See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/254063340

Mechanisms of Spiking and Humping in Keyhole

Welding

Article in IEEE Transactions on Components, Packaging, and Manufacturing Technology · March 2012
DOI: 10.1109/TCPMT.2011.2178412

CITATIONS READS

12 170

4 authors, including:

P. S. Wei Je Ku
National Sun Yat-sen University Universität Paderborn
103 PUBLICATIONS 884 CITATIONS 5 PUBLICATIONS 47 CITATIONS

SEE PROFILE SEE PROFILE

Tarasankar Debroy
Pennsylvania State University
282 PUBLICATIONS 7,579 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: P. S. Wei
letting you access and read them immediately. Retrieved on: 02 September 2016
IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012 383

Mechanisms of Spiking and Humping

in Keyhole Welding
P. S. Wei, Senior Member, IEEE, K. C. Chuang, J. S. Ku, and T. DebRoy

Abstract— The effects of power density, concentration of defects such as humping and spiking often occur in keyhole
volatile alloying element (magnesium, Mg), and welding speed on mode welding [11]–[14]. Humping is also a common defect
the mechanisms of fusion-zone defects, i.e., spiking and humping in arc-welding processes [15]–[19]. Spiking represents unex-
(and coarse rippling) during keyhole mode electron-beam welding
of Al 6061, Al 5083, and SS 304 are investigated experimentally. pected increase in the local penetration in periodic successions.
Spiking represents a sudden increase in penetration beyond the The instability of the keyhole leads to the formation of gas
average penetration line. Rippling exhibits rather regular, arc- bubbles that are often trapped within the weld metal when the
shaped topographic features on a solidified surface, whereas fluid freezes before the bubbles can escape [12], [20], [21].
humping shows an irregular surface contour consisting of a This gives rise to porosity and cold shuts, often near the root
series of swelled protuberance. These defects seriously reduce the
properties and strength of the joint. The quantitative variations of the weld. In view of practical and academic importance,
of humping, coarse rippling, and spiking defects with the beam- mechanisms of humping and spiking have been studied in the
focusing characteristics, volatile element, and scanning speed are past [11]–[21]. However, no unified mechanistic understanding
quite limited in the literature. The experiments in this paper for their formation has emerged from the previous work.
confirm that average pitch of the humps or coarse ripples is Different mechanisms of spiking have been proposed in the
approximately identical to that of the spikes. Thus, the frequency
of spiking can be determined from the observation of the weld literature. Tong and Giedt [21], [22] proposed from their X-ray
surface. The results show that the average pitches and amplitudes studies that spiking and humping (and rippling) formed from
of humping or coarse rippling and spiking increase with decreas- the oscillation of the keyhole were produced during keyhole
ing welding speed and increasing content of volatile element Mg mode electron-beam welding. They proposed that when the
from Al 5083. The frequency and amplitude of spiking, however, electron-beam struck the surface of the molten pool, the
are increased by lowering the focal-spot location. The measured
amplitude and frequency of spiking and humping and fusion- intense heating vaporized the material and the recoil pressure
zone depth are confirmed from good agreement with available exerted a downward force on the melt. This led the cavity to
scaled analysis. This paper provides quantitative results useful for grow latterly near the base, but constrict at a higher location.
understanding mechanisms of these defects, so that preventing The melt eventually rolled over or slid into the cavity, leaving a
spiking and humping during keyhole mode welding becomes large melt region beneath the electron beam and through which
possible.
the electron beam must once again penetrate. The severity of
Index Terms— Humping, rippling, spiking, weld defects. spiking and rippling thus depended on the frequency of the
oscillation of the cavity, welding speed, and the material to be
I. I NTRODUCTION welded.
Armstrong [23] suggested that spiking was determined by
K EYHOLE mode welding techniques offer signifi-
cant advantages in the construction of optoelec-
tronic packages involving optical transmission devices, laser
power density and beam power. By raising the focal spot above
the workpiece surface, the root became smoother and the weld
diodes, array–array and 3-D microelectronics, and bio- penetration decreased. Spiking seemed to be a direct function
microelectromechanical systems (MEMS) [1]–[10]. This of high concentration, and any factor that lowered the power
construction method provides reliable joining strength with density would reduce spiking as well as penetration. However,
high-speed and high-volume production. However, serious in their experiments, controlling the focus level was not
reproducible enough to be used as a reliable spiking control.
Manuscript received June 15, 2010; revised December 28, 2010; accepted The fluctuation of energy density caused by the collisions of
November 26, 2011. Date of publication January 12, 2012; date of current
version February 28, 2012. Recommended for publication by Associate Editor the electrons with ionized metal vapor can result in spiking.
S. K. Mondal upon evaluation of reviewers’ comments. Weber et al. [24] used a pinhole camera to observe
P. S. Wei and K. C. Chuang are with the Department of Mechanical and the process of electron-beam penetration. The impact point
Electro-Mechanical Engineering, National Sun Yat-Sen University, Kaohsiung
80424, Taiwan (e-mail: [email protected]; [email protected]). between the beam and metal was recorded from X-ray expo-
J. S. Ku is with the Institute of Materials Science and Engineer- sures. The beam struck the cavity periodically from the base
ing, National Sun Yat-Sen University, Kaohsiung 80424, Taiwan (e-mail: to half way up the cavity wall near the surface in a few
[email protected]).
T. DebRoy is with the Department of Materials Science and Engineering, milliseconds. When the beam struck the cavity base, the liquid
Pennsylvania State University, University Park, PA 16802 USA (e-mail: at the base was heated and pushed up the cavity wall. As the
[email protected]). liquid layer began to thicken, it ceased the upward movement
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. and finally fell back into the cavity as a closure. The beam
Digital Object Identifier 10.1109/TCPMT.2011.2178412 was intersected. Instead of being heated, lower portions of the
2156–3950/$31.00 © 2012 IEEE
384 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012

cavity cooled and even solidified. Spiking resulted when the arc welding. Humping was affected by the pool flow composed
closure-fallback mechanism failed so that the electron beam of three sreams, which were one from the filler metal and two
dwelled on the cavity base. from the base metal flowing around the sides of the pool. A
Arata et al. [25] used a cine-fluoroscopic technique to clearer video observation of humping in gas metal arc welding
study dynamic behaviors of the electron-beam welding cavity. was provided by Nguyen et al. [36]. Humping is periodic in
Spiking formed from the breakdown of force balance between nature. A portion of the weld connected the previous swelling
vapor pressure, hydrostatic pressure, and dynamic force of and the new tail of the weld pool. This portion of the weld
liquid layer. Spiking thus was closely related to the movement was part of the wall jet that supplied the swelling with molten
of wall-caves and wall-knobs taking place on the cavity wall. weld metal. At the tail of the weld pool, the molten weld metal
Schauer and Giedt [26] proposed a length ratio S H / h to accumulated to form a new swelling. Rapid solidification of
predict the tendency of spiking. The length H represents the wall jet choked off the flow of molten metal to the swelling.
the distance between the cavity base and the location where Humping increased in height as the solidification of the molten
surface tension and vapor pressure becomes equilibrium, deter- weld metal proceeded upward. Fluid flow and heat transfer
mined by the measured cavity temperature [27]. The length h for humping during a pulsed gas metal arc welding at a high
is the keyhole depth. Spiking did not occur when the parameter travel speed were numerically and quantitatively simulated
S was less than 0.3. Spiking occurred infrequently for S = 0.5 by Cho and Farson [37]. Two mechanisms responsible to
and became serious when S was greater than 0.5. the hump formation were identified. The first requirement
Tsukamoto and Irie [28] observed spiking phenomenon, was for a thin liquid channel generated by elongation of the
finding that as the focal spot was located at the cavity base molten metal deposit. The high transverse curvature associated
spikes readily took place. Melting and spiking were dominated with a narrow, thin channel caused a strong surface tension
by the periodic flow of the molten metal. Soon after the normal force that prevented backfilling of metal from any
periodic ejection of the molten metal [29]–[33], the beam accumulation at the back of the pool. The other requirement
cavity was clear of the molten metal and the electron beam for hump formation was pinching and rapid solidification of
was free to drill the bottom causing the spike. Wei and Ho [34] the thin channel, dividing the molten pool into front and rear
proposed that spiking in high-power-density beam welding sections. High-speed video observation images of the humping
resulted from the position of the focal spot with respect to the during keyhole mode laser-beam welding were also shown
cavity base in a shape of a paraboloid of revolution. Because by Kawahito et al. [38]. A humping was gradually formed
incident and reflected angles of an incident energy ray are at the rear end of the molten pool rather than at the laser-
identical, energy absorbed, which is dominated by specular irradiated position. The humping appeared to be caused by
reflections, can be geometrically shown to be affected by the superimposed production of melt backflow accompanied by
location of the focal spot. For example, when the focal spot laser-induced plume and higher surface tension induced by the
is above the cavity base, the energy absorbed in the central narrow molten pool width. Detailed descriptions of swelling
region is low because this region is irradiated only by the of the pool were also extensively provided by Arata [12] in
direct incident ray. Energy flux absorbed outside the central electron-beam keyhole welding.
region, however, jumps to high values from the additional In this paper, amplitudes and frequencies of spikes, humps,
energy rays after the first specular reflections. On the other and coarse ripples during keyhole mode electron-beam weld-
hand, when the focal spot is below the base, the energy flux ing for different beam-focusing characteristics, scanning
absorbed was similar to a Gaussian distribution. Depending on speeds, and concentration of volatile alloying elements are
the type of focusing, the central region thus can experience a measured. Understanding the mechanisms of the defects from
vast difference in the intensity. As a consequence, strong and successful comparisons between the measured and the analyt-
periodic melting can occur at the cavity base due to relative ical results is essentially required to avoid their occurrences
motion between the focal spot and the cavity base. in not only keyhole welding but also various arc welding or
Arata et al. [35] interestingly observed that the average pitch surface melting processes.
of the spikes was identical to that of humps (or coarse ripples)
in keyhole mode electron-beam welding. The mechanism pro- II. E XPERIMENTAL S ETUP AND M EASUREMENTS
posed by Tong and Giedt [21], [22], as discussed previously, In this paper, welding is conducted with electron beam
can be used for an interpretation of this observation. The due to its less disturbance and contamination in the vac-
periods of strong surface oscillations of the molten metal were uum chamber, and concentrated and steady energy flux. The
responsible for formations of both spiking and humping. The major components in the experimental setup are: 1) electron-
variation of frequencies of spikes and humps were from 2 to beam welder (Torvac CVE63B Cambridge Vacuum Eng.,
4 per mm. It was also found that the pitch of humps changed Cambridge, U.K.); 2) grinding/polishing machine (RotoPol-
a little by changing scanning speed, welding current and heat 22, Struers, Denmark); 3) abrasive cutting machine (ACM250,
input of electron beam. To understand the spiking mechanism, PlusOver Company, Taiwan); 4) digital camera (Fujifilm
a simultaneous study of humping thus is valuable. FinePix S602); and 5) ultrasonic cleaner (Branson 1210R-
Mechanism of humping was first studied by Bradstreet [15] MT, Energy Beam Sciences, Agawam, MA). Experimental
using a high-speed camera to observe humping phenomena, procedure is as follows.
associated with formation of a cavity at the front of the weld 1) Electron-Beam Welding: As illustrated in Fig. 1(a), a
and flow from the front to the rear of the pool in gas metal bead-on-plate weld was made in the vacuum chamber
WEI et al.: MECHANISMS OF SPIKING AND HUMPING IN KEYHOLE WELDING 385

TABLE I
N OMINAL C OMPOSITIONS (%) OF A LUMINUM A LLOYS AND
S TAINLESS S TEEL 304

Sample Al Si Cu Mn Mg Cr
Electron
Al 6061 97.9 0.6 0.28 0.15 1.0 0.2
beam
Al 5083 94.7 0.4 0.1 0.7 4.4 0.15
Al 1050 99.5 0.25 0.05 0.05 0.05 –
Al 1100 99.0 1.0(Si + Fe) 0.1 0.05 – –
SS 304 – 1.0 2.00 – 19

U
element Mg. Table I shows the nominal composition of
different alloys used in this paper [39]. It can be seen that
Vacuum chamber volatile element Mg in Al 5083 is 4.4%, which is much
greater than 1% in Al 6061, and 0.05% in Al 1050. The
effects of volatile element Mg on surface roughness and
spiking are revealed by comparing Al 5083 to Al 6061
or Al 1050. Carbon steel and SS 304 were also chosen to
(a) cover different properties of alloys. Each specimen was
cleaned in the ultrasonic cleaner with acetone to remove
grease, and scraped to remove oxides and debris prior
to welding.
3) Fixture: The effects of the focusing characteristics on
Test Focus the fusion-zone shape [11], [12], [40]–[45] can be
Electron beam
specimen Position(f)
investigated by arranging the specimen as a ladder, as
−40 mm
illustrated in Fig. 1(b). The surface where the most
bright spot occurs is referred to as the focal location
f  −20 mm f = 0. The focal-spot size can be considered as the
fusion width on the surface at f = 0. A more accurate
measurement of the beam size is definitely required.
Even though the spot size or the surface focus location is
80 mm f  0 mm
not measured, studying the effects of the focal location
on the fusion-zone shape subject to given scanning speed
f  20 mm and beam power is unaffected. As the electron beam
passes, the fusion zones in different specimens fixed on
Base
the ladder are produced by different focal locations. A
“V”-shaped ladder was also used to examine the effects
f  40 mm
of energy absorption on the fusion-zone shape. Provided
that energy attenuation due to absorption and scattering
250 mm in plasma from the electron gun to the keyhole surface
100 mm is negligible [11], [46], the fusion-zone shapes for the
specimen having the same level of the top surface on
both sides should be nearly the same. Schiller et al. [11]
(b) experimentally and theoretically showed that the sum of
external energy losses encountered with deep welding in
Fig. 1. Schematic sketch of (a) electron-beam welder and (b) fixture of a hard or soft vacuum and negligible scatter, in general,
specimen.
is about 5–10% for “dead-end” seams and 10–15% for
through-welded seams of beam power. External energy
with an electron beam having the maximum accelerating losses included absorption due to electron backscatter,
voltage of 60 kV and welding current of 50 mA. In this emerging vapor steams, emission of thermal or sec-
paper, an accelerating voltage of 55 kV and welding cur- ondary electrons, and X-radiation or heat radiation. The
rent of 20 mA were chosen. The degree of vacuum was observed fusion zones confirmed negligible effects of
maintained to be less than 4 ×10−5 torr (5.2 ×10−3 Pa). energy attenuation through the vacuum chamber.
The scanning speeds were chosen to be 15, 20, 25, and 4) Observation: After welding, the specimen was cut along
30 mm/s. the transverse and longitudinal planes, which are then
2) Specimens: Workpieces are aluminum alloy Al 6061 of polished by 100–2000 grit emery papers in different
10 mm in thickness, 50 and 100 mm in width and directions. The chemical etchant used to etch alu-
length, respectively. Al 5083 contains significant volatile minum alloys were hydrochloric acid (37.5% HCl and
386 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012

Electron Beam
or spiking tendency can be found to be
σ hs w
∼ (1)
h̄ ησ
where w is the width of the fusion zone, melting efficiency η
represents the fraction of beam power required to heat and melt
f
incoming solid along the joint line [14], [49], [50]. It can be
x evaluated from total energy balance in a triangular slab model
1
Solid nr proposed by Hashimoto and Matsuda [51]. They accounted
y for energy required for heating, melting, and evaporating the
h U incoming solid and weighting heat conduction transverse to
the welding direction using the simplified point source model.
Liquid Evaluating heat conduction in the scanning direction more
accurately, a modified melting efficiency is found to be
1
hs η=   2  (2)
w

1 + c2 1 + 2h̄
1
Pe + c1 Ste
Ste+1
Fig. 2. Sketch of physical model and coordinates. where Stefan number Ste ≡ c p (Tm − T∞ )/L, Peclet number
Pe ≡ U w/αs , c p is the specific heat at constant pressure, Tm
and T∞ are the melting and ambient or preheat temperatures
62.5% H2 O). Width, depth, spikes, ripples, or humps of respectively, L is the latent heat for solid–liquid phase transi-
the fusion zone were photographed by a digital cam- tion, αs is the solid thermal diffusivity, and empirical constants
era. Regardless of inherent oscillations of the keyhole c1 ≈ 0.5 and c2 ≈ 4 are selected by fitting measured depths
leading to vague distinctions, more rigorous definitions of the fusion zones, respectively [51]. Stefan number repre-
among coarse and fine ripples and humps are needed. It sents sensible heat-to-latent-heat ratio, whereas Peclet number
seems relevant to define the difference between coarse stands for welding speed-to-diffusion-speed ratio. The two
ripples and humps by the ratio of the pitch and fusion- terms in the parenthesis in the denominator of (2) represent
zone width. Humps in this paper are considered to be the the transverse heat conduction weighting between stationary
humping pitch-to-fusion zone width ratio around unity, and moving heat source. The square root is  referred to as the
whereas the ripples are for their pitch much less than side length of the fusion boundary. Because 1 + (w/2h̄)2 =

the fusion-zone width. w h̄ + (w/2) /(h̄w), the melting efficiency decreases with
2 2
increasing the ratio between the area of the fusion boundary
and transverse cross-sectional area of the fusion zone. It means
III. E STIMATION OF S PIKING A MPLITUDE
that an increase in the side area of the fusion boundary
Spiking is a consequence of periodic melting in the depth enhances conduction in transverse directions and reduces
direction, as schematically sketched in Fig. 2. The welding energy for melting the incoming solid in the welding direction.
is conducted by a focused electron beam having a beam In view of energy transfer in directions other than the scanning
power Q, scanning speed U , energy distribution parameter direction, melting efficiency is always less than unity. Melting
σ , and focal location f measured from the top surface in efficiency increases with Peclet number. Provided that welding
the direction toward the electrode. In most cases of keyhole speed or Peclet number Pe → 0, melting efficiency, η → 0 for
welding, the energy distribution parameter defines the region any given material. On the other hand, as Pe → ∞ melting
in which 95% of the total energy is deposited. The amplitude efficiency reaches a maximum asymptotic value. Therefore, we
of a spike is h s , whereas the average depth of the fusion zone can increase the Peclet number and reduce the Stefan number
is h̄. Incident flux should be balanced by heating and melting to enhance melting efficiency to reduce spiking tendency.
near the keyhole base in the vertical direction at the moment
when the incident energy impinging on the base. The depth IV. H UMPING OR R IPPLING A MPLITUDE
of a spike is proportional to the melting distance in a time Morphology of a free surface is determined by Young–
scale, which equals the energy distribution parameter divided Laplace equation governing interfacial normal stress balance
by welding speed [47]. Provided that the energy distribution between gas, liquid, and capillary pressures
parameter is 1 mm and welding speed 10 mm/s, time scale
for spiking or welding can be estimated to be 0.1 s, which ∂ 2ζ
p g = p + γ (3)
is of the same order of magnitude for that of the down- ∂x2
ward movement of wall-caves and wall-knobs with speeds of where pg and p are gas and liquid pressures, respectively.
20–30 cm/s in a keyhole depth of around 1 cm, as observed The last term is capillary pressure in which γ is the surface
by Arata et al. [25]. Modifying the equation from Giedt [47] tension, ζ and x are the free surface deformation toward the
by introducing a melting efficiency and energy for melting gas phase and coordinate along the opposite welding direction,
[48], the spiking amplitude-to-average fusion-zone depth ratio respectively. Bernoulli’s equation governing conservation of
WEI et al.: MECHANISMS OF SPIKING AND HUMPING IN KEYHOLE WELDING 387

mechanical energy is needed to determine the difference in along the joint plane. The decrease in the beam power-to-
pressures along the free surface [19] welding speed ratio thus decreases the transverse area, wh,
of the fusion zone. Absorbed energy flux by the keyhole wall
pe − pc = (1 − K loss)ρu 2c (4) is also strongly affected by the focusing characteristics and
where pe and pc are liquid pressures near the edge and center specular reflections by the wall, and energy escaped through
of the pool, respectively, ρ is the liquid density, and u c is the the keyhole opening to the surroundings [11], [12], as modeled
surface speed near the center of the pool. The loss coefficient from the exponential term on the left-hand side. That is,
K loss accounts for energy loss when the surface flow impinges absorbed energy is high, if the focal spot is near and below the
on the solidification front. Evaluating (3) near the center and base [34]. Absorbed energy decreases by raising the focal spot
edge of the pool, subtracting, and scaling give above the base. The value of correlation constant b1 is chosen
to 0.015 to fit experimental data for the fusion-zone depth. The
a
p e − p c ∼ γm 2 (5) effect of volatile alloying elements is determined by comparing
w welding of aluminum alloy Al 6061 with that of Al 5083
where deformation of the free surface (ζe − ζc ) can be scaled that contains significant concentration of volatile magnesium.
by the average amplitude of humps, a, within a spatial extent In view of high vapor pressure in the presence of volatile
scaled by the width of the molten pool. Combining (4) and elements, temperatures near the keyhole base can be lowered
(5) gives by several hundreds of Kelvin [27], [42]. This is attributed to
aγm  u w 2
= (1 − K )
c
(6) satisfaction of Young–Laplace equation near the base, see (3),
loss
ρα 2 α where gas pressure is an exponential function of temperature.
where α and γm represent thermal diffusivity, surface tension Energy conservation between incident flux and heat conduc-
at the melting temperature of the molten pool, respectively. tion thus significantly decreases the base temperature and
Equation (6) is first derived by Wei et al. [52] to determine any increases the fusion-zone depth, as experimentally and theoret-
shape of solidified surfaces including fine and coarse rippling ically confirmed by Schauer et al. [27] and Wei and Chow [42].
and humping, provided that liquid surface velocity can be Rippling and spiking of the fusion zone in Al 6061 in
found. Accounting for the effects of keyhole, the enhanced electron-beam welding with an under-focused focal location
liquid velocity in (6) can be expressed as of −40 mm and scanning speed of 15 mm/s are photographed
and shown in Fig. 3(a). The left and right photos show
 n
 1 ripples observed on the top surface and spikes along the joint
h̄ dγ q 2 3
uc = νw (7) plane, respectively. Comparing the ruler in photos, the top
w d T μk photo shows that there exist coarse ripples in an average
where index n is a positive empirical constant chosen to fit pitch of around 1 mm covered by fine ripples. Distinction
experimental data, q, dγ /d T , k, μ, and ν are, respectively, between coarse and fine ripples in the photo, unfortunately, is
incident flux, surface tension coefficient, thermal conductivity, not quite clear in this case. Spiking also exhibits irregularly
dynamic and kinematic viscosities of the liquid. Higher values periodic. The averaged frequency of coarse ripples, however,
of the index n reflect higher liquid speed, leading to serious is approximately identical to that of spikes, as can also be
humping and spiking. seen later. The measurements in this paper indeed confirm
those of Arata et al. [35]. Based on this principle, average
frequencies of both coarse ripples and spikes are measured and
V. R ESULTS AND D ISCUSSION
checked for comparison to make reliable results in this paper.
The roles of power density, welding speed, and volatile Because the focal-spot location is far below the keyhole base,
alloying elements on spiking and humping (and coarse the mechanisms of spiking formation proposed by Tsukamoto
rippling) defects are experimentally investigated during and Irie [28] and Wei and Ho [34], as described previously,
keyhole mode electron-beam welding. Importance of these are sufficient conditions. Spikes are accompanied by serious
factors can be revealed from conservation of energy through pores and cold shuts in lengths of around 0.5 to 1 mm. The
the cross-section of the melting front width and depth of the fusion zone are, respectively, around
 3 and 8 mm.
| f + h| 
ηηt Q exp −b1 ∼ wh̄Uρ c p (Tm −T∞ ) + L (8) As the focal spot increases to 20 mm below the top surface,
σ
Fig. 3(b) shows that, in addition to a decreased fusion-zone
which indicates that the absorbed energy is used for heating the depth, ripples and spikes become smoother. The decrease in
incoming solid at the ambient temperature to melting point and amplitude of spiking can be attributed to reduction of incident
changing phases from solid to liquid. In order to account for flux. The increased pitch of ripples or spikes can be interpreted
energy attenuation through the surroundings, a welding effi- from enhanced upward thermal convection, and decreased
ciency ηt defined as the percentage of energy absorbed by the oscillation of the keyhole. Referring to Fig. 3(a) and (c) shows
keyhole from the energy emitted from the electron gun is intro- that an increase in welding speed reduces spiking and humping
duced. The beam energy used for superheating and evaporating and the depth and width of the fusion zone. This is due to a
the liquid in the pool can also be taken into accounted through decrease in incident energy per unit length along the joint
welding efficiency. Since the unit of Q/U is J/m, an increase line. Decreases in amplitude and pitch of coarse rippling or
in welding speed indicates a decrease in energy per unit length humping can also be attributed to a decreased liquid speed,
388 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012

f  −40 mm, U  15 mm/s, Al 6061

f  −40 mm, U  15 mm/s, Al 6061

1 mm

1 mm

(a)

f  −20 mm, U  15 mm/s, Al 6061

f  −20 mm, U  15 mm/s, Al 6061

1 mm

1 mm

(b)

f  −40 mm, U  30 mm/s, Al 6061

f  −40 mm, U  30 mm/s, Al 6061

1 mm
1 mm

(c)

Fig. 3. Photographs showing rippling and spiking in welding Al 6061 (a) Ripples and spikes for f = −40 mm, U = 15 mm/s. (b) Ripples and spikes for
f = −20 mm, U = 15 mm/s. (c) Ripples and spikes for f = −40 mm and U = 30 mm/s.

see (6), resulted from a decrease in vapor pressure gradient or which increases the latent heat evolution and decreases the
thermocapillary force from a decreased temperature gradient heat conduction or temperature gradient from liquid to the
[52], [53]. To satisfy conservation of energy at the rear edge solidification front. However, it is contradictory to a well-
of the pool, increased welding speed increases the mass accepted fact that humping increases with welding speed in
flow rate experiencing phase transition from liquid to solid, arc welding [17], [18]. It can be interpreted as the increase
WEI et al.: MECHANISMS OF SPIKING AND HUMPING IN KEYHOLE WELDING 389

f  −20 mm, U  15 mm/s, Al 5083 f  20 mm, U  15 mm/s, Al 5083

1 mm

f  −20 mm, U  15 mm/s, Al 5083

1 mm

(b)

f  −20 mm, U  30 mm/s, Al 5083

1 mm

f  −20 mm, U  30 mm/s, Al 5083

(a)

(c)

Fig. 4. Photographs showing humping and spiking in welding Al 5083 (a) Humps and spikes f = −20 mm, U = 15 mm/s. (b) Spikes for f = 20 mm,
U = 15 mm/s. (c) Humps and spikes for f = −20 mm, U = 30 mm/s.

in incoming mass in the welding direction results in an average frequency [35]. The identical pitch of the humps
increase in humping volume and serious humping. However, and spikes points toward their common origin, namely both
the concavity of the free surface of the molten pool should not occur due to oscillations or instability of the keyhole owing
be large and liquid ejection is small. In keyhole welding, liquid to perturbations in any of the factors that affect the precarious
ejection in the keyhole is significant [29]–[32]. Humping thus balance of forces involving the surface tension, local pressure,
cannot be interpreted from conservation of mass viewpoint. and the gravitational force [12], [18], [21]. The width and
Serious humping and spiking appear in welding Al 5083, depth of the fusion zone are, respectively, about 3 and 12 mm.
as shown in Fig. 4(a). The average pitch of humps is found to The amplitude of the spikes can be about 2 mm, which is even
be around 2.5 mm. Humps are also covered by fine ripples. It 15% of the fusion-zone depth. Fig. 4(b) shows that frequency
is seen that the humps and spikes have also nearly identical and amplitude of spiking, fusion-zone width, and depth are
390 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012

4.0 3.5
Al 6061
3.5
3
3.0
2.5
ns (1/m)×10 −3

2.5

nr (1/m)×10 −3
2.0
2
1.5
1.5
1.0 yx
Al 6061
0.5 Al 5083 1 f  −40 mm
SS 304
f  −20 mm
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 f  0 mm
0.5
f  20 mm
nr (1/m)×10 −3
f  40 mm
0
Fig. 5. Comparison between numbers of coarse ripples or humps and spikes 15 20 25 30
per unit meter in Al 6061 and 5083, and SS 304. U (m/s)×10 3
(a)

1.2
Al 5083
decreased by raising the focal spot. Spikes are accompanied
by serious pores in lengths of about 0.5 to 1 mm. As welding 1
speed increases, humping, spiking, fusion-zone width, and
depth decrease, as shown in Fig. 4(c). It is interesting to see
that humps become regularly zigzag as the scanning speed 0.8
nr (1/m)×10 −3

increases. The reason for this phenomenon, however, is not

clearly understood at the present time. 0.6
The effects of volatile element on the defects can be
revealed by comparing their differences in welding of Al
5083 and 6061. Referring to Figs. 3(b) and 4(a) shows that 0.4
f  −40 mm
surface roughness, spiking, depth, and width of the fusion
f  −20 mm
zone are enhanced for Al 5083. Serious humping is attributed 0.2 f  0 mm
to increased gas and liquid velocities from increase in vapor f  20 mm
pressure gradient. Spiking results from an increase in absorbed f  40 mm
energy due to the focusing characteristics and multiple reflec- 0
tions in the increased depth of the keyhole. 15 20 25 30
Quantitative values characterizing humping or coarse rip- U (m/s)×10 3
(b)
pling and spiking affected by the beam-focusing character-
istics, volatile element, and welding speed are critical but Fig. 6. Number of coarse ripples or humps per unit meter as a function
limited. The measured frequencies or numbers of humps or of scanning speed for different focal locations in welding (a) Al 6061 and
coarse ripples per unit meter are identical to those of spikes for (b) Al 5083.
different focal locations and scanning speeds in Al 6061 and
5083, as well as SS 304, as shown in Fig. 5. It should be noted
that the average pitch in unit of meter between adjacent ripples is also decreased by lowering the focal spot and increasing
is the reciprocal of the number of coarse ripples per unit meter. welding speed. The effect of welding speed on the pitch of
The experimental results agree well with those obtained from humping, however, is less than volatile element content in
Arata et al. [35], as discussed previously. Available results Al 5083. The average amplitudes of spikes as a function of
of humping can be explained by the wave equation solution scanning speed for different focal locations in welding Al 6061
together with the Navier–Stokes equation [18]. and Al 5083 are shown in Fig. 7(a) and (b), respectively. The
The number of coarse ripples or humps per unit meter as corresponding average amplitude of spikes for Al 5083 can be
a function of scanning speed for different focal locations in 2.5 mm, in comparison with 0.5 mm in Al 6061. Therefore,
welding Al 6061 and Al 5083 are measured and presented sufficient conditions to increase amplitude of spiking are to
in Fig. 6(a) and (b), respectively. It is seen that average increase volatile element content, lower the focal spot, and
pitches of coarse ripples and humping increase significantly decrease welding speed. It is noted that the amplitude of
with volatile element content. In the case of a focal spot spiking increases for the local spot location far below the base
below the workpiece surface by 40 mm and welding speed rather than at the base. The reason for this is that absorbed
of 15 mm/s, the pitch of humps is 1.3 mm in Al 5083 by energy increases with the depth of penetration due to specular
comparison with 0.4 mm for Al 6061. The pitch of humping reflections.
WEI et al.: MECHANISMS OF SPIKING AND HUMPING IN KEYHOLE WELDING 391

0.6 0.6
Al 6061 f  −40 mm
f  −20 mm 0.5
0.5 y  0.00996 x−0.03009
f  0 mm
0.4
f  20 mm
0.4 f  40 mm
0.3

hs / h
hs (m)×10 3

Al 6061
0.2 Al 5083
0.3 Al 1050
Al 1100
0.1
(Giedt, 1991)
SS 304
0.2 0.0 (Giedt, 1991)
0 10 20 30 40 50 60 70
0.1 w
ησ

0 Fig. 8. Comparison between measured and scaled spiking tendency.

15 20 25 30
U (m/s)×10 3
3.5
(a)
y  0.0048 x+1.041
3
3.0
Al 5083 f  −40 mm
f  −20 mm
2.5
(m)×10 3
f  0 mm 2.5
f  20 mm
2 f  40 mm
2.0
nr
1
hs (m)×10 3

1.5 1.5

Al 5083
1 1.0
80 100 120 140 160 180 200 220
2 2 /3
3 /2
h ρ dγ qw2
0.5 w γm v dT kμ

0 Fig. 9. Comparison between measured and scaled pitch of humping.

15 20 25 30
U (m/s)×10 3
(b)
(1) was scaled from energy balance at the moment when
Fig. 7. Average amplitude of spikes as a function of scanning speed for the beam irradiated on the keyhole base. The successful
different focal locations in welding (a) Al 6061 and (b) Al 5083. prediction of spiking tendency implies that time scale for
strong oscillations, collapse, or closure-fallback mechanism is
the energy distribution parameter divided by welding speed.
The measured spiking tendency, characterized by the spik- Their role is to cause intermittent moments at which incident
ing amplitude-to-average fusion-zone depth ratio, versus the flux impinges on the keyhole base.
fusion-zone width divided by the energy distribution parameter Comparisons between the measured and the predicted aver-
and melting efficiency in welding of different aluminum alloys age pitches of humps and spikes versus dimensionless parame-
and SS 304 for different focal-spot locations and scanning ter governing incident energy and surface tension coefficient
speeds in this paper and available experimental data from in welding Al 5083 are shown in Fig. 9. In view of high gas
Giedt [47] are shown in Fig. 8. In order to confirm experi- flow induced by high vapor pressure gradient from volatile
mental data, (1) is used for comparison. It is found that the element, serious humps result from, for example, the enhanced
measured and predicted results agree quite well. To dissipate Kelvin–Helmholtz instability [42]. In this paper, the effects of
incident energy a decrease in energy distribution parameter or the volatile element on humping are included in the enhanced
area results in an increase of the amplitude of spiking. Spiking liquid surface speed, as can be seen from (7) with the index
tendency can also be enhanced by reducing welding speed, n = 0.75. The scaled results from (6) are found to agree with
leading to an increase in fusion-zone width and decrease experimental data. It also indicates that the factors controlling
in melting efficiency. Low melting efficiency indicates more serious humping or coarse rippling in Al 5083 containing
energy is used for drilling the base. Amplitudes of spikes volatile element are the same as those of alloys in the absence
can reach 40% of depth of the fusion zone. It is noted that of volatile alloying element.
392 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012

VI. C ONCLUSION or instabilities, which can enhance fluid flow can be

responsible for humping or rippling formation. On the
The following conclusions are drawn. other hand, spiking is apt to any factor that can enhance
1) Frequencies of humps, coarse ripples and spikes, ampli- energy absorption near the keyhole base.
tude of spikes, and fusion-zone depth as functions of the 9) Focusing the focal spot near the keyhole base is a
volatile element content, beam-focusing characteristics, sufficient condition for spiking formation.
and welding speed are measured. In view of identical 10) Factors controlling spiking or humping in alloys in the
frequency, both the spikes and humps or coarse ripples presence and the absence of a volatile alloying element
are measured to assure reliability and accuracy. The are the same.
measured spiking amplitude and humping pitch agree 11) More rigorous definitions among coarse and fine ripples
with scaled results under different conditions in welding and humps are still needed. Humps in this paper are
carbon steels, aluminum alloys containing minor or considered to be the humping pitch-to-fusion-zone width
significant volatile element Mg, and SS 304. ratio around unity, whereas the ripples are for their pitch
2) The amplitude and pitch of spiking and humping much less than the fusion-zone width.
increase with volatile element content. The serious
humping can be attributed to enhanced liquid flow,
induced by increased vapor pressure gradient. Spiking R EFERENCES
results from an increase in absorbed energy due to the [1] J. H. Lau and S. W. R. Lee, Chip Scale Package, CSP: Design, Materials,
focusing characteristics and multiple reflections in the Process, Reliability, and Applications. New York: McGraw-Hill, 1999.
[2] S. Jang, “Packaging of photonic devices using laser welding,” Proc.
induced deep penetration of the keyhole. SPIE, vol. 2610, pp. 138–149, Jan. 1996.
3) The amplitude and pitch of spiking and humping [3] C. M. Ho and Y. C. Tai, “Micro-electro-mechanical-systems (MEMS)
increase with decreasing welding speed, as a result of and fluid flows,” Ann. Rev. Fluid Mech., vol. 30, pp. 579–612, Jan. 1998.
[4] W. H. Cheng, W. H. Wang, and J. C. Chen, “Defect formation mecha-
an increase in incident energy per unit length along nisms in laser welding techniques for semiconductor laser packaging,”
the joint line. In view of an increase in temperature IEEE Trans. Compon. Packag. Manuf. Technol. Part B, vol. 19, no. 4,
gradient, increased gas flow due to gas pressure gradient, pp. 764–769, Nov. 1996.
[5] M. K. Song, S. G. Kang, N. Hwang, H. T. Lee, S. S. Park, and K.
or increased liquid flow due to thermocapillary force can E. Pyun, “Laser weldability analysis of high-speed optical transmission
also enhance humping or rippling. Spiking is serious for device packaging,” IEEE Trans. Compon. Packag. Manuf. Technol. Part
a deep keyhole. B, vol. 19, no. 4, pp. 758–763, Nov. 1996.
[6] W. S. Chang and S. J. Na, “Thermomechanical analyses of laser
4) The effect of welding speed on humping in keyhole precision joining for optoelectronic components,” IEEE Trans. Compon.
welding is contradictory to that in arc welding. Humping Packag. Technol., vol. 26, no. 2, pp. 349–358, Aug. 2003.
in arc welding increases with welding speed, because [7] T. Velten, H. H. Ruf, D. Barrow, N. Aspragathos, P. Lazarou, E. Jung,
C. K. Malek, M. Richter, J. Kruckow, and M. Wackerle, “Packaging
the increased incoming mass in the welding direction of bio-MEMS: Strategies, technologies, and applications,” IEEE Trans.
increases humping volume. The concavity of the free Adv. Packag., vol. 28, no. 4, pp. 533–546, Nov. 2005.
surface of the molten pool, however, should not be [8] Y. Wu, A. Zhang, J. Chun, and S. Li, “Simulation and experimental study
of laser hammering for laser diode packaging,” IEEE Trans. Compon.
large and liquid ejection is small in order to satisfy Packag. Technol., vol. 30, no. 1, pp. 163–169, Mar. 2007.
conservation of mass. In keyhole welding, liquid ejection [9] C. W. Tan, Y. C. Chan, N. W. Leung, and H. D. Liu, “Mechanical
in the keyhole is significant. Humping thus cannot be behavior and low-cycle shear fatigue life of the pure Ni laser-welded
interpreted from conservation of mass viewpoint. joints in optoelectronics packaging,” IEEE Trans. Adv. Packag., vol. 31,
no. 2, pp. 386–393, May 2008.
5) The frequencies of humping, coarse rippling and spik- [10] P. W. Fuerschbach, “Control of photonic package alignment with asyn-
ing, and the amplitude of spiking increase by lowering chronous laser spot welds,” IEEE Trans. Electron. Packag. Manuf., vol.
the focal spot from the top surface. The increased 31, no. 2, pp. 114–119, Apr. 2008.
[11] S. Schiller, U. Heisig, and S. Panzer, Electron Beam Technology. New
spiking amplitude can be attributed to enhanced incident York: Wiley, 1982.
flux. The increased frequencies of humping and spiking [12] Y. Arata, Plasma, Electron and Laser Beam Technology. Materials Park,
can be interpreted from a decrease in upward liquid OH: ASM, 1986.
[13] N. Taniguchi, M. Ikeda, I. Miyamoto, and T. Miyazaki, Energy-Beam
convection and strong oscillations of the keyhole. Processing of Materials. Oxford, U.K.: Clarendon Press, 1989.
6) Spiking tendency can be successfully scaled from energy [14] W. W. Duley, Laser Welding. New York: Wiley, 1999.
balance at the moment when the beam irradiated on [15] B. J. Bradstreet, “Effect of surface tension and metal flow on weld bead
formation,” Weld. J., vol. 47, no. 7, pp. 314-s–322-s, Jul. 1968.
the base. It implies that the time scale for the incident [16] E. Soderstrom and P. Mendez, “Humping mechanisms present in high
flux periodically impinging on the base is the energy speed welding,” Sci. Technol. Weld. Join., vol. 11, no. 5, pp. 572–579,
distribution parameter divided by welding speed, which Sep. 2006.
[17] T. C. Nguyen, D. C. Weckman, D. A. Johnson, and H. W. Kerr, “High
can be due to strong oscillations, collapse, or closure- speed fusion weld bead defects,” Sci. Technol. Weld. Join., vol. 11, no.
fallback mechanism of the keyhole. 6, pp. 618–633, 2006.
7) The identical frequency of the humps and spikes point [18] A. Kumar and T. DebRoy, “Toward a unified model to prevent humping
defects in gas tungsten arc welding,” Weld. J., vol. 85, no. 12, pp. 292-
toward a common origin, namely both occur due to the s–304-s, 2006.
same instability or waves on the keyhole wall. [19] P. S. Wei, “Thermal science of weld bead defects: A review,” J. Heat
8) Humping or rippling (amplitude) is susceptible to Transf., vol. 133, no. 3, pp. 031005-1–031005-22, 2011.
[20] M. Pastor, H. Zhao, and T. DebRoy, “Pore formation during continuous
enhanced surface liquid flow. Any factor such as inci- wave Nd:YAG laser welding of aluminum for automotive applications,”
dent flux, vapor pressure gradient, thermocapillary force, Weld. Int., vol. 15, no. 4, pp. 275–281, Jun. 2001.
WEI et al.: MECHANISMS OF SPIKING AND HUMPING IN KEYHOLE WELDING 393

[21] H. Tong and W. H. Giedt, “A dynamic interpretation of electron beam [47] W. H. Giedt, “A periodic melting model of high intensity electron beam
welding,” Weld. J., vol. 49, no. 6, pp. 259-s–266-s, 1970. welding,” in Proc. 5th Int. Conf. Model. Cast., Weld, Adv. Solidificat.
[22] H. Tong and W. H. Giedt, “Radiographs of the electron beam welding Processes-V, Davos, Switzerland, Sep. 1991, pp. 115–122.
cavity,” Rev. Sci. Instrum., vol. 40, no. 10, pp. 1283–1285, Oct. 1969. [48] P. S. Wei and J. Y. Ho, “Energy considerations in high-energy beam
[23] R. E. Armstrong, “Control of spiking in partial penetration electron beam drilling,” Int. J. Heat Mass Transf., vol. 33, no. 10, pp. 2207–2217, 1990.
welds,” Weld. J., vol. 49, no. 8, pp. 382-s–388-s, Oct. 1970. [49] G. Mladenov, K. Vutova, and S. Wojcicki, “Experimental investigation
[24] C. M. Weber, E. R. Funk, and R. C. McMaster, “Penetration mechanism of the weld depth and thermal efficiency during electron beam welding,”
of partial penetration electron beam welding,” Weld. J., vol. 51, no. 2, Vacuum, vol. 51, no. 2, pp. 231–233, 1998.
pp. 90-s–94-s, 1972. [50] P. W. Fuerschbach, “Measurement and prediction of energy transfer
[25] Y. Arata, E. Abe, and M. Fujisawa, “A study on dynamic behaviors of efficiency in laser beam welding,” Weld. J., vol. 75, no. 1, pp. 24-s–34-s,
electron beam welding (report I)—The observation by a fluoroscopic 1996.
method,” Trans. Jpn. Weld. Res. Inst., vol. 5, no. 1, pp. 1–9, 1976. [51] T. Hashimoto and F. Matsuda, “Effect of welding variables and
[26] D. A. Schauer and W. H. Giedt, “Prediction of electron beam welding materials upon bead shape in electron-beam welding,” Trans. Nat. Res.
spiking tendency,” Weld. J., vol. 57, no. 2, pp. 189-s–195-s, 1978. Inst. Metals, vol. 7, no. 3, pp. 22–35, 1965.
[27] D. A. Schauer, W. H. Giedt, and S. M. Shintaku, “Electron beam welding [52] P. S. Wei, C. Y. Chang, and C. T. Chen, “Surface ripple in electron-beam
cavity temperature distributions in pure metals and alloys,” Weld. J., vol. welding solidification,” J. Heat Transf., vol. 118, no. 4, pp. 960–969,
57, no. 5, pp. 127-s–133-s, 1978. 1996.
[28] S. Tsukamoto and H. Irie, “Effect of molten metal behavior on melting [53] P. S. Wei, Y. H. Chen, J. S. Ku, and C. Y. Ho, “Active solute effects on
process in electron beam welding—A study on electron beam welding surface ripples in electron-beam welding solidification,” Metall. Mater.
phenomena,” Q. J. Jpn. Weld. Soc., vol. 9, no. 3, pp. 348–354, 1991. Trans. B, vol. 34B, no. 4, pp. 421–432, 2003.
[29] T. Miyazaki, S. Yoshioka, and T. Kimura, “Ejection of molten material
produced by pulsed electron and laser beams,” Precis. Eng., vol. 10, no.
1, pp. 141–146, 1988.
[30] D. L. DeBastiani, M. F. Modest, and V. S. Stubican, “Mechanism of
material removal from silicon carbide by carbon dioxide laser heating,”
J. Amer. Ceram. Soc., vol. 73, no. 7, pp. 1947–1952, 1990.
P. S. Wei (SM’08) received the Ph.D. degree from
[31] B. S. Yilbas and M. Sami, “Liquid ejection and possible nucleate boiling
the Mechanical Engineering Department, University
mechanisms in relation to the laser drilling process,” J. Phys. D: Appl.
of California, Davis, in 1984.
Phys., vol. 30, no. 14, pp. 1996–2005, 1997.
He is currently a Professor with the Department
[32] X. He, J. T. Norris, P. W. Fuerschbach, and T. DebRoy, “Liquid metal
of Mechanical and Electro-Mechanical Engineer-
expulsion during laser spot welding of 304 stainless steel,” J. Phys. D:
ing, National Sun Yat-Sen University (NSYSU),
Appl. Phys., vol. 39, no. 3, pp. 525–534, 2006.
Kaohsiung, Taiwan. His current research interests
[33] J. J. Batteh, M. M. Chen, and J. Mazumder, “A stagnation flow-analysis
include magneto-fluid dynamics, heat transfer during
of the heat transfer and fluid-flow phenomena in laser drilling,” J. Heat
resistance spot welding, interfacial plasma physics,
Transf., vol. 122, no. 4, pp. 801–807, 2000.
formations of the defects such as humping, rippling,
[34] P. S. Wei and C. Y. Ho, “Beam focusing characteristics effect on energy spiking, porosity encountered in lasers, and electron-
reflection and absorption in a drilling or welding cavity of paraboloid of beam welding.
revolution,” Int. J. Heat Mass Transf., vol. 41, no. 21, pp. 3299–3308, Dr. Wei is a fellow of the American Welding Society in 2007 and the Amer-
1998. ican Society of Mechanical Engineers in 2000. He received the Outstanding
[35] Y. Arata, F. Matsuda, and T. Murakami, “Some dynamic aspects of Research Achievement Awards from the National Science Council in 2004
weld molten metal in electron beam welding,” Trans. Jpn. Weld. Res. and NSYSU in 1991, 2001, and 2004, and the Adams Memorial Membership
Inst., vol. 2, no. 2, pp. 152–161, 1973. Award from the American Welding Society in 2008. He held the position of
[36] T. C. Nguyen, D. C. Weckman, D. A. Johnson, and H. W. Kerr, “The the Xi-Wan Chair Professor of NSYSU in 2009.
humping phenomenon during high speed gas metal arc welding,” Sci.
Technol. Weld. Join., vol. 10, no. 4, pp. 447–459, 2005.
[37] M. H. Cho and D. F. Farson, “Understanding bead hump formation in
gas metal arc welding using a numerical simulation,” Metall. Mater.
Trans. B, vol. 38B, no. 2, pp. 305–319, 2007.
[38] Y. Kawahito, M. Mizutani, and S. Katayama, “Elucidation of high- K. C. Chuang received the M.S. degree from the
power fibre laser welding phenomena of stainless steel and effect of Department of Mechanical and Electro-Mechanical
factors on weld geometry,” J. Phys. D: Appl. Phys., vol. 40, no. 19, pp. Engineering, National Sun Yat-Sen University,
5854–5859, 2007. Kaohsiung, Taiwan, in 2008.
[39] ASM Committee on Aluminum and Aluminum Alloys, “Properties and He conducted experimental work for investigating
selection: Nonferrous alloys and pure metals,” in Metals Handbook, welding defects of spiking and humping. Currently,
vol. 2, 9th ed. Materials Park, OH: ASM, 1979. he is an Assistant Engineer with the Institute of
[40] G. I. Leskov and L. I. Zhivaga, “The melting of thick metal during Nuclear Energy Research, Taoyuan, Taiwan.
welding with a conical electron beam,” Avt. Svarka, vol. 31, no. 4, pp.
11–14, 1978.
[41] A. A. Kaydalov and O. G. Miklukhin, “Effects of convergence of
a conical electron beam on penetration depth during welding,” Avt.
Svarka, vol. 34, no. 3, pp. 37–41, 1982.
[42] P. S. Wei and Y. T. Chow, “Beam focusing characteristics and alloying
element effects on high-intensity electron beam welding,” Metall. Trans.
B, vol. 23B, no. 1, pp. 81–90, 1992. J. S. Ku is currently pursuing the Ph.D. degree with
[43] C. Y. Ho, “Fusion zone during focused electron-beam welding,” J. the Mechanical Engineering Department, National
Mater. Process. Technol., vol. 167, nos. 2–3, pp. 265–272, 2005. Cheng Kung University, Tainan, Taiwan.
[44] M. F. Modest, “Effects of multiple reflections on hole formation during He is an Assistant Professor with the Department
short-pulsed laser drilling,” J. Heat Transf., vol. 128, no. 7, pp. 653–661, of Product Design, Shu-Te University, Yanchao,
2006. Taiwan, and the Department of Education, National
[45] T. A. Palmer and J. W. Elmer, “Characterisation of electron beams at Sun Yat-Sen University, Kaohsiung. Taiwan. His
different focus settings and work distances in multiple welders using current research interests include experimental mea-
the enhanced modified Faraday cup,” Sci. Technol. Weld. Join., vol. 12, surements of laser, electron, arc-welding processes,
no. 2, pp. 161–174, Aug. 2007. crystal growth, heat treatment, and microsturcture
[46] W. W. Duley, “A comparison of keyhole absorption processes in laser analysis of workpieces.
and electron beam welding,” in Proc. Conf. Laser Electron Beam Weld., Mr. Ku is serving the Chairman of the Certified License Committee and the
Cutt., Surface Treatment-State Art, 1987, pp. 160–167. Training Committee, Taiwan Welding Society.
394 IEEE TRANSACTIONS ON COMPONENTS, PACKAGING AND MANUFACTURING TECHNOLOGY, VOL. 2, NO. 3, MARCH 2012

T. DebRoy received the Ph.D. degree from the

Department of Metallurgy, Indian Institute of
Science, Bengaluru, India.
He is currently a Professor with the Department
of Materials Science and Engineering, Pennsylvania
State University, University Park. He has edited four
books and published 280 papers on computational
materials processing, particularly in the application
of numerical transport phenomena and optimization
in welding. His papers have been cited over 3900
times in the literature.
Prof. DebRoy has been recognized by prestigious awards from the
American Welding Society (AWS), the ASM International, the American
Iron and Steel Society, the International Institute of Welding (IIW), the
American Vacuum Society, the University of Graz, and the Pennsylvania
State University. He is an Honorary Member and fellow of AWS, and a
fellow of the ASM International. His awards include the Yoshiaki Arata
Award of IIW, the Kenneth Easterling Best Paper Award from the University
of Graz and IIW, the 57th Comfort A. Adams Lecture Award of AWS,
and the Faculty Scholar Medal of Pennsylvania State University. He has
given 14 keynote and plenary lectures in international conferences, and has
given numerous invited lectures in many prestigious institutions in Australia,
Canada, China, Egypt, India, Japan, Sweden, Taiwan, Ukraine, and the U.S.
He is a Founding Editor of Science and Technology of Welding and Joining,
and serves as a Principal Reviewer of the Welding Journal and as the Chair
of the Research and Development Committee of AWS.