Investigating the Linear Thermal Expansion of Additively Manufactured Multi-Material Joining Between Invar and Steel

materials

Article Investigating the Linear Thermal Expansion of Additively Manufactured Multi-Material Joining between Invar and Steel

Alexander Arbogast 1,2,* , Sougata Roy 3,4 , Andrzej Nycz 2, Mark W. Noakes 2, Christopher Masuo 2 and Sudarsanam Suresh Babu 1,2 1 Department of Mechanical, Aerospace and Biomedical Engineering, University of Tennessee, Knoxville, TN 37916, USA; [email protected] 2 Manufacturing Science Division, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA; [email protected] (A.N.); [email protected] (M.W.N.); [email protected] (C.M.) 3 Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37830, USA 4 Department of Mechanical Engineering, University of North Dakota, Grand Forks, ND 58202, USA; [email protected] * Correspondence: [email protected]; Tel.: +931-267-3988

Received: 25 October 2020; Accepted: 9 December 2020; Published: 12 December 2020

Abstract: This work investigated the linear thermal expansion properties of a multi-material specimen fabricated with Invar M93 and A36 steel. A sequence of tests was performed to investigate the viability of additively manufactured Invar M93 for lowering the coefficient of thermal expansion (CTE) in multi-material part tooling. Invar beads were additively manufactured on a steel base plate using a fiber laser system, and samples were taken from the steel, Invar, and the interface between the two materials. The CTE of the samples was measured between 40 ◦C and 150 ◦C using a thermomechanical analyzer, and the elemental composition was studied with energy dispersive X-ray spectroscopy. The CTE of samples taken from the steel and the interface remained comparable to that of A36 steel; however, deviations between the thermal expansion values were prevalent due to element diffusion in and around the heat-affected zone. The CTEs measured from the Invar bead were lower than those from the other sections with the largest and smallest thermal expansion values being 10.40 µm/m-K and 2.09 µm/m-K. In each of the sections, the largest CTE was measured from samples taken from the end of the weld beads. An additional test was performed to measure the aggregate expansion of multi-material tools. Invar beads were welded on an A36 steel plate. The invar was machined, and the sample was heated in an oven from 40 ◦C and 160 ◦C. Strain gauges were placed on the surface of the part and were used to analyze how the combined thermal expansions of the invar and steel would affect the thermal expansion on the surface of a tool. There were small deviations between the expansion values measured by gauges placed in different orientations, and the elongation of the sample was greatest along the dimension containing a larger percentage of steel. On average, the expansion of the machined Invar surface was 42% less than the expansion of the steel surface. The results of this work demonstrate that additively manufactured Invar can be utilized to decrease the CTE for multi-material part tooling.

Keywords: additive manufacturing; multi-material joining; Invar; steel; coeﬃcient of thermal expansion; tooling

1. Introduction Invar is an Fe-Ni alloy that has an exceptionally low coeﬃcient of thermal expansion (CTE) over a broad range of useful temperatures [1]. The name Invar originates from the term invariable

Materials 2020, 13, 5683; doi:10.3390/ma13245683 www.mdpi.com/journal/materials Materials 2020, 13, 5683 2 of 18 because of the persistent dimensional stability of the alloy over varying temperatures [2]. Between the temperatures of 30 ◦C and 150 ◦C, the CTE of Invar is between 1.2 and 2.7 µm/m-K [3]. Although Fe-Ni alloys demonstrate anomalous thermal behavior with many different compositions (some including additional components such as titanium and manganese), Invar is typically manufactured with weight percentages of 64% Fe and 36% Ni to procure the lowest possible CTE for the binary alloy [4,5]. The thermal expansion attenuation, often referred to as the Invar-effect or magnetostriction, is a result of the magnetic state of the alloy [1,6]. The ferromagnetic state required for magnetostriction only exists below the Curie temperature of Invar [7]. This temperature is a function of the composition and impurities present in the Invar alloy, and when the Curie temperature is reached, the alloy transitions from a ferromagnetic state to a paramagnetic state and loses its anomalous thermal expansion properties [8]. The low thermal expansion of Invar makes the alloy a suitable choice for many precision performance and manufacturing applications. Invar is commonly used in pipes for liquified natural gas, bimetal devices, watch hair springs, and filters for electronics [7,9]. Moreover, molds, dies, and other forming tools can be manufactured with Invar to produce equipment that maintains exceptional dimensional tolerances through forming processes that involve cyclical temperature cycles. The ability to acquire such tight tolerances is extremely beneficial to processes like composite forming in the aerospace industry. With the initiative to reduce carbon emissions and produce lighter and more efficient aircraft, the aerospace industry has begun to utilize large numbers of lightweight composite components. When molding composite parts, dissimilarities between the CTEs of resins, fibers, and the mold material can lead to geometric distortion and induce interfacial stresses and strains in the finished part [10]. The typical CTE of carbon fiber-reinforced epoxy composites is between 2.9 and 3.6 µm/m-K [11]. This is much smaller than the thermal expansion of most alloys. The CTE for typical tool steels can range from 11 to 15 µm/m-K depending on the composition of the alloy [12]. Therefore, the geometric distortion produced by an Invar mold will be far less than that of a tool steel mold. For this reason, the dimensional stability, surface quality, and durability of Invar has led to the alloy’s success and enduring popularity for composite molds [13,14]. Typically,the materials used for hard tooling include alloys of aluminum and steel [15]. These materials are inexpensive and capable of withstanding rigorous high-production cycles, but the thermal expansion of these alloys during the cyclical heating and cooling processes engenders a loss of geometric precision. Although Invar meets many of the requirements of the ideal tooling material, the alloy is relatively expensive compared to other tooling alloys (approximately 16 times the cost of carbon steel) [16,17]. Conventionally, Invar tools are manufactured through a manual process involving up to 17 or more operations; these include but are not limited to forming, butt-welding, rough machining, finish machining, and thermal stress relieving [13]. The conventional manufacturing process requires skilled personnel and involves extensive lead times. Invar is a soft and gummy alloy which makes it rather difficult to machine [4]. The inefficiency of the conventional manufacturing methods associated with the Invar alloy are further emphasized through the large amounts of material often wasted through bulk-machining processes, the excessive wear on cutting tools, and slow cutting speeds [18]. To increase the efficiency of manufacturing Invar parts, measures must be taken to optimize the amount of Invar necessary to produce the required thermal expansion properties and reduce the amount of machining required to fabricate components. Additive manufacturing (AM) is a process where consecutive layers of material are added on top of one another to build components directly from a computer-aided design model. Since the development of AM is the 1980s, the process has become increasingly popular because it facilitates the production of geometrically complex components and reduces lead times and human interaction [19]. Multiple metal additive manufacturing techniques have been developed to introduce the benefits of AM to the metal manufacturing industry. These techniques include powder bed fusion, binder jetting, directed energy deposition (DED), and others. Each metal AM technique can produce parts that are near net shape—significantly reducing the amount of material that must be removed by machining. Materials 2020, 13, 5683 3 of 18

Recent studies have concluded that there is between a 75% to 90% production time and cost reduction when using additive manufacturing for tooling rather than conventional methods [20]. This is particularly useful for expensive materials such as Invar. Previous research has determined that selectively laser melted (SLM) Invar maintains thermal expansion properties similar to those of conventionally manufactured Invar [4,21]. Other studies have investigated the mechanical properties of SLM Invar and suggested process parameters to achieve a stable and refined microstructure with comparable mechanical properties to the traditionally fabricated alloy [22]. Powder-bed AM processes, like SLM, can produce high-resolution parts, but the build volumes of these systems are typically less than 0.3 m3 [23]. Additionally, powder feedstock is very expensive, so these processes are not ideal for building large-volume components (such as molds and dies) [24]. Directed energy deposition processes have higher production rates, lower capital cost, and a larger build volume than other metal AM processes; this makes DED the most efficient and economic method for producing large near net shape components [24,25]. In addition to the previously mentioned benefits of DED, the industrial robots often used for DED processes can be equipped with end effectors capable of depositing two or more materials in a single layer of a build. Multi-material metal additive manufacturing has been used to successfully fabricate parts with dissimilar metals or functionally graded materials to obtain user-defined material properties throughout the part. The powder-based DED processes, such as laser metal deposition, are generally used to fabricate functionally graded materials because the powder composition can be altered between layers [26,27]. Defining deposition efficiency as the percentage of material deposited that embodies the final product, powder-based DED has a 14% deposition efficiency which is much lower than the 100% deposition efficiency attainable with wire-based DED [28,29]. Wire-based processes are not yet capable of producing functionally graded materials, but they can deposit dissimilar metals to achieve local material properties throughout the part [30]. Dissimilar metal joining using high energy-density beams has been used to improve joint characteristics and reduce distortion in additively manufactured parts. Hinijos et al. utilized electron beam melting additive manufacturing to join Inconel 718 and 316 stainless steel [31]. They noted that the affinity of the dissimilar alloys to precipitate and form phases or precipitates can have undesirable effects on the weld quality or material properties [31]. Additionally, they observed cracks at the interface of the Inconel and stainless steel due to the dissimilarity between the CTE of the substrate and the filler material [31]. T. Abe and H. Sasahara used wire-arc based additive manufacturing to join stainless steel and a nickel-based alloy to produce a part with high heat and corrosion resistance [28]. Using energy dispersive X-ray spectroscopy, the authors found that the nickel alloy had been diluted with iron from the stainless steel at the interface of the two materials [28]. They claimed that the dilution would reduce the resulting materials heat and corrosion resistance, but the mechanical properties of the combined alloys proved comparable to those of the individual materials [28]. For Invar, the multi-material additive manufacturing process is a promising technique for reducing the amount of Invar needed to obtain the required thermal properties. Parts can be manufactured with an Invar shell and an inner metal core (such as low-carbon steel) to tailor the thermal expansion properties to specific applications while enhancing the economic feasibility of Invar fabrication. Multiple studies have analyzed methods for joining Invar and dissimilar alloys. Xu et al. used tungsten inert gas (TIG) welding to join Invar and carbon steel plates [32]. They noted that residual stresses caused by thermal expansion coefficient mismatches and second-phase precipitates from element diffusion are the two primary concerns associated with dissimilar metal joining involving Invar. They investigated the effects of temperature field distribution and its relation to element diffusion and concluded that a full penetration dissimilar weld joint between Invar and steel was obtained using their specified welding parameters. Li et al. successfully joined two Invar 36 plates using ER309 filler wire [33]. They used both laser welding and laser-arc hybrid welding to compare the microstructure, thermal expansion characteristics, and mechanical properties of the weld joints. The authors claim that the base material and the laser welded specimens match closely to the theoretical expansion Materials 2020, 13, 5683 4 of 18

Materials 2020, 13, x 4 of 19 expected from the Invar alloy, but the laser-arc hybrid welded specimen exhibits a sharp increase in CTE.The They authors extended claim this that explanation the base material and described and the thatlaser hybridwelded welding specimens resulted match inclosely an alloy to the with a smallertheoretical fraction expansion of the γ-(Fe,Ni) expected phase—the from the Invar phase allo responsibley, but the forlaser-arc the Invar hybrid effect. welded The mostspecimen popular 2γ-stateexhibits model a sharp for the increase Invar in eff CTE.ect proposed They extended by Weiss this statesexplanation that the and thermal described expansion that hybrid attenuation welding is a functionresulted of both in an the alloy ferromagnetic with a smaller high-volume fraction of statethe γ-(Fe,Ni) and antiferromagnetic phase—the phase low-volume responsible statefor the of the alloyInvar [1]. Li effect. et al. The reported most popular that the 2γ hybrid-state model welding for the process Invar resulted effect proposed in a decreased by Weiss Ni states percentage that the and the introductionthermal expansion of elements attenuation such as is C, a Mn, function and Cr. of Theboth reduction the ferromagnetic in nickel percentage high-volume and state the additionand of theantiferromagnetic alloying elements low-volume produced state a new of the phase alloy of [1]. FeCr0.29Ni0.16C0.06 Li et al. reported that resultingthe hybrid in welding a two-phase process alloy that hasresulted a smaller in a decreased fraction ofNiγ percentage-(Fe,Ni) and and did the not introd exhibituction the of sameelements thermal such expansionas C, Mn, and characteristics Cr. The reduction in nickel percentage and the addition of the alloying elements produced a new phase of as the single-phase γ-(Fe,Ni) alloy. FeCr0.29Ni0.16C0.06 resulting in a two-phase alloy that has a smaller fraction of γ-(Fe,Ni) and did notIn the exhibit present the same study, thermal Invar expansion sections of characteristics varying thicknesses as the single-phase were deposited γ-(Fe,Ni) atop alloy. a steel base plate and samplesIn the were present taken study, from Invar below sections the Invar of varyin beads,g thicknesses from the weld were interface, deposited and atop directly a steel frombase the printedplate bead. and samples The samples were taken were from used below to investigate the Invar beads, the thermal from the expansion weld interface, properties and directly throughout from the weldedthe printed specimen. bead. Energy-dispersive The samples were used X-ray to investigate spectroscopy the thermal (EDS) andexpansion electron properties micro probethroughout analysis (EPMA)the werewelded performed specimen. to Energy-dispersive analyze the composition X-ray spectroscopy of the alloy (EDS) along and the interfaceelectron micro regions probe of Invar and theanalysis base (EPMA) plate. The were study performed found thatto analyze steel samplesthe composition taken from of the under alloy thealong Invar the interface bead had regions CTEs that wereof typical Invar and for steel.the base The plate. diff usionThe study in and found around that steel the fusionsamples zone taken of from the weldunder increased the Invar bead the CTEs had for mostCTEs samples that takenwere typical from the for interfacesteel. The ofdiffusion the two in materials. and around The the samples fusion zone taken of directlythe weld from increased the Invar beadthe showed CTEs for a decreased most samples thermal taken expansion, from the interface and the of di theffusion two materials. had a smaller The samples effect when taken thedirectly samples from the Invar bead showed a decreased thermal expansion, and the diffusion had a smaller effect were taken farther from the fusion zone. In addition to finding the local thermal expansion properties when the samples were taken farther from the fusion zone. In addition to finding the local thermal of theexpansion printed specimen,properties of the the nominal printed surfacespecimen, expansion the nominal of a surface part was expansion analyzed of a to part determine was analyzed how well the partto determine would function how well in the low part thermal would function expansion in low applications. thermal expansion Invar beadsapplications. were depositedInvar beads on a steelwere base plate,deposited and on the a beadssteel base were plat facee, and milled the tobeads replicate were theface surface milled to of replicate a post-processed the surface additively of a manufacturedpost-processed tool. additively The part wasmanufactured then heated tool. in anThe oven, part andwas strainthen heated gauges in were an usedoven, toand measure strain the overallgauges thermal were expansion used to measure of the the combined overall thermal materials. expansion The surface of the expansioncombined materials. of the face-milled The surface Invar beadsexpansion was measured of the toface-milled be between Invar the beads expected was thermal measured expansions to be between for Invar the andexpected steel. thermal expansions for Invar and steel. 2. Materials and Methods 2. Materials and Methods Invar beads were deposited on two 9.5 mm-thick A36 steel base plates using a laser-hotwire additiveInvar manufacturing beads were process.deposited Theon two laser-hotwire 9.5 mm-thick process A36 steel uses base a plates laser end-eusing ffaector laser-hotwire held by an additive manufacturing process. The laser-hotwire process uses a laser end-effector held by an industrial robot. The laser end-effector is moved while wire is fed through a hot wire torch, and the industrial robot. The laser end-effector is moved while wire is fed through a hot wire torch, and the laser melts the filler wire to deposit a weld bead. A schematic of the laser hot wire process is provided laser melts the filler wire to deposit a weld bead. A schematic of the laser hot wire process is in Figureprovided1. in Figure 1.

Figure 1. Laser end-eﬀector used for the laser-hot wire additive manufacturing process. MaterialsMaterials2020, 202013, 5683, 13, x 5 of 195 of 18

Figure 1. Laser end-effector used for the laser-hot wire additive manufacturing process. Aperam Invar® M93 wire was used to deposit invar beads on both the ﬁrst and second steel plates. TheAperam chemical Invar composition® M93 wire was of the used wire to provideddeposit invar by thebeads vendor on both is shown the first in and Table second1. steel plates. The chemical composition of the wire provided by the vendor is shown in Table 1. Table 1. Chemical composition of Aperam Invar® M93 wire (wt.%). Table 1. Chemical composition of Aperam Invar® M93 wire (wt.%). Ni Fe Mn C Si S P Ni Fe Mn C Si S P 36 Balance 0.2–0.4 0.04 0.25 0.0015 0.008 36 Balance 0.2–0.4 ≤ ≤ 0.04 ≤ ≤ 0.25≤ ≤ 0.0015≤ ≤ 0.008

The firstThe first welded welded plate plate contained contained three three sections sections varying varying in in bothboth thethe number of of beads beads and and layers layers of Invar.of TheInvar. approximate The approximate dimensions dimensions of the of invarthe inva beadsr beads and and the the base base plate plate are are providedprovided in in Figure Figure 2a, and the2a, and physical the physical specimen specimen is shown is shown in Figure in Figure2b. The 2b. invarThe invar deposited deposited on on the the second second plate plate was was two two beads in width, and the weave amplitude was increased to 10 mm make the width of each bead beads in width, and the weave amplitude was increased to 10 mm make the width of each bead approximately 20 mm. The beads were spaced 19 mm apart two ensure a 1 mm overlap between approximately 20 mm. The beads were spaced 19 mm apart two ensure a 1 mm overlap between beads. beads. Two passes for each bead were performed, and the final height of the beads was Two passesapproximately for each 4 beadmm. The were plate performed, was cut to and 230 themm finalin length height and of 30 the mm beads in width, was and approximately the top of both 4 mm. The platebeads was on the cut tosecond 230 mmplate in were length machined and 30 mmto leave in width, a smooth and and the topflat of2.8 both mm-thick beads surface on the secondfor platethermal were machined expansion to analysis. leave a The smooth Invar andbeads flat on 2.8 both mm-thick plates were surface deposited for thermal using 1.2 expansion mm Invar analysis.M93 The Invarwire and beads a wire-based on both plates Lincoln were Electric deposited fiber laser using system. 1.2 mm Argon Invar was M93 used wire as and the ashielding wire-based gas Lincolnfor Electricthe fiberwelding laser process, system. and Argon the laser was power used was as the set shieldingto 5800 W. gasRobot for travel the welding speeds for process, welding and the thefirst laser powerand was second set to plates 5800 were W. Robot 21.2 mm/s travel and speeds 5.1 mm/s, forwelding respectively, the firstand weaving and second was platesused on were the second 21.2 mm /s and 5.1plate mm to /produces, respectively, a wider andbead. weaving was used on the second plate to produce a wider bead.

(a)

(b)

Figure 2. (a) Invar bead layout and Invar sample locations on A36 steel baseplate and measured dimensions (mm) of base plate and samples in each direction. (b) Welded invar beads on steel baseplate before sample removal. Materials 2020, 13, x 6 of 19 Materials 2020, 13, 5683 6 of 18 Figure 2. (a) Invar bead layout and Invar sample locations on A36 steel baseplate and measured dimensions (mm) of base plate and samples in each direction. (b) Welded invar beads on steel Samplesbaseplate before were sample machined removal. from four sections (indicated 1, 2, 3, and 4 in Figure3) on the ﬁrst welded plate. The dimensions of samples taken from the X, Y, and Z directions can be found in Figure2a, Samples were machined from four sections (indicated 1, 2, 3, and 4 in Figure 3) on the first andwelded the sampleplate. The distance dimensions from of the samples top of taken the weld from beadthe X, as Y, well and asZ thedirections distance can from be found theinterface in of the twoFigure materials 2a, and the can sample be inferred distance from from Figure the top3. of Section the weld 3 hasbead two as well layers as the of Invar,distance and from sections the 1 and 2 haveinterface one of layer the two of Invar.materials Samples can be inferred are referred from Figure to according 3. Section to 3 theirhas two material, layers of direction, Invar, and and number. Thesections label 1M93 and refers2 have to one samples layer of taken Invar. from Samp a two-layerles are referred Invar to bead. according M93 /toA36 their refers material, to samples taken fromdirection, the interfaceand number. between The label the M93 Invar refers bead to samples and the taken steel from substrate, a two-layer and Invar WA36 bead. refers M93/A36 to samples taken refers to samples taken from the interface between the Invar bead and the steel substrate, and WA36 from the substrate under the Invar bead. Samples taken from the pure steel region of the baseplate are refers to samples taken from the substrate under the Invar bead. Samples taken from the pure steel labeledregion of as the A36. baseplate M93, M93are labeled/A36, as and A36. WA36 M93,samples M93/A36, were and WA36 taken samples from 0.13 were mm, taken 0.45 from mm, 0.13 and 1.95 mm belowmm, 0.45 the mm, surface and of1.95 the mm Invar below bead, the respectively.surface of the A36 Invar steel bead, samples respectively. were takenA36 steel from samples the top surface of thewere steel. taken from the top surface of the steel.

FigureFigure 3. 3. A36A36 steel steel baseplate baseplate and andwelded welded Invar bead Invar samp beadle samplelocations locationsfor thermomechanical for thermomechanical analysis analysis andand corresponding cross-sectional cross-sectional views views of each of In eachvar and Invar steel and section steel with section the respective with the sample respective sample depths in relation to the material interface and top of the weld beads. depths in relation to the material interface and top of the weld beads. Selected samples from the first plate were studied using energy-dispersive X-ray spectroscopy Selected samples from the first plate were studied using energy-dispersive X-ray spectroscopy (EDS) to analyze elemental distribution along the weld metal and base metal interface regions. For (EDS)that, metallographic to analyze elemental specimens distribution were prepared along by the sectioning weld metal the andweld base interface metal regions interface using regions. a For that, metallographicprecision saw and specimens mounted wereon conductive prepared bakelite by sectioning. Later, they the weldwere prepared interface using regions standard using a precision sawmetallographic and mounted procedures on conductive of up to bakelite.1 μm diamond Later, paste they werepolish. prepared Scanning usingelectron standard microscopy metallographic procedures(SEM) and EDS of upspectroscopy to 1 µm of diamond the interface paste regions polish. wereScanning conducted electronusing a JEOL microscopy 6500 FEG-SEM (SEM) and EDS spectroscopy(JEOL USA, Inc., of thePeabody, interface MA, regions USA) at were 20 kV conducted equipped usingwith ED aAX JEOL and 6500 electron FEG-SEM back scatter (JEOL USA, Inc., Peabody,diffraction MA, (EBSD). USA) at 20 kV equipped with EDAX and electron back scatter diffraction (EBSD). Compositional analyses using EPMA were acquired on an electron microprobe (JEOL-JXA8200) Compositional analyses using EPMA were acquired on an electron microprobe (JEOL-JXA8200) (JEOL USA, Inc., Peabody, MA, USA) equipped with five tunable wavelength dispersive (JEOLspectrometers. USA, Inc., EDS Peabody, spectra were MA, acquired USA) equipped and processed with fiveusing tunable a Thermo wavelength Pathfinder dispersiveEDS (Thermo spectrometers. EDS spectra were acquired and processed using a Thermo Pathfinder EDS (Thermo Fisher Scientific., Waltham, MA, USA) system. The operating conditions were a beam energy of 15 keV and beam current of 50nA. Elements were acquired using EDS for Fe Kα, Ni Kα, and analyzing crystals LIFH for Mn kα, and TAP for Si kα. The counting time (both on peak and off peak) was 10 s for both Mn and Si. The off-peak correction method was linear for Mn and exponential for Si. Standard intensities were corrected for standard drift over time. The width of the line scans was 1µm to avoid capturing signals from sides which could affect the data. Materials 2020, 13, x 7 of 19

Fisher Scientific., Waltham, MA, USA) system. The operating conditions were a beam energy of 15keV and beam current of 50nA. Elements were acquired using EDS for Fe Kα, Ni Kα, and Materials 2020, 13, 5683 7 of 18 analyzing crystals LIFH for Mn kα, and TAP for Si kα. The counting time (both on peak and off peak) was 10 s for both Mn and Si. The off-peak correction method was linear for Mn and exponentialA Q400 thermomechanical for Si. Standard intensities analyzer were (TMA) corrected (TA Instruments, for standard Newdrift over Castle, time. DE, The USA) width was of usedthe to measureline scans thewas thermal 1μm to avoid expansion capturing for samples signals from taken sides from which the first could plate. affect The the initialdata. height of each sampleA was Q400 measured thermomechanical using the TMA’s analyzer thermal (TMA) expansion (TA Instruments, probe, and New the probe Castle, was DE, set USA) to expand was used from to measure the thermal expansion for samples taken from the first plate. The initial height of each the initial height of each sample. Samples were held at 40 ◦C for five min to maintain an isothermal sample was measured using the TMA’s thermal expansion probe, and the probe was set to expand temperature profile. The temperature was increased from 40 ◦C to 150 ◦C at a rate of 5 ◦C/min, from the initial height of each sample. Samples were held at 40 °C for five min to maintain an and thermal expansion measurements were taken at a frequency of 2 Hz. The coefficient of thermal isothermal temperature profile. The temperature was increased from 40 °C to 150 °C at a rate of 5 expansion α for each sample is given by: °C/min, and thermal expansion measurements were taken at a frequency of 2 Hz. The coefficient of thermal expansion α for each sample is given by: ∆L α = . (1) L0∆∆T = . (1) ∆ Using Equation (1), the CTE of each sample was calculated with the original length L0, change in length ∆UsingL, and Equation change (1), in temperature the CTE of each∆T recordedsample was from calculated the TMA. with the original length L0, change in Thelength second ∆L, and welded change plate in temperature and an invar ∆ referenceT recorded plate from were the TMA. used to analyze surface expansion of the machinedThe second Invar welded beads. plate The secondand an weldedinvar reference plate, the plate invar were reference used to plate,analyze and surface the experimental expansion setupof the is shown machined in Figure Invar4 .beads. The surface The second expansion weld ofed the plate, additively the invar manufactured reference plate, and machinedand the experimental setup is shown in Figure 4. The surface expansion of the additively manufactured and specimen was measured to determine how the combination of materials would affect the overall machined specimen was measured to determine how the combination of materials would affect the thermal expansion on the surface of a part. The thermal expansion on the surface of the steel and the overall thermal expansion on the surface of a part. The thermal expansion on the surface of the steel invar was measured using 12 strain gauges (forming six half-bridge circuits). and the invar was measured using 12 strain gauges (forming six half-bridge circuits).

(a)

(b)

FigureFigure 4. 4.(a ()a) Strain Strain gauges gauges bondedbonded to to the the machined machined Inva Invarr beads, beads, steel, steel, and and Invar Invar reference reference plate. plate. (b) (b)Invar Invar plates plates in in the the oven oven with with strain strain gauges gauges and and thermocouples. thermocouples.

StrainStrain gauges gauges are are sensors sensors that that have have a a change change inin resistanceresistance proportionalproportional to to the the strain strain induced induced on on thethe surface surface of of the the gauge. gauge. InIn thethe casecase of thermal expansion, expansion, this this strain strain is isalso also commonly commonly referred referred to as to asthe thermal thermal output output [34]. [34 The]. The 12 gauges 12 gauges were werebonded bonded in pairs in with pairs one with gauge one on gauge the analyzed on the specimen analyzed specimenand the andother the on otheran Invar on reference an Invar plate. reference When plate. the surface When expansion the surface is expansionmeasured with is measured a half-bridge with a half-bridgestrain gauge strain configuration, gauge conﬁguration, the thermal theexpansio thermaln will expansion be the difference will be the between diﬀerence the betweenindividual the individual thermal outputs of the strain gauges on the machined Invar beads and the Invar reference plate [34]. WK-00-125AD-350 strain gauges (Micro-Measurements—Vishay Precision Group, Malvern, PA, USA) were used to compare the thermal expansion of the machined Invar beads and the steel to the expansion of the Invar reference plate. The location and orientation of each strain gauge is Materials 2020, 13, x 8 of 19

thermal outputs of the strain gauges on the machined Invar beads and the Invar reference plate [34]. WK-00-125AD-350 strain gauges (Micro-Measurements—Vishay Precision Group, Malvern, PA, USA) were used to compare the thermal expansion of the machined Invar beads and the steel to the expansion of the Invar reference plate. The location and orientation of each strain gauge is shown in Figure 5. The printed invar extends to the edges of the baseplate in the physical specimen; however, space is left in Figure 5 to distinguish the baseplate from the printed beads. The WK-00-125AD-350 strain gauges were fully encapsulated K-alloy gauges that were chosen because of their minimal self-temperature compensation that matches the low thermal expansion coefficient of Invar [35,36]. The reference plate measured 152.4 mm × 152.4 mm × 2.5 mm, and the CTE of the plate met the American Society for Testing and Materials (ASTM) F1684 specifications [3]. M-Bond 610, a strain gauge adhesive with a maximum operating temperature of 260 °C, was used to apply the strain gauges to the steel, Invar, and the reference plate to ensure that the oven temperature did not exceed the maximum operating temperature of the adhesive [37]. K-type thermocouples were used to measure the temperature of each plate. A 1/16 in. hole was drilled in the base of the plates, and thermocouples were inserted into the holes. A National MaterialsInstruments2020, 13 , 56839211 temperature input module was used to record temperature data from the8 of 18 thermocouples, and the strain gauge measurements were recorded using a National Instruments 9237 shownbridge in completion Figure5. The module. printed Each invar module extends was to theinse edgesrted into of thea National baseplate Instruments in the physical CompactRIO specimen; backplane, and a LabVIEW program was used to collect the data. The wiring diagram for each module however, space is left in Figure5 to distinguish the baseplate from the printed beads. is shown in Figure 6.

Figure 5. Location and orientation of strain gauges on the machined invar beads, steel plate, and the Figure 5. Location and orientation of strain gauges on the machined invar beads, steel plate, and the invar reference plate. Dimensions of each plate and the interface between the steel and Invar. invar reference plate. Dimensions of each plate and the interface between the steel and Invar.

The WK-00-125AD-350 strain gauges were fully encapsulated K-alloy gauges that were chosen because of their minimal self-temperature compensation that matches the low thermal expansion coeﬃcient of Invar [35,36]. The reference plate measured 152.4 mm 152.4 mm 2.5 mm, and the × × CTE of the plate met the American Society for Testing and Materials (ASTM) F1684 speciﬁcations [3]. M-Bond 610, a strain gauge adhesive with a maximum operating temperature of 260 ◦C, was used to apply the strain gauges to the steel, Invar, and the reference plate to ensure that the oven temperature did not exceed the maximum operating temperature of the adhesive [37]. K-type thermocouples were used to measure the temperature of each plate. A 1/16 in. hole was drilled in the base of the plates, and thermocouples were inserted into the holes. A National Instruments 9211 temperature input module was used to record temperature data from the thermocouples, and the strain gauge measurements were recorded using a National Instruments 9237 bridge completion module. Each module was inserted into a National Instruments CompactRIO backplane, and a LabVIEW program was used to collect the data. The wiring diagram for each module is shown in Figure6. Materials 2020, 13, 5683 9 of 18 Materials 2020, 13, x 9 of 19

Figure 6. 6. NINI 9237 9237 strain strain gauge gauge module module and NI and 9211 NI thermocouple 9211 thermocouple module wiring module diagram. wiring diagram.

TheThe assemblyassembly was was placed placed inside inside a DX402C a DX402C oven oven (Yamato (Yamato Scientific Scientific Co., Ltd., Co., Santa Ltd., Clara, Santa CA, Clara, CA, USA) USA) (shown in Figure 4b), and ramped from 40 °C to 160 °C. The temperature of the oven was (shown in Figure4b), and ramped from 40 ◦C to 160 ◦C. The temperature of the oven was ramped at 20 ◦C ramped at 20 °C increments, and strain gauge data were taken at each temperature increment when increments, and strain gauge data were taken at each temperature increment when the thermocouples the thermocouples indicated a homogeneous temperature between the plates. After collecting the indicateddata, the CTE a homogeneous of the Invar surface temperature was determined between by: the plates. After collecting the data, the CTE of the Invar surface was determined by: = εsur f ace εre f , (2) α = ∆ − + α , (2) sur f ace ∆T re f where is the strain on the machined Invar bead surface, is the strain on the Invar wherereferenceεsur plate, f ace is and the strain is the on manufacturer’s the machined specification Invar bead fo surface,r the CTEεre of f theis the Invar strain reference on the plate Invar reference plate,[34]. The and measurementαre f is the acquired manufacturer’s from the specificationNI 9237 is the foramplitude the CTE ratio of of the the Invaroutput reference voltage plate [34]. The(measured measurement over the acquiredWheatstone from bridge) the and NI the 9237 excita is thetionamplitude voltage. The ratio voltage of ratio theoutput is converted voltage to (measured overa strain the differential Wheatstone with bridge) the following: and the excitation voltage. The voltage ratio is converted to a strain differential with the following: 2 = . (3) 2Vout εsur f ace εre f = . (3) As an additional test, the experiment was repeated− to measureVex the CTE of the Invar reference plateAs and an a additionallow-carbon test,steel theplate. experiment Two half-bridge was repeatedstrain gauge to circ measureuits were the used CTE to of measure the Invar these reference plate CTEs. The first half-bridge had one gauge mounted on the low-carbon steel plate and the other and a low-carbon steel plate. Two half-bridge strain gauge circuits were used to measure these CTEs. mounted on the Invar reference plate. The other half-bridge had both gauges mounted on the Invar Thereference first half-bridgeplate. This additional had one test gauge was mountedperformed onto confirm the low-carbon that the CTE steel of platethe reference and the plate other mounted onwas the within Invar the reference expected plate.values Theof 1.2 other and 2.7 half-bridge μm/m-K, and had the both A36 gauges steel plate mounted with a known on the CTE Invar reference plate.was used This as additional an experimental test was control performed [38]. to confirm that the CTE of the reference plate was within the expected values of 1.2 and 2.7 µm/m-K, and the A36 steel plate with a known CTE was used as an experimental3. Results and controlDiscussion [38 ]. 3.1. Thermomechanical Analysis and Energy Dispersive X-ray Spectroscopy on Different Samples 3. Results and Discussion The CTEs for samples in each section of the specimen (refer to Figure 3) are provided in Table 2. 3.1.The Thermomechanical average CTE of each Analysis section andis included Energy to Dispersive demonstrate X-ray how Spectroscopy the various thicknesses on Different of Samples Invar and the sample locations affected the overall thermal expansion of the section. The CTEs for samples in each section of the specimen (refer to Figure3) are provided in Table2. The averageTable 2. Average CTE of and each sample section coefficient is included of thermal to ex demonstratepansion (CTE) values how from the variousthermomechanical thicknesses of Invar and the sampleanalysis. locations affected the overall thermal expansion of the section.

Materials 2020, 13, 5683 10 of 18

MaterialsTable 2020 2., Average13, x and sample coeﬃcient of thermal expansion (CTE) values from thermomechanical10 of 19 analysis. CTE (μm/m-K) Section X1 X2 Y1CTE (µm/m-K) Y2 Z1 Z2 Average Section A36 11.92 X112.21 X2 Y112.96 Y212.67 Z1 11.34 Z2 Average12.53 12.27 WA36 A3613.06 11.9212.24 12.21 12.9612.44 12.6712.56 11.34 12.5313.95 12.2711.90 12.69 M93A36 WA3612.14 13.0613.14 12.24 12.4412.82 12.5612.08 13.95 11.9012.81 12.6915.14 13.02 M93 M93A36 10.40 12.14 9.07 13.14 12.82 9.03 12.08 9.08 12.81 15.146.46 13.022.09 7.69 M93 10.40 9.07 9.03 9.08 6.46 2.09 7.69 The A36 samples taken from the steel region of the baseplate had thermal expansion coefficients closeThe to the A36 value samples of 12.0 taken μm/m-K from the expected steel region for A36 of thesteel baseplate [12]. The had measured thermal expansion expansion data coe ﬃindicatecients closethat tothe the steel value samples of 12.0 µhadm/m-K isotropic expected thermal for A36 ch steelaracteristics [12]. The with measured an average expansion CTE data of indicate12.27 ± that0.58 theμm/m-K, steel samples (standard had deviation isotropic thermalis denoted characteristics by ±). The withA36 anCTE average values CTE are ofused 12.27 as a0.58 referenceµm/m-K, to ± (standardcompare deviationthe CTEs ismeasured denoted in by the). each The A36of the CTE Inva valuesr sections. are used The as WA36 a reference steel tosamples, compare taken the CTEs from ± measuredunder the inone-layer the each Invar of the bead, Invar behaved sections. similarly The WA36 to steelthe A36 samples, samples taken and from had under an average the one-layer CTE of Invar12.69 bead,± 0.73 behaved μm/m-K. similarly The CTEs to the for A36 the samplesWA36 samples and had are an averageplotted CTEagainst of 12.69the CTE0.73 ofµ them/m-K. A36 ± Thesamples CTEs in for Figure the WA36 7. samples are plotted against the CTE of the A36 samples in Figure7.

FigureFigure 7.7.CTE CTE comparisoncomparison betweenbetween thethe samplessamples takentaken fromfrom thethe toptop surfacesurface ofof thethe steelsteeland andsamples samples takentaken fromfrom 1.95 1.95 mm mm under under the the top top of of the the invar invar bead. bead.

TheThe WA36WA36 samplessamples havehave aa largerlarger variationvariation betweenbetween thethe CTEsCTEs measuredmeasured inin eacheach directiondirection thanthan thosethose ofof thethe A36A36 samples—assamples—as indicatedindicated byby thethe largerlarger standardstandard deviation.deviation. WhenWhen thethe WA36WA36 samplessamples werewere removed removed from from the the baseplate, baseplate, the the cutting cutting depth depth was was measured measured with with respect respect to the to top the of top the of plate. the Hence,plate. Hence, the top the face top of face each of sample each sample is ﬂush is with flush the with top the face top of face the baseof the plate. base Althoughplate. Although the weld the penetrationweld penetration depth depth should should remain remain relatively relatively consistent, consistent, the penetration the penetration depth depth was assumed was assumed to have to minorhave minor ﬂuctuations fluctuations along thealong length the oflength each of bead. each Depending bead. Depending on where on the where sample the was sample taken was from taken the bead,from certainthe bead, samples certain may samples have containedmay have a contained greater percentage a greater ofpercentage Invar than of others. Invar Also,than others. since steel Also, is consideredsince steel is to considered have isotropic to have thermal isotropic expansion therma properties,l expansion the properties, disparity between the disparity CTE valuesbetween can CTE be values can be considered a function of the sample location along the bead rather than the sample direction. In many of the previous studies conducted on dissimilar metal joining, the authors stressed the importance of element dissolution in the heat affected zone (HAZ) during the welding process—noting that it had measurable effects on the thermal and mechanical properties of the

Materials 2020, 13, 5683 11 of 18 considered a function of the sample location along the bead rather than the sample direction. In many of the previous studies conducted on dissimilar metal joining, the authors stressed the importance of element dissolution in the heat affected zone (HAZ) during the welding process—noting that it had measurable effects on the thermal and mechanical properties of the resulting alloys. While the WA36YMaterials and 2020 WA36Z, 13, x samples were taken from under the single Invar bead (section 1 in11 Figureof 19 3), the WA36X samples were taken from under the three beads in section 2. This was necessary because resulting alloys. While the WA36Y and WA36Z samples were taken from under the single Invar the single bead was not wide enough to take samples in the X direction. However, printing three beads, bead (section 1 in Figure 3), the WA36X samples were taken from under the three beads in section 2. instead of one, leads to a larger HAZ as well as a greater chance for element diffusion in section 2 of This was necessary because the single bead was not wide enough to take samples in the X direction. theHowever, specimen. printing The WA36X1 three beads, sample instead was of alsoone, takenleads to from a larger the startHAZ ofas thewell bead as a greater where chance the initial for laser powerelement must diffusion be higher in section to begin 2 of melting the specimen. the material. The WA36X1 This sample often leads was also to ataken greater from weld the start penetration of depththe bead and awhere higher the temperature initial laser power near must the ends be high ofer the to beads.begin melting These the factors material. lead This to aoften greater leads chance of dissolutionto a greater atweld the penetration interface of depth the materialsand a higher and temperature lend a potential near the explanation ends of the for beads. thegreater These CTE seenfactors in the lead WA36X1 to a greater sample. chance The of mechanisms dissolution at by the which interface diffusion of the contributes materials and to anlend increased a potential thermal expansionexplanation is discussed for the greater later. CTE seen in the WA36X1 sample. The mechanisms by which diffusion contributesFigure8 shows to an increased the secondary thermal scanningexpansion electronis discussed image later. of the weld metal-base plate interface region andFigure the 8 EDS shows line the scan secondary signals alongscanning the electron interface. image The of key the constituting weld metal-base elements plate were interface Si, Mn, Fe, region and the EDS line scan signals along the interface. The key constituting elements were Si, Mn, and Ni, but the weight percent of Si and Mn was negligible (below 1%). Ni percentage also dropped Fe, and Ni, but the weight percent of Si and Mn was negligible (below 1%). Ni percentage also close to zero after the interface and Si, Mn and Ni signals overlapped each other. The Ni percentage dropped close to zero after the interface and Si, Mn and Ni signals overlapped each other. The Ni belowpercentage 36% (the below optimal 36% (the percentage optimal percentage for the lowest for the CTE lowest in Invar) CTE in resulted Invar) resulted in a significant in a significant increase in CTEincrease of the in WA36X1 CTE of the sample. WA36X1 sample.

(a) (b)

FigureFigure 8. 8.(a )(a Secondary) Secondary scanning scanning electron im imageage of of interface interface region region and and (b) energy-dispersive (b) energy-dispersive X-ray X-ray spectroscopyspectroscopy (EDS) (EDS) line line scan scan signalssignals along along the the interface interface for for the the WA36X1 WA36X1 sample. sample.

TheThe M93 M93/A36/A36 samples samples (taken (taken from from the the interface interfac ofe of the the Invar Invar and and steel) steel) demonstrated demonstrated a lineara linear thermal thermal expansion similar to that seen in the WA36 samples. No signs of an Invar-like expansion expansion similar to that seen in the WA36 samples. No signs of an Invar-like expansion curve are seen in curve are seen in the first two sections of the samples. The average CTE of the M93/A36 samples is the first two sections of the samples. The average CTE of the M93/A36 samples is 13.02 1.12 µm/m-K. 13.02 ± 1.12 μm/m-K. A comparison of M93/A36 and A36 sample CTEs is shown in Figure ±9. A comparison of M93/A36 and A36 sample CTEs is shown in Figure9. Four out of six of the Invar sample CTEs in the M93/A36 section are larger than the sample CTEs from the steel region. The effects of dissolution are most noticeable in the M93/A36 section where samples were taken from the fusion zone of the weld. Like the WA36X1 sample that was taken from the end of the three beads, the M93A36Z2 sample on the opposite side of the beads shows a significant increase in thermal expansion. The samples taken closer to the center of the bead exhibit a smaller thermal expansion while those taken from the end regions show a sharp increase in the CTE. The samples in the Z direction were also half of the volume of the samples taken from the X and Y direction to ensure an equal area of material was taken from the fusion zone. Therefore, any increase in CTE that is caused by the diffusion process is likely to have a greater effect on the expansion of the

Materials 2020, 13, 5683 12 of 18 samples with a smaller volume. The average expansions of the WA36 and M93A36 sections indicate that the element diﬀusion had the greatest impact on increasing the CTE closer to the fusion zone of the weld.Materials 2020, 13, x 12 of 19

FigureFigure 9. CTE 9. CTE comparison comparison between between the the samples samples takentaken from the the top top surface surface of of the the steel steel and and samples samples takentaken from from the interface the interface between between the steelthe steel and and the invarthe invar bead bead (0.45 (0.45 mm mm below below the the top top of the of the Invar Invar bead). bead). Figure 10 presents the secondary electron micrograph of the weld interface region along with the EDS lineFour scan capturedout of six fromof the the Invar M93A36Z2 sample CTEs sample. in the It canM93/A36 be observed section thatare thelarger Ni than percentage the sample for this CTEs from the steel region. The effects of dissolution are most noticeable in the M93/A36 section sample is even lower (average weight percentage 28%) with a slightly higher Fe percentage compared where samples were taken from the fusion zone of the weld. Like the WA36X1 sample that was to that of the WA36X1 sample due to dissolution of iron during the fabrication process. The low Ni Materialstaken 2020from, 13 ,the x end of the three beads, the M93A36Z2 sample on the opposite side of the13 ofbeads 19 percentageshows a resulted significant in aincrease signiﬁcant in thermal increase expansion. in the CTE The forsamples the M93A36Z2 taken closer sample. to the center of the bead exhibit a smaller thermal expansion while those taken from the end regions show a sharp increase in the CTE. The samples in the Z direction were also half of the volume of the samples taken from the X and Y direction to ensure an equal area of material was taken from the fusion zone. Therefore, any increase in CTE that is caused by the diffusion process is likely to have a greater effect on the expansion of the samples with a smaller volume. The average expansions of the WA36 and M93A36 sections indicate that the element diffusion had the greatest impact on increasing the CTE closer to the fusion zone of the weld. Figure 10 presents the secondary electron micrograph of the weld interface region along with the EDS line scan captured from the M93A36Z2 sample. It can be observed that the Ni percentage for this sample is even lower (average weight percentage 28%) with a slightly higher Fe percentage compared to that of the WA36X1 sample due to dissolution of iron during the fabrication process. The low Ni percentage resulted in a significant increase in the CTE for the M93A36Z2 sample.

(a) (b)

FigureFigure 10. 10.(a ()a Secondary) Secondary scanningscanning electron electron im imageage of ofinterface interface region region and and(b) EDS (b) EDSline scan line signals scan signals along the interface for the M93A36Z2 sample. along the interface for the M93A36Z2 sample. The CTEs measured from the two-layer Invar section are much lower than those measured in the previous sections. The average M93 sample CTE is 7.69 ± 3.02μm/m-K. A comparison of the CTEs for samples taken from the M93 and A36 sections is provided in Figure 11.

Figure 11. CTE comparison between the samples taken from the top surface of the steel and samples taken from 0.13 mm below the top of the Invar bead.

The M93 section was the only section where all the sample CTEs remained below the CTE of the A36 steel samples. Since these samples were taken from above the weld interface, they were less likely to experience the effects of diffusion observed in the other sections of the specimen. The CTEs of samples generally decreased from one end of the bead to the other. As observed from the previous section, the M93X1 sample with the highest CTE was taken from the end of the bead. Conversely, the

Materials 2020, 13, x 13 of 19

(a) (b)

MaterialsFigure2020, 1310., 5683(a) Secondary scanning electron image of interface region and (b) EDS line scan signals13 of 18 along the interface for the M93A36Z2 sample.

TheThe CTEsCTEs measured measured from from the the two-layer two-layer Invar Invar section section are are much much lower lower than than those those measured measured in the in previousthe previous sections. sections. The averageThe average M93 M93 sample sample CTE CTE is 7.69 is 7.693.02 ± 3.02µm/μm-K.m/m-K. A comparison A comparison of the of CTEsthe CTEs for ± samplesfor samples taken taken from from the M93the M93 and and A36 A36 sections sections is provided is provided in Figure in Figure 11. 11.

FigureFigure 11. 11.CTE CTE comparisoncomparison betweenbetween thethe samplessamples takentaken fromfromthe thetop topsurface surfaceof ofthe thesteel steeland andsamples samples takentaken from from 0.13 0.13 mm mm below below the the top top of of the the Invar Invar bead. bead.

TheThe M93M93 sectionsection waswas thethe onlyonly sectionsection wherewhere allall thethesample sampleCTEs CTEsremained remained below belowthe theCTE CTEof ofthe the A36A36 steel steel samples. samples. Since Since these these samples samples were were taken taken from from above above the weld the interface, weld interface, they were they less were likely less to experiencelikely to experience the eﬀects the of dieffectsﬀusion of observed diffusion in observed the other in sections the other of thesections specimen. of the The specimen. CTEs of The samples CTEs generallyof samples decreased generally from decreased one end from of one the beadend of to the the bead other. to the As other. observed As observed from the from previous the previous section, thesection, M93X1 the sample M93X1 with sample the with highest the CTE highest was CTE taken wa froms taken the from end ofthe the end bead. of theConversely, bead. Conversely, the M93Z2 the sample had the lowest CTE and was also taken from the end of the bead. The sample with the lowest

CTE was never located on the end of a bead in any of the previous sections. Since this section had two layers with three beads per layer, the broader distribution of sample CTEs is likely an effect of the heat exposure patterns during the welding process. Each sample contained enough Invar to maintain a CTE lower than that of steel, but the section of the beads closer to the X-direction samples was likely exposed to more heat than the section of the bead near the Z-direction samples. The diffusion that resulted from the increased heat exposure altered the thermal expansion properties and increased the CTE of the samples in a similar manner to that seen in the M93A36 section. However, the volume of the samples that was affected by the diffusion was not large enough to increase the CTE past that of steel. Figure 12a,b shows secondary image and EDS line scan signals along the interface for M93Z2 and M93X2 samples. It can be observed that the M93Z2 sample contained comparatively higher Ni percentage (~average 33%) due to the decreased dissolution in this section of the weld bead. Similar to previous cases, M93X2 showed a smaller Ni percentage than other samples in this section which resulted in an increase in CTE for that sample. Table3 presents the average Ni weight percentage and resulting CTE of those specific samples. It can be observed that, increasing Ni percentage closer to the optimal value (36%) helps to maintain a lower CTE value in those samples. Materials 2020, 13, x 14 of 19

M93Z2 sample had the lowest CTE and was also taken from the end of the bead. The sample with the lowest CTE was never located on the end of a bead in any of the previous sections. Since this section had two layers with three beads per layer, the broader distribution of sample CTEs is likely an effect of the heat exposure patterns during the welding process. Each sample contained enough Invar to maintain a CTE lower than that of steel, but the section of the beads closer to the X-direction samples was likely exposed to more heat than the section of the bead near the Z-direction samples. The diffusion that resulted from the increased heat exposure altered the thermal expansion properties and increased the CTE of the samples in a similar manner to that seen in the M93A36 section. However, the volume of the samples that was affected by the diffusion was not large enough to increase the CTE past that of steel. Figure 12a,b shows secondary image and EDS line scan signals along the interface for M93Z2 and M93X2 samples. It can be observed that the M93Z2 sample contained comparatively higher Ni percentage (~average 33%) due to the decreased dissolution in this section of the weld bead. Similar to previous cases, M93X2 showed a smaller Ni percentage than other samples in this section which resulted in an increase in CTE for that sample. Table 3 presents the average Ni weight percentage Materialsand2020 resulting, 13, 5683 CTE of those specific samples. It can be observed that, increasing Ni percentage closer14 of 18 to the optimal value (36%) helps to maintain a lower CTE value in those samples.

(a) (b)

FigureFigure 12. Secondary 12. Secondary scanning scanning electron electron image image of interface of interf regionace region and EDSand lineEDS scanline signalsscan signals of (a )of M93Z2 (a) and (b)M93Z2 M93X1 and samples. (b) M93X1 samples.

TableTable 3. CTE 3. CTE versus versus weight weight percentage percentage ofof nickel.nickel.

SampleSample Percent Percent Ni Ni CTE CTE ((µμmm/m-K)/m-K) M93X1 30% 10.38 M93X1 30% 10.38 M93Z2 33% 2.11 M93Z2 33% 2.11 WA36X1WA36X1 29%29% 13.59 13.59 M93M93/A36Z2/A36Z2 28%28% 15.59 15.59

In each of the previous sections, element diffusion between the welded alloys was said to be the Inmechanism each of the responsible previous for sections, the deviations element and diff incrusioneases between in sample the CTEs. welded The alloys thermal was expansion said to be the mechanism responsible for the deviations and increases in sample CTEs. The thermal expansion coefficient of Invar is minimized with a weight percentage of 36% nickel [2]. A slight change in the composition can significantly affect the thermal expansion properties of the alloy. This effect is most noticeable with a decreased nickel percentage where (depending on the temperature that the alloy is studied) a loss of only a couple of percent of nickel can increase the alloy’s CTE far past that of steel [2]. Therefore, since the optimal weight percentage of nickel was altered by diffusion during the welding process, certain sections of the specimen had samples that demonstrated CTEs above those measured from the steel. The effects of the altered material composition were most prevalent near the fusion zone and were not as noticeable on samples taken farther away from the fusion zone. EPMA analysis was performed to provide further explanations on how the composition and the CTE of the samples was affected by diffusion.

coefficient of Invar is minimized with a weight percentage of 36% nickel [2]. A slight change in the composition can significantly affect the thermal expansion properties of the alloy. This effect is most noticeable with a decreased nickel percentage where (depending on the temperature that the alloy is studied) a loss of only a couple of percent of nickel can increase the alloy’s CTE far past that of steel [2]. Therefore, since the optimal weight percentage of nickel was altered by diffusion during the welding process, certain sections of the specimen had samples that demonstrated CTEs above those measured from the steel. The effects of the altered material composition were most prevalent near the fusion zone and were not as noticeable on samples taken farther away from the fusion zone. EPMA analysis was performed to provide further explanations on how the composition and the CTE of the samples was affected by diffusion.

3.2. Chemical Composition Study Using Electron Probe Micro Analyzer (EPMA) Although energy-dispersive X-ray spectroscopy could reveal the key mechanism acting behind observed differences in CTE of different samples, EPMA analysis was conducted to validate that trend of elemental dissolution during the AM process. Back-scattered electron (BSE) images along Materialswith2020 corresponding, 13, 5683 EPMA scan signals are presented in Figure 13. Two samples (M93A36Z2 and15 of 18 M93Z2) with same orientation (Z direction) but from two different sections (section 2 and 3) were selected for this study. Note that these two samples showed a significant difference in their CTE showsnumbers. that the M93A36Z2Figure 13 shows sample that had the comparativelyM93A36Z2 sample lower had Ni comparatively and higher Fe lower compared Ni and to higher that ofFe the M93Z2compared sample. to As that a result, of the theM93Z2 gap sample. between As Ni a result, and Fe the signals gap between in the plotNi and is widerFe signals for thein the M93A36Z2 plot is samplewider as compared for the M93A36Z2 to that ofsample the M93Z2 as compared sample to that (refer of the to EPMAM93Z2 signalssample (refer in the to plotsEPMA of signals Figure in 13 ). Higherthe dissolution plots of Figure of Fe 13). inthe Higher M93A36Z2 dissolution sample of Fe resulted in the M93A36Z2 in an increase sample in CTEresulted and in vice-versa an increase for inthe M93Z2CTE sample. and vice-versa Also, the for Ni the percentage M93Z2 sample. started Also, dropping the Ni andpercentage the Fe percentagestarted dropping started and increasing the Fe percentage started increasing from around 300 μm before the actual interface region for the from around 300 µm before the actual interface region for the M93A36Z2 sample. On the other hand, M93A36Z2 sample. On the other hand, both Ni and Fe percentages were more stable and uniform both Ni and Fe percentages were more stable and uniform until very close to the interface region for until very close to the interface region for the M93Z2 sample. This also indicates that the dissolution the M93Z2happened sample. in the This M93A36Z2 also indicates sample thatmore the severely dissolution as compared happened to the in M93Z2 the M93A36Z2 sample resulting sample in more a severelysignificantly as compared higher to CTE the M93Z2for that samplesample. resulting in a signiﬁcantly higher CTE for that sample.

(a) (b)

Figure 13. Back-scattered electron image and electron probe micro analyzer (EPMA) line scan signals of (a) M93A36Z2 and (b) M93Z2 samples. 3.3. Surface Expansion The CTE of the machined Invar surface is dependent on the expansion of both the Invar and the steel baseplate underneath the Invar beads. Since the CTE of steel is greater than that of Invar, it was expected that the measured CTE would be between the known values for A36 steel and Invar M93. The CTEs measured from the machined Invar surface and the steel baseplate are shown in Table4. The strain gauge labels indicate the positioning of the gauge and correlate with the locations pictured in Figure5.

Table 4. Strain and CTE of machined Invar beads and steel.

Strain Gauge Test Material Change in Strain from 40 ◦C–160 ◦C(µm/m) CTE (µm/m-K) SG1 Invar 529.22 5.62 SG2 Invar 759.17 7.54 SG3 Invar 417.51 4.68 SG4 Invar 866.02 8.43 SG5 A36 Steel 1227.21 11.44 SG6 A36 Steel 1213.55 11.33 Materials 2020, 13, 5683 16 of 18

The average CTE on the surface of the machined Invar was 6.57 1.72 µm/m-K. Variations between ± the measured CTEs are distinguishable between the orientations of the gauges. The two gauges placed along the long dimension of the test specimen measured a larger thermal expansion than the gauges placed along the short direction. The CTEs measured from the steel baseplate maintain consistent magnitudes in both directions. SG1 and SG3 are oriented along the width of the specimen (refer to Figure5) and have equal lengths of Invar and steel along this orientation. SG2 and SG4 are oriented along the length of the specimen where there is a larger length of steel than Invar. The larger portion of steel along the length of the specimen contributes to the higher CTE measured by gauges oriented in this direction. Overall, the thin layer of machined Invar decreased the surface expansion of the part by about 42%. Additional experimentation was performed to measure the CTE of the reference plate and the low-carbon steel plate to further conﬁrm the results of the strain gauge measurements. The measured CTE values for the reference plate and the low-carbon steel are shown in Table5.

Table 5. Strain and CTE of reference specimen and steel plate.

Strain Gauge Test Material Change in Strain from 40–160 ◦C(µm/m) CTE (µm/m-K) SG7 Reference Invar 73.82 1.82 SG8 Low-carbon Steel 1283.04 12.00

The 1.82 µm/m-K CTE measured for the reference plate is between values 1.2 µm/m-K and 2.7 µm/m-K specified by the manufacturer and ASTM standard F1684-06 [3]. Additionally, the measured CTE for low-carbon steel plate is within the known range for steel of 11 µm/m-K to 12.5 µm/m-K [12].

4. Conclusions This study utilized a ﬁber-laser welder to print Invar beads on a steel baseplate. The steel, Invar, and the interface between these two materials were analyzed to investigate the thermal expansion characteristics of the printed specimen. An additional experiment was performed to measure the surface expansion of a machined specimen composed of the two materials. The key ﬁndings of this study are as follows:

Steel material under printed Invar maintains thermal expansion properties comparable to those of • normal steel. There were minor deviations from the expected CTEs that were seen in samples taken from the ends of weld beads. The expected cause of these dissimilarities is the increased elemental diffusion associated with longer durations of heat exposure. The fusion zone between the steel and the Invar had the highest average CTE. The effects of • diffusion were most noticeable in the fusion zone region, and the composition of the region was altered to produce undesirable thermal expansion characteristics. Variation in weight percent of nickel due to dissolution showed direct correlation with the resulting CTE of different samples. The samples taken from the Invar beads showed a decreased thermal expansion compared to the • other sections. The diffusion was noticeable in this region, but it did not increase the sample CTEs past that of steel as seen in the fusion zone region. The effects of diffusion in the Invar region are seen through the decreases in CTE along the length of the region. The sections of this region that are were exposed to heat for a longer time had more diffusion and a larger CTE. Two layers of Invar were required to reduce to reduce the CTE of the specimen. The surface expansion of the machined Invar beads was between the expected expansion values • for the Invar beads and the substrate. The expansion on the surface of the thin layer of Invar was 42% less than the expansion on the surface of the steel. Multi-material tooling with Invar is a viable option to decrease the CTE for tools. Future work • should focus on determining the printed thickness of Invar needed to optimize the decrease in thermal expansion. Materials 2020, 13, 5683 17 of 18

Author Contributions: Conceptualization, A.N., M.W.N., S.S.B.; methodology, A.A., A.N., S.R., C.M.; validation, A.A., S.R.; formal analysis, A.A., S.R.; investigation, A.A., S.R.; data curation, A.A., S.R., C.M.; writing—original draft preparation, A.A.; writing—review and editing, A.A., S.R.; supervision, A.N., M.W.N.; project administration, A.N., M.W.N., S.S.B.; funding acquisition, A.N., S.S.B. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by Office of Energy Efficiency and Renewable Energy | Advanced Manufacturing Office | United States Dept. of Energy, under Contract DE-AC05-00OR22725 with UT-Battelle, LLC. Acknowledgments: The authors would like to acknowledge the support of collaborating partners Lincoln Electric, Wolf Robotics on this project and access to wire-arc additive manufacturing setup at the Manufacturing Demonstration Facility in Oak Ridge National Laboratory. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Van Schilfgaarde, M.; Abrikosov, I.A.; Johansson, B. Origin of the Invar effect in iron-nickel alloys. Nat. Cell Biol. 1999, 400, 46–49. [CrossRef] 2. Guillaume, C.E. Invar and Its Applications. Nat. Cell Biol. 1904, 71, 134–139. [CrossRef] 3. ASTM F1684-06: Standard Specification for Iron-Nickel and Iron-Nickel-Cobalt Alloys for Low Thermal Expansion Applications; ASTM International: West Conshohocken, PA, USA, 2016. 4. Qiu, C.; Adkins, N.J.; Attallah, M. Selective laser melting of Invar 36: Microstructure and properties. Acta Mater. 2016, 103, 382–395. [CrossRef] 5. Harner, L.L.; Frantz, E.L. Thermally Stable Super Invar and Its Named Article. U.S. Patent Application No. US07/132,702, 1 August 1989. 6. Shiga, M. Invar Alloys. Materials Science and Technology 2006, 1, 340–348. [CrossRef] 7. Wohlfarth, E.P. The Invar Problem. IEEE Trans. Magn. 1975, 11, 1638–1644. [CrossRef] 8. Legendre, B.; Sghaier, M. Curie temperature of nickel. J. Therm. Anal. Calorim. 2011, 105, 141–143. [CrossRef] 9. Pipes and Tubes for Lng Transfer Lines: Welding Processes. Available online: https://www.spacematdb.com/ spacemat/manudatasheets/Invar%20M93.pdf (accessed on 27 August 2019). 10. Potter, K.D.; Campbell, M.; Langer, C.; Wisnom, M.R. The Generation of Geometrical Deformations Due to Tool/Part Interaction in the Manufacture of Composite Components. Compos. Part A Appl. Sci. Manuf. 2005, 36, 301–308. [CrossRef] 11. Abberger, S.L. Advanced Composite Molds—A New Use for Invar. Presented at The Invar Effect: A Centennial Symposium, Cincinnati, OH, USA, 7–8 October 1996. 12. Cverna, F. ASM Ready Reference: Thermal Properties of Metals; ASM International: Russell Township, OH, USA, 2002. 13. Wimpenny, D.I.; Gibbons, G.J. Metal spray Invar tooling for composites. Aircr. Eng. Aerosp. Technol. 2000, 72, 430–439. [CrossRef] 14. Stewart, R. New mould technologies and tooling materials promise advances for composites. Reinf. Plast. 2010, 54, 30–36. [CrossRef] 15. Composites 101: Tooling. Available online: https://www.compositesworld.com/articles/tooling (accessed on 28 August 2019). 16. Mahamood, R.M.; Akinlabi, E.T. Laser metal deposition of functionally graded Ti6Al4V/TiC. Mater. Des. 2015, 84, 402–410. [CrossRef] 17. Abberger, S.L. Invar Truths and Rumors 2.0. SAMPE J. 2015, 51, 7–11. 18. Gibb, S. An Introduction to Invar. College of Optical Sciences, University of Arizona, Tucson. 2006. Available online: https://wp.optics.arizona.edu/optomech/wp-content/uploads/sites/53/2016/10/GibbPresentation.pdf (accessed on 28 August 2019). 19. Wong, K.V.; Hernandez, A. A Review of Additive Manufacturing. ISRN Mech. Eng. 2012, 2012, 1–10. [CrossRef] 20. Yigit, I.E.; Khan, S.A.; Lazoglu, I. Robotic Additive Manufacturing of Tooling for Composite Structures. Presented at The 18th International Conference on Machine Design and Production (UMTIK), Eski¸sehir, Turkey, 3–6 July 2018. 21. Harrison, N.J.; Todd, I.; Mumtaz, K. Thermal expansion coefficients in Invar processed by selective laser melting. J. Mater. Sci. 2017, 52, 10517–10525. [CrossRef][PubMed] Materials 2020, 13, 5683 18 of 18

22. Wei, K.; Yang, Q.; Ling, B.; Yang, X.; Xie, H.; Qu, Z.; Daining, F. Mechanical Properties of Invar 36 Alloy Additively Manufactured by Selective Laser Melting. Mater. Sci. Eng. A 2020, 772, 138799. [CrossRef] 23. Kouraytem, N.; Li, X.; Tan, W.; Kappes, B.; Spear, A. Modeling process-structure-property relationships in metal additive manufacturing: A review on physics-driven versus data-driven approaches. J. Phys. Mater. 2020, 23, 1917–1928. [CrossRef] 24. Simunovic, S.; Nycz, A.; Noakes, M.; Chin, C.; Oancea, V. Metal Big Area Additive Manufacturing: Process Modeling and Validation. Paper presented at the NAFEMS World Congress, Stockholm, Sweden, 11–13 June 2017. 25. Ding, D.; Shen, C.; Pan, Z.; Cuiuri, D.; Li, H.; Larkin, N.; Van Duin, S. Towards an automated robotic arc-welding-based additive manufacturing system from CAD to finished part. Comput. Des. 2016, 73, 66–75. [CrossRef] 26. Carroll, B.E.; Otis, R.A.; Borgonia, J.P.; Suh, J.-O.; Dillon, R.P.; Shapiro, A.A.; Hofmann, D.C.; Liu, Z.-K.; Beese, A.M. Functionally graded material of 304L stainless steel and inconel 625 fabricated by directed energy deposition: Characterization and thermodynamic modeling. Acta Mater. 2016, 108, 46–54. [CrossRef] 27. Bobbio, L.D.; Otis, R.A.; Borgonia, J.P.; Dillon, R.P.; Shapiro, A.A.; Liu, Z.-K.; Beese, A.M. Additive manufacturing of a functionally graded material from Ti-6Al-4V to Invar: Experimental characterization and thermodynamic calculations. Acta Mater. 2017, 127, 133–142. [CrossRef] 28. Abe, T.; Sasahara, H. Dissimilar metal deposition with a stainless steel and nickel-based alloy using wire and arc-based additive manufacturing. Precis. Eng. 2016, 45, 387–395. [CrossRef] 29. Ding, D.; Pan, Z.; Cuiuri, D.; Li, H. Wire-feed additive manufacturing of metal components: Technologies, developments and future interests. Int. J. Adv. Manuf. Technol. 2015, 81, 465–481. [CrossRef] 30. Roy, S.; Shassere, B.; Yoder, J.; Nycz, A.; Noakes, M.; Narayanan, B.K.; Meyer, L.; Paul, J.; Sridharan, N. Mitigating Scatter in Mechanical Properties in Aisi 410 Fabricated Via Arc-Based Additive Manufacturing Process. Materials 2020, 13, 4855. [CrossRef][PubMed] 31. Hinojos, A.; Mireles, J.; Reichardt, A.; Frigola, P.; Hosemann, P.; Murr, L.E.; Wicker, R.B. Joining of Inconel 718 and 316 Stainless Steel using electron beam melting additive manufacturing technology. Mater. Des. 2016, 94, 17–27. [CrossRef] 32. Xu, P.; Huang, C.; He, J. Study on temperature field for invar alloy during TIG welding. Int. J. Adv. Manuf. Technol. 2009, 42, 242–249. [CrossRef] 33. Li, G.; Gao, M.; Chen, C.; Zhang, C.; Zeng, X.Y. Characterisation comparison of laser and laser-arc hybrid welding of Invar 36 alloy. Sci. Technol. Weld. Join. 2013, 19, 30–37. [CrossRef] 34. Fox, J.N. Measurement of thermal expansion coefficients using a strain gauge. Am. J. Phys. 1990, 58, 875–877. [CrossRef] 35. General Purpose Strain Gages—Linear Pattern; Micro-Measurements, VPG: Malvern, PA, USA, 2016. 36. Stress Analysis Strain Gages.; Micro-Measurements, VPG: Malvern, PA, USA, 2015. 37. Instruction Bulletin B-130: Strain Gage Installations with M-Bond 43-B, 600, and 610 Adhesive Systems; Micro-Measurements, VPG: Malvern, USA, PA, 2016. 38. ASTM A108-18: Standard Specification for Steel Bar, Carbon and Alloy, Cold-Finished; ASTM International: West Conshohocken, PA, USA, 2018.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations.

© 2020 by the USA Government. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).