Open Access
Add to BibSonomy
Abstract
The feasibility of in situ quantitative multielemental analysis and production failures detection by laser induced breakdown spectroscopy (LIBS) has been demonstrated during direct energy deposition process in additive manufacturing. Compact LIBS probe was developed and equipped with the laser cladding head installed at industrial robot for real-time chemical quantitative analysis of key components (Ni, W) during the synthesis of high wear resistant coatings of nickel alloy reinforced with tungsten carbide particles. Owing to non-uniform distribution of tungsten carbide grains in the upper surface layer the only acceptable choice for LIBS sampling was made to the melt pool at growing clad. Laser ablation at powder particles above melt pool was insignificant for LIBS plasma properties due to low intensity and low probability of plasma breakdown at powder particles. No impact of LIBS sampling on cladding process and clad properties was observed according to optical and scanning electron microscopies. The feasibility of in situ LIBS quantitative elemental analysis of key components (tungsten and nickel) has been demonstrated during the cladding process. LIBS analysis results were in good agreement with offline measurements by electron energy dispersive X-ray spectroscopy and X-ray fluorescence spectroscopy. Finally, LIBS technique was demonstrated to be a good tool for real-time detection of cladding process failures (poor laser beam quality, undesirable variation of components concentrations).
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Additive manufacturing triggered a new era in metal parts production industry enabling rapid prototyping directly from digital models, superior flexibility for producing internal structures which cannot be fabricated with traditional machinery and unique capability to produce parts with designed gradients of elemental composition [1–7]. However, high quality production in high-value applications where component failure cannot be tolerated (aero and space industry) requires an additive manufacturing process control in real time and/or in situ by sensing systems [8–10].
Several approaches have been suggested toward in situ sensing and control in additive manufacturing [8,9,11–18]. Among the most popular techniques are high speed photography both in visible (VIS) and infrared (IR) spectral regions [8,9]. Optical and infrared pyrometry were frequently utilized to control the additive process by melt pool temperature and dimensions measurements [11–13]. For example, Smurov’s group [11,13,14] demonstrated that cladding defects can be identified by temperature measurements with infrared cameras. Single point pyrometer was suitable to detect major failures while two-dimensional brightness temperature maps were more useful for additive manufacturing process optimization. Later they utilized optical pyrometry combined with computer modelling for detailed study of melt pool dynamics during laser cladding [12]. Hofman et al. developed an optical feedback system to control the cladding laser power and to ensure the quality of cladded layers [19,20]. Recently, high speed X-ray imaging was demonstrated as a new powerful approach for real-time study of laser powder bed fusion [15,16]. Unfortunately, this technique utilization is limited due to low availability and high cost of X-ray synchrotron instrument.
Optical spectroscopy is a low cost but powerful alternative for real-time monitoring of additive manufacturing processes [17,18]. Optical spectroscopy technique has been successfully utilized for many years in case of welding process monitoring. For example, Szymanski et al. [21] suggested optical spectroscopy technique for welding monitoring. They observed that fluctuations of laser plume electron density and temperature are good indicators for welding quality in case of CO2 laser welding. Ancona et al. [22] measured the plasma temperature during CO2 laser welding of stainless steel and discovered that welding quality correlated with the plume temperature. Sforza et al. [23] distinguished low and high quality process online by comparing plume emission in IR, VIS and UV (ultraviolet) spectra regions. You et al. [24] utilized six advanced sensors in visual, ultraviolet and X-ray spectral ranges for detailed study of welding process. However, optic spectra were not obtained and plasma emission was spectrally compared by wide range bandpass filters. Liu et al. utilized optical spectroscopy [25] for real time monitoring of laser hot-wire cladding. According to this technique the fed wire was heated by an external current so less laser power was needed to melt the wire. When cladding process failed, a strong arcing was observed and arc emission spectrum was traced by spectrometer as indicator of process failure. Ya et al. [26] suggested optical spectroscopy to estimate clad dilution by monitoring of laser plume electronic temperature. Pekkarinen et al. [27] optimized powder feeding during laser scanning cladding procedure and demonstrated that poor feeding conditions can be detected with optic spectroscopy. Recently, Stutzman et al. [18] combined optical emission spectroscopy and optical plume imaging for real-time failures detection during directed energy deposition processing. They demonstrated that isolated flaws can be identified by monitoring optical emission of plume at cladding spot.
In this study we developed an instrument for in situ multielemental analysis during additive manufacturing process. Real time elemental analysis offers a unique opportunity to provide control and feedback coupling during additive manufacturing of compositional graded parts (parts with required gradients of chemical composition) or high-value applications where component failure cannot be tolerated (aero-space industry). Up to now, in situ or real time elemental analysis has not been demonstrated for additive manufacturing. The experiments presented here were focused on feasibility demonstration of laser induced breakdown spectroscopy (LIBS) for in situ and real-time quantitative elemental analysis and failure detection during composite samples synthesis by coaxial laser cladding. Coaxial laser cladding technique is a direct energy deposition procedure where metal powder flow is melted by powerful continuous wave laser. Both metal and composite materials can be utilized for three-dimensional parts printing or repairing by laser cladding technique. In this study we have demonstrated the feasibility of in situ elemental analysis and failure detection during synthesis of high wear resistant coatings (nickel alloy reinforced with tungsten carbides).
2. Experimental
2.1 Coaxial laser cladding setup
High wear resistant coatings of Ni-alloy reinforced with tungsten carbide particles were synthesized by coaxial cladding technique. The experimental cladding setup (Fig. 1) was based on a continuous wave ytterbium doped fiber laser (1064 nm, 2 kW, YLS-5 by IPG Photonics). Two channel powder feeder (PF-2/2, GTV) was utilized to produce aerosols with required ratio of nickel alloy and tungsten carbide (NiFeBSi and WC by Hoganas Inc., Table 1). Aerosols were transported by carrier gas (argon, 99.99%) and then were mixed before entering the coaxial cladding head (YC-50, Precitec) which was installed on an industrial robot arm (IRB-2600, ABB). The cladding laser spot was 2 mm in diameter but individual clad dimensions depend on numerous parameters. Typically, individual synthesized clad was of 2 mm high and of 4 mm width (laser power 1.2 kW, 6 and 1 g/sec flows of nickel alloy and tungsten carbide, cladding head movement speed 4 mm/sec). Individual clads were deposited on a flat steel (Steel Fe37-3FN) plate as well as at a 120 mm diameter steel tube.
Fig. 1 Scheme of the coaxial laser cladding head equipped with the laser induced breakdown spectroscopy (LIBS) probe for in situ elemental analysis. Digital camera (left bottom) was synchronized to nanosecond laser pulse for LIBS plasma imaging.
Table 1. Elemental composition (wt. %) of the steel substrate and powders used for laser cladding.
Scanning electron microscopy (SEM) was carried out using a TESCA VEGA LMH microscope equipped with an energy dispersive X-ray (EDX) microanalysis system (AZtecEnergy, Oxford Instruments). The EDX detector was calibrated using standard samples of Ni (99.99%), Mn (99.99%) and Fe (99.999%). A hand-held X-ray fluorescence spectrometer (Thermo Scientific Niton XL2) was utilized for onsite samples measurements.
2.2 LIBS system
Owing to possible complex design of printed parts, the developed LIBS probe should be installed at the cladding head to prevent possible line-of-sight shadowing by growing part or robotic arm itself. The smallest accessible distance between cladding spot (melt pool) and LIBS instrument head should be beyond 30 cm in order to prevent optics damage by backscattered hot powder particles or melt droplets. The maximum allowed mass of LIBS probe was ~2 kg in order fit the requirements for the robotized arm. The developed LIBS probe must quantitatively analyze both light and heavy elements in real time. In order to get analytically meaningful LIBS results powerful laser pulses are needed to ablate the sample and produce high temperature LIBS plasma. Summarizing both requirements, the developed LIBS probe must be light and compact but should provide nanosecond laser pulses of few dozens of mill joules. Recently we demonstrated that laser pulses delivery by fiber optics cannot be utilized for LIBS analysis due to low level of affordable nanosecond pulses energy (<30 mJ/pulse) and poor divergence of laser beam at the fiber output [28]. A low weight diode-pumped Nd:YAG laser (400 g head, 1064 nm, 5 ns, 130 mJ/pulse) was designed to fulfill the requirements for the low weight LIBS probe capable for standoff analysis.
The experiment setup scheme for cladding head and LIBS probe is shown in Fig. 1. The LIBS probe was based on the mentioned above compact Nd:YAG laser. Laser beam was focused through a pierced aluminum coated mirror by a single quartz lens to the 0.5 mm spot at the sample surface. The sampling spot location can be precisely adjusted by LIBS probe alignment with a help of mounting screws. The plasma emission was collected by the aluminum coated mirror and a quartz lens (F = 70 mm) to the entrance of a fiber cable. The chosen 180-degree backscattering optical scheme should provide improved reproducibility of LIBS signals during possible lens-to-sample distance variation. In order to minimize fluctuations of plasma plume location, a special design of fiber cable was chosen: seven quartz fibers arranged in a round at the fiber input and in line at the fiber output. Fiber optics output was fitted to the entrance slit of the Czerny-Turner grating spectrometer (Shamrock 303i, Andor, spectral resolution better than 0.03 nm) equipped with the intensified CCD camera (iStar, Andor). Plasma spectra were acquired with 5 μs exposure and 0.9 μs delay to skip strong continuous emission and get spectra with well resolved atomic/ionic lines. Argon gas (99.9% purity) was continuously flowed through the LIBS probe. The digital CMOS camera (aca1920-40um, Basler) equipped with color filters was utilized for LIBS plasma imaging. The camera was installed perpendicularly to the LIBS probe plane and was synchronized to LIBS pulse to get plasma image during laser ablation pulse. Spectrometer, laser power supply, synchronizing electronics and controlling computer were located in a safe box five meters away from the laser cladding setup. The LIBS probe was synchronized to the managing computer of the laser cladding setup thus LIBS measurements can be carried out automatically at any desired moment during the cladding program run.
3. Results and discussion
The examples of plasma photo and LIBS spectrum induced at an individual clad (nickel alloy reinforced with tungsten carbides) are shown in Fig. 2. The visual LIBS plasma dimensions were 2x3 mm while laser spot was only 0.5 mm in diameter. Tungsten carbide particles had 80-100 μm diameters so only a few particles can be ablated at LIBS sampling spot. LIBS spectrum in 362-375 nm spectral region was full of atomic nickel and iron lines. LIBS signal was defined as an atomic/ionic line net integral with background correction.
Fig. 2 Scanning electron microscopy image of individual clad cross-section (a) (nickel alloy reinforced with tungsten carbide grains), laser induced plasma image (b) and laser induced breakdown spectrum (c) for offline measurement at room temperature.
In LIBS, the choice of optimal spectral region and spectral lines depends on several factors including lines spectral interference, transition probabilities, detector sensitivity and possibility of line self-absorption. Additionally, it is conventionally to use an internal standardization by comparing the analytical line intensity with that of the matrix component of the sample. Procedure of internal standardization in LIBS compensates pulse - to - pulse variations in the amount of ablated matter and in the excitation characteristics of plasma. Different spectral ranges and atomic/ionic lines were utilized for LIBS which are summarized in Table 2: 190-210 nm – good choice for LIBS analysis due to non-spectrally interfered lines for C, W, Ni, Cr, Fe; 360-380 nm – good choice for measurement of both nickel atomic lines intensity and plasma temperature (by Boltzmann plot method with non-resonant iron lines); 489-509 nm – preferred for single pulse LIBS measurements for cladding process detection due to non-spectrally interfered lines for Ni, W and Fe and high sensitivity of the detector (intensified CCD camera).
Table 2. Atomic and ionic lines constants from NIST database [29]: wavelength, transition probability, degeneracy of upper level, energy of upper level (Ek) and energy of lower level (Ei).
Laser clad dimensions (height and width) are not always the same during cladding process so we have estimated the influence of lens-to-sample distance variation on LIBS measurements. Individual clad was deposited on a flat steel plate and then cooled to room temperature. Clad was continuously sampled by LIBS probe at 10 Hz in coaxial and perpendicular directions with 0.5 mm step (see schemes in Fig. 3(a)) thus laser spots were slightly overlapped for two consecutive shots. Laser beam profile projection and the laser peak power density at the cylindrical clad surface will be different during “perpendicular sampling” thus the laser plasma properties and emissivity might be different. Both LIBS signal (Ni I 361.93 nm line integral) and plasma temperature were quantified during coaxial and perpendicular movements and results are presented in Fig. 3. LIBS plasma temperature was estimated by Boltzmann method with non-resonant iron atomic lines (Fe I 370.93, 372.76, 373.49, 374.56 and 376.55 nm) [30]. According to Fig. 3 the chosen 180-degree backscattering scheme for LIBS probe provided good repeatability for both LIBS signal and plasma temperature even in case of “perpendicular sampling”. Minor influence of clad surface geometry on laser plasma temperature and LIBS signals illustrated that quantitative elemental analysis results will not be affected by the clad geometry.
Fig. 3 Reproducibility of laser induced breakdown spectroscopy (LIBS) measurements for coaxial and perpendicular sampling of an individual clad: sampling schemes (a); intensity of Ni I 361.93 nm line (b) and plasma temperature (c) during coaxial and perpendicular mapping of single clad at room temperature.
Typically, nanosecond infrared laser pulses (1064 nm, 10 ns) induced the LIBS plasma within a few nanoseconds. After that moment the Bremsstrahlung absorption in laser plume almost shields the 1064 nm pulse tail from sample surface [31–33]. Cladding process driven by a powerful continuous wave laser can induce low density thermal plasma near the melt pool or hot solidified clad surfaces [21,22]. Supposing that laser breakdown and LIBS plasma can be induced in such low density thermal plasma one can assume that sample surface will be shielded so clad material ablation by nanosecond pulse can be strongly influenced. To estimate low density thermal plasma properties we acquired emission of the melt pool and compared it with the LIBS plasma emission (Fig. 4). The intensity of melt pool thermal emission was rather weak compared to LIBS plasma (103-106-fold difference). According to the laser plasma modelling (at the NIST ASD interface for Laser-Induced Breakdown Spectroscopy [29]) for low temperature plasma (2000 K) the most strong nickel lines (clad matrix) should appear in 335-360 nm range while argon lines (shield gas) can be traced in 415-425 nm spectral window. However, we have not observed any lines in these spectral windows for spectrum acquired at melt pool surface. Absence of atomic lines indicated that plasma near the surface had low density, thus its potential impact during laser ablation by LIBS probe should be minimal if possible to detect.
Fig. 4 Laser induced breakdown plasma and melt pool emission spectra comparison in wide range (a) and 362-375 nm spectral window (b). According to laser plasma modeling by NIST ASD interface the strongest nickel atomic lines should be observed in spectral window 362-375 nm for plasma at 1800-2200 °C temperature.
The optimal choice of LIBS sampling spot during the cladding process is not straightforward due to a number of factors including high temperature in melt pool, temperature gradients at nearby areas and possible impact of powder flows melted in cladding spot. A systematic study was carried out to reveal the optimal sampling location for in situ LIBS measurements. LIBS probe had rather small sampling spot (ellipse 0.3x0.5 mm) compared to dimensions of the melt pool (ellipse 3x4 mm). Any location at cladding spot and 20 mm away can be analyzed by LIBS: melt pool, hot solidified clad. In order to optimize the LIBS sampling location we have installed LIBS probe separately from the cladding head. The idea was to produce individual linear clad which center will ‘pass’ through the stationary LIBS sampling spot. LIBS probe automatically captured plasma spectra at 10 Hz rate thus all areas of clad were sampled continuously (Fig. 5(a)): before cladding spot – inside melt pool – hot solidified clad – cooling solid clad. Plasma spectra were measured at 360-380 nm spectral region to quantify both LIBS signal (Ni I 361.93 nm integral) and electronic plasma temperature for every LIBS measurement. In order to quantify surface temperature of melt and solidified hot clad surface we utilized a disappearing-filament pyrometer equipped with a CMOS camera (aca1920-40um, Basler). Figure 5 illustrated the dependence of plasma emissivity (atomic line integral) and electronic temperature as a function of different clad areas. The plasma temperature and emissivity were higher for the melt pool LIBS sampling compared to the solidified clad ablation. Atomic line emission enhancement was explained by greater electronic temperature.
Fig. 5 Different areas sampling by LIBS probe during cladding process. LIBS probe was installed separately from cladding head and clad was synthesized through the LIBS sampling spot thus “through the melt pool LIBS measurements” were made in single experiment run: experiment scheme (a); LIBS signal for line Ni I 361.93 nm (integral with background correction) (b), laser plasma temperature (c) and clad surface temperature (d).
LIBS signals reproducibility is important parameter defining the accuracy of LIBS chemical analysis. We compared three cases of LIBS sampling spots on signals reproducibility: melt pool, hot solidified clad and clad at room temperature. For first two cases the LIBS measurements were made in real time while for the last case it was done offline. The precision of robotized arm positioning was better than 100 μm thus the same spots were probed after the clad cooling to room temperature. According to Fig. 6 results the poorest reproducibility (defined as relative standard deviation, RSD) was determined for melt pool sampling. Interestingly, but LIBS signal reproducibility for hot solidified clad was better than that for the cooled clad. The poor reproducibility of LIBS measurements for the melt pool sampling cannot be explained by melt surface topography variations due to low effect of lens-to-sample distance variation as it was demonstrated above (Fig. 2). The increased variability of LIBS measurements can be attributed to laser breakdown at powder particles above cladding spot as well as to local temperature fluctuations at melt pool surface.
Fig. 6 LIBS signals (integral of Ni I 361.93 nm line) reproducibility (defined as relative standard deviation, RSD) during online measurements with laser ablation in melt pool (a), just solidified hot clad (b) and offline measurements of solid clad cooled to room temperature (c).
According to best practices for high wear resistant coating synthesis nickel particles must be fully fused in cladding spot while tungsten carbide particles should not (to prevent formation of secondary carbides with lower hardness). This is accomplished by careful control of cladding conditions, higher melting temperature and greater dimensions for tungsten carbide particles. At the tail of cladding spot, the solid tungsten carbide particles will be more effectively scattered compared to small and almost fused nickel alloy particles. This will result in nickel enrichment in 20-50 μm upper surface layer of clad while deeper layers had uniform distribution of WC particles (see Fig. 7(a)). This can be a problem for any express analysis technique including XRF and LIBS. Typical depth for single shot LIBS analysis is less than a micron. In order to estimate such a problem we synthesized the clad with varied concentrations of WC particles (Fig. 7(b)) and LIBS measurements in melt pool was made during the cladding. Synthesized sample was cooled to room temperature and sampled by LIBS probe again. Finally, the upper surface of this clad was grinded by diamond wheel to remove ~200 μm surface layer and then sampled by LIBS probe. Good precision of robotized arm movement (<100 μm) assured that the same spots were sampled by LIBS in these experiments. The comparison of LIBS signals in Fig. 7 clearly demonstrated that preferential enrichment of nickel in the upper surface layer will interrogate LIBS results. Consequently, online LIBS analysis must be carried out by melt pool sampling only.
Fig. 7 LIBS measurements for clad synthesized with varying concentrations of matrix (NiFeBSi) and reinforced grains (WC): (a) – scanning electron microscopy image of top clad cross-section cut in the center along the cladding trajectory; (b) - programed flows of tungsten carbide (WC) and NiFeBSi powders; (c) - in situ LIBS sampling at clad melt pool; (d) - offline LIBS sampling of solid clad at room temperature; (e) – offline LIBS sampling after surface grinding.
When laser plasma is induced by nanosecond pulse the expanding plume generates a shockwave. Shockwave itself as well as the sound wave induced by shockwave push away the material near the ablation spot resulting in melt/liquid splash [31,33]. Melt pool splashes induced by LIBS measurement can impact cladding process resulting in low quality product. Another impact factor is that expanding LIBS plasma can prevent powder particles reaching the cladding spot thus clad dimensions will be change. However, rough estimations of LIBS sampling impact showed that this influence should not be strong enough due to three reasons: low energy of nanosecond laser pulse, the low duty cycle (period between LIBS measurements was 0.1 sec) and high flow of powder material in cladding spot (few grams flow per second compared to few nanograms ablated by LIBS). Supposing that LIBS sampling can have impact on clad properties we synthesized the clad with and without laser ablation pulses (Fig. 8). We have compared clad shape as well as clad cross-section by optical microscopy and scanning electron microscopy and have found no impact of LIBS sampling on clad properties.
Fig. 8 Top-view optical microscopy photo for individual clad (a) and scanning electron microscopy image for lengthwise clad cross section (b) in case of melt spot LIBS sampling (left side) and without any ablation (right side).
The positive correlation of LIBS signals and tungsten concentration in Fig. 7 can be explained by laser ablation which is induced at the particles in the flow rather than melt pool surface. Supposing that laser ablation at powders particles above melt pool can strongly influence LIBS measurements we have systematically studied it with the CMOS camera imaging. Camera was triggered before LIBS pulse in order to visualize LIBS plasma and nearby areas during laser ablation. To suppress blackbody emission from the melt pool we used 50 μs gate and added the color filters transparent in 350-450 nm spectral range. Still low intensity melt pool emission can be traced in images (Fig. 9). No emission of melted or high temperature particles was detected by CMOS camera when it acquired images without LIBS probe pulses. The locations of melt pool surface and laser beam path are shown in Fig. 9(a). Supposing that ablation on the particles can be observed only at the laser beam optical path three characteristic areas were chosen: ‘laser plasma’ – plume induced at melt pool; ‘laser beam’ – micro plasmas induced at particles; LIBS plasma reflection by metal particles. The ‘laser plasma’ signal was defined as a sum of pixels at area defined as ‘laser beam’ in image. Laser breakdown at particles can be induced only along LIBS probe laser beam (‘laser beam’ area in Fig. 9(a) and can be traced as a single or few spikes arranged in a line. All other bright spikes are due to reflection of laser plasma emission by particles (sum of pixels above ‘laser plasma’ area was defined as ‘reflected by particles’). The example of image where laser plasma breakdown at particles can be observed is shown in Fig. 9(b). To visualize plasmas induced at particles, emission reflected by particles and melt emission the images (a) and (b) are shown with laser plasma saturated but for signals calculation original images were not scaled by intensity. All images were processed automatically in home-made software written in LabVIEW environment. Nickel alloy and tungsten carbide powders flows were varied during the experiment in order to estimate influence of different flows on probability of particles ablation by LIBS probe. Plasma imaging revealed that laser breakdown at powder particles is of low probability under the experimental conditions used. Moreover LIBS plasma emission signal was not influenced (Fig. 9(d)) by variation of powder flows (see video Visualization 1). When tungsten carbide flow was increased then the ‘breakdown at particles’ signal enhanced as well as the ‘reflected by particles’ signals.
Fig. 9 Images of the LIBS plasma induced at melt pool surface and powder particles: (a) – scheme of LIBS laser beam, melt pool and laser plasma; (b) – laser plasma photo when laser ablation at powder particles (breakdown at particles) was taking the place (note, that laser plasma emission can be effectively reflected by particles); (c) – scheme of controlled powder flows of nickel alloy (NiFeBSi) and tungsten carbide (WC) particles; (d) – LIBS plasma emission signal (sum of pixels at area defined as ‘laser plasma’ in image (a)); (e) – laser plasma emission (breakdown at particles) in case of laser ablation at powders particles (sum of pixels at area defined as ‘laser beam’ in image (a)); (f) – plasma emission signal for particles which reflected the LIBS plasma emission (sum of pixel above ‘laser plasma’ area shown in image (a)). To visualize plasmas induced at particles, emission reflected by particles and melt emission the images (a) and (b) were scaled by intensity but for signals calculation original images were not scaled by intensity. Laser breakdown at powder particles was presented in the video Visualization 1.
To quantitatively estimate influence of breakdown at particles on LIBS plasma emission a correlation analysis was carried out for these three signals (Fig. 10). We have marked the data for greater flow of tungsten carbide particles with cyan color while low WC concentration data are shown with green color. The ‘reflected by particles’ signal was positively correlated with the laser plasma emission illustrating that these “off LIBS laser beam” spikes are due to reflection of plasma emission by particles. Supposing that laser ablation at powder particles will decrease laser pulse energy delivered to melt pool surface one can expect negative correlation between these two signals. However, a positive correlation between laser plasma emission and ‘breakdown at particles’ signals was observed for flows with different concentrations of tungsten carbides. This indicated that ‘breakdown at particles’ signal is partially due to the ‘reflection’ rather than due to the micro plasmas formation at particles surfaces. Summarizing, we have not determined impact of laser breakdown at particles on LIBS plasma induced at melt pool during cladding process. The poorer reproducibility of the LIBS measurements during melt pool ablation can be attributed to the local fluctuation of surface temperature in melt pool because of material flows and solid particles impingement in the liquid metal [34,35]. To improve reproducibility for in situ and offline LBS measurements each spectrum was summed by ten consecutive laser pulses (RSD = 9%).
Fig. 10 Correlations of emission signals for the laser plasmas induced at melt pool surface and powder particles (a) and emission signals for laser plasma induced at melt surface and plasma light reflected by particle in flow (b). Data for greater flow of tungsten carbide particles (scheme in Fig. 9) are marked with cyan color while low concentration WC flow is shown with green color.
Online detection of cladding process failures is of great demand for improving quality of metal 3D printed parts: when failure occurs the printing process can be aborted or repaired by further processing. Any instrument capable to detect cladding process failures will benefit the additive manufacturing. In order to estimate capabilities of LIBS measurements for cladding process failures detections we have simulated two problems. The first typical failure is poor quality of cladding process due to insufficient power density and/or low quality beam profile at cladding spot. For example, backscattered particles can induce defects at the safety-glass installed inside cladding head to protect the focusing optics. Defects at safety glass (cracks, melted particles, dielectric covering damages) will transform the cladding laser beam profile resulting in non-uniform beam profile and lower power density at cladding spot. The second failure is of high importance for compositionally graded materials synthesis - failure with the powder feeding for one or more components during multiple ingredient cladding. In situ control of elemental composition during cladding process is a necessary requirement for high quality production of compositionally graded parts.
According to the first type failure we have synthesized individual clads with the same parameters except different safety-glasses installed in the cladding head: new and defected safety-glass are shown in Fig. 11. Defect at safety-glass surface transforms the laser beam profile and decreased the power density at the cladding spot. This resulted in low efficiency of powder capturing in the cladding spot and poor quality of the individual clad. During cladding process LIBS probe continuously measured and the results of different signals were compared in Fig. 11. In case of cladding with the defected safety-glass the manifold increase of iron lines intensity was observed during LIBS measurements. This was observed due to partially ablation of substrate surface through defects in clad. Nickel line intensity did not indicate the cladding process failure as well as tungsten-to-nickel lines intensities ratio. The last parameter is a relative measure of WC to Ni powders ratio thus absence of parameter variations demonstrating the absence of problems with the powder flows. LIBS plasma imaging by CMOS camera was also carried out to estimate how plume emissivity will changed during cladding process failures. According to Fig. 11 LIBS plasma emission signal (defined as sum of pixels in plume image area) was rather insensitive for detecting such type failures.
Fig. 11 Cladding failure (defected safety glass in cladding head) detection by LIBS measurements: (a) and (b) – new and defected safety glass photographs as well as corresponding laser clads photographs; (c) – Fe I 489.63 nm line integrals; (d) –Ni I 508.10 nm line integrals corresponding to clads photograph; (e) – ratio of W I 505.34 nm to Ni I 508.10 nm lines integrals; (f) - LIBS plasma emission signal defined as sum of pixels in plume image area for photos acquired by CMOS camera.
LIBS probe was calibrated for quantitative elemental analysis. A series of clads with different tungsten carbide and nickel matrix concentrations were synthesized and cooled to room temperature. Then clad surfaces have been grinded and analyzed by X-ray fluorescence spectroscopy and energy dispersive X-ray spectroscopy as a reference techniques (both techniques provide the same analysis results for major components). Calibration curves for tungsten, nickel and iron were constructed for LIBS probe.
The feasibility of in situ elemental analysis was demonstrated for a clad with a variation of tungsten carbide concentration. The powder feeder was programmed for two-level concentrations of the tungsten carbide flows and LIBS measurements were carried out at 10 Hz repetition rate during cladding process (Fig. 12). Results of LIBS analysis during cladding process are presented in left part of Fig. 12. To verify LIBS results synthesized clads were cooled to room temperature, grinded and then analyzed by a hand-held XRF spectrometer (Niton XL2 by Thermo scientific). A good agreement was achieved between online LIBS and offline XRF measurements for tungsten and nickel according to Fig. 12 results.
Fig. 12 The in situ LIBS quantitative elemental analysis (left part) and cladding failures detection (right part) during additive manufacturing process (a) – programmed flows of tungsten carbide (WC) and Ni-alloy (NiFeBSi) powders. (b) and (c) – LIBS analysis for nickel and tungsten during melt pool sampling (XRF analysis results are marked with the dotted rectangular). It should be noted that sampling spots for LIBS and XRF was rather different – 0.5 and 5 mm respectively. (d) – top-view photograph of clads synthesized without failures (orange and blue) and failure with tungsten carbide powder flow (violet).
In situ LIBS measurements can be a good solution for cladding failures detection during production of compositional graded materials. To simulate failures with tungsten carbide powder flow we have twisted one of the flexible transport tubes to decrease the WC flow. In a single experiment we have synthesized the clad with the required designed of tungsten carbide concentration (left part of Fig. 12 – clad length coordinate <1000 mm) and with the WC flow failure (right part of Fig. 12 – coordinate >1000 mm). When the failure occurred the LIBS probe has identified the tungsten concentration decrease. Clad morphology also changed during WC flow failure (Fig. 12(d)) but the powder flow failure can be traced only by elemental analysis. Online elemental analysis by LIBS technique can be effectively used for controlling quality of cladding process in future. For example, LIBS technique can be utilized for providing 3D maps of elemental composition during production of compositionally graded materials and parts with the predesigned elemental gradients. Elemental 3D maps can be essential for improving printed parts quality as well as predicting lifecycle duration for parts exploited at high value applications.
4. Conclusions
In this study we have demonstrated the feasibility of in situ quantitative elemental analysis during direct energy deposition process in additive manufacturing. Coaxial laser cladding technique was utilized for high wear resistant coatings (NiFeBSi nickel alloy reinforced with tungsten carbide grains) synthesis. Laser induced breakdown spectroscopy (LIBS) was suggested for in situ and real time elemental analysis of clad as well as cladding process failures detection. Compact LIBS probe was developed and equipped the laser cladding head installed at industrial robot for real-time chemical quantitative analysis of key components (tungsten and nickel). It was demonstrated that LIBS signal was not influenced by lens-to-sample distance fluctuations under studied experiment conditions.
LIBS sampling was optimized for different cladding areas (melt pool vs hot solidified clad). Melt pool ablation by LIBS probe provided the higher atomic/ionic lines plasma emission while the better LIBS signals reproducibility was achieved during hot solidified clad sampling. Owing to non-uniform distribution of tungsten carbide grains in the upper surface layer it was demonstrated that the only acceptable choice for LIBS sampling location was the melt pool.
LIBS probe was calibrated for tungsten and nickel analysis with samples characterized by conventional techniques (X-ray fluorescence spectroscopy, energy dispersive X-ray spectroscopy). A good agreement between online LIBS analysis and offline XRF measurements was achieved stimulating further applications for LIBS as a control tool during additive manufacturing process. Additionally, LIBS probe can be used to detect cladding process failures. For example, LIBS analysis was capable to detect problems with cladding process induced by poor laser beam quality due to defected safety-glass. Finally, LIBS probe demonstrated the feasibility of cladding process failures detection triggered by powder components flow misalignment. This failure resulted in undesirable variation of tungsten-to-nickel elemental ratio and hence the unpredicted properties of the printed part.
Funding
Russian Science Foundation (agreement Nº 16-19-10656).
References
1. D. C. Hofmann, J. Kolodziejska, S. Roberts, R. Otis, R. P. Dillon, J.-O. Suh, Z.-K. Liu, and J.-P. Borgonia, “Compositionally graded metals: A new frontier of additive manufacturing,” J. Mater. Res. 29(17), 1899–1910 (2014). [CrossRef]
2. A. Bandyopadhyay and B. Heer, “Additive manufacturing of multi-material structures,” Mater. Sci. Eng. Rep. 129, 1–16 (2018). [CrossRef]
3. T. DebRoy, H. L. Wei, J. S. Zuback, T. Mukherjee, J. W. Elmer, J. O. Milewski, A. M. Beese, A. Wilson-Heid, A. De, and W. Zhang, “Additive manufacturing of metallic components - Process, structure and properties,” Prog. Mater. Sci. 92, 112–224 (2018). [CrossRef]
4. A. J. Pinkerton, “Lasers in additive manufacturing,” Opt. Laser Technol. 78, 25–32 (2016). [CrossRef]
5. C. Leyens and E. Beyer, 8 - Innovations in laser cladding and direct laser metal deposition, in: J. Lawrence, D.G.B.T.-L.S.E. Waugh (Eds.), Woodhead Publ. Ser. Electron. Opt. Mater., Woodhead Publishing, 2015: pp. 181–192.
6. M. Zenou and Z. Kotler, “Printing of metallic 3D micro-objects by laser induced forward transfer,” Opt. Express 24(2), 1431–1446 (2016). [CrossRef] [PubMed]
7. Q. Li, A. P. Alloncle, D. Grojo, and P. Delaporte, “Generating liquid nanojets from copper by dual laser irradiation for ultra-high resolution printing,” Opt. Express 25(20), 24164–24172 (2017). [CrossRef] [PubMed]
8. G. Tapia and A. Elwany, “A Review on Process Monitoring and Control in Metal-Based Additive Manufacturing,” J. Manuf. Sci. Eng. 136(6), 060801 (2014). [CrossRef]
9. S. K. Everton, M. Hirsch, P. Stravroulakis, R. K. Leach, and A. T. Clare, “Review of in-situ process monitoring and in-situ metrology for metal additive manufacturing,” Mater. Des. 95, 431–445 (2016). [CrossRef]
10. N. Shamsaei, A. Yadollahi, L. Bian, and S. M. Thompson, “An overview of Direct Laser Deposition for additive manufacturing; Part II: Mechanical behavior, process parameter optimization and control,” Addit. Manuf. 8, 12–35 (2015). [CrossRef]
11. M. Doubenskaia, P. Bertrand, and I. Smurov, “Pyrometry in laser surface treatment,” Surf. Coat. Tech. 201(5), 1955–1961 (2006). [CrossRef]
12. I. Smurov, M. Doubenskaia, and A. Zaitsev, “Comprehensive analysis of laser cladding by means of optical diagnostics and numerical simulation,” Surf. Coat. Tech. 220, 112–121 (2013). [CrossRef]
13. P. Bertrand, I. Smurov, and D. Grevey, “Application of near infrared pyrometry for continuous Nd: YAG laser welding of stainless steel,” Appl. Surf. Sci. 168(1-4), 182–185 (2000). [CrossRef]
14. M. Doubenskaia, P. Bertrand, and I. Smurov, “Optical monitoring of Nd:YAG laser cladding,” Thin Solid Films 453–454, 477–485 (2004). [CrossRef]
15. C. Kenel, P. Schloth, S. Van Petegem, J. L. Fife, D. Grolimund, A. Menzel, H. Van Swygenhoven, and C. Leinenbach, “In Situ Synchrotron X-Ray Diffraction and Small Angle X-Ray Scattering Studies on Rapidly Heated and Cooled Ti-Al and Al-Cu-Mg Alloys Using Laser-Based Heating,” JOM 68(3), 978–984 (2016). [CrossRef]
16. C. Zhao, K. Fezzaa, R. W. Cunningham, H. Wen, F. De Carlo, L. Chen, A. D. Rollett, and T. Sun, “Real-time monitoring of laser powder bed fusion process using high-speed X-ray imaging and diffraction,” Sci. Rep. 7(1), 3602 (2017). [CrossRef] [PubMed]
17. W.-W. W. Liu, Z.-J. J. Tang, X.-Y. Y. Liu, H.-J. J. Wang, and H.-C. C. Zhang, “A Review on In-situ Monitoring and Adaptive Control Technology for Laser Cladding Remanufacturing,” Procedia CIRP. 61, 235–240 (2017). [CrossRef]
18. C. B. Stutzman, A. R. Nassar, and E. W. Reutzel, “Multi-sensor investigations of optical emissions and their relations to directed energy deposition processes and quality,” Addit. Manuf. 21, 333–339 (2018). [CrossRef]
19. J. T. Hofman, D. F. de Lange, B. Pathiraj, and J. Meijer, “FEM modeling and experimental verification for dilution control in laser cladding,” J. Mater. Process. Technol. 211(2), 187–196 (2011). [CrossRef]
20. J. T. Hofman, B. Pathiraj, J. van Dijk, D. F. de Lange, and J. Meijer, “A camera based feedback control strategy for the laser cladding process,” J. Mater. Process. Technol. 212(11), 2455–2462 (2012). [CrossRef]
21. Z. Szymanski, J. Kurzyna, and W. Kalita, “The spectroscopy of the plasma plume induced during laser welding of stainless steel and titanium,” J. Phys. D Appl. Phys. 30(22), 3153–3162 (1997). [CrossRef]
22. A. Ancona, V. Spagnolo, P. M. Lugarà, and M. Ferrara, “Optical sensor for real-time monitoring of CO2 laser welding process,” Appl. Opt. 40(33), 6019–6025 (2001). [CrossRef] [PubMed]
23. P. Sforza and D. de Blasiis, “On-line optical monitoring system for arc welding,” NDT Int. 35(1), 37–43 (2002). [CrossRef]
24. D. You, X. Gao, and S. Katayama, “Data-driven based analyzing and modeling of MIMO laser welding process by integration of six advanced sensors,” Int. J. Adv. Manuf. Technol. 82(5-8), 1127–1139 (2016). [CrossRef]
25. S. Liu, W. Liu, M. Harooni, J. Ma, and R. Kovacevic, “Real-time monitoring of laser hot-wire cladding of Inconel 625,” Opt. Laser Technol. 62, 124–134 (2014). [CrossRef]
26. W. Ya, A. R. Konuk, R. Aarts, B. Pathiraj, and B. Huis in ’t Veld, “Spectroscopic monitoring of metallic bonding in laser metal deposition,” J. Mater. Process. Technol. 220, 276–284 (2015). [CrossRef]
27. J. Pekkarinen, A. Salminen, V. Kujanpää, J. Ilonen, L. Lensu, and H. Kälviäinen, “Powder cloud behavior in laser cladding using scanning optics,” J. Laser Appl. 28(3), 032007 (2016). [CrossRef]
28. V. N. Lednev, A. E. Dormidonov, P. A. Sdvizhenskii, M. Ya. Grishin, A. N. Fedorov, A. D. Savvin, E. S. Safronova, and S. M. Pershin, “Compact diode-pumped Nd:YAG laser for remote analysis of low-alloy steels by laser-induced breakdown spectroscopy,” J. Anal. At. Spectrom. 33(2), 294–303 (2018). [CrossRef]
29. Y. Ralchenko and A. E. Kramida, J. Reader, N. Team, NIST Atomic Spectra Database (version 5.4), (2018). http://physics.nist.gov/asd.
30. C. Aragón and J. a. Aguilera, “Characterization of laser induced plasmas by optical emission spectroscopy: A review of experiments and methods,” Spectrochim. Acta Part B 63, 893–916 (2008).
31. A. De Giacomo and J. Hermann, “Laser-induced plasma emission: from atomic to molecular spectra,” J. Phys. D Appl. Phys. 50(18), 183002 (2017). [CrossRef]
32. T. A. Labutin, V. N. Lednev, A. A. Ilyin, and A. M. Popov, “Femtosecond laser-induced breakdown spectroscopy,” J. Anal. At. Spectrom. 31(1), 90–118 (2016). [CrossRef]
33. D. W. Hahn and N. Omenetto, “Laser-Induced Breakdown Spectroscopy (LIBS), PART I: Review of Basic Diagnostics and Plasma-Particle Interactions: Still-Challenging Issues Within the Analytical Plasma Community,” Appl. Spectrosc. 64(12), 335–366 (2010). [CrossRef] [PubMed]
34. P. A. Hooper, “Melt pool temperature and cooling rates in laser powder bed fusion,” Addit. Manuf. 22, 548–559 (2018). [CrossRef]
35. H.-S. Tran, J. T. Tchuindjang, H. Paydas, A. Mertens, R. T. Jardin, L. Duchêne, R. Carrus, J. Lecomte-Beckers, and A. M. Habraken, “3D thermal finite element analysis of laser cladding processed Ti-6Al-4V part with microstructural correlations,” Mater. Des. 128, 130–142 (2017). [CrossRef]
