Accepted Manuscript Directional solidification of a Sn-0.2Ni solder alloy in water-cooled copper and steel molds: Related effects on the matrix micromorphology, nature of intermetallics and tensile properties Marcella G.C. Xavier, Clarissa B. Cruz, Rafael Kakitani, Bismarck L. Silva, Amauri Garcia, Noé Cheung, José E. Spinelli PII: S0925-8388(17)32337-X DOI: 10.1016/j.jallcom.2017.06.329 Reference: JALCOM 42398 To appear in: Journal of Alloys and Compounds Received Date: 20 March 2017 Revised Date: 28 June 2017 Accepted Date: 30 June 2017 Please cite this article as: M.G.C. Xavier, C.B. Cruz, R. Kakitani, B.L. Silva, A. Garcia, Noé. Cheung, José.E. Spinelli, Directional solidification of a Sn-0.2Ni solder alloy in water-cooled copper and steel molds: Related effects on the matrix micromorphology, nature of intermetallics and tensile properties, Journal of Alloys and Compounds (2017), doi: 10.1016/j.jallcom.2017.06.329. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Directional solidification of a Sn-0.2Ni solder alloy in water-cooled copper and steel molds: related effects on the matrix micromorphology, nature of intermetallics and tensile properties Marcella G. C. Xaviear, Clarissa B. Cruzb, Rafael Kakitanbi , Bismarck L. Silvac , Amauri Garciab , Noé Cheungb, José E. Spineall,i* a Department of Materials Engineering, Federal Unsiivtye rof São Carlos UFSCar, 13565-905 - São Carlos B, SraPz,il. b Department of Manufacturing and Materials Enginnege,r iUniversity of Campinas UNICAMP, 13083-860 - Canmapsi, SP, Brazil. c Department of Materials Engineering, Federal Urnsiivtye of Rio Grande do Norte-UFRN, 59078-970 - Na RtaNl,, Brazil. Abstract The present investigation is focused on, firstly, performing transient directional solidification experiments with a Sn-0.2wt.% Ni solder alloy using two different substrates as mold sheets separating the alloy casting from the cooling fluid: copper and low carbon steel. Secondly, the examination of the obtained microstructures is carried out highlighting not only the micromorphology aspects of the formed β-Sn phase but also the nature and the shape of the intermetallic compounds (IMCs) developed. The purpose of this research work is to verify the influences that different substrate materials may have on the alloy solidification kinetics, resultant microstructures and tensile properties of the Sn-0.2 wt.%Ni solder. The microstructure characteristics may be correlated with thermal solidification parameters such as the eutectic cooling rate and eutectic growth rate along with a qualitative evaluation of Fe and Cu dissolutions into the alloy. The results display that the dissolution of Cu into the Sn-Ni alloy provided the prevalent growth of the (Cu,Ni)6Sn5 fiber-like eutectic phase along the length of the casting. Other than, the Cu-containing Sn-Ni alloy allowed the growth of high-velocity β- Sn cells only for very high cooling rates, associated with positions closer to the bottom of the alloy casting. Farther positions are characterized by a complex growth of β-Sn dendrites. On the other hand, for the alloy solidified against the steel mold, a predominance of the non-equilibrium NiSn4 eutectic phase with plate-like shape has been identified by SEM/EDS and XRD. In this case, the predominant growth of β-Sn cells associated with the presence of the plates of the NiSn4 IMC allowed lower tensile strength and higher ductility to be attained. Keywords: Sn-Ni alloys, solders, directional solidificatiomn,ic rostructure, intermetallics, tensile propert i es. * Corresponding author: Tel: +55 16 3351-8548, F+a5x5: 16 3364-5404. E-mail address: spinelli@ufscar.br 1 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 1. Introduction With the move toward lead-free electronic produ acnts i,ncreasing number of manufacturers are preparing their machinery to adopt alternaatinvde sustainable lead-free solder alloys. This is driven by the legal restrictions on the usage aodf -lbeased solders. Based on that, a great deaflo ortfs e f has been placed on replacing the traditional Snso-Pldbe rs with other Sn-based alloys. Many studies have examined different alloys over the past y eparrosp,osing ternary and quaternary near-eutectic alloys based on Sn, Cu, Ag and/or Ni, which are nesxitvely used in the electronics industry. While alloys concerning the Sn-Cu [1-3] and Sn-Cu-Ag ][ 4s-y6stems have been comprehensively investigated and understood, there has beenr leitstlea rch emphasizing the solidification behavio r of Sn-Ni [7] and Sn-Cu-Ni [8-10] solder alloys. Furrt hinevestigations into the understanding of the mechanical behavior of these alloys, as well aisr rtheeliability, are ongoing. In the case of the Sn-Ni system, even though noilniberqi um NiSn phases are commonly reported in heat treated Sn-Ni couples, Belyakodv Gaonurlay [7] showed that the metastable N4 iSn also forms during the solidification of Sn-rich SNni -alloys. This study examined a series of Sn-Ni alloys containing 0-0.45 wt% Ni. It is also streds stheat the NiS4n phase may form as either a primary or a eutectic phase. Despite the general preva loefn Scne-NiSn4, eutectic regardless the solidification condition used, i.e., 115-120K/s; 1.7-2.6K/s; 0.52 -K0/s or 0.008-0.022K/s, a small fraction of Sn- Ni3Sn4 was observed to occur for hypereutectic componssit i(o0.3, 0.37, 0.39 and 0.4 wt.% Ni). Despite suggesting that the eutectic Sn-N4 ihSans easier growth kinetics than the eutectic S3nS-nN4,i the critical thermal solidification parameter gonvienrg the growth of the metastable phase has nont bee determined. The shape of the N4i Spnhase was recognized as plate-like whilst th3eS nN4 iparticles grow with a needle-like morphology. It is important to underline that there are no retepdo studies to date in the literature, regarding the effects of a variety of solidification condintiso neither on the morphology nor on the microstructure length-scale of thβe-S n phase in Sn-Ni alloys. This knowledge is qiumitpeo rtant since such attributes may govern the solder mechanicoaple prrties together with the IMCs forming in as- soldered Sn-Ni joints. A particular focus should g biveen on the understanding of the stability of the 2 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT solidification front during liquid-to-solid transrfmo ation, which could lead to the formationβ o-fS n cells, eutectics oβr -Sn dendrites as resultant morphologies. Another important reason to investigate Sn-Ni asl loreymains on the fact that Ni is a very common substrate in electronic packaging, whiche sta akdvantage in industrial applications due to the significant slower growth kinetics between Ni annd tShan between Cu and Sn. As lead-free solders become increasingly preva qleunets, tions have arisen about copper dissolution into lead-free alloys during wave-sorilndge process. The presence of copper in the molten alloys may induce sluggish behavior during soldge roinperations. This is caused by a buildup of a copper-tin intermetallic (CuSn), which is densearn t hthe molten lead-free solder. In the case of bS n-P solders, the copper-tin intermetallic floats annd bcae easily removed [11]. Other disadvantage is related to the rise of the melting temperature opkroevd by the increase in the alloy copper contesn t. A a result, the use of lead-free solder alloys maqyu ire more solder container maintenance due tor thei high copper dissolution rates. Another level of cceorn encompasses the need for running comparisons between different substrate materials holding S na-lNloiys and their effects on the final microstrurect u of the alloy. The soundness of these alloys ase sr omldaterials depends on the morphology ofβ t-hSen matrix and on the distribution/nature of phasesro suunrding it. In contrast to what has been devoted to resear Schn -oCfu alloys, microstructural evolutions and mechanical properties of the eutectic Sn-Cdu- flreeae solder have been extensively studied up to date [12,13]. Most of the investigations on thete ectuic Sn-Cu to date stated the growth of microstructures with either dendritic pattern ocro an figuration containing eutectic cells. For gro wth velocities lower than 0.35 mm/s, the prevalencteh eo fe utectic two-phase cells is noted, which asroe al called eutectic colonies [10,14]. In addition toa tt,h Felberbaum and collaborators [15] observed a transition from fully eutectic to cellular eutec twicith the increase in the growth velocity. These authors affirm that this change is a consequen cse gorfegation of impurity elements, particularly Pb. Among the minute additions of elements in Sn-Couy asl,l the Ni-containing compositions deserve attention. The microstructures of Ni moedi fai lloys with 500 and 1000ppm of Ni were observed to develop simultaneously the eutectic,N (Ci)6uSn5 phase and the N3Sin4 primary intermetallic for specimens solidified under faosot lcing conditions [16]. 3 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT A systematic study to understand the effects loidf isfiocation thermal parameters - such as eutectic growth rate (EV) and eutectic cooling rateṪ E() - on the microstructure evolution of the eutectic Sn-Ni alloy has not been performed soto f athr e best of our knowledge. There is a common sense that the solidification thermal parametenrstr ocol the final arrangement of the eutectic strurec tu of a certain alloy [17-21], and some reports in littheerature indicate the prevalence of N4i SIMnC under fast cooling conditions of Sn-Ni solders [T7h].erefore, it seems essential to expand such knowledge for Sn-Ni solders by examining the otlhikerl y effects in the final microstructure that m ay be induced by the solidification cooling rate. The present research work is focused on the chearirzaacttion of microstructure features along the length of directionally solidified (DS) Sn-0w.2t. % Ni alloy castings, with the use of two diffenrt e mold materials (substrates), i.e., Cu and steeel tsh seeparating the alloy casting from the coolliunigd f during solidification. Experimental interrelatioonfs cellular and primary dendritic spacingλsc, (λ1) and cooling rate Ṫ( E) are envisaged giving emphasis on the influen cteh eo ftype of substrate. The nature and morphology of the intermetallic particles aerete drmined for each tested condition, and the combined influences of thβe- Sn matrix length scale and of the type/distribnu toiof IMCs on the tensile mechanical properties of the Sn-0.2 wt. %Ni soilsd eerx amined. 2. Experimental procedure Directionally solidified (DS) specimens of the S.n2- 0wt.% Ni alloy were attained by using a water-cooled solidification setup, which allows pkieneg transient heat flow regime during the processing. Details of the whole directional sofilcidaition system, including the necessary equipm ents for running the experiments, can be seen in F igT.w 1o. different solidification experiments were run with this alloy. One of them by using the electrtiocl ycopper bottom-part mold and the other solidifying the alloy into the low-C (SAE 1020) estl emold. The surface of the bottom-part molds has been finished with a 1200 grit SiC abrasive paApenr i.n set image of the bottom-part mold coupled with the DS system can be observed in Fig. 1. 4 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Bottom-part mold 3mm-thick Fig. 1. Unidirectional solidification system fora ntrsient heat flow regime coupled with the neces sary devices. In both tested conditions performed in this reshe,a trhce same melt treatment was adopted, i.e., firstly, the molten alloy was homogenized in anu icntdion furnace during 60 minutes, being transfe rred to the mold inside the DS system to be naturalolyl ecdo. Secondly, a remelting procedure was run until the desired melt temperature was attained, i.em.,p eteratures of about 10% above the eutectic temperature. At this stage, the furnace windingsre w deisconnected and at the same time the water flow at the bottom of the container was initiaterodu tgh a water supply tube, initiating the cooling procedure. Several J-type thermocouples were einds elratterally along the length of the casting whereas their tips were placed in the center of ctohnetainer so that the cooling curves along solidification could be acquired. The temperatuimree-t records were stored to be evaluated suitably. The thermocouples were positioned at up to eighsti tipoons from the heat-extracting surface at the bottom of each DS casting. Post mortem longitud sineacltions of the solidified specimen containing the thermocouple tip were obtained in order to rdmeitnee the real position of each thermocouple. In the solidification setup, heat is directionaellxyt racted only through the bottom-part mold, which is subjected to continuous water-cooling ndgu rsiolidification. This promotes upward vertical directional growth of the alloy structures. Thei dsiofilcation system consists of the stainless-sstepelitl 5 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT mold with an internal diameter of 60 mm, a heigfh 1t 5o7 mm and a wall thickness of 5 mm. A coating made of alumina is applied into the lateral innyelirn cdrical mold surface with a view to providing a thermal insulating layer between molten alloy anodld m, thus minimizing radial heat losses. The following etching solution was used to revehael at lloy macrostructure on the longitudinal section of the casting: 2mL Hl;C 10g FeCl3 and 100mL H2O. A light etching procedure with the solution of 92% (vol.) CH3OH, 5% (vol.) HNO3 and 3% (vol.) HCl was applied during 5-10s so that microstructural characteristics regarding the molropghies of the β-Sn matrix could be revealed. In order to assess the 3D-morphology of the formed sI MinC both Sn-0.2 wt.% Ni solder alloy DS castings, a deep etching procedure was carriebdy o ust ing a solution of 5% NaOH and 3.5% orthonitrophenol in distilled H2O. So, the Sn-Ni samples were immersed into thuet isoonl heated at 60oC for 30 minutes. Microstructures were investiga utesidng light microscopy with a coupled optical image processing system Olympus, GX51 (Olympus JCaop.a, n) and using a Field Emission Gun (FEG) - Scanning Electron Microscope (SEM) Phil(iXpsL 30 FEG) coupled to an Energy Dispersive Spectroscope - EDS (Oxford Link ISIS 300). EDS isntvigeation was attractive since the manner that the formed phases have been revealed in the mriuccrotusrte befitted the application of this techniq ue. Either coarse or quite dense arrangements of eicu tIMecCt s were found which may assure that most of the X-ray signal captured by the EDS-detector rse tfoe rthe aimed phases. The X-ray diffraction (XRD) patterns have been oinbetad by a Siemens D5000 diffractometer with a 2-theta range from 20° to 90°, Cαu Kradiation and a wavelengtλh,, of 0.15406 nm. The triangle method was employed to measure both tihmea pryr dendritic arm spacingλ 1() and the cellular spacing λ(c) on transverse sections of the directionally sifoielid castings [22]. At least 40 measurements were performed for each selectedio pno asliot ng the length of the DS castings. The eutectic spacingsλ )( were measured on the transverse sections byn ttehrec ei pt method . The transverse specimens were prepared accord isnpge tcoifications of the ASTM Standard E 8M/04 and tested in an Instron 5500R machine atrta ain s rate of about 1 × 1-30 s-1. In order to ensure reproducibility of the tensile results, three spmeecni s were tested for each selected position an d the ultimate tensile strength, the yield tensile strtehn agnd elongation-to-fracture have been determfionre d specimens extracted from different positions altohneg l ength of the castings. 6 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 3. Results and discussion 3.1. Effect of mold material on growth rate/ cooling rate and solder/substrate heat transfer coefficient Fig. 2 shows the cooling curves corresponding eto t hthermocouples positioned along the length of the DS Sn-Ni alloy castings. These exmpenrital curves were used to determine the experimental evolutions of the solidification thearlm parameters, i.e. eutectic growth rateE) ( aVnd eutectic cooling rateṪ (E), during solidification of the Sn-0.2 wt.% Ni ayll ocastings. If positions closer to the bottom of the castineg caornsidered and if similar positions in the casting are considered for comparison purposecas,n i tb e noted that the slopes of the curves are quit different from each other, e.g., P=4mm for the ls mteoeld and P=5mm for the copper mold, as can be seen in Fig. 2. Thus, it is worth nothing that dites tphe use of a cooling fluid and 3mm thick mold sheets, the material of the mold is shown to ha svieg naificant effect upon the cooling conditions during solidification of the Sn-Ni solder. Fasteoro cling is associated with the very beginning of solidification into the copper mold. As a matte rf aocft, copper molds are well-known by their high thermal conductivity which may assist heat to baen strferred from the metal/mold interface. Another point to be realized refers to the higher abilift yc opper to be dissolved by Sn based alloys thatn th offered by steels [23]. The dissolution of coppeeer mss to increase the physicochemical affinity between the neighboring layers at the metal/motledr fiance, consequently increasing the thermal conductance at such interface. The cooling curves of Fig. 2 have been used toi dpero evxperimental plots of position (P), from the metal/mold interface, and the correspogn dtiimne (tE) of the eutectic front passing by each thermocouple, i.e., P=fE()t, thus permitting experimental fitting functionos bte obtained. A time- derivative of this fitting function was carried o suot that the eutectic growth rateE ()V could be calculated in the form of V=fE()t. By replacing Et=f(P) with tE inside the equation for EV, a resultant equation of the form EV=f(P) has been obtained. The eutectic cooling (rṪaEt)e w as also determined along the length of the castings, by considerineg t hthermal data recorded immediately after the passage of the eutectic front by each thermoco uTphle .values of VE andṪ E were plotted in Fig. 3 7 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT against the position P in both Sn-0.2 wt.% Ni a lDloSy castings so that the influence of the mold material could be noticed. The thermal gradienFti gin. 3c is obtained through a simple relation: G= Ṫ/V. 300 300 Sn-0.2 wt.% Ni Positions from the bottom of the casting: Positions from the bottom of the casting: 280 [Cu mold] 7mm 9mm 15mm 280 3 mm 8 mm 12 mm 17 mm 20mm 25mm 45mm 47 mm 58 mm 70 mm 260 60mm 75mm 260 T T E E240 240 220 220 200 200 180 180 160 160 140 140 120 120 100 100 80 80 Sn - 0.2wt.%Ni 60 60 [Steel mold] 40 40 0 100 200 300 0 100 200 300 Time (s) Time (s) (a) (b) Fig. 2. Experimental cooling curves registered ndgu ruinsteady-state directional solidification of the Sn-0.2wt.%Ni solder alloy in water-cooled (a) corp apned (b) steel molds.E T is the eutectic temperature, which is 231ºC. 1.8 Sn-0.2wt.%Ni [Cu] 1.6 V = 5.8 (P)-0.7 E 1.4 Sn-0.2wt.%Ni [Steel] V = 3.8 (P)-0.7 1.2 E 1.0 0.8 0.6 0.4 0.2 0.0 0 10 20 30 40 50 60 Position (mm) (a) 8 Temperature (°C) AC ECutectic growth rate - V (mm/s)E EPTED MA Temperature (°C)NUSCRIPT ACCEPTED MANUSCRIPT 32 30 Sn-0.2 wt.% Ni [Cu] 28 T = 390.3 (P)-1.4 26 E 24 Sn-0.2 wt.% Ni [Steel] 22 T = 21.9 (P)-0.8 E 20 18 16 14 12 10 8 6 4 2 0 0 10 20 30 40 50 60 Position (mm) (b) 24 Sn-0.2 wt.% Ni [Cu] 22 G = 67.8 (P)-0.66 20 E Sn-0.2 wt.% Ni [Steel] 18 G = 26.6 (P)-0.66 E 16 14 12 10 8 6 4 2 0 0 10 20 30 40 50 60 70 80 Position (mm) (c) Fig. 3. Experimental tendencies related to (a) cetiuct egrowth rate, (b) cooling rate and (c) thermal gradient for the solidified Sn-0.2wt.%Ni soldero ayl leither in copper or in steel substrates. The heat transfer efficiency between a molten a allnody a substrate is rendered by the solder/substrate interfacial heat transfer coeefnfitc,i hi, which is considered a key parameter controlling not only the solidification kinetics but also thees ur ltant microstructures [24-25]. Recently, experimental power relations oif tho time (hi = a.t -m) have been proposed for the Sn- 0.7wt.% Cu (-0, 500 and 1,000 ppm Ni) alloys solidified againsth b Cot-steel [16] and a Cu bottom-part molds [26]. The same ‘m’ exponent of 0.03 was determined foyr a allnoy examined and conditions. However, the multipliers ‘a’ for the copper substrate are abtohuret e times higher when compared with those de rived for the carbon steel substrate. Similar behaviosr b heaen observed (Figs. 4 and 5) in the simulated 9 ACCE Thermal gradient - G (°C/mm)E Cooling rate - T (°C/s)EPTED MANUSCRIPT ACCEPTED MANUSCRIPT thermal profiles for the determination oi fi nh the present work, applying the same procedusrec rdibeed in a previous study by some of the present aut[h2o7r]s. The thermal-physical properties, e. g. thel rma conductivity, density, specific and latent heatesd u fsor the simulations were taken from references [28,29] for tin, since the addition of only 0,2%. ,e i. 2000 ppm, of Ni into tin (about the limit of impurities specified in grade “A” tin) , may notf eacf t considerably the thermal properties. In adodni,t i the aim of the simulations is to qualitatively coamrep the hi profile with the use of different molds. It seems that the exponent 0.03 is associated wi tphu trhee tin or tin alloys with small addition of alloying elements. The determined coefficient “na ”t hie copper mold experiment is found to be three times higher than that of the steel mold experim, tehnuts indicating a better wettability of the al lony the copper substrate. The straight relation betw hei aend wettability has been reported in reference [27] showing that the determination oi fi sh an alternative method to evaluate the wettayb wiliht ich is an important parameter considered in solderinge psrsoecs. The considerably higher experimental eutectic growth and cooling rates observed in 3F ifgo.r the Sn-Ni alloy solidified into Cu mold may be due to the higher heat transfer efficiency csoprorending to this condition. 300 300 250 250 200 200 150 150 Positions from metal/mold interface: 7 mm 9 mm Positions from 100 15 mm 100 metal/mold interface: 20 mm 3 mm 25 mm -0.03 8 mm Sn - 0.2wt.%Ni 45 mm h= 12,000.t 12 mm 60 mm i Sn - 0.2wt.%Ni 17 mm 75 mm 47 mm simulated [Copper mold] 58 mm h= 3,500.t -0.03 [Steel mold] i 50 70 mm50 simulated 0 10 20 30 40 50 0 20 40 Time [s] Time [s] (a) (b) Fig. 4. Experimental cooling curves and numericimalu slations used for determining the transiein- t h profiles of the Sn-0.2 wt.% Ni alloy for (a) copp aenrd (b) steel molds . 10 Temperature [oC] ACCEPTED Temperature [°C] MANUSCRIPT ACCEPTED MANUSCRIPT 14000 12,000.t-0.03 (copper mold) 3,500.t-0.03 (steel mold) 12000 10000 3800 3600 3400 3200 3000 0 10 20 30 40 50 Time [s] Fig. 5. Time-dependenti phrofiles for copper and steel molds. 3.2. Macrostructure, microstructure and experimental growth laws Fig 6 and Fig. 7 show optical micrographs and trheelairtive positions along the length of the Sn-Ni alloy castings, which are indicated by arr.o Awlsigned columnar grains growing in opposite way regarding the heat extraction direction chaerraizcet the entire macrostructures. The transverse microstructures of the alloy solidified against cthoepper mold are characterized by very fine cefl ls o the β-Sn phase considering positions closer to the msuorlfda ce, followed by the formation of a narrow transition zone and finally by βa- Sn dendritic network. The cellular region is sho two nbe associated with cooling rates higher than 5.5 K/s, i.e., feolra rtive positions in the casting lower than P=20 mm. The darker areas are formed by the eutectic mi.x ture 11 AC h [W.m-2.K-1]iCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 6. Macrostructure of the Sn-0.2wt.%Ni soldlelor ya casting directionally solidified against theu C mold with some typical related microstructures erespernting features found along length of the ca.s ting P is the position from the cooled surface of thset incag. The observed microstructural transition in Figto, o6k place for a relatively high cooling rate regime, i.e., it can be considered a reverse tyf pce lolular/dendritic transition since thβe-S n phase having cellular morphology prevailed for fast conogl irates, whereaβs- Sn dendrites are developed for slower cooling rates. This type of occurrence iasi nasgt the common features found on directionally 12 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT solidified microstructures, which established tmhaict rostructural transitions may follow plane fro>n t cells> dendrites occurrences with the increashee in g trowth rate [21]. Reverse transition from dendrites to cells has also been observed in tshee ocfa the rapidly solidified Sn-0.7 wt.% Cu by impulse atomization [30]. In this case, high-speuetde ctic cells were found to prevail for the powsd er of size smaller than 30µ0m . It is stressed that the reverse cells/dendtrriatenss ition for the Sn-0.7 wt%Cu alloy occurs for a cooling rate of about 1 K2/.s0 (for cooling rates > 12.0K/s the cellular growth prevails), which is around 2 times highearn t hthe critical value found for the Sn-0.2 wt.% Ni alloy. The lower range of cooling rates along the lenogf tthh e DS alloy casting solidified against the steel mold (compared with that of the copper mno ldF iig. 3) restricted the morphology of the Sn-rich matrix to low cooling rates cells along the enctiares ting length (see Fig. 7). Even for positions ecrlo to the cooled bottom of the casting, where highoeorl incg rates are operative, tβh-eS n matrix has a cellular morphology. The stabilization of the celallru structure in the higher cooling rates regioenm sse to be caused by the diffusion of Fe from the smteoelld towards the casting, thus affecting the local composition of regions closer to the casting suer.f aCcanté et al. [31] reported a similar role ofa Fse an element capable of favoring the growth of cienl lAs l-Ni alloys. An Al-1.0wt.%Ni alloy directionally solidified in a solidification appaturas like that used in the present study, presean ted dendritic matrix from bottom to top of the DS cansgt.i With the addition of 1.0wt%Fe to this alloy (i.e., an Al-1wt.%Ni-1.0wt.%Fe alloy) the resulti nDgS casting was shown to have a fully cellular matrix, i.e., the addition of Fe anticipated thaen tsrition from dendrites to low cooling rates ce lls. 13 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 7. Macrostructure of the Sn-0.2wt.%Ni soldlelor ya casting directionally solidified against the steel mold with some typical related microstrucstu rreepresenting features found along length of the casting. P is the position from the cooled surfoafc teh e casting . The dissolution of Cu in the molten alloy duringli dsiofication of the Sn-0.2 wt.% Ni alloy against the copper mold has affected the natutrhee o ef utectic phase located in the intercellular and interdendritic regions (dark areas of microstruecstu srhown in Fig. 6). The expected Ni-rich eutectic phases were replaced with the (Cu6,SNni)5 phase, which was identified by both the morpho laongdy 14 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT EDS-SEM analyses, as can be seen in Figs. 8 aCnud6 S9n. 5-based Cu-Ni-Sn IMC is often reported in Sn-based solder materials [32, 33]. This ternarays peh and the C6Sun5 binary IMC have similar crystal structure, in which the Ni atoms are arranged rceinpgla the Cu atoms, giving rise to this Cu-Ni-Sn ternary, i.e., the (Cu,N6iS) n5 IMC. This eutectic phase typically exhibits a rloikde- shape as depicted in Fig. 8B-D. The NiSn4 eutectic phase is reported to have no detectaobluleb islity for copper [7], whereas the tables inside Fig. 9 show appreciamboleu nat of copper. Since the eutectic IMC is a Ni- rich phase, the rod-like phase shown in Fig. 8h eis ( Ct u,Ni)6Sn5. The main microstructural characteristics of the0 S.2n -wt. % Ni alloy solidified into the copper mold, observed by SEM, are shown in Fig. 8. Afteer pd etching procedure, the specimens extracted at positions (P) closer to the cooled bottom of thloey a clasting (P=5mm and P=20mm) exhibit very fine Sn-rich cells enveloped by quite dense layers toefc etiuc (Cu,Ni)6Sn5 rods. Such rods become significantly coarse because of the decrease i nc othoeling rate for positions in the casting fartfhreorm the cooled surface. The spacing between the rotidc lpeasr (eutectic spacingλ,E ) varied considerably along the length of the DS casting: from 100nm0 t0on 9m (Fig. 9C/D). It can be observed in Fig. 8 that dense and fine (Cu,N6Si)n5 rods have been developed. The SEM-EDS results shown in Fig. 9 confirm thew gtrho of the primary N3iSn4 and the eutectic (Cu,Ni6)Sn5 constituting the microstructure of the Sn-Ni a.ll oTyhe presence of dissolved copper from the mold is more prevalent at posit icolnoser to the cooled bottom of the casting, which induced the growth of large polygonal3 SNni4 crystals as can be seen in Fig. 9A. These IMC s are reported to have Cu content around 11 at. % [7e]. oTther microstructures are governed by the rod- like morphologies of (Cu,N6iS) n5 organized in bundles. 15 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 8. Representative high-magnification SEM msictruoctures showing both thβe-S n high-velocity cells at positions closer to the bottom of the S.2nw-0t.%Ni alloy casting and the (Cu,N6Si)n5 intermetallic particles: (A)/(B) P=5mm; (C)/(D) P0=m2m and (E)/(F) P =60mm. (water-cooled Cu mold). 16 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 9. SEM images at the positions (A)/(B) P=5m(Cm); P=20mm and (D) P=60mm from the water- cooled surface of the Sn-0.2wt%Ni alloy castingth ien Cu mold. They are related to EDS measurements emphasizing the composition of thsee psh wahich are the N3Sin4 hexagonal rods at the image (A) and the (Cu,N6iS) n5 intermetallic particle in (B), (C) and (D ). 17 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 10a and Fig. 10b show the experimental scsa ftoter rcellular,λ c, and primary dendriticλ, 1, spacings as a function of the eutectic cooling arantde the eutectic growth rate, respectively. The results concerning the experiments in both moldrse wgeraphed together. Points are average spacing values along with their standard deviations forh e paocsition monitored in the DS casting. The incere as in either cooling rate or growth rate induces ar edaesce in the microstructural spacing regardles s the experimental condition observed. As found for o tdhiererctionally solidified Sn-based solder alloys [26, 34], the experimental microstructural growathw las a function of cooling rate can be represe nted by a power function having a -0.55 exponent. The adopted multipliers in the growth power funncsti oindicate thaλt 1 is around 2 times larger than λc if a certain cooling rate or a certain growth riast econsidered. This is due to the different morphologies of the Sn-rich matrix found for boxthp eriments considering different mold materials. Other studies dedicated to the effects of matrixrp mhology on the length-scale of the microstructure reported similar observations around the celluola-dr-etndritic transitions [34]. For instance, Eshenlm a and co-authors [35] noted a significant increas teh ein microstructural spacings at the beginningh eo f t transition range from cells to dendrites and hattvreib auted such coarsening to an increase in amdpeli tu of the formed stalks. The dendritic array has prevailed for the cases oinfg u copper mold while low cooling rate cells prevailed for the steel-cooled specimens. lTatheeral perturbations during the growth of the primary dendrite trunks induce the growth of secaoryn darms. The growth of these secondary structures can increase the distance between jtahcee andt primary trunks, resulting in larger spacing between them. This kind of microstructural formna tiso not appropriate for the growth of cells, wh ich remains typically finer that dendrites. Despite ltahreger cooling rates associated with the DS Sn-Ni casting solidified in the Cu-cooled mold (see F3ibg)., higher microstructural spacings characterhizee t length scale of thβe -Sn phase when compared with the correspondinge sv aoluf the DS casting solidified in the steel-cooled mold. The power function exponent -1.0 characterizede tuhte ctic spacingλ,E , variation with the eutectic growth rate, as can be observed in Ficg.. T1h0e classical growth law proposed by Jackson and Hunt for eutectics is restricted to binary yasll o[36] and does not take into account the growf th o 18 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT ternary eutectics. Thus, the traditional -1/2 prsoepdo by these authors seems to be not suitablhee f or t eutectic growth of the Sn-0.2 wt.% Ni alloy continagin the ternary (Cu,N6iS) n5 phase. An alternative experimental interrelation consindge rλiE vs. G -1/2xV-1/4 is proposed with fair to good compatibility with the experimental scatte rs haoswn in Fig. 10d. The growth in atomic scale of the equilibrium eutectic phase 3NSin4 is faceted and forms a highly-branched struct3u7re]. [In addition to that, the non-equilibrium Sn-Ni4S enutectic in Sn-Ni alloys is of non-faceted/face ttyepde, exhibiting regions with almost perfect lamellarg anlmi ent in addition to other regions of irregular eutectic. Sn-0.2wt.%Ni [Cu mold] λ = 49 (T ) -0.55 1 E 102 Sn-0.2wt.%Ni [Steel mold] λ = 22 (T ) -0.55 C E Dendrites Cells 101 Transition Cells zone with cells and dendrites 100 100 101 102 Cooling rate, T (°C/s) E (a) Sn-0.2 wt.% Ni [Cu mold] λ = 11.7 (V ) -1.1 1 E 102 Sn-0.2 wt.% Ni [Steel mold] λ = 5.5 (V ) -1.1 C E 101 100 10-1 100 Eutectic growth rate, V (mm/s) E (b) 19 A CelCl spacinCg / Primary dendrite Cell spacing / Primary dendrite arm spacing, λE / λ (µm)C 1 arm spacing, λ / λ (µm)C 1PTED MANUSCRIPT ACCEPTED MANUSCRIPT 101 Sn-0.2 wt.% Ni [Cu mold] Sn-0.2 wt.% Ni [Cu mold] λ = 0.19 (V ) -1.0 λ = 0.8 (G -1/2 . V -1/4) E E 1 E E E10 100 100 10-1 10-1 10-2 10-2 10-1 100 10-1 100 Eutectic growth rate, V (mm/s) G -1/2 x V -1/4E E E (c) (d) Fig. 10. Experimental evolutions of microstructu srpaal cings along the length of the Sn-0.2wt.%Ni alloy casting: (a)λ 1 vs. cooling rate; (b)λ 1 vs. eutectic growth rate and (c,λdE) vs. eutectic growth rate and λ vs. G -1/2E E xV -1/4E . 3.3. Tensile properties vs. cellular and primary dendritic arm spacings Fig. 11 shows the experimental evolutions of ultiem taensile strengthσ (u), yield tensile strength σ( y) and elongationδ (),considering the DS Sn-0.2 wt. % Ni alloys cassti nsgolidified against two different mold materials, i.e., Cu and steeal.l l-HPetch type correlations were adopted to repnrte se all the experimental scatters. In geneσrau la, ndσ y increase with decreasinλg1, c for both Sn-Ni solder alloy castings. Fineλr 1,c allows better distribution of the reinforcing IM Cthsroughout the microstructure. On the other hand, opposite tenidees nwcere found for the elongation-to-fracture with λ1,c. Higher mechanical strength is associated wieth s tahmples solidified against the copper mold. This can be probably explained by the prevalen ceeit hoef r very fine Sn-rich cells reinforced by sub- micrometric scaling (Cu,N6iS) n5 rods or by more complex morphology of tβh-ep hase found for positions farther from the bottom of the alloy cinags.t Due to the intrinsic complexity associatedh w it the dendritic growth, a homogeneous distributiodn alnternation of the reinforcing hard phase, ti.hee., (Cu,Ni)6Sn5, is attained. The non-equilibrium NiS4n eutectic phase exhibiting a plate-like shape cea sne ben through the SEM microstructures in Fig. 12. The chemis tprireovided by the SEM-EDS analyses in Fig. 13 confirmed the presence of such phase within thec etiuc tmixture for the DS casting solidified aga inst 20 Eutectic spacing, λ (µm) E ACCEPTED MA Eutectic spacing, λ (µm)ENUSCRIPT ACCEPTED MANUSCRIPT the water-cooled steel mold. The development orfs ceo Na iSn4 eutectic plates [38] seems to have contributed to the lower strength values observne Fdi gi . 11. Higher values of ductility are observine d this case (see Fig. 11c). These results may bec iastesdo with the refinement of the cell spacing combined with the higher fracture toughness of N piShanses as reported in the literature [39]. The fracture toughness reported for the Ni-rich phass 4e. 2i2 MPa/ m while that referring to the C6Sun5 phase is 2.68 MPa/m . Considering that Vickers microhardness of thIMesCes is about 370HV for a load of 10g, one can qualitatively infer that tnhter ini sic ductility of the NiSn phase is higher, contributing to the results provided in Fig. 11cr tfhoe Sn-Ni alloy. Similar behavior on ductility againλst-1/2c was observed for DS the Sn-0.7wt%Cu(-xNi) alloys specimens [26]. Those, which are fully formed btye ectuic cells, also permitted an increase in dutyc tili to be perceived with the increase in the celluplarc sing. 38 Sn-0.2 wt.% Ni [Cu] 36 σ = 43.8 (1/λ 1/2) + 13.8 u c,1 34 Sn-0.2 wt.% Ni [Steel] 32 σ = 29.1 (1/λ 1/2) + 13.1 u c 30 28 26 24 22 20 18 16 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 1/(λ )1/2 ( µm ) - 1/2 c,1 (a) 21 AC Ultimate tensile strength, σu, (MPa)CEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 36 Sn-0.2 wt.% Ni [Cu] 34 σ = 46.4 (1/λ 1/2) + 10.3 y c,1 32 Sn-0.2 wt.% Ni [Steel] 30 σ = 39.8 (1/λ 1/2) + 8.3 y c 28 26 24 22 20 18 16 14 12 10 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 1/(λ )1/2 ( µm ) - 1/2 c,1 (b) 75 70 Sn-0.2 wt.% Ni [Cu] 65 δ = -20 (λ -1/2) + 20.5 c,1 60 Sn-0.2 wt.% Ni [Steel] 55 δ = -95 (λ -1/2) + 61.6 c 50 45 40 35 30 25 20 15 10 5 0 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 1/(λ )1/2 (µm) - 1/2 c,1 (c) Fig. 11. Correlations between tensile mechanicoapl eprrties and the cellularλ -c /primary dendritic -λ 1 spacings for the Sn-0.2wt.%Ni alloy DS castingsid sifoield against Cu and steel cooled moldsσ: (u a) vs. (λc,1)-1/2; (b) σy vs. (λ )-1/2 c,1 and (c)δ vs. (λ )-1/2c,1 . 22 AC Elongation-to-fracture, δ (%) Yield tensile strength, σy (MPa)CEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 12. Representative high-magnified SEM micruocstturres found for the DS Sn-0.2wt.%Ni alloy casting solidified against the steel cooled moldp heamsizing the presence of the N4i Sinntermetallic particles: (A)/(B) P=5mm; (C)/(D) P=10mm and (E)/ (PF =15mm (P is the position from the metal/mold interface). 23 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Fig. 13. SEM images at the positions (A) P=5mm; P(B=)10mm and (C) P=15mm from the water- cooled surface of the Sn-0.2wt%Ni alloy castingid sifoiel d against the steel substrate. Each SEM microstructure is associated with EDS measurem wenhtisc,h reveal the composition of the Ni4S n phase in all cases. Fig. 14 shows elemental SEM-EDS maps for a binanr-y0 .S2 wt.% Ni alloy specimen (P=5mm) from the DS casting solidified against sthteee l mold. Sn contrast (in green) is more associated with the Sn-rich matrix. Ni (in red) eaaprps concentrated in the non-equilibrium N4i Sn IMC, while Fe (in blue) seems to be more homogesnleyo duistributed in the microstructure. According to Belyakov and Gourlay [40], Fe2S fnorms during the solidification of most commerc Piabl-free 24 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT solders due to impurity iron. Under equilibrium cdoitnions, the first solid to be formed in the Snh-r ic Sn-Ni alloys is the FeS2 nwith the L+ FeS2n+Ni3Sn4 three-phase equilibrium occurring at 270°C. FeSn2 particles may operate as heterogeneous nuclesaiteiosn f or NiSn4, being completely engulfed by the NiSn4 in most of the cases. In the present investig,a thioen iron content noted in the EDS-SEM results probably originated from the interactiontw beeen the steel bottom-part mold and the molten alloy. It seems that the Fe in the pre-existingn F2 epShase has been incorporated during the growth of the NiSn4 eutectic plates. A homogeneous distribution ocf aFne be seen in Fig. 14. P=5mm SnL NiL FeK P=5mm SnL NiL FeK Fig.14. Elemental chart of a sample at the pos i5timonm of the DS Sn-0.2 wt.% Ni alloy casting solidified against the steel mold: SEM-EDS anal y(sMeasgnification 20,000x). 4. Conclusions The following conclusions can be drawn from thes pernet experimental investigation: 1-It was shown that the by changing the bottom opfa trhte water-cooled mold from steel to copper, the solder/mold heat transfer efficienecyp,r ersented by a metal/mold heat transfer coeffti cien (hi) increased about 3.4 times. This has reflectesdig ini ficant differences in the solidification conogli rates Ṫ( E) of regions in the DS casting closer to the co osluerdface of the casting. For instance, for a position of about 5mm from the cooled surface oef ctahsting,Ṫ E increased from about 7°C/s with the steel mold to about 31°C/s when a copper mold wsaesd .u 25 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT 2- The microstructure of the Sn-0.2wt.%Ni alloy w sahsown to be characterized by significant differences in the morphology of thβe-S n matrix. The used of the copper mold induced o tchceurrence of high cooling rates cells foṪrE > 5.5C/s, followed by the growth of dendrites w tihthe decrease in the cooling rate, thus characterizing a reverse celnlsd/drites transition. For the experiment with theee ls t mold, theβ -Sn matrix was shown to have a cellular morphol(ologwy cooling rate cells) that prevailed along the entire DS casting, i.e. the range of reimxpeental cooling rates was not high enough to ined uc the usual dendrites/cells transition. 3- Both copper and steel molds have their elem(eCnuts a nd Fe) being incorporated into the alloy composition affecting mainly the initial potiosins near the metal/mold interface inducing the formation of polygonal N3Si n4 crystals and FeS2 nparticles, respectively. These latter particleasy pthl e role of substrates for the nucleation of plate- lNikieSn4 particles. The dissolution of Cu in the molten alloy during solidification of the Sn-0.2 wt.% Nlil oay against the copper mold has affected the nea tur of the eutectic phase located in the intercellaunladr interdendritic regions. The expected Ni-rich eutectic phases were replaced with the (Cu6S,Nn5i) phase, which was shown to exhibit a rod-like shape. 4- Experimental growth laws have been proposedti nregl athe length scale of the phases characterizing the microstructure of the Sn-0.2wNti. %alloy (microstructural spacings) to solidification thermal parameters (cooling and gtrho wrates). These experimental equations can be used in the control of the soldering process byi pmualanting solidification processing parameters s uch as the cooling rate in the preprogramming of ar deeds mi icrostructural arrangement. 5- Hall-Petch type correlations were proposed irnegla tthe tensile properties with the length scale of theβ -Sn matrix, for both experimental conditions; corp apned steel moldσ. u andσ y were shown to increase significantly with the decreans λe1 ,ic The tensile responsσeu (andσ y) associated with different microstructural morphologies such c aeslls, dendrites and rod-like, plate-like intermetallics has shown that the highest tensriolefi lpe is obtained when the dendrites and cells are 26 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT reinforced with rod-like intermetallics morpholo(g(yC u,Ni)6Sn5). Concerningδ , the best observed behavior was shown to be associated with cellso usunrdred with N3iSn4 plate-like intermetallics. Acknowledgements The authors acknowledge the financial support pdreodv iby FAPESP (São Paulo Research Foundation, Brazil: grant 2015/11863-5) and CNPhqe- TBrazilian Research Council. References [1] A.F. Abd El-Rehima, H.Y. Zahrana, Investigat iofn microstructure and mechanical properties of Sn-xCu solder alloys, J. Alloys Compd. 695 (201676) 63–367. [2] Li Yang, Yaocheng Zhang, Jun Dai, Yanfeng J iJnign,guo Ge, Ning Zhang, Microstructure, interfacial IMC and mechanical properties of Sn–C0u.–7xAl (x = 0–0.075) lead-free solder alloy. Mater. Des. 67 (2015) 209–216. [3] S.A. Belyakov, J.W. Xian, K. Sweatman, T. Nimshuira, T. Akaiwa, C.M. Gourlay, Influence of bismuth on the solidification of Sn-0.7Cu-0.05Ni-i/xCBu joints. J. Alloys Compd. 701 (2017) 321- 334. [4] U. Böyük, N. Maraşli, Dependency of eutectic spacings and microhasrsd noen the temperature gradient for directionally solidified Sn–Ag–Cu le-fardee solder. Mater. Chem. Phys. 119 (2010) 442- 448. [5] P.D. Pereira, J.E. Spinelli, A. Garcia. Combdi neeffects of Ag content and cooling rate on microstructure and mechanical behavior of Sn-Ags-Colud ers, Mater. Des. 45 (2013) 377–383. [6] W.R. Osório, D.R. Leiva, L.C. Peixoto, L.R. Gcaiar, A. Garcia. Mechanical properties of Sn-Ag lead-free solder alloys based on the dendriticy a arnrad Ag3Sn morphology, J. Alloys Compd. 562 (2013) 194-204. [7] S.A. Belyakov, C.M. Gourlay, NiSn4 formation rdinug the solidification of Sn-Ni alloys. J. Alloys Compd. 701 (2017) 321-334. [8] R. Schueller, N. Blattau, J. Arnold, C. Hillm,a Snecond generation Pb-free alloys, J. SMTA 23 (2010) 18-26. 27 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT [9] C.M. Gourlay, K. Nogita, J. Read, A.K. Dahlen,t eIrmetallic formation and fluidity in Sn-rich Sn- Cu-Ni alloys, J. Electron. Mater. 39 (2010) 56-69. [10] T. Ventura, C.M. Gourlay, K. Nogita, T. Nishuimra, M. Rappaz, A.K. Dahle, The influence of 0- 0.1 wt.% Ni on the microstructure and fluidity lethn gof Sn- 0.7Cu-xNi, J. Electron. Mater. 37 (2008) 32-39. [11] W. C. W. Ng, K. Sweatman, T. Akaiwa, T. Nishuimra, M. Sato, C. Gourlay, S. Belyakov, Dissolution in Service of the Copper Substrate ooldf eSr Joints, 2016 IEEE 18th Electronics Packaging Technology Conference (EPTC), 388-393. [12] A. Bogno, J. E. Spinelli, C. R. M. Afonso, H.e nein, Microstructural and mechanical properties analysis of extruded Sn–0.7Cu solder alloy. J. Mr. aRtes. Technol. 2015 (4) 84-92. [13] J.W. Xian, S.A. Belyakov, C.M. Gourlay, Conltlirnog bulk Cu6Sn5 nucleation in Sn0.7Cu/Cu joints with Al micro-alloying, J. Electron. Mate4r.5 (2015) 69-78. [14] I.T.L. Moura, C.L.M. Silva, N. Cheung, P.R. uGlaort, A. Garcia, J.E. Spinelli, Cellular to dendritic transition during transient solidificanti of a eutectic Sn-0.7wt%Cu solder alloy. Mater. Chem. Phys. 132 (2012) 203-209. [15] M. Felberbaum, T. Ventura, M. Rappaz, and A D.Ka.hle, Microstructure Formation in Sn-Cu-Ni Solder Alloys, JOM 63 (2011) 56-59. [16] B.L. Silva, N. Cheung, A. Garcia, J.E. Spini,e Tllhermal parameters, microstructure, and mechanical properties of directionally solidifiend– S0.7 wt%Cu solder alloys containing 0 ppm to 1000 ppm Ni, J. Electron. Mater. 42 (2013) 179– 191. [17] J.E. Spinelli, D.M. Rosa, I.L. Ferreira, A. rGciaa, Influence of melt convection on dendritic spacings of downward unsteady-state directionaollliyd isfied Al–Cu alloys. Mater. Sci. Eng. A 383 (2004) 271-282. [18] P. Donelan, Modelling microstructural and maencihcal properties of ferritic ductile cast iron. Mater. Sci. Technol. 16 (2000) 261-269. [19] P.R. Goulart, J.E. Spinelli, N. Cheung, A. Gcia,r The effects of cell spacing and distributiof n o intermetallic fibers on the mechanical propertife hs yopoeutectic Al–Fe alloys. Mater. Chem. Phys. 119 (2010) 272-278. 28 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT [20] I.T.L. Moura, C.L.M. Silva, N. Cheung, P.R. uGlaort, A. Garcia, J.E. Spinelli, Cellular to dendritic transition during transient solidificanti of a eutectic Sn 0.7wt%Cu solder alloy. Mater. Chem. Phys. 132 (2012) 203-209. [21] M.A. Pinto, N. Cheung, M.C.F. Ierardi, A. Gaiar,c Microstructural and hardness investigation of an aluminum-copper alloy processed by laser me, lMtinagter. Character. 50 (2003) 249-253. [22] M Gunduz, E Çadirli, Directional solidificatnio of aluminium–copper alloys. Mater. Sci. Eng. A 327 (2002) 167-185. [23] T. Takemoto, M. Takemoto, Dissolution of stlaeisns steels in molten lead-free solders. Solder Surf. Mt. Tech. 18 (2006) 24-30. [24] M. Krishnan, D.G.R. Sharma, Determination hoef tinterfacial heat transfer coefficient h in unidirectional heat flow by Becks nonlinear estiimona tprocedure. Int. Commun. Heat Mass Transf. 23 (1996) 203-214. [25] F. Bertelli, C.H. Silva-Santos, D.J. BezerNra. ,C heung, A. Garcia, An Effective Inverse Heat Transfer Procedure Based on Evolutionary Algorit htom Dsetermine Cooling Conditions of a Steel Continuous Casting Machine. Mater. Manuf. Proc3e0s,s(.2 015) 414-424. [26] B. L. Silva, N. Cheung, A. Garcia, J. E. Splliin, eSn–0.7wt%Cu–(xNi) alloys: Microstructure– mechanical properties correlations with solder/tsruabtes interfacial heat transfer coefficient. J.o Aylsl Compd. 632 (2015) 274-285. [27] W. L.R. Santos, B. L. Silva, F. Bertelli , EJ. Spinelli, N.Cheung, A. Garcia. An alternative thermal approach to evaluate the wettability odf esor lalloys. Appl. Therm. Eng. 107 (2016) 431-440. [28] E A Brandes & G B Brook, Smithells Metals Rrefnece Book -7th Edition, Imprint: Butterworth- Heinemann, 2013. [29] R. D. Pehlke, A. Jeyarajan, H. Wada, Summaf rTyh oermal Properties for Casting Alloys and Mold Materials, University of Michigan,1982. [30] G. R. Moreno, B. L. Silva, A. Bogno, H. Hen,e Jin. E. Spinelli, Microstructure-property relatsio n in as-atomized and as-extruded Sn-Cu (-Ag) solldloeyrs a. J. Alloys Compd. 680 (2016) 259-267. 29 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT [31] M.V. Canté, C. Brito, J.E. Spinelli, A. Garc, iIanterrelation of cell spacing, intermetallic compounds and hardness on a directionally soldid iAfile-1.0Fe-1.0Ni alloy. Mater. Des. 51 (2013) 342-346. [32] M. A. A. Mohd Salleh, C. M. Gourlay, J. W. Xni,a S. A. Belyakov, H. Yasuda, S. D. McDonald and K. Nogita, In situ imaging of microstructurerm foation in electronic interconnections. Sci. Re p. 7 (2017) 40010. [33] H. Lee and K. Huang, Effects of Cooling Rante t hoe Microstructure and Morphology of Sn- 3.0Ag-0.5Cu Solder. J. Electron. Mater. 45 (201862)- 190. [34] D.M. Rosa, J.E. Spinelli, I.L. Ferreira, A. rGciaa, Cellular/dendritic transition and microstrurcet evolution during transient directional solidificoanti of Pb–Sb alloys. Metall. Mater. Trans. A 39 (82)0 0 2161-2174. [35] M.A. Eshelman, V. Seetharaman, and J.W. Trii,v Cedellular spacings—I. Steady-state growth. Acta Metall. 36 (1998) 1165-1174. [36] KA Jackson, JD Hunt, Lamellar and rod eute gctriocwth. Trans. Metall. Soc. AIME 236 (1966) 1129-1142. [37] R. Elliott, Eutectic Solidification, London,u Btterworth, 1983. [38] C. Schimpf, P. Kalanke, S.L. Shang, Z.K. LAiu. ,L eineweber, Stacking disorder in metastable NiSn4. Mater. Des. 109 (2016) 324-333. [39] G. Ghosh, Elastic properties, hardness, adnedn itnation fracture toughness of intermetallics relevant to electronic packaging. J. Mater. Res(2.1090 4) 1439-1454. [40] S.A. Belyakov, C.M. Gourlay, Role of Fe imptuiersi in the nucleation of metastable N4i,S n Intermetallics 37 (2013) 32–41. 30 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT Figure Captions Fig. 1. Unidirectional solidification system fora ntrsient heat flow regime coupled with the neces sary devices. Fig. 2. Experimental cooling curves registered ndgu ruinsteady-state directional solidification of the Sn-0.2wt.%Ni solder alloy in water-cooled (a) corp apned (b) steel molds.E T is the eutectic temperature, which is 231ºC . Fig. 3. Experimental tendencies related to (a) cetiuct egrowth rate and (b) cooling rate for the solidified Sn-0.2wt.%Ni solder alloy either in coeprp or in steel substrates. Fig. 4. Experimental cooling curves and numericimalu slations used for determining the transiein- t h profiles of the Sn-0.2 wt.% Ni alloy for (a) copp aenrd (b) steel molds . Fig. 5. Time-dependenti phrofiles for copper and steel molds. Fig. 6. Macrostructure of the Sn-0.2wt.%Ni soldlelor ya casting directionally solidified against theu C mold with some typical related microstructures erespernting features found along length of the ca.s ting P is the position from the cooled surface of thset incag. Fig. 7. Macrostructure of the Sn-0.2wt.%Ni soldlelor ya casting directionally solidified against the steel mold with some typical related microstrucstu rreepresenting features found along length of the casting. P is the position from the cooled surfoafc teh e casting . Fig. 8. Representative high-magnification SEM msictruoctures showing both thβe-S n high-velocity cells at positions closer to the bottom of the S.2nw-0t.%Ni alloy casting and the (Cu,N6Si)n5 intermetallic particles: (A)/(B) P=5mm; (C)/(D) P0=m2m and (E)/(F) P =60mm. (water-cooled Cu mold). Fig. 9. SEM images at the positions (A)/(B) P=5m(Cm); P=20mm and (D) P=60mm from the water- cooled surface of the Sn-0.2wt%Ni alloy castingth ien Cu mold. They are related to EDS 31 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT measurements emphasizing the composition of thsee psh wahich are the N3Sin4 hexagonal rods at the image (A) and the (Cu,N6iS) n5 intermetallic particle in (B), (C) and (D ). Fig. 10. Experimental evolutions of microstructu srpaal cings along the length of the Sn-0.2wt.%Ni alloy casting: (a)λ 1 vs. cooling rate; (b)λ 1 vs. eutectic growth rate and (c,λdE) vs. eutectic growth rate and λE vs. G -1/2xV -1/4E E . Fig. 11. Correlations between tensile mechanicoapl eprrties and the cellularλ -c /primary dendritic -λ 1 spacings for the Sn-0.2wt.%Ni alloy DS castingsid sifoield against Cu and steel cooled moldsσ: (u a) vs. (λ )-1/2c,1 ; (b) σy vs. (λc,1)-1/2 and (c)δ vs. (λ -1/2c,1) . Fig. 12. Representative high-magnified SEM micruocstturres found for the DS Sn-0.2wt.%Ni alloy casting solidified against the steel cooled moldp heamsizing the presence of the N4i Sinntermetallic particles: (A)/(B) P=5mm; (C)/(D) P=10mm and (E)/ (PF =15mm (P is the position from the metal/mold interface). Fig. 13. SEM images at the positions (A) P=5mm; P(B=)10mm and (C) P=15mm from the water- cooled surface of the Sn-0.2wt%Ni alloy castingid sifoiel d against the steel substrate. Each SEM microstructure is associated with EDS measurem wenhtisc,h reveal the composition of the Ni4S n phase in all cases. Fig.14. Elemental chart of a sample at the pos i5timonm of the DS Sn-0.2 wt.% Ni alloy casting solidified against the steel mold: SEM-EDS anal y(sMeasgnification 20,000x). 32 ACCEPTED MANUSCRIPT ACCEPTED MANUSCRIPT RESEARCH HIGHLIGHTS • Higher hi coefficients were obtained by using Cu as substrate. • Low cooling rate cells were developed for Sn-0.2Ni against low C steel mold. • High cooling rate cells were obtained for Sn-0.2Ni against copper mold. • Higher strength is obtained when the β-Sn phase is reinforced with (Cu,Ni)6Sn5. • Cells surrounded with Ni3Sn4 plate-like intermetallics refer to higher ductility. ACCEPTED MANUSCRIPT