–MD hybrid model. Reference  reports MD simulations of liquid-assisted laser micromachining.
To study the possible Coulomb explosion mechanism in laser metal ablation by a model, the electron continuity equation and the heat transfer equations for electrons and the lattice can be solved in the target, together with the equation calculating the surface electron emission flux due to photoemission and thermionic emission . If needed, the emitted electron motion can be simulated with the particle-in-cell method . The electrical field in the entire domain can be calculated by solving the Poisson equation [212,234].
5.3 Applications and Summary of Challenges and Future Work.
Laser micromachining has numerous current or potential applications in biomedical, electronics, photonics, aerospace, automobile, and other areas [265–271]. Some specific examples may include (but are certainly not limited to) stent cutting, hole drilling for circuit boards, inkjet printer nozzles, aeroengines and diesel fuel injectors, thin film scribing, production of micro-optics, and surface texturing for tribological property enhancement [265–271]. Lots of research work has been conducted in the field of laser micromachining. However, challenges still exist. Some examples will be briefly discussed here. For fundamental research, additional modeling and experimental studies are certainly desirable to further advance the basic understanding of laser-matter interactions and material responses in laser micromachining or ablation. On the other hand, for practical industrial applications, sufficiently reliable, accurate (in a sufficiently wide range) and computationally efficient models would benefit rapid parameter selection, process design and optimization, and reduce costs. Further work would still be beneficial in this aspect. One important goal for future developments is to make laser micromachining have a better combination of high productivity, good quality, and low cost. Cost reduction and productivity enhancement for micromachining using ultrafast lasers and quality improvement for micromachining using nanosecond or longer-pulsed lasers can potentially help achieve the goal.
6 Laser Shock Peening
Many engineering parts require a desired level of fatigue life. While fatigue life can depend on many attributes of finished part quality, it is now well known that residual stresses are one of the key factors that affect fatigue life. In order to achieve a desired state of residual stresses, the manufactured parts often must undergo another finishing process, such as shot peening.
Laser shock peening (LSP) is an emerging competitive technology as a method of introducing compressive residual stresses into the surface of metals to improve fatigue and corrosion properties. Compared with the traditional shot peening, which has been widely used for the past six decades, LSP can produce a deeper plasmatic deformation depth (more than 1 mm versus below 0.25 mm for shot peening) and a higher magnitude of residue stresses, without leaving a roughened surface . This deeper and higher compressive residual stresses in turn result in longer fatigue life as illustrated in Fig. 21. In addition, LSP can provide higher corrosion resistance.
Laser peening was first found and investigated in the early 1960s [273,274], and initial feasibility studies with prototype facilities were carried out at Battelle Laboratory [275–277]. However, it was not commercialized for years due to the lack of a reliable, high repetition rate, and high average power laser. The first commercial application was developed at GE Aircraft Engines (Cincinnati, OH) in 1997 to mitigate the foreign object damage on fan blade leading edges for a military aircraft turbine engine . Research about laser shock peening can be divided into fundamental research and application research. The former is mainly concerned with the understanding of the basic physics in the process, and the latter is to conduct case studies of laser shock peening under different parameters and configurations, and to apply it to different materials.
It was found that LSP carried out under “confined” regime configurations can induce pressures on the target surface four times higher and 2–3 times longer than that under direct regime configurations . In LSP, the target material is coated with an opaque “sacrificial” coating layer, which serves to protect the target material from any thermal damage caused by laser irradiation. A transparent “confinement” overlay, namely, water, glass, and oil, is then placed over the surface of the material and a high energy laser pulse is irradiated on the target surface through the transparent overlay. This irradiation results in the ablation of the sacrificial coating layer and the generation of high-pressure plasma in the confinement region. This plasma also causes a high-pressure shockwave to propagate into the target material, which imparts compressive residual stresses into the material, predominantly in the surface region . A schematic of the laser shock peening process is shown in Fig. 22. A short pulse laser (typically about 1–20 ns) with a high power density (typically more than 1 GW/cm2) is irradiated through the confinement layer onto the coating layer surface. The expansion of the produced plasma is restricted by the confinement layer, thus producing higher pressures with longer durations.
For LSP under the “confined” regime, water is often chosen as the confinement material, because it is much easier to apply than solid materials and also because it has a higher threshold for dielectric breakdown than many crystalline materials (e.g., quartz) , therefore allowing the application of laser pulses with higher intensities.
The very first attempt to use the shock waves induced by a high-power Nd-glass pulsed laser was made by Fairland et al.  in 1972 to improve the mechanical properties of 7075 aluminum alloy. Since then, LSP has been applied to various engine materials including Al7075-T7351 , IN718 , Ti6Al4 V , and Al2024-62 . The initial challenge in using LSP for production was its slow speed due to the low repetition rate of the Nd-glass laser (about one cycle per 8 min.) and the time taken to apply and remove the coating. These hurdles have been overcome with the more compact and higher repetition lasers, mostly nanosecond Nd-YAG lasers although ultrafast lasers have also been used, and the introduction of a system for rapid coating application and removal by the Rapid Coater Processing System . Since then, many industrial applications of LSP have been reported, particularly for the fatigue life improvement of various aircraft components such as blades, gears, and wheels . In general, it has been reported that fatigue life can be extended 5–7 times with the application of LSP. In particular, the efforts made by LSP Tech, MIC, and GE greatly expanded the applications of LSP in the industry. In recent years, its applications have expanded to other industries such as orthopedic, nuclear, power, and automotive industries. Significant improvements in fatigue life of turbine blades after LSP have been reported for thermal power plants made of martensitic steel or Ti6Al4 V alloy [289,290]. LSP has also been applied to improve the fatigue life of automotive components such as leaf springs made of SAE 9260 spring steel . Several studies have also been reported for the reduction of stress corrosion cracking [292,293].
Some other efforts have been made to perform laser shock peening without coating [294–296] and warm laser shock peening [297–299]. Laser shock peening without coating eliminates the need to add and remove coatings and can be performed on objects immersed in water. Karthik and Swaroop provided a review on LSP without coating . The advantages of LSP without coating include time- and cost-savings due to the elimination of adding and removing the protective coating layer. LSP without coating is usually performed with low laser power to avoid excessive ablation of the target material. While most LSP is performed with a large beam diameter to cover the large area, Zhang and Yao also worked on the microscale LSP [301–303], which allows for treating small areas with superior resolution down to 100 µm.
Noting that the compressive stress generated by surface processing is not stable at high working temperature [304–306], warm laser shock peening (WLSP) was proposed by Ye et al.  to improve the residual stress stability of 4140 steel by dynamic strain aging (DSA) and dynamic precipitation (DP). The diffusion of carbon and nitrogen atoms during DSA and generation of precipitation during DP help to pin the dislocation movement. The advantages of warm laser shock peening combined with post-heat treatment include enhanced mechanical properties and extended fatigue life via the higher stability of residual stresses during cyclic loading, which has been well documented by Liao et al. . Additive manufacturing with laser shock peening has been developed to generate 3D metal components with hybrid nanostructures consisting of graphene nanocrystal and metal composite layers, with much better mechanical properties .
6.3 Predictive Modeling.
In LSP with a confinement layer, the shock waves and compressive stresses induced in the solid target depend on the pressures of the confined plasma. Therefore, the modeling of the whole LSP process should start with the modeling of laser ablation and the relevant physical processes of the confined plasma. However, most of the published models have not included all the relevant physical processes [279,309,310], and they typically contain several adjustable free variables (e.g., the fraction of laser energy absorbed and the ratio of plasma thermal energy to internal energy), which have to come from experiments under the same conditions. At the same time, measurement of plasma fields being developed under the shot laser peening period and assessing its interaction with laser beams is a very challenging task. The published experimental research in thermal aspects up to now is mainly the measurement of confined plasma pressure [311–314], the observation, and transmission measurement of breakdown plasma . The plasma pressure is generally obtained indirectly through back surface velocity measurement, which can be made through a velocity interferometer system for any reflector (VISAR) technique based on the Doppler shift of a laser beam , or through a polyvinylidene fluoride (PVDF) technique based on piezoelectric relationship, or through an electromagnetic (EMV) technique based on electromagnetic induction effect . The measurements of some other important parameters of confined plasma under common laser peening configurations, such as plasma temperature, electron and ion density, or plasma-water interface reflectivity, have not been found in literature. Furthermore, because most existing models are not self-closed, some of the parameters must be determined experimentally, which often involves an elaborate setup and difficult measurements.
A model that can predict the shock pulse generation in a confined medium during LSP was first introduced by Fabbro and coauthors [310,313]. The physics of the confinement effects of the plasma generated has been studied extensively in the literature. Fabbro et al.  proposed a 1D analytical model considering the plasma pressure generated by a glass transparent overlay. Zhang et al.  improved upon this model later. However, the major shortcomings of this model were that it had multiple free variables that had to come from experimental measurements. Wu and Shin  developed a self-closed thermal model that can predict plasma pressure under water confinement. The standout advantage of this model is that it has no free variables and has included the most important physical phenomena including the water evaporation, ablation of the sacrificial layer, plasma ionization and expansion, laser-plasma interaction, etc. The validity of the model was demonstrated by comparing the predictions with various experimental results observed in literature for different parameters like wave length, shock pressure and pulse width and found to be in good agreement. They subsequently developed one-dimensional  and two-dimensional  hydrodynamic models to capture the complete physics of the LSP process in conjunction with the properties calculated by the Quotidian Equations of States (QEOS). These models were able to predict material's response to laser irradiation and the resultant shock wave for LSP in a confined regime without any experimental calibration. It was shown that the 2D hydrodynamic axi-symmetric model would be more desirable than a 1D model when the focused laser-beam diameter is sufficiently small. Wu and Shin  also developed the model that can predict the water breakdown during the laser-beam transmission in water, which could elucidate why the maximum laser-induced shock wave pressure saturates when the laser power density is sufficiently high.
In a three-dimensional finite element analysis model, magnitude and distribution of stress due to shock peening is predicted, Wei and Ling  used a three-dimensional finite element model with the shock profile determined by Fabbro et al.'s model given above to predict the magnitude and distribution of stress and analyzed the role of various parameters like pulse width, power density and overlap, etc., on the shock peening. Many similar efforts have been made in predicting residual stresses using finite element modeling [323–327]. Recently Ayeb et al.  predicted the residual stress profile using the artificial neural network. Cao et al.  carried out a parametric study of residual stresses using FEM in overlapping LSP of 4140 steel using a more physics-based model presented by Wu and Shin . Molecular dynamics simulation has also been used to study the shock wave and material interactions. Ren et al.  used molecular dynamics simulation modeling to study the amorphization of NiTi under ultra-high strain rate dynamic loading.
6.4 Future Research.
Laser shock peening is a critical process in the applications requiring improved fatigue life of engineering components. In addition, it can be used for applications requiring enhancement of abrasion and corrosion resistance . These effects are achieved due to the refined grains as well as a lot of deformation twins and persistent slid bands as a result of LSP . Compared with other competing processes such as shot peening and liquid peening processes, LSP can provide surface treatment of selected areas. Some areas that need to be further addressed in the future is use of different media as the confinement media that can further increase the shock wave pressure at the surface, use of different protective layer materials and bonding methods the part surface as these have been shown to have significant effects on the resultant residual stresses. Confining medium with high evaporation temperature and a suitable way of elevating the target temperature are needed for industrial implementation of warm temperature LSP. Additional modeling work is also required to predict the resultant microstructure evolution including grain refinement and precipitate formation.
7 Laser Welding
Laser welding has emerged as an attractive alternative to the conventional welding processes such as resistance spot welding or arc plasma welding. The precisely controllable energy input in laser welding provides many advantages in terms of process speed, cost and flexibility. Laser welding can be performed using a robot that can provide the remote and flexible positioning of a laser beam, which will greatly simplify the set-up and provide the flexibility of operation. Laser welding is a very versatile process and is used in several industries such as automotive, aerospace, ship building, bridge construction, nuclear, biomedical, electronics, etc. Laser welding can be categorized into conduction welding and keyhole welding depending on the laser energy density and laser beam absorptivity. Typically the transition to keyhole welding occurs around 106 W/cm2, but it also depends on the substrate temperature. Figure 23 illustrates how a transition from conduction welding mode to keyhole welding mode can occur even with the constant power and scanning speed as the substrate temperature rises due to heat accumulation. Once a keyhole is formed, it causes a large deep hole into the substrate and creates strong recoil pressure driving the molten material to an upward direction. The keyhole can reach the depth to width ratio up to the order of 100. In keyhole welding, the laser beam is absorbed into the target via Fresnel absorption with multiple reflections, thus resulting in much higher laser energy absorption up to 90% and it becomes very energy efficient.
When the filler material is supplied by melting the base material or is of identical composition to the base material, it is called autogenous welding. Laser welding can also be performed with various filler materials to enhance the welding quality and joint strength. In addition, hybrid laser welding can be also performed by using simultaneously a laser and an electrical arc such as tungsten inert gas (TIG), plasma arc or metal inert gas (MIG) and has been shown to improve the welding efficiency. Laser welding can be applied to many different types of materials including metals and plastics, and also to dissimilar materials in butt welding or lap welding configurations. In addition, the laser welding can be carried out for bonding transparent material and opaque material by transmitting the laser beam through the transparent material and creating bonding by melting the opaque material, which is called transmission welding.
Laser welding is currently used in numerous applications, ranging from micro welding of electronics and medical devices to macro welding of automotive parts and dies and tools in heavy manufacturing industries. With the drop of high power, reliable laser cost, it is projected that the laser welding market is going to show a steady growth rate over 5% over the next five years .
In the automotive industry, welding is extensively used for body-in-white (BIW), needing as many as 2000 to 5000 spot welds, which has been traditionally done by resistant spot welding . However, a number of problems exist with the resistance spot welding of galvanized steel sheets, such as the long time required for welding, high maintenance cost of electrodes, and the tendency of zinc coating sticking to electronics . As the automotive industry is heading to more lightweight structures, other materials such as aluminum and magnesium alloys are emerging as candidates to replace galvanized steels . As BIW accounts for approximately 27% of the vehicle weight, the use of these lightweight materials is expected to reduce the total weight of vehicles. For these materials, however, the problems associated with resistance spot welding worsen. Laser welding has been replacing resistance spot welding to overcome some of these problems. Besides BIW, laser welding has been applied to engine parts, transmission parts, alternators, solenoids, fuel injectors, fuel filters, fuel cells, etc.
In the aerospace industry, laser welding has been used to join various superalloys such as nickel-based alloys and titanium-based alloys. Ti6Al4 V alloys are commonly used for static and rotating components in turbine engines . Inconel 718, on the other hand, is popularly used for components of aero engines and gas turbines, which are running at high temperatures. Since these alloys are quite expensive, welding offers the potential for reducing the consumption of material compared with subtractive manufacturing processes. Aluminum alloys are also very popularly used in the aerospace industry and there are selected cases where laser welding can provide competitive advantages over other welding techniques such as friction stir welding.
Recently, the welding of dissimilar materials has been gaining more popularity because it can reduce the part cost and design flexibility. Despite its attractive features, welding of dissimilar materials bring challenges because of improper wetting of material, differences in thermal expansion coefficients, and formation of brittle intermetallic phases, which are known to weaken the weld joint strength. Therefore, a large number of studies have been made to address these issues; including welding of aluminum to steel, aluminum to titanium alloys , copper to steel , stainless steel to titanium alloys , titanium to nickel alloys [341,342] and dissimilar stainless steels .
Laser welding has been used for joining polymers and plastics. CO2 lasers have been primarily used for welding of plastic components in the early days because the laser energy can be easily absorbed in its long wavelength of 10.6 microns. Most plastics are transparent at infrared wavelength but opaque at the long wavelength. In recent years, transmission laser welding has become a viable way of welding plastics with the internal absorber. This provides a way of heat absorption only at the weld interface and minimizes the heat-affected zone as shown in Fig. 24 . Moskvitin et al.  provided an overview of laser welding of plastics in recent years.
Laser welding involves multiple complex physical mechanisms. In order to properly model laser welding processes, one must consider the laser-beam absorption, particularly Fresnel absorption via multiple reflections if a keyhole is formed, fluid flow in the molten pool, recoil pressure generation based on temperature, vapor, and plasma formation with their interaction with incident laser, assist gas effects, keyhole shape tracking, etc., as illustrated in Fig. 25. Once the surface of the material being welded is irradiated by a high-intensity laser beam, the target material will melt and some of the molten material will evaporate. Evaporation will generate recoil pressure on the keyhole wall, which is a liquid/gas interface, while a strong thermo-capillary force is also generated on the keyhole wall due to the high gradient of the surface temperature. The recoil pressure and the thermo-capillary force will drive the flow of the liquid material outward and hence a keyhole is formed. Once a keyhole is formed, the laser beam will go through multiple reflections on the keyhole, and thus more laser energy is captured inside the keyhole, which is called Fresnel absorption, thereby resulting in a sudden increase in the total laser energy absorbed. The evaporated material will form plume, some of which will turn into plasma, and scatter and absorb the incident laser energy, thereby reducing the laser energy reaching the target surface. Furthermore, the assist gas, its type and velocity, affects the formation of the keyhole and laser beam attenuation. Therefore, a high fidelity keyhole laser welding model must be able to capture all these physical mechanisms to accurately predict the keyhole formation.
Numerical modeling techniques have become increasingly involved in the investigation of keyhole dynamics to reduce the cost of experimentally designing welding conditions and also to gain a deeper understanding of the process dynamics. The most challenging issue in keyhole modeling is not the transport phenomena of conduction and convection but the understanding of laser-matter interaction and the determination of keyhole shape.
Some early studies assumed predefined geometry for the keyhole. Lankalapalli et al.  estimated the penetration depth based on 2D heat conduction in a conical keyhole. Kaplan  introduced a method of calculating an asymmetric keyhole profile with the concept of iterative point by point absorption of laser energy by using Rosenthal's theoretical model of moving heat source on an infinite plane, but no fluid motion was considered. Ye and Chen  and Chen and Wang  assumed a cylinder keyhole and studied the coupled heat transfer and Marangoni force-driven flow of weld pool. A similar model was also adopted by Cho and Na  with a source term including an adjustable parameter. These models however cannot capture the dynamic change of keyhole.
Some researchers [351–353] tried to determine the keyhole shape dynamically based on the energy balance on the keyhole wall. However, the effect of convection, which can dominate the energy balance on the keyhole wall as shown by Semak and Matsunawa, , was ignored in the consideration, which may introduce error to the prediction.
Subsequent models attempted to track the dynamic evolution of keyhole shape using an extra governing equation, which is usually coupled with hydrodynamic equations. Two methods are chosen for this purpose: the level-set (LS) and volume-of-fluid (VOF) methods. The LS-based models are capable of simultaneously simulating the full three-dimensional keyhole wall motion as well as the heat transfer and fluid flow and have been used to investigate the effects of various interfacial phenomena, including evaporation, free surface evolution, and multiple reflections [355–358]. On the other hand, the VOF method also serves the purpose and has been used to investigate the issues regarding keyhole growth, keyhole stability, and porosity formation [359–366].
In keyhole welding processes, the keyhole boundaries are formed when the material vaporizes. The thickness of this transition region from liquid to vapor is usually very thin, the order of several mean-free paths. Such a narrow layer of phase transition is called the Knudsen Layer. Therefore, one needs to regard this evaporation interface as sharp since there is a sudden jump in temperature, pressure, and physical properties. In the LS-based models, a diffusive interface is assumed across the keyhole wall, and the temperature, velocity, and physical properties are smoothed out across this transition region. The boundary conditions of energy and pressure are incorporated into the models as extra source terms in the transition region. In the VOF-based models, the temperature, velocity, and physical properties of the interface cells are usually the averages of different phases according to their volume percentages, and the boundary conditions of energy and pressure are also treated as extra source terms for the interface cells. These techniques are called “numerical smearing” and are utilized to avoid the difficulties when dealing with the discretization in interface cells, the neighbors of which may be located on the other side of the interface and therefore have vastly different states and properties. Pang et al.  theoretically revealed that a parasitic mass flow can be generated if “numerical smearing” techniques are utilized, thus resulting in producing unphysical fluid motions near the keyhole.
Pang et al.  used the sharp interface method to overcome this problem and showed a more promising result for the prediction of keyhole evolution. The model was successful in predicting the keyhole transition from stable to unstable state when the welding speed increased, and multiple characteristic flow patterns in an unstable keyhole were captured. Unfortunately, the model only took into account the phenomena in the liquid and solid phases, while the physics in the vapor region was not considered.
The above review of the modeling techniques shows that the sharp interface model based on level-set equation provides an advantage to capture complicated jump conditions across a moving sharp interface. This technique is promising as it can better consider the physics associated with the “sharp” keyhole wall. A sharp interface keyhole model that comprehensively accounts for the physics in both the liquid and gas phases was introduced by Tan and coauthors [367,368] with experimental validations of predictions. Their model was able to capture the interplay of the multiple reflections and plume attention and predicted the temperature and shape of the keyhole wall more accurately. Hong and Shin  further expanded this model to include the interface gap and its effects in the resultant molten material flow and weld pool geometry.
Modeling of laser welding of plastics has also been carried out. Some approaches used include the Monte Carlo model using molecular dynamics simulation at meso-scale , and thermal modeling using FEM [371–373] without consideration of fluid motion. Modeling of resultant thermally induced stresses has been conducted using FEM by some researchers [374–376]. However, these models ignored melting and fluid motions during welding. Recently, Ai et al.  using the computational fluid dynamics model considering melting and fluid flow were able to show that the molten pool with a vortex flow pattern formed only in the plastic side is significantly affected by welding parameters and accurately predict the fusion zone with the occurrence of porosity.
Since laser welding involves the complex dynamics of molten pool formation and keyhole evolution, many attempts have been made to monitor laser welding process conditions to achieve reliable results. Many different types of sensors have been used for this purpose, including cameras, photodiodes, spectrometer, acoustic sensor, pyrometer, and plasma charge sensor. You et al. provides a comprehensive overview of welding process monitoring methods used up to 2013 . They conclude that non-contact optical sensing is the most ideal real-time monitoring technology for laser welding. Each of the sensors provides its respective advantages and limitations. Vision systems provide a lot of information about the weld pool characteristics [378,379] and appearance of weld seam surface such as seam tracking [380–382]. Photodiodes are relatively cheap and also provide rich information of high-frequency features coming from the thermal radiation of metallic vapor and molten pool surface. They have been used to detect defects [383,384]. Spectrometers can provide the information on the species coming out of the weld pool and potentially plasma temperature, thus providing indirect information about the weld condition . Acoustic sensors have also been used with limited success in detecting welding quality [386–388] but are sensitive to the distance and environment.
In more recent years, some promising results have been reported. Luo and Shin developed a weld pool monitoring system using a coaxially mounted camera, an optical bandpass filter and an auxiliary green laser to monitor weld pool boundary  and subsequently used to estimate keyhole geometry and welding defects . Zhang et al.  extended this work to detect porosity during welding using a deep learning method. Zhang et al.  used a scheme similar to Luo and Shin  and proposed a visual monitoring system that can estimate the presence of keyhole or penetration and weld sim width using a coaxially mounted camera. Chen et al.  used a high-speed camera and extracted multiple features from the plume and spatters to construct a classifier for welding quality using a support vector machine. The use of multiple sensors via sensor fusion and machine learning has also been attempted to monitor welding conditions and quality. Gao et al.  tried to monitor the disk laser welding process using the integration of photodiode and visual sensing. Such efforts to monitor welding process conditions are continuing.
Laser welding holds promise to replace conventional welding processes such as resistance spot welding. Despite that, the challenges are remaining. One critical area to be addressed is to minimize the reduction of joint strength after welding, since it involves melting and re-solidification, which often results in coarse grains and phase changes in the weld joint. Possible approaches may include, but not limited to, adding gaps , new filler material, and post-processing by heat treatment or warm laser shock peening . Another area of growing interest is welding of dissimilar materials such as Fe + Al and metal + polymers, hybrid welding process using multiple welding energy sources, numerical modeling to predict the resultant microstructure and phases such as secondary phases and intermetallics, multi-sensor monitoring via deep learning methods and signal processing. Data-driven science can also be applied to the design and optimization of laser welding processes as well as the prediction of resultant microstructure and mechanical properties.
8 Laser Direct Processing of Nanomaterials and Their Heterostructures
In today's manufacturing of thin film devices, such as photovoltaics and touch panels, various post-heat treatment techniques are often used to refine the microstructures, such as grains, and dislocations, and thereby reduce crystal defects. Among those processes, rapid thermal annealing (RTA)  has been widely used due to its relatively low cost and uniform crystallization. However, it has problems of non-selective heating, low speed, quasi-static temperature changes, etc. Zone melting crystallization (ZMC) [398,399] has been applied to thin films after chemical vapor deposition (CVD) deposition. However, due to the high processing temperature, the substrate is limited to those with high melting temperatures with relatively low processing speed (10 ∼ 20 mm/min). In addition, ZMC has the problem of poor control of grain size and internal structure, need for costly vacuum/inert gas systems, and general unsuitability for large-area substrates. Most present-day device inefficiencies in converting photons to electrons are due to defects in the thin-film's crystalline structure. In the last decade, laser direct irradiation of nanoparticles has been applied in many materials systems, ranging from insulators, semiconductors to conductors.
Laser direct writing of semiconductor nanoparticles: A laser printing technique has been developed for precise deposition of silicon nanoparticles, in which the size of Si nanoparticles and their crystallographic phase can be manipulated . Si nanoparticles produced by femtosecond laser printing were initially in an amorphous phase (a-Si), followed by a second femtosecond laser pulse to convert them into a crystalline phase (c-Si). This method has been applied to direct writing of dielectric nanoparticles for resonant optical responses for applications in nanoantennas, nanolasers, and metamaterials. Laser crystallization of nanoparticle inks has been a promising technique to create crystalline functional thin films, including electro-optical materials, photovoltaic absorbent, transparent conductive layer, semiconductor oxide for high power transistors, leading to fast top-down manufacturing of high-quality crystalline materials on a substrate with large lattice mismatch. It has a great potential in scalable production because it is performed at high speed, low temperature, and ambient processing conditions. Laser crystallization of solution-based nanoparticles has been applied to transparent conductive layers [401,402], thin-film solar cells [403,404].
Laser sintering of metal nanoparticle inks: Laser sintering of inkjet-printed metal nanoparticles has been developed for all-inkjet-printed flexible electronics , in which the low-temperature metal deposition as well as high-resolution patterning can be realized to overcome the resolution limitation of the current inkjet direct writing processes. Air-stable high-resolution organic field-effect transistor (OFET) by selective laser sintering of inkjet printed  have been made in ambient pressure and room temperature without any photolithographic steps or vacuum deposition processes. Nonvacuum, maskless fabrication of a flexible metal grid transparent conductor has been realized by low-temperature selective laser sintering of nanoparticle ink . The metallic grid transparent conductors with high transmittance (>85%) and low sheet resistance (30 Ω/sq) can be produced on glass and polymer. A molecular dynamics simulation study has been conducted to investigate the solid-state neck growth mechanisms in low-energy laser sintering of gold nanoparticles . A drop-on-demand laser sintering of silver nanoparticles has been employed for electronics packaging, in which a “dry” laser-sintering method consists of ink-jet printing of metal nanoparticles with dispersants and solvents, short preheating to remove organic substances in the ink and laser-beam irradiation under atmospheric conditions with the flow of argon gas.
Laser joining of nanowires: Compared with metal nanoparticles, metal nanowires provide better electrical conductivity due to their highly crystalline and low defects. However, the interfaces between the nanowires are critical to reduce the resistance of electron transportation. Currently, large-scale and selective nanoscale integration of nanowires is challenging due to high cross nanowire junction resistance. Nanowire-network with high electrical properties are promising for many high-performance materials and devices, including solar cells, thermoelectrics, sensors, transistors, and transparent electrodes. A light-induced plasmonic nanowelding technique  was developed to assemble metallic nanowires into large interconnected networks. Pulsed laser processing of semiconductor nanowires  has been realized to form low resistance interfaces between silicon nanowires for flexible semiconductor thin films, which is promising for many functional electronic devices. Pulsed laser processing has been utilized to realize crystalline joining of silver nanowires  without affecting other regions of the network. The method can be applied to roll-to-roll printed AgNWs percolating networks on a flexible substrate. The excellent optoelectronic performance of AgNW/PET (polyethylene terephthalate) has achieved sheet resistance of ∼ 5 Ω/sq at transmittance (91% @λ = 550 nm).
Laser joining of heterostructured nanomaterials: Nanojoining of heterostructured materials to form good interfaces between 2D crystals, nanowires, and nanoparticles has become important in many applications in electronics, optoelectronics, and plasmonics. An optically controllable targeted nanohealing technique has been developed by utilizing the plasmonic-enhanced photothermal effect , by which the nanogaps between two silver nanowires (NWs) is healed by incident laser irradiation. A low-cost method  for welding semiconductors and metal nanowires (NWs) has been developed utilizing a plasmon-enhanced photothermal effect, by which different types of heterojunction-based (single Schottky junction and back-to-back Schottky junctions) electronic nanodevices can be fabricated by welding various combinations of silver and ZnO NWs on two gold electrodes using continuous wave laser (λ = 532 nm) shots. It was found that after laser-induced nanojoining, the junction formed between the ZnO nanowire (NW) and the gold electrode demonstrates Schottky barrier-like behavior, whereas the junctions between silver-and-gold and silver-and-ZnO behave like ohmic contacts. Recently, laser-induced nanoscale integration has been applied for nano-joining of 2D crystals and nanoparticles (such as QDs, nanodiamonds, and plasmonic nanocrystals) on various substrates. With laser-induced transfer, the integration of the 2D crystals on nanomaterials and various substrates is much better than the traditional ways. Laser deposition of graphene on silver nanowires as a protective layer for high power irradiation . Laser-induced layer integration of 0D/1D nanomaterials with 2D materials has been utilized to manufacture photodetectors  and energy storage devices [416,417]. Laser shock inducted integration of graphene with plasmonic nanoparticles has been used for the fabrication of biosensing and chemical sensing devices [418,419].
8.3 Future Directions.
Lasers are ideal tools for the integration and joining of nanomaterials and their heterostructures. They offer some important advantages in comparison with other techniques to realize control of changes in temperature, microstructure, phase, and defects in the nanomaterials and their interfaces. In the meanwhile, high-resolution patterned processing could be realized under ambient conditions. In the future, additive process could be combined with a subtractive process to enable laser processing to treat functional nanomaterials to replace many traditional fabrication processes that have to use multiple deposition and lithography processes in cleanroom. Lasers will be used in roll-to-roll manufacturing process to directly print functional nanomaterials with large scale, high speed, and low cost. With the widely tunable laser processing parameters, the chemical compositions, microstructures, and defects can be well controlled for optimal physical properties in the wide range of functional devices.
9 Large-Scale Patterning of Functional Nanomaterials
Large-scale micro/nanoscale patterning of functional nanomaterials, such as metals, insolators, semiconductor, and 2D materials, to control their energy band gap and surface chemical property has been getting more attention for many applications [420,421], such as plasmonic systems, optical field regulation, optoelectronics, energy harvesting, chemical sensing, and biosensing [422–425]. In order to fabricate structures with high spatial resolution patterns, many techniques have been implemented, including directed assembly , nanoimprinting , and multiple electron-beam lithography, nonlinear laser lithography , thermochemical nanolithography , and scanning probe lithography , femtosecond laser direct writing  and laser-beam interference , and laser-induced periodic surface structures (LIPSSs) . Many factors need to be considered in order to select the appropriate nanolithography method, including processing speed, manufacturability of overlay structures, quality control of the micro/nanopatterns, large-area uniformity, cost and resolution of the patterns limited by optical diffraction limit, repeatability, and the processing accuracy.
Femtosecond Laser plasmonic lithography (FPL): Controllable energy deposition to achieve subwavelength patterning through the FPL strategy has been recognized in recent years . The key factor in this process is the reversible nature of plasma waves, which makes it possible to obtain feedback at local positions and in turn control the energy distribution to achieve the formation of nanostructures . This method has been applied in various materials, including metals, semiconductors, and insulators [436–439]. Recently, femtosecond laser plasmonic lithography (FPL) was presented as large-area manufacturing technique to generate nanoscale patterned photoreduction of graphene oxide (GO) films on a silicon substrate . Periodic grating structure (∼680 nm period) can be induced at a centimeter scale. It was found that the laser-induced gradient reduction of GO film produces a gradient dielectric permittivity (DP) with the maximum DP at the surface and a smaller DP at deeper thicknesses, which allows excitation of TE-mode surface plasmons (TE-SPs) . The interference between the incident light and the excited TE-SPs leads to a periodic LIPSS with orientation parallel to the laser polarization direction .
Laser imprinting-based nanolithography: Nanoimprint lithography is a useful method to fabricate nanopatterns on polymers [428,442], metallic glasses . Nanoscale metallic structures have becoming important in a variety of fields such as plasmonics , electronics , and biosciences . Direct nanoimprinting of crystalline metals has been very difficult because of the limitations on formability, due to fluctuations of plasticity at the nanoscale , size effects in plasticity , and grain size effects . These limitations can be circumvented by heating the sample close to melting temperature  or using a superhard mold and very high pressure . However, the large-scale manufacture of metallic structures with high fidelity and high crystallinity represents a substantial challenge. A scalable nanomanufacturing technique—laser shock imprinting—has been developed by exerting inhomogeneous 3D strain field to metallic thin film or nanomembranes, where the laser shock–induced pressure is applied on metal thin layers, 1D nanowires [452,453], 2D nanocrystals , on top of a nanomold to generate 3D nanoshaping [455,456]. Laser-shock imprinting (LSI) can be used for mass production of quasi-3D nanostructure [457,458] and nano-pattering [459,460]. LSI is a nonlithographic 3D microfabrication technique. The feature sizes are decided by the nanomold and not limited by the wavelength of the laser in UV lithography-based pattering. The process could be completed by one-step laser irradiation, not by layered deposition/etching in traditional microfabrication techniques. The capability to manipulate 2D crystals’ physical property could be significantly extended  due to the tunable strain engineering controlled by designing the shape of the nanomolds . In addition, the nanomold could be fabricated on large area by nanolithography widely used in microsystems, such as E-beam lithography and laser interference lithography. The beam size of the LSI could be as large as centimeters, which is feasible to realize wafer-to-wafer or roll-to-roll process.
9.3 Future Directions.
Laser-induced nanopatterning, as first developed by the semiconductor industry, has been utilized for nanolithography of silicon, silica glass, or polymers to produce 2D nanophotonic devices. In the last two decades, 3D direct writing with femtosecond laser direct writing has been developed to introduce multiphoton absorption during polymerization, which increases the resolution of the patterning much smaller than the optical limit of the laser beams. In the future, 3D laser nanolithography on other optical materials needs to be developed with high resolution and large scale. In addition, the mechanism of how the femtosecond laser generates ablation and shock wave and induces lattice damage, crack propagation, and residual stress/strains in the crystals will need to be discovered. In order to make the nanopatterning process scalable, roll-to-roll process, multibeam techniques, and parallel lithography methods such as LIPSS, FPL, and laser imprinting will need to further be developed to generate uniform and controllable nanopatterns on various materials in large area. Inline metrology and monitoring techniques will need to be developed to control the quality of the nanoscale patterns accurately.
The purpose of this paper is to highlight the use of lasers for various manufacturing processes and also describe the relevant technical developments in each of the processes covered. This paper is not intended to be a comprehensive review of this vast field of laser applications, but rather to highlight the key laser-based manufacturing processes, their benefits, and related technical issues. As shown, the use of lasers has spawned many new manufacturing processes, which could significantly improve the way of manufacturing engineering products and enabling manufacturing producing parts that were not feasible by traditional manufacturing processes. Along with the key applications and technical development in each laser-based manufacturing process, some future challenges were described in each section to guide readers. While many additional technical innovations need to be made to fully realize the potential benefits of laser-based manufacturing processes, with the rapid advancement of laser technologies, it is expected that laser applications in manufacturing and materials processing will greatly increase in the future.
Conflict of Interest
Dr. Benxin Wu is the inventor of the granted US patent on ultrasound-assisted water-confined laser micromachining (Ref.  of this paper).
Data Availability Statement
The data sets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request. No data, models, or code were used for this paper.