Laser cladding process parameters

Functional properties and the quality of laser coatings are strongly dependent on chemical composition and microstructure as will be discussed later in section 1.5. Chemical composition is dictated not only by the material selection but also by the control of dilution, which can be defined geometrically or compositionally latter one giving a little bit higher values [27, 72]. Both the microstructure and dilution together with other relevant process results illustrated in chart (Figure 5) created by Ollier et al. [73], and modified by the author, are dependent on laser cladding process parameters, which are chosen on the basis of coating and base materials, the desired coating thickness and available laser characteristics.

Consequently, laser cladding parameters can be divided into process and material parameters [74]. Process parameters include actual process parameters, which are variable, fixed laser parameters dictated by the choice of laser and optics and parameters related to the feeding of coating material. Among process parameters, laser power (P), traverse speed (Vb) and feed rate (f) are the principal parameters since they have the largest effect on the process results as will be explained below. By using classical power density – interaction time field, laser cladding is typically carried out in the range of 100–1000 W/mm² and interaction times less than 1 s as displayed in Figure 6. Among material parameters, beside surface condition, the most important ones include thermophysical properties, which comprise thermal diffusivity (α(T)), thermal conductivity (k(T)), density (ρ), melting temperature (Tm), coefficient of thermal expansion (αCTE), latent heat of fusion (Lm) and specific heat capacity (c(T)) among which α(T), k(T), ρ, αCTE and c(T) are temperature dependent.

Figure 5. Blown powder laser cladding parameters [18, 73].

Process results

Since the dilution and bead or coating geometries vary with different process parameters, it is essential to understand the relationship of the process parameters to the optimal product quality. Considerable experimental and theoretical efforts have been carried out to study this.

Low diluted coating with fusion bond, the ultimate goal to be achieved, can be obtained simply by feeding the optimal amount of material (g/mm or g/mm²) to laser energy used (J/mm or J/mm²) for a given clad/base material combination. Process parameter window mapped, for instance, against W/mm and g/min for this is rather large, and expands as a function of Vb, in blown powder cladding of Stellite on mild steel assuming that geometrical dilution is allowed, for instance, up to 10%, which is a value typically used to distinguish cladding from the laser alloying process [23, 75]. As in this case, operating windows are always restricted by the dilution, aspect ratio and power limits. What comes to the influence of process parameters on bead or coating geometries, height above the original base material can be increased by increasing f to the limit where the fusion bond is lost while keeping other parameters constant [76]. Simultaneously dilution gets lower as theoretical J/g value decreases [77], whereas thickness of the heat-affected zone (HAZ) surprisingly slightly increases [76].

10 100 1000 10000

0 0.5 1 1.5 2 2.5 3 3.5

Interaction time (s) Power density (W/mm2 )

Nd:YAG HPDL CO2

Figure 6. Blown powder laser cladding values (Co-, Ni- and Fe-based alloys on Fe-based base materials because their thermophysical properties do not vary a lot) collected from the literature and mapped against the theoretical power density and interaction time. Note that the power density axis is represented on a logarithmic scale [18, 45-47, 76-83].

Dependencies between height and f [82] and powder feed rate per unit length are linear in blown powder cladding [83]. If powder feed rate is, however, increased beyond the certain threshold value, cross-section of the coating above the original base material surface start to increase at accelerating rate due to multiple scattering of laser beam, which takes place in dense powder cloud coaxially fed into the laser beam. This improvement in absorption enhances the process energy efficiency as explained in Ref. [74]. Other straightforward ways to adjust the bead or coating height is to increase or decrease Vb while keeping other parameters constant. Higher Vb generates thinner and slower Vb thicker bead. Even if theoretical J/g value remains constant, increasing Vb tends to increase dilution as reported in Refs. [38, 76, 77]. This is particularly true for thin beads produced by high traverse speeds.

With higher bead heights, dilution is in practice independent of Vb [75]. Height of a single bead cannot be, however, increased infinitely by any means because the aspect ratio

(width/height (w/h)) of the bead affects the formation of inter-run pores between adjacent tracks and thus hinders the formation of defect-free large surface area coating. It was shown by Steen [23] that, in general, aspect ratio (w/h) of single bead should be higher than 5 to avoid such pores or the contact angle (α) defined in Figure 7 should be higher than 100º [83].

Talking about continuous large area coating, the coating height (hc) must satisfy hc<2/3rl, where rl is beam radius, in order to avoid inter-run pores [30]. As the height of the bead is strongly dependent on f and Vb, increasing Vb or decreasing f can increase the aspect ratio [77]. Using wider beam would do the same. Example of this was given in Ref. [27], where 50 mm bead width was obtained by using linear scanning optics. Pelletier et al. [74] and de Damborenea et al. [82] noticed that cross-section and height increases linearly as a function of interaction time, i.e. when Vb decreases. Simultaneously, the depth of HAZ becomes larger since heat input (J/mm) increases. Overlapping defined by bead width and inter-track advance can be used as well to adjust the coating height. By decreasing the inter-track advance, coating height increases [40]. Simultaneously the surface smoothness (periodic variation in coating height) in direction perpendicular to cladding direction decreases minimizing the need for post-machining and increasing material efficiency [22, 84]. At the same time heat-treating effects on previous beads, however, increase. The effect of the third principal process parameter, P, is clear. Bead height increases when P is increased and other parameters are kept constant. Simultaneously, depth of HAZ and dilution become larger [76].

Bead width is primarily dictated by the spot size. However, with low Vb [85] or with high preheat temperature bead width can be wider than the spot size. This is because sufficient energy is available to melt the clad material situated at a distance wider than the beam itself [85]. For the same reason, preheating the base material increases dilution and widens the HAZ [86]. Furthermore, it was observed that bead width increases linearly with P in coaxial cladding [83]. The higher the Vb, the narrower is the single bead. This dependence is also linear [40, 83, 85]. Komvopoulos and Nagarathnam [76] noticed further that the bead width increases when f is increased.

Even if the examples shown here follow mainly the linear relationships between principal process parameters and bead or coating geometries, non-linear dependencies are common when there is change, for example, in absorption as in multiple scattering of laser beam in dense powder cloud. In addition to this, absorption can change due to changes in angle between incident beam and inclined plane (surface of the leading edge of the melt pool) it impinges, which depends on bead thickness and overlapping. Vb is another such factor, which may cause some non-linear dependencies since heat conduction losses to the base material decrease as Vb is increased to extraordinary high levels of 10 000 mm/min, for instance, at the expense of bead and spot width. It may also affect absorption and powder catchment efficiency since narrower beads produced by high Vb let more power and powder to impinge on solid base material. An important question, which arises from the relationships between coating thickness and process parameters, is: is it faster to clad coating with certain thickness by applying two thin consecutive layers instead of one thick? Weerasinghe and Steen [40]

noticed that by doubling Vb, bead height (and bead cross-section) decreased by a factor of more than 2 while keeping other parameters constant. This meant that the cladding was energetically more efficient with the slow traverse speeds (~200-800 mm/min). This also suggests that it is faster to clad coating with certain thickness by applying one layer instead of two. This was also confirmed later by the mathematical model developed by Hoadley et al.

[22]. Situation, however, changed when Vb was increased from ~800 to ~1200 mm/min.

Cladding with the latter speed was now more efficient [40]. The most appropriate way to find

these relationships and simultaneously to reveal the operational window for the process is to create process parameter maps, which usually include vast amount of data presented in simple and concise manner. Creation of such maps involve large amount of cladding experiments where the principal parameters are varied up to their limits. To diminish experimental work, one of the main parameters can be adjusted “on the fly” as in Ref. [83], where the single bead of Ni-based self-fluxing alloy was clad around the rotating carbon steel rod. As a result of 175 analyzed cross-sections, process parameter map displayed in Figure 7 was created.

Figure 7. Process parameter map for single bead in coaxial laser cladding. Powder feed rate per unit length increases on lower and bead height on higher horizontal axis from right to left. Solid and dashed curves represent limits for dilution and cross-section area, respectively.

Vertical solid line determines the contact angle required for defect-free continuous coating. D is geometrical dilution: (D) = Am / (Ac + Am). Operating window is restricted by dilution and contact angle limits [83].

1.3.1.1 Modelling

Since laser cladding process involves three principal process parameters for given beam properties, finding optimal ones for a given coating/base material combination to generate low diluted and fusion bonded coating with desired thickness necessitates usually a series of cladding trials accompanied with some basic metallography. To reduce the amount of experimental work, to help to select the optimal parameters and improve the understanding of the process, modelling has been widely practised. At least two approaches have been utilized;

physical models involving large amount of numerical calculations and models based on experimental data. The first approach requires considerations of heat and mass transfers;

energy, momentum and mass balance including all the relevant losses [30, 73, 84];

interactions between beam and powder cloud [29, 30]; changes in absorption due to change in incidence angle of beam [30, 87] and multiple reflections in powder cloud; fluid flow due to Marangoni convection [87, 88]; surface tension [31, 84, 89] and phase transformations. Heat

transfer and accompanied temperature fields can be calculated by solving 3D heat conduction equation for moving heat source. This necessitates the knowledge of temperature dependent material properties (k(T), c(T) and ρ) for clad and base materials and net absorption of the laser beam, i.e. how much heat is provided by the beam itself through the powder cloud and how much via preheated powder particles, which participate the bead formation. More realistic melt pool temperatures and temperature fields can be obtained if losses of convection and radiation, energy distribution modified by energy brought by powders [33], Lm

(endothermic reaction in melting, exothermic reaction in solidification) and possible exothermic reactions associated with phase formations are taken into account. Heat transfer via Marangoni fluid flow, and powder feeding’s influence on it [90], should not be neglected either since it is well known that heat conducts faster in melt pool than predicted by heat conduction equations. In realistic modelling of bead dimensions, at least, powder catchment efficiency, deformation of the melt pool due to powder and gas impact [87], mass transfer due to Marangoni fluid flow and temperature dependent surface tension (and parameters, which influence on it) of the molten clad layer and gravity forces [91] should be known. All these considerations lead to a very complex set of coupled equations, which should be solved three-dimensionally. For these reasons, number of assumptions and simplifications has to be often made reducing the level of correspondence between the mathematical model and reality. Lack of data about temperature-dependent material properties is another factor, which have retarded the creation of reliable ‘universal’ simulation tool to optimize cladding parameters.

In the earliest attempt to model the blown powder laser cladding, 2D physical model developed by Weerasinghe and Steen [29] estimated the bead dimensions and temperature field in longitudinal direction (along the bead length) when the process parameters were given. They noticed, for instance, that melt pool becomes elongated at high traverse speeds.

Attenuation of the laser beam and heat provided by the preheated particles were considered.

Later Hoadley and Rappaz [32] proposed also 2D model to define the melt pool shape in transverse cross-section parallel to cladding direction. Accurate melt pool shape and process parameters’ influence on it was needed to simulate the solidification velocities and microstructure formation, which are more closely discussed later in section 1.5.1. The model calculated also the clad height when the processing parameters were given and predicted the laser power required to deposit a given clad thickness. Temperature dependent thermophysical properties were used for both the clad and the base material. Latent heat of fusion was included, too. Picasso et al. [30] established 3D model for predicting the process speed and the powder feed rate for the given laser power, beam diameter, geometry of powder stream and clad height. It also gave information about the process energy and powder efficiencies.

Their model considered the change in absorption due to bead height (melt pool shape) related change in incidence angle of circularly polarized beam. Kaplan and Groboth [80] created model based on energy and mass balance equations, which permitted the calculation of temperature fields and bead dimensions in transverse cross-section of the bead (perpendicular to cladding direction). Bead dimensions in plane perpendicular to cladding direction were also calculated in Ref. [92]. This model was developed further to commercial simulation software (LAVA) for coaxial laser cladding, which is capable to compute the bead geometries in plane perpendicular to cladding direction when the process and material parameters are given [89, 91]. The required input parameters include laser power, beam diameter, powder feed rate, powder jet focus diameter, traverse speed, particle size and speed, distance between the coaxial nozzle and work piece as well as melting temperatures of base and clad materials. As input data, absorption of laser beam to base material/melt pool and powder particles should be known as well. Output data consists of bead width and height, maximum temperatures on clad

and base materials, powder loss and melt depth below the surface of base material (geometrical dilution) as well as depth of HAZ.

Another comprehensive model for coaxial laser cladding was described by Han et al. [90], who developed model to simulate the melt pool dynamics and dimensions, temperature fields and the geometry of the bead when the process parameters were given. Their results indicated, for instance, that the particle injection has a significant effect on the melt pool flow. Particle injection caused an increase in the maximum fluid flow velocity from 4800 to 18 000 mm/min compared to the situation without powder feed. Furthermore, it was noted that the melt pool temperature increased when the laser power was increased while keeping other parameters constant. The higher the powder feed rate, the lower the melt pool temperature due to attenuation of laser beam. Laser power and powder feed rate influenced also on melt pool length. That is, higher laser powers generated longer and higher powder feed rates shorter melt pools. Cho et al. [93] showed the importance of Lm in modelling by analyzing the time dependent thermal fields with or without it. In simulations where Lm was considered, maximum melt pool temperature decreased together with melt pool and HAZ dimensions as compared with the situation where Lm was neglected.

Laser cladding process modelling, that involves computing the temperature fields, melt dynamics and bead or coating geometries as discussed above, can be supplemented with material modelling. For example, Martukanitz and Babu [94] and Babu et al. [95] attempted to couple heat transfer and fluid flow (process modelling) with thermodynamic and kinetic models (material modelling) to predict the microstructural evolution and stability of WC in different matrix materials. Moreover, 3D temperature distributions can be also supplemented with thermo-mechanical calculations by finite-element model (FEM) to reveal the development of residual stress distributions. These stress and strain fields are computed with the aid of some commercial software like for instance ABAQUS, ANSYS and Sysweld.