Tu

Ma

b

a

ARRAA

KUATDWB

1

pt(uttaotBpottvs

(

U

8

0d

Journal of Materials Processing Technology 211 (2011) 1650– 1657

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

j o ur nal ho me p age : www.elsev ier .com/ locate / jmatprotec

hermal transients during processing of materials by very high powerltrasonic additive manufacturing

.R. Sriramana,∗, Matt Gonsera,1, Hiromichi T. Fujii a,2, S.S. Babua, Matt Blossb

Department of Materials Science and Engineering, The Ohio State University, Columbus, OH 43221, USAEdison Welding Institute, Columbus, OH 43221, USA

r t i c l e i n f o

rticle history:eceived 26 October 2010eceived in revised form 11 April 2011ccepted 8 May 2011vailable online 13 May 2011

a b s t r a c t

Dynamic recrystallisation at interfaces has been suggested as the bonding mechanism in the joining ofmetallic tapes, during very high power ultrasonic additive manufacturing. To understand the reasonsfor such occurrence of dynamic recrystallization, thermal transients from the interface regions wererecorded during processing of aluminum alloy (3003 and 6061 series) and 11 000 copper tapes undersimilar conditions. Measurements in 3003 Al were also carried out for different processing parameters.Measured peak temperatures were seen to increase with increase in shear strength of the material and

eywords:ltrasonic additive manufacturingdiabatic heatinghermal transientsynamic recrystallisationelding

ultrasonic vibration amplitude. The observations have been rationalized based on interfacial heating atasperities due to adiabatic plastic deformation.

© 2011 Elsevier B.V. All rights reserved.

onding

. Introduction

Ultrasonic additive manufacturing (UAM) is a rapid prototypingrocess by which near-net shaped 3-D parts can be fabricated fromhin metallic foils or tapes directly from a computer aided designCAD) model (White, 2003). The process is based on the principle ofltrasonic seam welding, a solid-state joining process. It involveshe application of ultrasonic vibrations (20 kHz frequency) to theape through a sonotrode that vibrates laterally at the imposedmplitude under a given normal force and rolls along the lengthf the tape at a chosen speed (Kong et al., 2004a). This bonds theape to the one below to produce a seam weld between them.onding is attributed to the rapid “scrubbing action” that takeslace between the faying surfaces, which disrupts and dispersesxide layers between them promoting nascent metal–metal con-act (White, 2003). The sonotrode is given a surface texture so as

o enable it grip the tape at any given instant and subject it toibrations. This alternate engaging and disengaging action of theonotrode with respect to the tape leaves a rough imprint on the

∗ Corresponding author. Tel.: +1 614 292 3654.E-mail addresses: ramanujam.10@osu.edu, mrsriraman@yahoo.com

M.R. Sriraman).1 Now with College of Engineering & Engineering Technology, Northern Illinoisniversity, De Kalb, IL 60115, USA.2 Visiting Researcher from Tohoku University, 28 Kawauchi, Aoba-ku, Sendai 980-

576, Japan.

924-0136/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.jmatprotec.2011.05.003

tape surface along its seam. This is the surface onto which thenext layer is welded and so on. This “additive” process is usuallycombined with machining operations as well (White, 2003). Such“subtractive” processes allow for the fabrication of internal featuresinto the part as it is being built. Typically, the first tape is weldedonto a base plate, which can then be machined off from the part, ifdesired. The UAM process thus offers the possibility of producingparts of complex shapes and designs (Graff, 2010). Embedding ofoptical fibers between layers and even sensors into the parts is nowpossible (Janaki Ram et al., 2007).

With the low power capabilities of the machines available thusfar, the application of UAM has however been mostly limited toaluminum only (Graff, 2010). Even in aluminum, bonding has beenfound to be inadequate (Schick et al., 2010). Thus, with a viewto improve the bonding quality and extend this technology toother materials (for similar metal welding), a prototype to whatis termed the “very high power ultrasonic additive manufactur-ing” (VHP UAM) machine has been designed and constructed (Graff,2010). (To make a distinction between the processes, future refer-ence to UAM in the paper will pertain to the low power process.)Since vibration amplitude and normal force constitute the mostimportant parameters to the process, the power capability of themachine has been increased manifold (rated at 9 kW) to deliver

higher amplitudes (up to 52 �m) under larger force levels (up to15 kN). Such high amplitudes of the sonotrode have been possibleusing two transducers (as opposed to the earlier UAM machineswhich had just one) on either side of the sonotrode, operating in

M.R. Sriraman et al. / Journal of Materials Proce

Fm

tof

tfo1(ti(tcwwTCltAsrI1Mtr

Fo

ig. 1. Schematic of the double transducer–sonotrode system used in the VHP-UAMachine.

andem under push–pull mode, and at 180◦ out of phase with eachther. A schematic of this is shown in Fig. 1. An example of a partabricated by UAM technology can be seen from Fig. 2.

Since the process involves high-speed scrubbing between tapeso be joined, localized heating and temperature increase due toriction/deformation are expected at the interfaces. The occurrencef recrystallisation observed in ultrasonic metal welding (Phillips,960), UAM (Mariani and Ghassemieh, 2010), and in VHP UAMSriraman et al., 2010a) without any external heating suggests thatemperatures generated can be quite significant. Sometimes, local-zed melting in dissimilar metal welding has also been reportedWeare et al., 1960). The importance of thermal measurementsherefore cannot be overstated. Several investigations have beenarried out to measure the temperatures at the weld zone in spotelding/UAM. Weare et al. (1960) in their investigations on spotelding of copper to Monel measured temperatures of up to 505 K.

he copper–Monel junction itself was used as a thermocouple.opper was however seen to turn red hot and evidence of its

ocalized melting was also noted especially during long weldimes under conditions of high amplitudes and low force levels.ccording to the investigators (Weare et al., 1960), the mea-ured temperature was lower probably because of the thermo emfesponse of molten Cu–Monel being different from solid Cu–Monel.n another experiment as part of the same study (Weare et al.,960), “hot spots” were observed when metals such as Al, Cu, and

o were let to “slide” ultrasonically over a glass plate suggesting

hat temperatures in Al and Cu reached at least 793 K based on theed light emission and 1673 K in Mo based on white light emission.

ig. 2. An aluminum part fabricated by ultrasonic additive manufacturing (courtesyf Graff, 2010).

ssing Technology 211 (2011) 1650– 1657 1651

Hazlett and Ambekar (1970) in their studies on iron–constantanspot welding, measured a temperature of 447 K, again based on thethermo emf generated between the dissimilar metals themselves.They attributed this value to being an average temperature overthe contact area and not the maximum possible. de Vries (2004)in his experiments on 6061 Al-T6 spot welding, recorded peaktemperatures between 448 and 498 K using IR camera, dependingupon the weld energy and contact pressures chosen. The tempera-ture was seen to increase steeply to a peak value in about 40 ms ofthe weld cycle and then register a decrease. In a more recent study,Cheng and Li (2007) measured temperatures of up to 523 K using“thin film thermocouple arrays” during welding of copper to nickel.A sharp initial rise in temperature was seen here again, which wasfound to then level off. Temperature measurements in UAM havebeen carried out only on 3003 Al-H18 and using Type K thermo-couples. Kong et al. (2004b) recorded peak temperatures in therange of 343–423 K depending upon processing conditions. Mea-surements were made during embedding of shape memory alloyNiTi fibers between Al layers. Yang et al. (2009), in a subsequentstudy reported a temperature rise in the range of 341–371 K basedon the processing parameters used. In a more recent investigation,Schick et al. (in press) measured temperatures of up to 438 K whenprocessing was done under a preheat temperature of 338 K.

Thus, increase in temperature during ultrasonic welding/ UAMhas been observed by all investigators although there seems to bea variation in the magnitudes measured depending upon the tech-nique adopted and welding/processing conditions. Interestinglyhowever, there has been no research done, be it in spot weldingor additive manufacturing, to investigate the temperature risein different materials and relate it to the material deformationcharacteristics. Similarly, studies have also not been attemptedto determine and understand the effect of each of the principalprocessing parameters viz. vibration amplitude, normal force, andtravel speed in inducing a temperature rise. The goals of this inves-tigation therefore were to measure and compare the temperaturerise occurring during processing of different materials such as 3003Al-H18, 6061 Al-H18, and 11 000 Cu-Hard, by VHP UAM, as well asbetween various processing conditions in a single material (chosento be 3003 Al-H18). Since metallurgical bonding of layers in VHPUAM, from materials investigated thus far (Sojiphan et al., 2010 on3003 Al-H18, Sriraman et al., 2010b on 6061 Al-H18, and Sriramanet al., 2010a on 11 000 Cu-Hard), is attributed to dynamic recrys-tallisation (DRX) across the interfaces, knowledge of the interfacialthermal cycles during processing was considered crucial for exten-sion of this technology to other alloy systems. Thermocouples (TypeK) were considered here for measurements for the reasons thatthey are easily embedded between layers and temperatures couldbe recorded from the interior of the interfacial weld region, whereheating is expected to be a maximum. It was however recognizedright at the outset that “capturing” the actual magnitude of thetemperatures generated in a highly localized manner varying at apossible time resolution of 5e−5 s (1/20 kHz) might not be feasible.Hence, the objective of this work was primarily to make compar-isons in the data between materials/processing conditions and seeif any trends are revealed, rather than seek “actual” temperatures.

2. Experimental procedure

2.1. Thermal measurements

Single strand individually sheathed Type K thermocouple wires

(42 AWG) measuring 70 �m bare wire diameters were chosen forrecording the temperatures. Wires of this thickness are more easilyembedded between layers and are expected to have a fast timeresponse (compared to thicker gages). The sheathing was to ensure

1652 M.R. Sriraman et al. / Journal of Materials Processing Technology 211 (2011) 1650– 1657

Table 1Processing parameters in 3003 Al-H18.

Vibration amplitude (�m) Normal force (kN) Travel speed (mm/s)

26 3.2 35.526 4.0 35.526 5.6 25.526 5.6 35.526 5.6 51.026 6.7 35.531 5.6 35.536 5.6 35.5

tanTttaoiIDtiw

2

Haop3bTwtpcosssaidaAtmbmpprwc6ePb

Fig. 3. Schematic showing plan view of how the thermocouple was placed on thetape for embedding (a). (b) and (c) Illustrations of the embedding pass and the buildpass, respectively.

Table 2Processing of 6061 Al-H18 and 11 000 Cu-Hard.

Material Vibrationamplitude(�m)

Normal force(kN)

Travel speed(mm/s)

Typical thermal profiles from an embedding and subsequentbuild pass are presented in Fig. 4. During an embedding pass, aninitial increase in the temperature (up to about 40%) was usually

36 6.7 30.5

hat the temperature would be picked up only from the twistednd spot welded portion (of about 10 mm length) of the wires (andot through any electrical short circuiting outside of this region).hermocouples prepared thus were then tested for electrical con-inuity and were then connected to the data acquisition device runhrough a personal computer. Temperature data were collectedt a sampling rate of 10 000/s during processing to an accuracyf ±1 K (Boulware, 2010). The recorded data were then plottednto thermal profiles using commercial data processing softwaregorProTM, and “smoothed” to determine the peak temperatures.ue to low-pass filtering done by the signal conditioner, much of

he high frequency noise was eliminated during data acquisitiontself. The smoothed peak temperature values thus were usually

ithin ±1% of the acquired peak temperature data.

.2. VHP UAM processing conditions

Before thermal measurements were commenced on 3003 Al-18, three layers of the material in the form of tapes (150 �m thicknd 25.4 mm wide) were first welded over a length of 150 mmn a 3003 All-H14 base plate to form the “base layers. These wererocessed at 26 �m vibration amplitude, 5.6 kN normal force, and5.5 mm/s travel speed that was known to produce good bondingased on earlier research of the authors (see Sojiphan et al., 2010).he purpose of the base layers was to make sure the thermocouplesere well rested and flush with the tape. The processing condi-

ions chosen for thermal measurements were based on the abovearameters and other combinations of amplitude, force, and speedentered around these parameters (Table 1). Following the weldingf base layers, the thermocouple connected to the data acquisitionystem was placed on the tape at the center of the seam. This ishown in Fig. 3. The next tape was bonded onto it at one of the cho-en processing conditions (see Table 1). The temperature data werecquired (as described in Section 2.1) from “start” to “finish” dur-ng welding of this layer. Since the thermocouple was embeddeduring this operation, this is termed as “embedding pass”. The startnd end times corresponded to the start and end of seam welding.lthough the welding process itself was run through a computer,

he start and end times for thermal measurements were onlyanually controlled. The corresponding times therefore could not

e precisely synchronized. Subsequent to the embedding pass, twoore layers (referred to as “build passes”) were welded at the same

arameters and temperatures recorded as before. The embeddingass and the build pass are illustrated in Fig. 3. This process wasepeated for all other processing conditions. A similar procedureas adopted for 6061 Al-H18 and 11 000 Cu-Hard for just one pro-

essing condition each as shown in Table 2. Based on earlier work on061 Al-H18 (Sriraman et al., 2010b) and 11 000 Cu-Hard (Sriraman

t al., 2010a), these parameters were known to yield good bonding.rocessing of both these materials was also done on 3003 Al-H14ase plate. No external heating was provided in any of these cases.

6061 Al-H18 31 5.6 35.511 000 Cu-Hard 36 6.7 30.5

3. Results and discussion

3.1. Examples of thermal profiles

Fig. 4. Typical thermal profiles from embedding and build passes (start times arenot the same). A shoulder is seen during embedding of thermocouple.

M.R. Sriraman et al. / Journal of Materials Proce

Fpm

s(wafmawt(tttct

3

picptaatiipiwaom6hmrgU

rev1

ig. 5. Comparison of peak temperatures in different materials under similarrocessing conditions. Higher temperature increase is seen with higher strengthaterials.

een before it rapidly rose to peak value (see Fig. 4). This humpshoulder) is probably an effect of vibrations induced in the tape onhich the thermocouple is placed for embedding as the sonotrode

pproaches it. Such vibrations could lead to some heating due toriction between the (un-embedded) thermocouple and the tape. In

easurements from subsequent build passes, such a shoulder wasbsent (see Fig. 4). A rise in temperature to peak value is expectedhen the sonotrode is right on top of the thermocouple. The peak

emperature recorded is usually the highest during its embeddingsee Fig. 4). However, this was not always the case, and is attributedo variations in contact area under the sonotrode and between theapes at any instant. It is also seen from Fig. 4 that the duration ofhe transient is less than 0.5 s. Although this could depend upon theontact area between tapes and the travel speed, it was found to beypically of this order only.

.2. Effect of material processed on temperature rise

The peak temperatures measured in 3003 Al-H18 for tworocessing parameters are shown in Fig. 5. Measurements made

n 6061 Al-H18 and 11 000 Cu-Hard under similar processingonditions are also shown in the same figure (see Fig. 5). The tem-eratures in 6061 Al-H18 and 11 000 Cu-Hard are clearly higherhan in 3003 Al and consistently observed for both embeddingnd subsequent build passes (see Fig. 5). Given that 6061 Al-H18nd 11 000 Cu-Hard have higher yield strength (or shear strength)han 3003 Al-H18, this denotes that the temperature generateds dependent on the material shear strength. In other words, thiss evidence of plastic deformation heating. The involvement oflastic deformation heating (in addition to frictional heating)

n ultrasonic metal welding/UAM has been noted earlier, in theorks of Yadav (2001) on 1100 Al, de Vries (2004) and Siddiq

nd Ghassemieh (2008) on 6061 Al, and Zhang and Li (2009)n 3003 Al, but mainly through modeling. This has also beenentioned in the research of Mariani and Ghassemieh (2010) on

061 Al. It is perhaps for the first time that such a phenomenonas been experimentally verified as shown above. Such heatingust have contributed to interfacial DRX observed in the past

esearch of the authors. However, the earlier works have sug-ested frictional heating to be the dominant mechanism in theAM process.

In order to rationalize the correlation seen here between mate-

ial shear strengths and corresponding temperature rise, a simplestimation of the temperature rise was sought based on the con-ersion of plastic work into heat under adiabatic conditions (Dieter,986). With the VHP UAM process involving dynamic shearing

ssing Technology 211 (2011) 1650– 1657 1653

between layers, the corresponding equation from Dieter (1986) canbe written in the following form:

�T = �a�aˇ

�C(1)

where �T here represents the temperature rise occurring everycycle over a given volume of a material, �a and �a are the “shearstress amplitude” and “strain amplitude”, respectively, ̌ is the frac-tion of plastic work converted into heat, � is the density, and C isthe specific heat capacity of the material.

From the above equation, it is obvious that �T is proportionalto the material strength. Based on the hypothesis that bondingbetween tapes is essentially the formation of a series of microwelds at the contacting asperities between them, the strain ratesexperienced at these asperities can be of the order of 104–105/s(Sriraman et al., 2010a). With such rapid dynamic straining, both�a and �a are expected to change almost every cycle during theweld period (“dwell time”) due to both, the Bauschinger effect andthe effect of a temperature rise from the previous cycle (Dieter,1986). This could result in the magnitude of �T also changingwith every cycle, with an accompanying change in microstructureover the particular region. However, for the purpose of estimatingthe temperature rise here, during VHP UAM of Al alloys and Cubased on the above phenomenological model, both �a and �a wereassumed to be constant. They were given as:

�a = 2�0 (2)

�a = A

h(3)

where �0 is the shear strength of the material (asperity) at roomtemperature, A is the total shear (lateral) displacement (equal tothe vibration amplitude imposed) of the “ridge” of the asperity andh is the height of the ridge with respect to the corresponding “val-ley”. Considering the asperities on the tapes to be an imprint of thesonotrode surface texture of Ra = 7 �m (Sriraman et al., 2010a), thetotal height h was taken to be 14 �m. Although the fraction of plas-tic work dissipated as heat ̌ could depend upon the plastic strainsand strain rates involved in the deformation process (Ravichandranet al., 2002), a constant value of 0.95 (from Dieter, 1986) has beenassumed for all materials. Substituting for the above and appropri-ate values of � and C (from Brandes and Brook, 1999), �a (from Rooy,2010 and Robinson, 2010 for Al alloys and Cu, respectively) and �a

in Eq. (1), estimates of �T were made for all materials/processingconditions. These are presented in Table 3. It must be mentionedthat since data for 6061 in H18 temper were not readily availablefrom literature, the shear strength was determined based on thematerial hardness and comparing it with the hardness of 3003Al-H18. The strengths of 3003 Al-H18 and 6061 Al-H18 wereassumed to be in the same proportion as their hardness.

It is seen from Table 3 that the estimated �T values/peaktemperatures also increase with increasing shear strength of thematerial as was observed with measured temperatures. It is alsonoted (see Table 3) that the estimated �T values are significantlyhigher than the measured temperatures. This could either be aneffect of overestimated temperature rise or a lower measured tem-perature. Given that the estimated homologous temperatures (seeTable 3) fall into the regime of DRX, especially in Al alloys (Prasadand Ravichandran, 1991), the former is unlikely. The measuredtemperatures therefore, must have been lower. As mentionedearlier, this was to be expected. The temperature picked up by thethermocouple could well be an average integrated value (to a reso-lution of 0.1 ms) from a “larger” volume of material (corresponding

to the thermocouple wire diameter). [Heating in VHP UAM, on thecontrary, is expected to be highly localized occurring at the asperitylevel (order of few microns) and varying at 5e−5 s.] During suchtemperature integration, heat would be expected to diffuse away

1654 M.R. Sriraman et al. / Journal of Materials Processing Technology 211 (2011) 1650– 1657

Table 3Estimated and measured temperatures (from embedding passes) in different materials.

Material Processing condition Estimated Measured

�T (K) Tpa (K) Tp/Tm

b �T (K)c Tp/Tmb

3003 Al-H18 31 �m–5.6 kN–35.5 mm/s 459 479 0.51 349 0.406061 Al-H18 31 �m–5.6 kN–35.5 mm/s 484 504 0.54 367 0.423003 Al-H18 36 �m–6.7 kN–30.5 mm/s 488 508 0.54 387 0.4411 000 Cu-Hard 36 �m–6.7 kN–30.5 mm/s 553 573 0.42 420 0.32

a Assuming RT = 293 K.

daTo

mTlbwsgttttdeTtt

ibwlesc“istvstustte�itaimn(cnfie

Fig. 6. Estimated temperature rise vs. measured temperature rise (based on embed-ding passes) in different materials showing a correlation between them. Labels 1 and3 correspond to 3003 Al-H18 processed at vibration amplitudes of 31 �m and 36 �m,

b From Brandes and Brook (1999).c From peak temperatures and assuming RT = 293 K.

ue to conduction resulting in lower readings. The observation of much lower measured homologous temperature in copper (seeable 3), whose thermal conductivity is 1.5 times higher than thatf aluminum (Brandes and Brook, 1999), supports this hypothesis.

With regard to the temperature estimates, although the esti-ated homologous temperatures for Al alloys look acceptable (see

able 3), the values for copper (see Table 3) seem to be on theower side (based on Prasad and Ravichandran, 1991). This coulde related to the shear strains in Cu being actually higher thanhat has been assumed [based on Eq. (3)]. Copper, with its lower

tacking fault energy than aluminum (Dieter, 1986), can exhibitreater localized plastic flow (Sriraman et al., 2010a) comparedo Al, for the same applied vibration amplitudes. This is probablyhe reason for the model [based on Eq. (1)] predicting a loweremperature rise. It is however interesting to note that despitehe “underestimation” of strain, and with the ‘�C’ term [in theenominator of Eq. (1)] for Cu being higher than for Al, the �Tstimate is still higher in Cu by virtue of its higher shear strength.his, in a sense, validates the approach of the authors here inrying to understand the temperature rise in VHP UAM based onhe model of plastic work conversion to heat.

The estimated �T values presented however are only approx-mate. They are based on many assumptions, with asperity heighteing one of them. A change in the asperity dimensions consideredould alter the estimated �a and �T values. Based on the imprint

eft by the sonotrode, a variation in asperity height is indeedxpected spatially. It could therefore be argued that multipletrains (and strain rates) are generated during VHP UAM thatould lead to different levels of temperature increase over theweld region” at any given instant (and consequently variationsn microstructures and bonding as well). The material propertiesuch as �a and �a have also been assumed to remain constanthroughout the weld cycle for simplification. Likewise, a constantalue of ̌ = 0.95 has also been assumed for all materials. This wouldtill be a fair assumption to make for a conservative estimate, givenhat the fraction of plastic work converted to heat can almost reachnity for high plastic strains in materials that are not particularlytrain-rate sensitive (Ravichandran et al., 2002) as considered inhis investigation. Such high strains are expected in VHP UAM athe asperity level as determined from Eq. (3) (see also Sriramant al., 2010a). Furthermore, approximations in the estimation ofT also arise due to vibration amplitude being solely considered to

nfluence the shear strains [see Table 3 and Eqs. (1)–(3)]. It is clearhat other parameters namely normal force and travel speed wouldlso have had an effect in inducing a temperature rise (see follow-ng section) and consequently increased the estimated values in all

aterials. However, such effects of normal force and speed couldot be described based on this phenomenological model [see Eqs.1)–(3) and Table 3]. This probably explains why the data point

orresponding to 3003 Al processed at 36 �m amplitude, 6.7 kNormal force, and 30.5 mm/s travel speed (see Table 3) does nott into what otherwise seems to be a linear correlation betweenstimated and measured values (as illustrated in Fig. 6).

respectively. Label 2 pertains to 6061 Al-H18 while label 4 refers to 11 000 Cu-Hard(see Table 3).

3.3. Effect of processing parameters on temperature rise

Fig. 7 shows the effect of vibration amplitude, normal force, andtravel speed (based on Table 1) on temperatures measured in 3003Al-H18. It is observed that an increase in amplitude increases thepeak temperature (Fig. 7a) while an increase in speed decreasesit (Fig. 7c) almost linearly. (The “trendlines” in each of these fig-ures have been drawn based on the embedding pass only.) With anincrease in force, on the other hand, the variation in temperatureappears to follow a parabolic behavior (Fig. 7b). These observationsare consistent between embedding and subsequent build passes. Inthe case of 26 �m amplitude, 5.6 kN normal force, and 25.5 mm/stravel speed however, the measurement could be made only for theembedding pass as the thermocouple broke during the subsequentbuild pass (Fig. 7c).

The effect of amplitude is seen to be the strongest among allthe three parameters (see Fig. 7a). A 40% increase in amplituderaises the temperature by almost an equal proportion (about 36%).This is expected since amplitude is considered to have the great-est influence on the dynamic interfacial plastic shear strains atthe contacting asperities. From Fig. 7a, it is also seen that thereis probably a threshold level to the amplitude. A linear extrap-olation of data to lower amplitudes would suggest an amplitudevalue at which no temperature rise may be induced. Normal force,on the other hand seems to have a relatively modest effect (seeFig. 7b). In the initial stages (3–5 kN) with a 25% increase in force,the temperature increases but only by about 11%. A saturation level

is reached beyond 5 kN. With higher force, the temperature risebecomes smaller. The results suggest that an “optimal” force levelexists to induce maximum temperature rise. This needs some dis-

M.R. Sriraman et al. / Journal of Materials Proce

Fig. 7. Measured peak temperatures in 3003 Al-H18 as a function of processingpa

cwboataFmRtbdiitr

Sojiphan et al., 2010 and Fujii et al., submitted for publication) are

arameters: (a) vibration amplitude, (b) normal force, and (c) travel speed. Vibrationmplitude is seen to have the most marked effect.

ussion. The purpose of a static normal force in UAM (or ultrasonicelding) is to facilitate scrubbing of asperities between layers to

e welded as they move relative to each other under the influencef vibrations. If the force is such that optimal scrubbing conditionsre induced, higher temperatures are attained. Force levels lowerhan this could cause sliding (rather than scrubbing) of the tapesnd hence lower shear strains and temperatures as observed inig. 7b. On the other hand, higher force levels could stifle relativeotion between layers due to high contact stresses (Kukalov and

ack, 2009) causing excessive “sticking” leading to a decrease in theemperature rise as was observed (see Fig. 7b). Such behavior haseen noted in spot welding too (Phillips, 1960). Again, although aecrease in travel speed increases the peak temperature, the effect

s relatively smaller (see Fig. 7c). For a 100% decrease in speed, the

ncrease in temperature is about 18%. The effect of travel speed onemperature rise is also expected since a lower speed would cor-espond to a higher “dwell time” over the “weld region”. This will

ssing Technology 211 (2011) 1650– 1657 1655

increase the number of shear cycles within the region leading togreater “accumulation” of heat and hence a higher temperature rise.

Although plotted separately, it is clear that the processingparameters could mutually interact with each other. The varia-tion observed in �T with respect to amplitude, for instance (seeFig. 7a), is for the chosen force level of 5.6 kN and travel speed of35.5 mm/s. A different force level (for the same speed of 35.5 mm/s,say) is expected to influence the process and the temperature risedifferently for these amplitude regimes (see Fig. 7a) dependingupon whether scrubbing of faying surfaces is promoted or slid-ing/sticking conditions are encountered. Likewise, the change in �Tarising from a variation in speed (for the same force level of 5.6 kN)could also be different, depending upon the operating amplitude.Similar scenarios with regard to the effects of amplitude/speed onthe variation in �T with respect to force (different from that seenin Fig. 7b) and the effects of amplitude/force on the variation in �Twith respect to speed (different from Fig. 7c) are conceivable.

3.4. Implications of the current results

Results presented in earlier sections demonstrate that thermaltransients in VHP UAM are related to the material (strength) andprocessing parameters. Higher temperatures were recorded formaterials with higher shear strength confirming the occurrenceof plastic deformation heating during this process. This result isof significance considering that such a phenomenon in ultrasonicwelding/UAM research has been demonstrated through exper-imentation. The �T estimates are also consistent with plasticdeformation heating. Although past researchers (see Section 3.2)are in general agreement about the occurrence of heating inultrasonic metal welding/additive manufacturing, by both frictionand plastic deformation, it is still an ongoing debate as to whatfraction of the heat comes from frictional work. Yadav (2001)based on his finite element analysis and simulation of thermaltransients in ultrasonic welding of 1100 Al, suggests that the heatis primarily generated by friction and is more than twice thatof heating through plastic work. Zhang and Li (2009), based ontheir finite element model indicate that heat generation due tofriction in ultrasonic welding of 3003 Al-H18 is 5 times larger thanplastic deformation during the initial periods of welding, with thedifference increasing even further with increasing weld times.

While the authors of this paper recognize that both frictionaland deformation heating can and will occur due to relative motionbetween surfaces, it is possible that the former is dominant inthe “initial stages” of the process – involving disruption of theoxide layers between the contacting asperities and the bringingof nascent metal surfaces into contact. It is hypothesized that oncenascent metal–metal contact is established, and interfacial plasticdeformation initiated, there would be a rapid “take over” of themechanism from the classic Coulomb sliding friction conditions tothat of “sticking” (based on Dieter, 1986). Based on the principle ofmetal working (Dieter, 1986), once sticking is established, and norelative motion between the mating surfaces is possible, subsurfacedeformation starts to occur. Such a phenomenon has been noted inhot working (Dieter, 1986) and could exist in VHP UAM as well (seeSriraman et al., 2010a for discussion on the similarity between thetwo processes). The observations of a dynamically recrystallisedmicrostructure spanning over a width of about 20 �m across theinterface region in 3003 Al-H18 builds (Fujii et al., submitted forpublication), and of “interaction zones” (or “thermo-mechanicallyaffected zones”) extending into the bulk of the layers of such buildsas evidenced by microstructural gradients (from the research of

supportive of this theory (of subsurface deformation).Further, based on the analytical formulation given by Dieter

(1986), frictional heating in a material would essentially be

1 Processing Technology 211 (2011) 1650– 1657

intpH

tiIftc

arbta33uRUpTberibafliauTs

aattmaa

bbaThthph3tuempgpb

m

656 M.R. Sriraman et al. / Journal of Materials

nversely related to the ‘�C’ term [with friction coefficients beingearly equal in all materials (from Blau, 2010)]. If such heating washe dominant mechanism all through the process, a lower tem-erature rise should have been noted in Cu compared to Al alloys.owever, this was not the case.

It is therefore postulated that the extent of heating derivedhrough plastic deformation (by the time bonding is accomplishedn VHP UAM) could be significantly larger than frictional heating.t would be interesting to describe the VHP UAM process based onrictional conditions leading to sticking and subsurface deforma-ion conditions beyond. This is however beyond the scope of theurrent paper.

In as far as the effect of processing parameters is concerned,mplitude is seen to have a significant bearing on the temperatureise. A similar effect of amplitude has been noticed with respect toonding as well in UAM. Kong et al. (2004a), based on their peelesting of 3003 Al-H18, found peel loads to increase with increasingmplitudes. Hopkins (2010) from his tensile and shear tests on003 Al-H18 also observed a similar behavior. Again, builds of003 Al-H18 made by VHP UAM were found to have minimal voidsnder higher amplitudes of processing (Sojiphan et al., 2010).esults from preliminary qualitative peel experiments on VHPAM of 3003 Al-H18 have also pointed to better bonding whenrocessing is done at higher amplitudes (Sriraman et al., 2009).he investigations (Sriraman et al., 2009) further revealed pooronding or lack of bonding when relatively low amplitudes weremployed. A similar phenomenon is also suggested in the currentesearch with regard to temperature rise (see Fig. 7a and discussionn Section 3.3). The temperature rise during the process could thuse indicative of the bonding quality. Higher temperatures gener-ted could cause more softening through DRX, enhanced plasticow, and greater degree of bonding. Such a behavior was observed

n ultrasonic spot welding too by Weare et al. (1960) who found correlation between “weld strength” (based on breaking loadnder transverse tension) and temperature increase in 1100 Al.hey noted that a minimum temperature increase at the interfaceeems to be required before significant bonding can occur.

Although vibration amplitude is the primary driver to a temper-ture rise and metallurgical bonding in VHP UAM, its influence in

given material could reach saturation. Under amplitudes morehan what is “required” for bonding, there could be a tendencyowards greater heat dissipation into the bulk of the tapes. Some

icrostructural observations indicative of this phenomenon havelready been made as part of the overall research of the currentuthors (Sojiphan et al., 2010).

With plastic deformation heating (under adiabatic conditions)elieved to be the governing mechanism for temperature rise,onding of higher strength materials by VHP UAM would requiredequate plastic shear strains/strain rates to be induced in them.his would mean the use of increased amplitudes supported byigher force levels (than what has been used in this study) inheir processing. In other words, the higher the material strength,igher would be the ultrasonic energy required through therocessing parameters. (It was seen that processing of copperad to be done under higher amplitude and force levels than003/6061 aluminum.) The phenomenal stages to bonding canhus be represented by a schematic as shown in Fig. 8. If requiredltrasonic energies are not available to process the material,xternal heating could be resorted to, which would help lower theaterial strength facilitating easier processing. With the asperities

laying a crucial role in bonding, it is also conceivable that asperityeometries be modified to enhance plastic shear strains and thus

romote bonding. Such surface modification has been found to beeneficial in roll bonding (Liu et al., 2008).

Although results from the current paper using Type K ther-ocouples have clearly demonstrated the effect of material/

Fig. 8. Phenomenological stages to bonding in VHP UAM.

processing parameters on thermal transients in VHP UAM, itwould be worthwhile to see if these trends are corroborated byanother measurement procedure. IR thermography, a non-contacttechnique, could be attempted as an alternative method. Suchmeasurements would provide information about surface ther-mal fields only (which again may require calibration), but thedifferences between materials/processing conditions may still berevealed. Capturing the actual magnitude of the temperature riseeven by this method would however remain a challenge.

4. Conclusions

Thermal transients measured during VHP UAM show higherpeak temperatures in high-strength copper compared to low-strength aluminum alloys under similar processing conditions. Thisis indicative of a direct correlation between plastic deformationheating and interfacial temperature rise. Occurrence of dynamicrecrystallisation associated with bonding is attributed predomi-nantly to such heating. Vibration amplitude has the strongest effectin inducing higher temperatures, due to enhanced dynamic plasticshear strains at the asperities with increasing amplitudes. Basedon the observations, bonding in a given material could be relatedto the temperature rise induced in it.

Acknowledgements

The authors sincerely thank the Ohio Department of Develop-ment for funding this research work through the Third FrontierWright Projects Program. The support received from Dr. Karl Graff,Matt Short, Mark Norfolk, Paul Boulware and other colleagues fromEWI is greatly appreciated. Kind assistance from the staff/studentsof OSU is also gratefully acknowledged.

References

Blau, P.J., 2010. Friction, Lubrication and Wear Technology, ASM Handbooks Online,vol. 18. ASM International, http://products.asminternational.org/hbk/index.jsp.

Boulware, P., 2010. Edison Welding Institute, private communication.

Proce

B

C

d

DF

GH

H

J

K

K

K

L

M

PP

R

M.R. Sriraman et al. / Journal of Materials

randes, E.A., Brook, G.B., 1999. Smithells Metals Reference Book, seventh ed.Butterworth–Heinemann, Oxford.

heng, X., Li, X., 2007. Investigation of heat generation in ultrasonic metal weldingusing micro sensor arrays. Journal of Micromechanics and Microengineering 17,273–282.

e Vries, E., 2004. Mechanics and mechanisms of ultrasonic metal welding. Ph.D.Thesis. The Ohio State University, Columbus, p. 175.

ieter, G.E., 1986. Mechanical Metallurgy, third ed. McGraw-Hill, New York.ujii, H.T., Sriraman, M.R., Babu, S.S. Quantitative evaluation of bulk and interface

microstructures in Al-3003 alloy builds made by very high power ultrasonicadditive manufacturing. Metallurgical and Materials Transactions A, submittedfor publication.

raff, K., 2010. Edison Welding Institute, private communication.azlett, T.H., Ambekar, S.M., 1970. Additional studies of interface temperature and

bonding mechanisms of ultrasonic welds. Welding Journal 49, 196-s–200-s.opkins, C.D., 2010. Development and characterization of optimum process param-

eters for metallic composites made by ultrasonic consolidation. M.S. Thesis. TheOhio State University, Columbus, p. 90.

anaki Ram, G.D., Robinson, C., Yang, Y., Stucker, B.E., 2007. Use of ultrasonic consol-idation for fabrication of multi-material structures. Rapid Prototyping Journal13, 226–235.

ong, C.Y., Soar, R.C., Dickens, P.M., 2004a. Optimum process parameters for ultra-sonic consolidation of 3003 Al. Journal of Materials Processing Technology 146,181–187.

ong, C.Y., Soar, R.C., Dickens, P.M., 2004b. Ultrasonic consolidation for embeddingSMA fibers within aluminum matrices. Composite Structures 66, 421–427.

ukalov, M., Rack, H.J., 2009. Control of 3003-H18 aluminum ultrasonic consolida-tion. Journal of Engineering Materials and Technology 131, 021006-1–021006-6.

iu, J., Li, M., Sheu, S., Karabin, M.E., Schultz, R.W., 2008. Macro and micro-surfaceengineering to improve hot roll bonding of aluminum plate and sheet. MaterialsScience and Engineering A 479, 45–57.

ariani, E., Ghassemieh, E., 2010. Microstructure evolution of 6061 O Al alloy duringultrasonic consolidation: an insight from electron backscatter diffraction. ActaMaterialia 58, 2492–2503.

hillips, A.L., 1960. Ultrasonic Welding. American Welding Society, New York, p. 12.rasad, Y.V.R.K., Ravichandran, N., 1991. Effect of stacking fault energy on the

dynamic recrystallisation during hot working of FCC metals: a study using pro-cessing maps. Bulletin of Materials Science 14, 1241–1248.

avichandran, G., Rosakis, A.J., Hodowany, J., Rosakis, P., 2002. On the con-

version of plastic work into heat during high-strain rate deformation.In: Furnish, M.D., Thadhani, N.N., Horie, Y. (Eds.), Proceedings of theTwelfth International Conference of the American-Physical-Society Topical-Group-on-Shock-Compression-of-Condensed-Matter, vol. 620, pp. 557–562,Atlanta, USA.

ssing Technology 211 (2011) 1650– 1657 1657

Robinson, P., 2010. Properties and Selection: Nonferrous Alloys and Special-Purpose Materials, ASM Handbooks Online, vol. 2. ASM International,http://products.asminternational.org/hbk/index.jsp.

Rooy, E.L., 2010. Properties and Selection: Nonferrous Alloys and Special-Purpose Materials, ASM Handbooks Online, vol. 2. ASM International,http://products.asminternational.org/hbk/index.jsp.

Schick, D.E., Hahnlen, R.M., Dehoff, R., Collins, P., Babu, S.S., Dapino, M.J., Lippold,J.C., 2010. Microstructural characterisation of bonding interfaces in aluminum3003 blocks fabricated by ultrasonic additive manufacturing. Welding Journal89, 105-s–115-s.

Schick, D.E., Babu, S.S., Foster, D., Short, M., Dapino, M.J., Lippold, J.C. Transient ther-mal response in ultrasonic additive manufacturing of aluminum 3003. RapidPrototyping Journal, in press.

Siddiq, A., Ghassemieh, E., 2008. Thermomechanical analyses of ultrasonic weldingprocess using thermal and acoustic softening effects. Mechanics of Materials 40,982–1000.

Sojiphan, K., Sriraman, M.R., Babu, S.S., 2010. Stability of microstructure in Al 3003builds made by very high power ultrasonic additive manufacturing. In: Bourell,D.L., Crawford, R.H., Seepersad, C.C., Beaman, J.J., Marcus, H. (Eds.), Proceedingsof the Twenty First International Solid Freeform Fabrication Symposium. Austin,USA, pp. 362–371.

Sriraman, M.R., Babu, S.S., Short, M., 2010a. Bonding characteristics during veryhigh power ultrasonic additive manufacturing of copper. Scripta Materialia 62,560–563.

Sriraman, M.R., Fujii, H., Gonser, M., Babu, S.S., Short, M., 2010b. Bond characteri-zation in very high power ultrasonic additive manufacturing. In: Bourell, D.L.,Crawford, R.H., Seepersad, C.C., Beaman, J.J., Marcus, H. (Eds.), Proceedings of theTwenty First International Solid Freeform Fabrication Symposium. Austin, USA,pp. 372–382.

Sriraman, M.R., Babu, S.S., Short, M., Graff, K., 2009. Characterization of bonds in veryhigh power ultrasonic additive manufacturing. In: Proceedings of the SecondSymposium on Ultrasonic Additive Manufacturing Technology , Columbus, pp.E-68–E-72.

Weare, N.E., Antonevich, J.N., Monroe, R.E., 1960. Fundamental studies of ultrasonicwelding. Welding Journal 39, 331-s–341-s.

White, D.R., 2003. Ultrasonic consolidation of aluminum tooling. Advanced Materialsand Processes, 64–65.

Yadav, S., 2001. Thermo-mechanical analysis of ultrasonic metal welding for rapidmanufacturing. M.S. Thesis. Tufts University, Medford, pp. 44–45.

Yang, Y., Janaki Ram, G.D., Stucker, B.E., 2009. Bond formation and fiber embedmentduring ultrasonic consolidation. Journal of Materials Processing Technology 209,4915–4924.

Zhang, C.(S.), Li, L., 2009. A coupled thermal–mechanical analysis of ultrasonic bond-ing mechanism. Metallurgical and Materials Transactions B 40, 196–207.