Investigation of erosion temperature in micro-blasting

Author’s Accepted Manuscript Investigation of erosion temperature in microblasting Ruslan Melentiev, Fengzhou Fang

www.elsevier.com/locate/wear

PII: DOI: Reference:

S0043-1648(18)31107-4 https://doi.org/10.1016/j.wear.2018.12.073 WEA102644

To appear in: Wear Received date: 4 September 2018 Revised date: 19 November 2018 Accepted date: 21 December 2018 Cite this article as: Ruslan Melentiev and Fengzhou Fang, Investigation of erosion temperature in micro-blasting, Wear, https://doi.org/10.1016/j.wear.2018.12.073 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Investigation of erosion temperature in micro-blasting Ruslan Melentiev1, Fengzhou Fang1,2* 1

2

Centre of Micro/Nano Manufacturing Technology (MNMT-Dublin), University College Dublin, Dublin, Ireland

State Key Laboratory of Precision Measuring Technology and Instruments, Centre of Micro/Nano Manufacturing Technology (MNMT), Tianjin University, Tianjin, China * Corresponding email: [email protected] Abstract

Micro-blasting is a non-conventional subtractive micro-manufacturing technology based on erosion localisation and intensification. The thermal aspects of erosion are rarely discussed in the literature and commonly neglected in the analysis of erosion mechanism and quality control of machined surface. This study uses analytical, numerical and experimental approaches to explore erosion temperature in relation to micro-blasting of carbon steels. The research shows that, in terms of temperature, airborne erosion by solid particles is a double-natured process; the surface temperature may increase up to a few thousand degrees Celsius after a single impact, yet it dissipates at the subsurface over several microseconds. Melted and adhered drops of the metallic substrate were found in optical images of abrasive particles and SEM images of a machined surface. EDX elemental analysis confirmed that a thin metal layer was deposited on the particle’s interface. Despite this, several FEM models and measurements taken by a thermocouple and IR camera—showed that the steady-state temperature is negligible. The heat generated during each impact has sufficient time to dissipate, leaving a residual temperature of a few degrees before the next impact at the same site. This heat accumulation could elevate the temperature to a few hundred degrees, particularly when the workpiece is small, although convective cooling by air flow effectively dominates this effect. In general, small and slow abrasives are recommended to reduce the temperature. Application of alumina particles smaller than 27 μm to blast low-carbon steels produces minimal negative consequences. Graphical abstract:

Steady state temperature of the eroded spot Instant temperature in the single impact of solid particle Solid particles erosion

Keywords: solid particle erosion, impact temperature, micro-blasting, abrasive jet machining.

1. Introduction Micro-blasting is a non-conventional subtractive micro-manufacturing technology that recently branched off from abrasive jet machining (AJM). AJM is based on erosion localization and intensification. In micro-blasting, a compressed flow of dry air accelerates abrasive micro-particles to impact the target material, diminishing the volume of the workpiece until it reaches the required geometry. Material is removed by means of cutting or ploughing in a ductile substrate or fracturing in a brittle substrate through plastic deformation or crack propagation, respectively. Micro-blasting is applied in the manufacturing of semiconductors, electronic devices, LCDs, micro mould dies, frictional surfaces with specific patterns for machinery and bioimplants, etc. Among the various advantages of AJM, including its negligible thermal effect, small cutting force, high machining versatility and high flexibility, cost efficiency, energy- and resource-efficiency, insensitivity of the machining accuracy to vibrations or temperature changes, slow tool wear, improvement of metallic surface properties via, for example, work hardening and lack of a machining signature and other aspects, recently reviewed in [1], the thermal effect is the least studied and subjected to contradictions. Similar to other tribological processes, solid particle erosion is a combined process; the mechanical load is related to the secondary thermal, chemical and physical reactions between the counterparts [2]. Physical reactions were extensively studied by Finnie, Coffin, Manson, Bitter, Neilson and Gilchrist, Evans, Hutchings and Levy and other scholars since the late 20th century. Their understanding of erosion serves as a foundation for contemporary methods of erosion process control. The chemical reactions that occur during solid particle erosion are described by chemisorption theory. They are mainly based on the surface hydroxylation effect, which was used to develop an elastic removal mode for fluid jet polishing of surfaces to Angstrom-scale roughness (e.g. Ra = 1.8 Å), as described by Peng et al. [3]. In contrast to the other two reactions, there have been almost no studies on the thermal aspect of erosion, that gains no controlled advantages to erosion-based machining technologies. The thermal conductivity and temperature resistance of the target can be important parameters for the erosion process. Localised deformation induced by the impact of particles and the adiabatic conditions prevailing at high strain rates may produce high temperatures. Since high temperatures affect the hardness and fracture toughness of the target, the thermal field has an influence on the erosion resistance of parts [4]. A significant increase in the erosion rate of ceramics

was observed when it was heated to above 800 °C, presumably because the grain boundary softened [5]. In other words, the thermal aspect of erosion may reveal an additional factor that must be considered for process control. High erosion temperatures can have pernicious effects on, for instance, the surface treatment of biomedical implants. A popular type of stainless steel for bone substitution, 316 LVM, includes martensite fragments after sand-blasting [6,7]. Because martensite is ferromagnetic, it limits the application of magnetic resonance imaging for clinical diagnosis. In addition, titanium-based alloys, which are commonly used to fabricate artificial hip joints [8], were reported to exhibit parabolic oxidation kinetics above 600 °C [9]. Corrosion and wear in artificial implants are a known synergic couple [10]. Further, the magnesium-based alloys used in state-of-the-art biodegradable implants [11] are flammable when exposed to oxygen in their molten state, which occurs above 600 °C [12]. Because of the properties of the aforementioned materials, erosion-based machining technologies should be controlled to reduce the effect of thermal reactions. In 1989, Hutchings and Levy [13] reviewed studies on the thermal effects of erosion of ductile metals and discovered two contradictory results: no surface softening occurred, but there was evidence of recrystallization in the erosion debris and close to the surface regions. They revealed that the high temperature at a single impact is more likely to be caused when large particles are used at a high velocity. Significant heat accumulation at a spot being machined can occur only at a very high particles flow rate, and under normal AJM conditions, a steady state temperature does not have any negative consequences. Momber’s [14] brief overview of the thermal phenomena that occur during shot blasting showed that, in the contact zone, the temperature can increase by 200–380 °C. However, Maeda et al.’s [15] theoretical analysis revealed that the instant surface temperature of a steel substrate may increase by up to 1500 °C when a 50-μm steel ball hits the substrate at 222 m/s. Wakuda et al. [16] reported the ductile response of a ceramic substrate to impacts by SiC particles, which is typical only at elevated temperatures [5]. Zhang et al. [17] presented SEM micrographs of melted surfaces (elongated stringers, rounded ends and half-bubbles) of alumina ceramics after airborne erosion by garnet sand, revealing that the garnet sand itself melted and splashed. According to this evidence, while there are multiple arguments supporting the hypothesis that erosion is a hot process, direct temperature measurements during erosion indicate the opposite. The purpose of this study is to examine temperature during solid particles erosion. First, the initial and boundary conditions of heat generation during impact of a single particle are analysed in detail for theoretical determination of the instant surface temperature produced by a single impact. Second, empirical evidence that could support the theoretical results is considered. Third, the steady state temperature of the eroded spot generated by the abrasive flow is simulated and measured. If the steady-state thermal field calculated from the cumulative temperature produced by multiple impacts converges with the measured thermal field, then the theoretical instant temperature produced by a single impact is valid. 2. Theoretical study of thermal field 2.1. Determination of initial and boundary conditions The first step of temperature calculation is to determine the source of energy and which proportion of it is converted into heat. The initial energy source is the kinetic energy of the particle, . The multifaceted nature of the impact of two bodies is related to various phenomena and results in energy dissipation; only a fraction of is converted into heat. Hutchings [18] assumed this fraction to be 0.8 for metals. Shipway and Weston [19] accepted 0.8 for aluminium and 0.7 for polymers. Maeda et al. [15], calculating the shot-peening temperature, assumed 0.7. In this study, we attempt to derive a reasonable fraction of and as it is shown below, the fraction is much smaller. The energy balance during a single impact can be presented as follows ([14] complemented): ,

(1)

where is the initial kinetic energy; , , and are the particle radius, mass, density and velocity, respectively; and are the amounts of energy that turn to plastic and elastic work in particles; and are the amounts of energy that turns to plastic and elastic work in workpieces; and are the surface energy and energy loss associated with mechanical activation and light emission; and is the energy loss associated with noise and vibrations. The last three terms are negligibly small and can be ignored in this quantitative analysis. Considering that the hardness of the particle is considerably higher than the target material, can also be disregarded. Only energy of plastic work in the workpiece ( ) is converted to heat. In part, the elastic response of both the workpiece ( ) and the particle ( ) creates a rebound effect; the initial kinetic energy of the particle ( ) is converted into elastic energy and then, during the rebound phase, back into kinetic energy ( ). The coefficient of restitution ( ) can be used to describe the rebound effect. Complete plastic response ( ) results in full energy absorption by the target material during impact, and complete elastic response ( ) results in . It is obvious that depends on the mechanical characteristics of the material and particle, the particle velocity and the impingement angle. The empirical dependency of for a metal ball impacting a flat plate at moderate velocity was determined previously [20]. Unfortunately, direct measurement of the coefficient of restitution for angular micro-particles at 100–200 m/s presents a considerable experimental challenge. Since we do not have a more accurate determination of , the one mentioned earlier [20] is accepted. After cancelling and in eq. 1, the amount of kinetic energy that turns to plastic work in a workpiece becomes the following [21]: (

)

[

(

( ) (

)

(

) )]

(2)

where H is the substrate hardness, , , and are the Poisson's ratio and Elastic modulus of the workpiece and projectile particle, respectively. The expression in the last rounded brackets on the right-hand side of the equation is also known as . Eq. 2 gives a higher value for soft materials than for hard; the high hardness-to-modulus ratio of soft materials allows for more of the particle’s kinetic energy to be transformed into rebound kinetic energy compared to materials with a low hardness-to-modulus ratio. Particle mass can be determined based on the mean particle size and density . Particle velocity may be found with the theoretical models of Li et al. [22]. Gillstrom and Jarl [23] measured the energy dissipated during shot blasting of steel rod and concluded that “The maximum energy dissipated as heat in the rod is estimated to 39% of the energy in the shots assumed to hit the rod”. In the same analysis, Gillstrom and Jarl reported that only 70% of the applied energy has been identified in the rod. If these numbers can be interpreted as (which is realistic) then the fraction of energy converted to heat for steels. This estimation gives for around 30% lower amount of energy than those being assumed previously. In this case, the density of heat flux (W/m2) can be expressed as follows:

,

(3)

where is the power of a single particle’s impact (W), is the contact area (m2), is the contact radius (m) and is the contact time of the impact (s). Maeda et al. [15] assumed that the contact area is the projected area of the particle. However, multiple shot-peening and sand-blasting experiments clearly showed that the penetration depth during impact is considerably shallower than the radius of the particle. Thus, the real contact area is smaller than the projected area of the particle. As far as the analysis considers the materials with a low ductility (SS, Ti-alloys and ceramics), the indentation depth is small and the springback effect is strong. Thus, the application of fully-plastic model of the semi-brittle substrate response is incorrect. The application of elastic-plastic analysis of Papini and Spelt [24] requires for the separate numerical procedure. Perhaps, the best combination of simplicity and adequality for this thermal analysis can be achieved by assuming a purely elastic substrate response. According to Hertz contact theory, the contact radius can be determined as follows: ( Where is the particle`s radius and follows [14]:

)

(4)

is the contact force. The contact force generated by the impact of a spherical particle can be calculated as (

)

(

)

(5)

where is the factor of elasticity given in the last rounded brackets on the right-hand side of eq. 2. The total impact time time for elastic and plastic interactions:

consists of the contact

(6) A solution for elastic contact time was proposed by Timoshenko and Goodier (1970) (here, the solution is presented in its modified form): (

)

(7)

The equation for plastic contact time was suggested by Chaudri and Walley (1978): (

)

(8)

Since there is no accurate solution for dynamic hardness ( ), static hardness ( ) will be used. When the contact time and radius are calculated, the density of heat flux can be obtained through eq. 3. The advantage of this equation is that the energy fraction, contact time and radius are connected to blasting conditions and substrate properties, while in the previous solution [15] all of them were assumed. The cooling time ( ) is the statistical time interval between two collisions at the same site within the jet spot: , ( ) (9) where

( )

is the local impact number (particles/second): ,

( )

(10)

where is the radial direction (m), is the particle flow rate (g/s), is the normalized function of particle distribution across the jet and is the area of the jet spot (m2), which is always bigger than the aperture of the jet. Due to natural jet divergence, particles move diagonally from the jet’s centreline. During AJM, the particles’ divergence angle ( ) is typically about 7° [25,26] for nozzles ranging from 0.36–1.5 mm and pressure levels ranging from 0.4–0.7 MPa. By using the local jet radius ( ) as a function of stand-off distance (SOD), the area of the jet spot can be determined as follows: [

( )]

(11)

In line with a previous study [25], determines the border after which the erosion depth is less than 10% of the centreline blasting depth. Since cylindrical nozzles generate an axisymmetric normal distribution of abrasive flow density, the function of particles’ distribution with the 10% border is as follows: [ ] ),

( where was fitted to provide ( ) ( ) thermal field in workpiece during AJM can be considered.

(10% from the centreline). When ,

(12) and

are determined from eq. 3, 6 and 9 the

2.2. Analytical solution of the thermal field during impact of a single particle When the impact spot is several orders of magnitude smaller than the jet spot, nearby impact spots are heated to approximately the same temperature and the heat is conducted only to the depth of the substrate (z-axis). Thus, a one-dimensional solution can be applied for each point within the jet spot. A one-dimensional heat conduction equation with a local heat flux density of (W/ ) is as follows [27]: ,

(13)

where is time (s), is the distance from the heated surface, ( ) is the thermal diffusivity ( ), is the thermal conductivity coefficient ( ), is the thermal capacity or specific heat ( ) and is the density of the workpiece ( ). When the boundary conditions are constant, the surface heat flux is and surface heating ( ) starts from , the complementary error function,

( ), can be used to solve the diff usion equation. The derivations are known [27] and omitted here for the sake of brevity. The localized temperature (°C) at z distance from the heated surface at a determined moment of heating ( ) can be determined using the following equation: √ ( ) { [ ( ) ] } (14) √ √ √ √ To determine the surface temperature during heating, ( ), all parts of the equation with can be omitted. After the heating time ( ), the heat flux becomes 0 during the cooling time ( ), and the temperature generated in near-surface regions spreads deeper than the surface due to the temperature gradient. Following Maeda et al. [15], the cooling phase can be written as follows: √ ( ) )) ( ) ( ( (√ ) √ (15) ) √ √ ( where ( ) is the integral of ( ). Eq. 14 and 15 show how temperature changes in a workpiece during a single impact of a spherical particle and during the cooling process, respectively. The heating–cooling cycle can be described by the Heaviside step function. A determined number of cycles can be introduced using the square wave function. However, neither of these functions are described to maintain brevity. Both equations were written in Excel spreadsheet to obtain the analytical prediction of the temperature rise in a single impact. 2.3. Numerical solution of the thermal field at the macro scale. As shown in the results and discussion, the cooling time ( ) is 4-5 orders of magnitude longer than the impact time ( ). A numerical simulation of the temperature generated at the jet spot for one second requires around time steps, which is unaffordable in terms of the computational costs, especially when a two- or three-dimensional solution is desired. To simulate the thermal field during a prolonged blasting time, the heat flux ( ) of single impact was multiplied by the total number of impacts at the jet spot within a unit of time. Based on the Gaussian density of particles’ distribution in the jet ( ), the impact frequency falls as radial distance increases. Indicated leads to the Gaussian distribution of the heat flux density ( ( ) ) over the jet radius ( ): (

)

(16)

Two conditions were modelled. The geometry of the first model imitates a thermocouple juncture with a 0.6 mm spherical head. The second model is a parallelepiped with a length and width that are sufficiently large to maintain the ambient temperature at the side walls. The thickness of the plate was 0.5 mm to concentrate heat at the surface by limiting conduction in the z-direction. Heat flux occurs on the upper surface of the circular area with a density of and a blasting duration of , as shown in Figure 1. Simultaneously, the sphere and the upper surface of the plate undergo ( ), where and convective cooling due to the fast flow of cooled air. The convective heat flux is are the temperature of the surface of the workpiece and the air temperature, respectively. The convective heat transfer coefficients for the horizontal plate ( ) and sphere ( ) during external forced cooling by turbulent flow were calculated as described on p. 70 -76 in [28]. Simulations were performed with COMSOL Multiphysics 5.3. The mesh was extra fine with additional refinement at the upper wall. In total, each model included around 120.000 elements. Mesh refinement to 200.000 elements produced difference only in the third digit after the decimal point. The input data used in analytical and numerical calculations is given in the experimental section (Table 1). Different SODs were selected to achieve a small machining spot ( ) for aimed erosion of a thermocouple head (0.6 mm in diameter) and a large machining spot ( ) on the steel plate for accurate IR camera observation. To achieve comparable heat flux density at different SODs, the flow rate during blasting of the plate was increased by four times the original value. Model 1: Thermocouple 0.42 mm

Model 2: SS 316L Plate Erosion-heating spot

Erosion-heating spot SS 316L plate Heat flux density q, W/m2 9.2·105 9.8·105 0.97 mm

Cooling surfaces (All of them) Insulated surfaces

Cooling surface

Figure 1. Conditions of the FEM models This study makes the following assumptions:  The particle contacts with the workpiece as a sphere with a plane  The substrate response is purely elastic.  Energy loss due to particle deformation and fracturing, light emission, noise, vibrations and surface energy are neglected.  Material removal and heat withdrawal into a chip are not considered.  The heat flux produced by a single impact is uniform during the contact time and within the contact radius.  The thermo-mechanical properties of the system are constant.

3. Experimental investigation of thermal field

The majority of experimental results reviewed in previous studies ([13] and [14]) were obtained using the thermocouple method. The essential drawback of thermocouples is their operational inertia. The response time of the thinnest commercially available thermocouples (Ø25–80 μm) is barely 1 ms, which is still two to five orders of magnitude longer than the impact time during shot/sand/powder-blasting or other airborne erosion conditions. In other words, the thermocouple registers the temperature long after the impact event. Thus, the thermocouple method is able to identify only the steady state temperature generated by multiple impacts during a prolonged period of time. Unfortunately, there is no way to directly measure temperature during a single impact by a solid particle. Indirect markers of high temperature are the change of microhardness and microstructure of the target surface, chip morphology and chemical composition of the interface of impingement bodies. However, as an indicator of impact temperature, microhardness is unreliable. Despite the structural changes in the impact zone, the metallic phase transitions can result in either material strengthening or softening, depending on the initial structure, erosion intensity and cooling rate. Thus, no solid conclusions about temperature can be made based on a surface’s microhardness. The microstructure of an eroded surface is of greater interest as an indicator since high temperatures typically produce recognisable surface features such as rounded edges and micro-cavities, see an example [29]. If a chip is small enough and a significant amount of heat is drawn into it, then the high temperature and melting may affect the chip’s geometry. Other evidence of a high contact temperature includes elemental deposition and adhesion of impacted bodies. Since abrasives have high temperature resistance, for instance around 2000 °C for corundum, the fusible target material is more likely to be deposited onto abrasives. The instant temperature was examined based on the microstructure and results of elemental analysis of an eroded surface and chip. The workpiece was a flat washer made from low carbon steel with a 5-μm-thick galvanised zinc (Zn) coating. A Zn coating was selected because its low melting temperature and diamagnetic properties allowed Zn and iron (Fe) chips to be easily separated with a magnetic field. This method has two advantages: 1) it allows for study of pure substrate chips without considering previous surface contamination and oxidation and 2) it enables more reliable conclusions if both metals are found at the interface of abrasive particles. Erosion tests were conducted using a Crystal Mark Inc. MV- 2 micro-blaster with adjustable air pressure and powder flow. During the tests, 50-μm white alumina particles were emitted through a 0.36 x 8 mm cylindrical nozzle with a flow rate of 3 g/min at 0.7 MPa. The mixture of particles and workpiece chips were collected from a sealed blasting minichamber and the chips of diamagnetic Zn coating and ferromagnetic Fe substrate were separated by a neodymium magnet. The surface of the magnet was observed using a Keyence VHX-5000 optical microscope. Then, the collected chips, blasted surface and used and unused particles were observed using FEI – QUANTA SEM 3D FEG and exposed to an EDX probe. The steady state temperature was measured in two ways. First, a K-type thermocouple with a PTFE welded 0.6-mm tip operating in a temperature range of -75–260°C and first-class tolerance according to IEC 584 was directly exposed to an abrasive jet. Second, a thin stainless-steel 316L plate (20 x 20 x 0.5 mm) was blasted and observed through an infrared thermal image camera (Fluke Ti9) and infrared spectrometer, as shown in Figure 2. The setup parameters are shown in Table 1. Table 1. Blasting conditions used in experiments and simulations. Parameter

Exp. 1

Exp. 2

Target Thermocouple SS 316L plate Absolute pressure 0.7 MPa Abrasive particles 50-μm Al2O3 Particle flow rate 0.04 g/s 0.17 g/s Nozzle length 8 mm Nozzle diameter 0.36 mm Stand-off distance 0.5 mm 5 mm Particle velocity 120 m/s 140 m/s Air velocity 300 m/s Air flow temperature 14 °C Room temperature 21 °C Blasting time 10 s

Micrometre positioner Multimetre

Nozzle

IR camera Workpiece IR Spectrometer

Thermocouple head

Dust absorber

Figure 2. Experimental setup 4. Results and discussion 4.1. Instant temperature

The temperature increase that occurs after a single impact is shown in Figure 3 - Figure 8. These results were obtained by applying the analytical procedure given in section 2.2 to the erosion of SS 316L with averaged mechanical and thermal properties. Particle size and velocity play a significant role in determining the maximum of the instant temperature produced by a single (Figure 3). The heating rate (ΔT/Δt) is very high, varying from 1·1011 to 1·1010 °C/s. Interestingly, the heating rate is insensitive to particle size and velocity but is closely related to heating time (Figure 4). In other words, the heating rate is independent of the amount of kinetic energy and is mainly governed by the size and duration of the elastoplastic response of the substrate. However, the maximum temperature is strongly governed by kinetic energy. The role of conductivity grows as the temperature increases, and the cooling rate remains the same regardless of particle size. The analysis shows that the temperature caused by a single impact of a spherical particle is significantly higher than that measured by Oka et al. [30]. However, the convergent theoretical and experimental data shown in Figure 5 demonstrates that the first response of a thermocouple mounted to the reverse side of 20-μm iron leaf observed by Oka et al. was registered 0.3 s after the impact, when the temperature dissipated to around 60 °C. Note, the only fitting factor is the time (abscissa). Most of the input data was taken from the experimental details in [30]. The temperature just 1–2 μm below the surface was several times lower than on the surface (Figure 6). Thus, while the 0–1-μm sub-surface layer may undergo a phase transformation, the 1–3-μm sub-surface layer may endure only high thermal tempering or low annealing. The trigger point for steels (727 °C) can be easily exceeded when large particles are used at a high velocity. Considering that the cooling rate is extremely high (ΔT/Δt is up to 800 °C/μs), the FCC lattice of β-iron will become a BCC lattice of α-iron with a martensitic grid. Multigner et al. [6,7] and Barriuso et al. [31] reported grain size refinement and the presence of α’-martensite down to 30 μm below the surface of 316 LVM steel blasted with 0.75–2-mm angular Al2O3 particles under 0.35 MPa. Recrystallisation can be facilitated by high temperatures or the deformation gradient. However, the critical deformation gradient, which is necessary for the formation of strain-induced α’-martensite, decreases as temperature increases. By applying the developed analytical model to the blasting conditions in a study of Barriuso et al. [31] (316 LVM blasted with 1-mm Al2O3 at 50 m/s), once can see that the impact temperature in the 30-μm subsurface layer falls from 2830°C to 130°C. The combination of a high deformation gradient and elevated temperature during blasting sufficiently explains the formation of α’-martensite in austenitic steels, which was observed in [6,7,31]. The effect of particles’ size and velocity on the maximum temperature at 1 μm below the surface in the current conditions is shown in Figure 7. The time scale of heat dissipation can be seen in Figure 8. At 1 μs, the surface temperature falls below 300 °C. Hutchings and Levy’s expectation that ‘the temperature pulse from each impact has time to decay to negligible proportions before the next impact on the same site’ [13] is thus confirmed. In this analysis, the residual temperature before the next impact at the same site does not exceed 10°C for a wide range of materials and blasting conditions. Even at such extremally high flow rate as 10 g/min of 27-μm Al2O3 particles emitted through the 360-μm nozzle to steel target at 1 mm stand-off distance, the cooling time is still for 4 orders of magnitude longer than the heating time. The residual temperature in the aforementioned conditions is only 5.2°C. The cumulative effect will be discussed in the next section. It can be objected, that the peak temperature should be lowered due to heat convection and radiation, which are not included in the eq. 14 and 15. To analyse the contribution of these factors, we performed a separate numerical simulation of the instant temperature produced by a single impact. The result is illustrated on the left side of the graphical abstract of this paper. Interestingly, but both convection and radiation produce an insignificant effect on the maximum temperature. The amount of heat transferred to surrounded air per impact time is unexpectedly small. For the impact conditions given on Figure 6, the mechanisms of radiation and heat convection withdraw the first 1°C of surface temperature at 2 μs after the impact. Worth to notice that the simulated impact time was only 0.2 μs. It implies that the heat conduction is the main mechanisms of heat dissipation in a single particle`s impact. The comparison of the results obtained with the numerical simulation to those from the analytical model (eq. 14-15) shows no more than 2% divergence for a wide range of impact conditions. 2000 ti = 0.1 μs ti = 0.2 μs ti = 0.3 μs

dp = 50 µm, vp = 160 m/s

3000

dp = 27µm, vp = 160 m/s

Temperature, T (C)

Temperature, T (C)

3500

dp = 12 µm, vp = 160 m/s

2500

dp = 5 µm, vp = 160 m/s

2000 1500 1000

1500 1000 500

500 0 1.E-09

1.E-07

2.E-07

3.E-07

4.E-07

0 1.E-09

5.E-07

Time, t (μs) Figure 3. Effect of particle size and velocity on the instant temperature produced by a single impact Measured by Oka et al. Calculated Target material Particle size Particle velocity Distance to surface

100

4.E-07

6.E-07

8.E-07

Time, t (μs) Figure 4. Effect of contact time on the instant temperature produced by a single impact

2500 Temperature, T (oC)

Temperature, T (C)

1000

2.E-07

Iron (HV = 1.22 GPa) 3.18 mm, steel ball 100 m/s 20 μm

2000 z = 0 μm

1500

Material Air pressure Nozzle d x L Particle size Particle velocity

SS 316 L (2.6 GPa) 0.7 MPa 0.36 x 8 mm 50 μm Al2O3 120 m/s

1000 z = 1 μm

500

z = 2 μm

10 0

0.5

1

1.5 Time, t (s)

2

2.5

3

0 0.0

0.2

0.4

0.6

Time, t (μs)

0.8

1.0

Figure 5. Comparison of the predicted temperature produced by a single impact with that measured by Oka et al. [30]

Figure 6. Dissipation of the instant temperature below the surface

727 50 27 0

12 50

100

150 Particle velocity, vp (m/s)

3 200

Figure 7. Temperature at 1 μm below the surface depending on the particle’s size and velocity

Temperature, T (oC)

α-Fe to γ-Fe transformation

1454

Particle size, dp (μm)

Temperature, T (C)

2500 t = ti = 0.09 μs (impact time)

2000 1500

T = 4 oC (bottom line) at t = tc = 5800 μs (cooling time untill next impact)

1000 t = 0.18 μs

500

t = 1 μs

0 0

1 2 3 Distance from surface, z (μm)

4

5

Figure 8. Instant temperature produced by a single impact depending on the distance from the surface

The metallic chips that stick to the magnet’s surface are black (see Figure 9) and thus are notably different from the colour of the substrate material. The straightforward explanation for these black chips’ colour is soot produced by thermal decomposition or burning of the metal. Unfortunately, SEM observation does not support either of these reasons. It is also difficult to make any conclusions about temperature based on the chips’ morphology. Nevertheless, markers of high temperature are visible on the blasted surface. SEM images with a magnification of 1000x and below do not demonstrate any evidence of high temperature, while fine SEM images with 15.000x magnification reveal interesting features at the lower end of the micro scale. In addition to obvious extruded zones with striations, the edges of local asperities were rounded (see Figure 10). Furthermore, multiple spherical features attached to the surface were identified to be crystallised drops of metal. These features are similar to those obtained by thermally assisted blasting at 600 ºC, as shown in Figure 14d in a previous study [29], and to those on an annealed surface of a sintered specimen, as shown in Figure 12a in another study [32]. Both referred surfaces were processed at elevated temperatures and demonstrate the similar morphological elements (rounded edges and spherical drops) to those observed on Figure 10. It is obvious that the surface features in Figure 10 were obtained at the elevated temperature as well. Another phenomenal feature was observed on the surface of a magnet: non-magnetic particles of white alumina oxide (99,9%) exhibited magnetic behaviour after blasting (Figure 9). Tarasenko et al. [33] stated that some of the magnetisations of alumina particles can be due to the presence of inherent magnetic impurities. However, unused particles were scanned by the neodymium magnet, and the surface of magnet remained clean. The magnetic particles’ behaviour could also be explained by a triboelectric effect. The intensive friction produced during an impact may generate an electrostatic charge in abrasive particles, resulting in magnetisation. Since the generation of triboelectricity does not require electric conductivity in frictional counterparts, we reproduced the experiment, replacing the steel target with one made of pure aluminium, and we observed no chip or abrasive particles on the surface of magnet. The incidence angle was reduced to 45° to increase the tangential component of impact force that results in increased friction, even so, no magnetic activity was observed in the worked-out mixture. Further, the particles’ magnetisation could be due to melting and splashing of metal or evaporation and deposition on the surface of abrasives. The separated black dots on particles’ surface can be seen in optical images of the magnet. Presumably, these dots are the adhered splatter of the melted substrate. It can be argued, that such deposition could be from plastic material transfer. Although, the plastically transferred chip would be taken away from the alumina particle by the magnet, that is not the case. The adhesion of these elements is strong enough to draw the whole abrasive particle and keep it on the magnet`s surface during experimental manipulations. Figure 9 also shows that the magnetic alumina oxide changed in colour from white to dark grey. This may be due to thermal decomposition of the metal target and deposition on the particles. In SEM analysis, the hue of the observed object is related to its conductivity. The colour of the pure surface of alumina particles is monophonic, like in [34] for instance, while surface impurities create more contrast picture, similar to those in thesis [35]. Typical impurities include non-abrasive residue, lead, water-soluble contaminants and oil. Here, SEM images of unused alumina particles and those collected from the magnet after blasting at the same voltage and current settings are shown in Figure 11. Conductive impurities are spread across the surface of alumina particles. EDX analysis of the surface revealed abundant Zn and Fe (Figure 12), which are the coating and base of the target material, respectively. According to Figure 3, the maximum temperature is lower for particles with lower kinetic energy. Thus, smaller and slower particles should not cause melting or evaporation. The experiment was reproduced with 27-μm Al2O3 particles blasted at 120 m/s. Some alumina particles were still present on the magnet’s surface, but most were larger than the mean size of the fraction. The observations discussed in this paragraph confirm that solid particle erosion produces a high instant temperature.

Neodymium magnet

Mixture of workedout particles and workpiece chip

Figure 9. Collected ferromagnetic chips and optical image of the magnet’s surface

Figure 10. SEM image of chips (left) and the surface (right) eroded by 50-μm Al2O3 particles at 120 m/s

Figure 11. SEM image of 50-μm Al2O3 particles before and after impacting the metal substrate

KV: 10.00 TILT: 0.00 TAKE-OFF: 34.64 AMPT: 1.6 DETECTOR TYPE: SDD APOLLO XV RESOLUTION: 129.40

KV: 10.00 TILT: 0.00 TAKE-OFF: 34.64 AMPT: 1.6 DETECTOR TYPE: SDD APOLLO XV RESOLUTION: 129.40

Figure 12. EDX analysis of alumina particles before and after impacting the metal substrate

4.2. Established thermal field Under the erosion conditions listed in Table 1 for a thermocouple, 113 particles hit the same site each second. The residual surface temperature accumulated after each impact may establish a steady state elevated temperature below the machining spot. A FEM model of thermocouple heating due to exposure to an abrasive jet shows that the temperature increases to 200 °C and gradually decreases along the wires. When the model includes forced external cooling, the temperature barely exceeds 30 °C. During blasting for 10 s, the thermocouple remains at a comparable temperature. The experimental and numerical results for thermocouples are illustrated in Figure 13. The sudden increase in temperature during the first few seconds in all three tests is associated with an unstable flow rate. Numerous particles conglomerate in the mixing chamber below the orifice plate when the blaster is in stand-by mode and are instantly directed to the nozzle when the blasting process begins. The total number of particles that impact the target during the first second can be several times higher than the average per minute, thus creating more powerful heat flux. In the next few seconds, the heat generated at the beginning is dissipated to a steady state level, which is around 30 ºC under these conditions.

70

Test 1 Test 2 Test 3

Temperature, T (C)

60 50 40 30 20 10 0 0 a) Optical image of the eroded thermocouple head

c) Simulation of thermocouple heating

2

4

6 Time, t (s)

8

10

b) Temperature of the thermocouple (measured)

d) Simulation of thermocouple heating and convection cooling

12

Figure 13. Thermocouple temperature produced by erosion (experimental and numerical) The machined spot on the SS 316L plate was approximately equal to that predicted by eq. 11, which was used in the FEM model. The eroded footprint can be seen in Figure 14a. Observation with an IR camera also showed a sharp temperature increase to around 50 °C during the first second. For the following 10 s of erosion, the temperature was near 30 °C, as shown in a thermal screenshot (Figure 14b). The numerical model leads to very similar results. The analysis shows that, in case of a stable flow rate, the maximum temperature (30 °C) is almost established in one second (Figure 14c and 12d). Since the air flow is colder than the ambient temperature, the edges of the plate become colder that starts to facilitate conduction in a few seconds after the beginning of blasting process. Without convection, the steel plate continues to accumulate heat and the maximum temperature grows linearly after the first second. This experiment was reproduced with thicker steel plates, and no heating was observed when the thickness exceeded 4 mm.

b) Thermal image taken by an IR camera

Temperature, T (oC)

a) Optical image and topography of the eroded spot

Distance from the centre of the machined spot, y (mm) c) Simulated thermal field

d) Simulated temperature across the plate

Figure 14. Temperature of the stainless steel plate produced by the erosion during a prolonged time (experimental and numerical)

Conclusions Airborne erosion by solid particles is both a hot and cold process. Ambiguity regarding the subject can be clarified by distinguishing the temperature at the micro scale from that at the macro scale. The first case is the instant temperature increase produced by a single impact, which is dissipated over several μs within a subsurface region that is a few μm thick. This temperature is sufficient to initiate melting and phase transformation in the thin subsurface layer of the metal substrate. The maximum temperature is proportional to the amount of energy converted to head, but inverse to contact radius and time. The heating rate is independent of impact energy and mainly governed by the impact time and conduction of the selected material. The cooling rate immediately after the impact (ΔT/Δt < 800 °C/μs for steels) is for six order of magnitude higher than that, which is sufficient for martensite formation. The main mechanism of surface cooling is the heat conduction. The surface radiation and convection contribute negligibly to the heat dissipation during the impact time. The developed analytical model demonstrates an excellent convergence (within 2%) with numerical simulation. Material properties have a significant influence on maximum temperature, its depth distribution and rate of dissipation. These dependencies can be effortlessly predicted by the analytical model presented in this paper. In general, small and slow abrasives are recommended to prevent local austenisation and, consequently, martensite formation. This study shows that use of alumina particles below 27 μm to blast steels produces a minimal thermal effect. The second case is the steady state temperature, which is the residual temperature accumulated after each impact at the same site. This temperature field is established across the erosion spot during the first few seconds. During micro abrasive jet machining, steady-state erosion temperature cannot result in annealing or intensified oxidation, even under aggressive erosion conditions (e.g. 10 g/min through a 360-μm nozzle). Two numerical models were developed and verified experimentally. The thermal field distribution in the workpiece is determined by its geometry and thermal properties, the abrasive jet’s power and the air jet’s characteristics. When at least one of the dimensions of the workpiece is at the micro scale, external convection by air flow plays a significant role in process of heat withdraw. The findings of this study complement existing knowledge about the nature of erosion. Further quantitative estimation of the integrity of eroded surfaces is required to correlate it with theoretical temperatures for practical application in quality control strategies for abrasive jet machining.

Acknowledgements We are grateful to Maeda et al. [15] for their thermal analysis of shot peening and Isupov for the ideas about empirical verification of instant temperature he discussed in his doctoral thesis. We also appreciate N. Zhang for providing thermal measuring equipment and H. G. Zhang for useful discussions. Funding This work was supported by the Science Foundation Ireland (No. 15/RP/B3208) and the National Natural Science Foundation of China (Nos. 51320105009 & 61635008). References [1] [2] [3] [4] [5] [6]

[7]

[8]

[9] [10] [11]

[12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

R. Melentiev, F. Fang, Recent advances and challenges of abrasive jet machining, CIRP J. Manuf. Sci. Technol. 22 (2018) 1–20. doi:10.1016/j.cirpj.2018.06.001. M.-N. Barkoula, J. Karger-Kocsis, Processes and influencing parameters on the solid particle erosion of polymers and their composites, J. Mater. Sci. 37 (2002) 3807–3820. W. Peng, C. Guan, S. Li, Material removal mode affected by the particle size in fluid jet polishing., Appl. Opt. 52 (2013) 7927–33. doi:10.1364/AO.52.007927. D. Patel, P. Tandon, Experimental investigations of thermally enhanced abrasive water jet machining of hard-to-machine metals, CIRP J. Manuf. Sci. Technol. 10 (2015) 92–101. doi:10.1016/j.cirpj.2015.04.002. M.A. Tetsuya Senda, Kazuyoshi Arai, Sand erosion of alumina ceramics at elevated temperatures, J. Ceram. Soc. Japan. 109 (2001) 254–259. doi:10.2109/jcersj.109.1267_254. M. Multigner, S. Ferreira-Barragáns, E. Frutos, M. Jaafar, J. Ibáñez, P. Marín, M.T. Pérez-Prado, G. González-Doncel, A. Asenjo, J.L. González-Carrasco, Superficial severe plastic deformation of 316 LVM stainless steel through grit blasting: Effects on its microstructure and subsurface mechanical properties, Surf. Coatings Technol. 205 (2010) 1830–1837. doi:10.1016/J.SURFCOAT.2010.07.126. M. Multigner, E. Frutos, J.L. González-Carrasco, J.A. Jiménez, P. Marín, J. Ibáñez, Influence of the sandblasting on the subsurface microstructure of 316LVM stainless steel: Implications on the magnetic and mechanical properties, Mater. Sci. Eng. C. 29 (2009) 1357–1360. doi:10.1016/J.MSEC.2008.11.002. B. Jetté, V. Brailovski, C. Simoneau, M. Dumas, P. Terriault, Development and in vitro validation of a simplified numerical model for the design of a biomimetic femoral stem, J. Mech. Behav. Biomed. Mater. 77 (2018) 539–550. doi:10.1016/j.jmbbm.2017.10.019. P. Kofstad, High-temperature oxidation of titanium, J. Less Common Met. 12 (1967) 449–464. doi:10.1016/0022-5088(67)900173. J. Komotori, B.J. Lee, H. Dong, P.A. Dearnley, Corrosion response of surface engineered titanium alloys damaged by prior abrasion, Wear. 250 (2001) 1239–1249. doi:10.1016/S0043-1648(01)00753-0. N. Hort, Y. Huang, D. Fechner, M. Störmer, C. Blawert, F. Witte, C. Vogt, H. Drücker, R. Willumeit, K.U. Kainer, F. Feyerabend, Magnesium alloys as implant materials-Principles of property design for Mg-RE alloys, Acta Biomater. 6 (2010) 1714–1725. doi:10.1016/j.actbio.2009.09.010. N. Tomac, K. Tonnessen, F.O. Rasch, Formation of Flank Build-up in Cutting Magnesium Alloys, CIRP Ann. - Manuf. Technol. 40 (1991) 79–82. doi:10.1016/S0007-8506(07)61938-6. I.M. Hutchings, A. V. Levy, Thermal effects in the erosion of ductile metals, Wear. 131 (1989) 105–121. doi:10.1016/00431648(89)90248-2. A. Momber, Blast cleaning technology, 2008. doi:10.1007/978-3-540-73645-5. K.I. H. Maeda, N. Egami, C. Kagaya, N. Inoue, H. Takeshita, Analysis of particle velocity and temperature distribution of struck surface in fine particle peening, J. Jpn. Soc. Mech. Eng. (in Japanese). 67 (2001) 306–312. doi:10.1248/cpb.37.3229. M. Wakuda, Y. Yamauchi, S. Kanzaki, Material response to particle impact during abrasive jet machining of alumina ceramics, J. Mater. Process. Technol. 132 (2003) 177–183. doi:10.1016/S0924-0136(02)00848-8. Y. Zhang, Y.B. Cheng, S. Lathabai, Erosion of alumina ceramics by air- and water-suspended garnet particles, Wear. 240 (2000) 40–51. doi:10.1016/S0043-1648(00)00335-5. I.M. Hutchings, A model for the erosion of metals by spherical particles at normal incidence, Wear. 326 (1980) 858. doi:10.1111/mec/14011. P.H. Shipway, D.P. Weston, Thermal effects in blasting and erosion of polymeric materials, J. Mater. Process. Technol. 209 (2009) 6161–6167. doi:10.1016/j.jmatprotec.2009.03.006. K.L. Johnson, Contact mechanics, Cambridge University Press, Cambridge, UK, 1987. doi:https://doi.org/10.1002/zamm.19890690713. B.F. Levin, K.S. Vecchio, J.N. DuPont, A.R. Marder, Modeling solid-particle erosion of ductile alloys, Metall. Mater. Trans. A. 30 (1999) 1763–1774. doi:10.1007/s11661-999-0175-9. H.Z. Li, J. Wang, J.M. Fan, Analysis and modelling of particle velocities in micro-abrasive air jet, Int. J. Mach. Tools Manuf. 49 (2009) 850–858. doi:10.1016/j.ijmachtools.2009.05.012. P. Gillström, M. Jarl, Replacement of pickling with shot blasting for wire rod preparation, Scand. J. Metall. 33 (2004) 269–278. doi:10.1111/j.1600-0692.2004.00692.x. M. Papini, J.K. Spelt, The plowing erosion of organic coatings by spherical particles, Wear. 222 (1998) 38–48. doi:10.1016/S0043-1648(98)00274-9. M. Achtsnick, High Performance Micro Abrasive Blasting, Ph.D. Thesis, Technical University of Berlin, 2005. J.M. Fan, H.Z. Li, J. Wang, C.Y. Wang, A study of the flow characteristics in micro-abrasive jets, Exp. Therm. Fluid Sci. 35 (2011) 1097–1106. doi:10.1016/j.expthermflusci.2011.03.004. H.S. Carslaw, J.C. Jaeger, Conduction of Heat in Solids, Oxford University Press, Oxford, 1959. B. Vath, The Heat Transfer Module User’s Guide, Comsol. November (2013) 394. A.T. Fry, P. Lovelock, N. Smith, M. Gee, A.J. Gant, An innovative high temperature solid particulate erosion testing system, Wear. 376–377 (2017) 458–467. doi:10.1016/j.wear.2016.12.064. Y.I. Oka, M. Matsumura, T. Kawabata, Relationship between surface hardness and erosion damage caused by solid particle

impact, Wear. 162–164 (1993) 688–695. doi:10.1016/0043-1648(93)90067-V. S. Barriuso, M. Jaafar, J. Chao, A. Asenjo, J.L. Gonzalez-Carrasco, Improvement of the blasting induced effects on medical 316 LVM stainless steel by short-term thermal treatments, Surf. Coatings Technol. 258 (2014) 1075–1081. doi:10.1016/j.surfcoat.2014.07.027. M. Ahangarkani, K. Zangeneh-Madar, S. Borji, Z. Valefi, Microstructural study on ultra-high temperature erosion mechanism of infiltrated W-10 wt%Cu composite, Int. J. Refract. Met. Hard Mater. 67 (2017) 115–124. doi:10.1016/j.ijrmhm.2017.05.009. R. Tarasenko, A. Orendacova, O. M, M. Kajnakova, A. Feher, Experimental Study of Magnetic Properties of Alumina Ceramics, Acta Phys. Pol. A. 118 (2010) 1067–1068. C. Stenson, M.C. Meyer, B. Twomey, R. Lupoi, Particle speed analysis in CoBlast, Procedia CIRP. 37 (2015) 77–82. doi:10.1016/j.procir.2015.08.004. H. Wensink, Fabrication of microstructures by Powder Blasting, University of Twente, 2002. http://core.ac.uk/download/pdf/11465378.pdf.

[31]

[32] [33] [34] [35]

Highlights:   

Temperature rise and decay occur as a result of a single impact of a solid particle Steady-state temperature and thermal field expansion occur in eroded workpieces Thermocouple and IR temperature measurements are taken during micro-blasting