Journal of Microscopy, Vol. 217, Pt 2 February 2005, pp.

167– 173
Received 25 June 2004; accepted 29 November 2004

Microstructural characterization of autogenous laser welds on

Blackwell Publishing, Ltd.

316L stainless steel using EBSD and EDS

J. KELL*, J. R. TYRER*, R. L. HIGGINSON† & R. C. THOMSON†

*Wolfson School of Mechanical and Manufacturing Engineering and †Institute of Polymer Technology
and Materials Engineering, Loughborough University, Leicestershire, LE11 3TU, UK

Key words. 316L stainless steel, electron backscatter diffraction, energy

dispersive spectroscopy, laser welding, microstructure.

The inherent rapid solidification makes the use of lasers very

Summary
attractive because completed welds contain a range of micro-
This research is concerned with autogenous welding of 316L structures with metastable and stable phases, minimum segrega-
stainless steel and the microstructure generated by such a tion and fine grain sizes which often improve the mechanical
process. Autogenous welding does not require a filler material properties. A Schaeffler diagram (Schaeffler, 1974) and the
and in this case relies on an initial shallow melt phase to modified version from Kotecki & Siewert (1992) can be used to
maintain a conduction limited weld. Essentially, a high power calculate the primary solidification mode, and predict the
laser beam traverses the substrate, with the beam shaped by ferrite content. In the case of the 316L stainless steel used in
conventional optics, which produces a Gaussian irradiance this work, chromium and nickel equivalents can be calculated
distribution; or with a diffractive optical element, used to using the following expressions:
produce a uniform irradiance distribution.
Initial results have shown that due to the nature of the heating Nieq = Ni + 35C + 20N + 0.25Cu
cycle, complex microstructures are developed. These fine, com-
plicated microstructures cannot be satisfactorily resolved and Creq = Cr + Mo + 0.7Nb
quantified using standard optical microscopy techniques. Elec-
tron backscatter diffraction (EBSD) and energy dispersive spec- David et al. (1987) have shown that rapid cooling has a profound
troscopy (EDS) have been carried out on a number of different effect on the weld microstructures of stainless steels, which
microstructures prepared using a range of welding parameters. makes predictions from conventional constitution diagrams
It is demonstrated that the simultaneous determination of the unreliable. They showed that ferrite content upon solidification
chemistry and crystallography is a very useful tool for rapid initially increased with traversing velocity, and then signifi-
identification of the different phases formed on solidification as cantly tailed off with increasing velocity, leaving an austenitic
a consequence of varying welding procedures. solidification structure. Gou & Kar (1999) outlined a proposed
mathematical model for predicting microstructural evolution
Received 25 June 2004; accepted 29 November 2004

in a 304 stainless steel weld metal and illustrated a cellular

Introduction
microstructure. A grain size of approximately 7 µm was reported,
Laser welding is a process that melts and joins metals by heating although no further metallography was carried out. Lippold
them with a laser beam. The laser beam is produced using a gas (1994) showed, in a study of solidification behaviour and
mix of CO2, N2 and He, which is continually excited by electrodes, cracking susceptibility of pulsed laser welds, that the primary
to produce a collimated photon beam. The laser welding process solidification mode of 316L stainless steel was fully austenitic,
has proved to be a viable alternative to the more accepted methods with no residual ferrite content whatsoever, and a well defined
of welding, such as gas tungsten arc welding (GTAW). The auto- solidification subgrain structure. David et al. (1979) confirmed
motive industry in particular has successfully utilized laser from transmission electron microscopy and energy dispersive
welding for a number of applications. X-ray analysis studies that diffusion occurs during the δ → γ
The general attributes of a laser beam weld are that the transformation, and also suggested that the microsegregation
relative heat input is low compared to other welding processes, (in 308 stainless steel) on dendrite arms is δ-ferrite, whereas
and therefore the cooling rate is relatively high (Locke et al., 1972). the remaining region is γ-austenite. The EDX scans supported
this suggestion by showing an increase in chromium and a
Correspondence: Dr John Tyrer. E-mail: [email protected] corresponding reduction in nickel in the δ-ferrite dendrite core

© 2005 The Royal Microscopical Society

168 J. K E L L E T A L .

Table 1. Chemical composition of 316L stainless steel in %weight.

Element C Cr Mn Mo N Ni P S Si Fe

%weight 0.020 16.820 1.214 2.353 0.042 10.200 0.026 0.002 0.038 69.247

area. Similar studies have been reported by Lyman (1979) for a

304 stainless steel. In the present work, a combination of electron
backscatter diffraction (EBSD) and energy dispersive spectroscopy
(EDS), has been used to successfully characterize laser weld micro-
structures in 316L stainless steel, which cannot be satisfac-
torily understood with optical microscopical techniques.

Experimental methods
The material used in this study was a 316L stainless steel
sheet of 1 mm thickness, the composition of which is shown in
Table 1. Prior to welding, each sheet was chemically cleaned
in a bath of 50% concentrated hydrochloric acid and 50%
water, washed in alcohol and dried under hot air.

Laser operation
The laser used was a 1 kW CO2 Coherent Everlase S48 (Coherent,
UK Ltd, Ely, U.K.), fitted with a coaxial focusing head. The
tracking velocity (generated with an x-y Computer Numeri- Fig. 1. A schematic diagram showing the laser beam delivery and coaxial
cally Controlled, CNC, platform) can be changed by modify- assist gas nozzle.
ing the control programme, from 1 mm min−1 to 6 m min−1 at
increments of 1 mm min−1. The workpiece was clamped surface at different submicrometre depths. When subjected to
between two plates of 5 mm mild steel with a 10 mm thick a raw beam (diameter ∼20 mm), each small mirror creates a
Duratec 750 insulation board beneath the substrate. This specific wave front. When these are combined, they interfere
arrangement was fixed to the CNC table and allowed several with each other to create the premeditated beam shape at the
linear welds of 30 mm length to be completed before the sheet reconstruction plane. In this case the beam shape used was a
was removed. The plate was allowed to return to ambient tem- 1.25 × 1.25 mm uniform irradiance distribution.
perature before the next weld was performed. The zinc selenide
(ZnSe) lens had a diameter of 29 mm and focal length of
Sectioning/polishing
63.5 mm. A schematic of arrangement of the optics is shown in
Fig. 1. The stand off distance (lens–workpiece distance) can be The weld samples were sectioned in the transverse plane (end
set by positioning the nozzle on the workpiece using a microm- on), i.e. perpendicular to the weld direction, and mounted in
eter attached to the flight tube, which creates a datum point to conducting Bakelite, and were ground using SiC paper from
which the stand off distance was measured. The focal distance 800 to 2400 grit. The samples were then polished with both a
of the lens was set at 1 mm below the nozzle exit, so a larger 6 µm and a 1 µm diamond suspension. The mounted samples
beam spot is produced by raising the focusing head with the were chemically etched using Kallings Reagent (distilled water
attached micrometer. The laser power was monitored via an 40 mL, copper chloride 2 g, hydrochloric acid 40 mL, ethanol
internal meter and was set via the excitation current, to (80%) 40–80 mL) for approximately 4 s. The microstructures
between 125 W and 1 kW. The workpiece was aligned using were imaged using bright-field, polarized light and dark-field
burn paper. Compressed air was used as an assist gas which optical microscopy using a Reichart MEF3.
exits directly from the nozzle onto the workpiece with a flow
rate from 15 L min−1 to 50 L min−1.
Scanning electron microscopy
A novel way of beam shaping for laser welding has been
developed and implemented into a laser system (Kell et al., Immediately before examination in the SEM, final polishing
2004). The beam can now be shaped using a device called a was carried out using colloidal silica for approximately 30 min.
diffractive optical element (DOE). This is a small mirror com- The samples were then characterized using a LEO 1530VP
prising of thousands of 6 µm squares which are set below the scanning electron microscope (SEM) fitted with a field emission

© 2005 The Royal Microscopical Society, Journal of Microscopy, 217, 167– 173
 M I C RO S T RU C T U R A L C H A R AC T E R I Z AT I O N O F AU TO G E N O U S L A S E R W E L D S 169

Fig. 2. Modified Schaeffler diagram. The red lines indicate the intercept
for 316L stainless steel (Kotecki & Siewert, 1992).
Fig. 3. Thermodynamic equilibrium calculation for the composition of the
gun. The microscope is fitted with an EDAX Pegasus analytical steel used in this work showing the predicted high temperature phase stability.
system which allows simultaneous chemical EDS to be carried
out during the EBSD scan. All maps were produced using the air as assist gas. The parabolic weld profile expected when using a
OIM analysis package (OIM, EDAX UK, Cambridge, U.K.). Gaussian irradiance distribution can clearly be seen. The fluctua-
tions on the surface are due to the assist gas (in this case air), which
also accounts for the oxide scale on the surface. EDS data showed
Results and discussion
that the oxide layer was predominantly chromium oxide.
The chemical composition presented in Table 1 was used to Figure 5 is a higher magnification image of the interfacial
calculate values of Creq = 19.173 and Nieq = 11.72. These values region between the weld metal and the substrate for the region
are used in the revised Schaeffler diagram (Fig. 2, marked in illustrated by the box in Fig. 4. This interfacial region shows the
red) (Kotecki & Siewert, 1992) to return a ferrite prediction grains at the edge of the fusion zone growing towards the centre
(Welding Research Council ferrite number) of 6%. This value of the weld, before entering the bulk of the weld metal, where a
falls in the Ferrite–Austenite area, indicating that ferrite may be cellular microstructure is evident. Also the size of the substrate
present upon solidification under normal welding conditions. grains (∼20 µm) can be seen in the lower right area of the image.
Equilibrium thermodynamic calculations for the steel com- Figure 6(a) shows a higher magnification of the cellular
position used in this work were performed using the software structure near the weld surface, with Fig. 6(b) showing the
package mtdata (Davies et al., 1989) in conjunction with a same region using dark-field illumination.
database specific for iron-based alloys (Sundman, 2001). The EBSD and EDS were used as analysis tools throughout this
predicted high temperature phase stability is shown in Fig. 3, study. From the optical micrographs, it can be seen that each
and also supports the possibility of small amounts of δ-ferrite cell is approximately 3 µm, therefore using a step size of 1 µm
forming initially on solidification. would show any orientation changes between cells across
Figure 4 shows a transverse cross-section of a weld produced the scan area. It was found that, instead of each cell existing
with the following parameters: power 900 W, traverse velocity as a single entity, it was actually part of a larger grain of
400 mm min−1, stand off distance 73.93 mm, with 20 L min−1 of approximately 200 µm. This is shown clearly in Fig. 7, where

Fig. 4. Weld overview, transverse cross-section of a

weld performed at power 900 W, transverse velocity
400 mm min−1, and a stand off distance of 73.93 mm.
Box indicates the region shown in Fig. 5 at higher
magnification.

© 2005 The Royal Microscopical Society, Journal of Microscopy, 217, 167– 173
170 J. K E L L E T A L .

Fig. 5. Interfacial region between weld metal and substrate metal of the
same weld sample within the box illustrated in Fig. 4.

the large grains that make up the weld microstructure are

illustrated. The smaller grains (approximately 20 µm diameter)
in the lower left corner of the image are those of the substrate.
The backscatter detector in the SEM could also be used to image
the cellular structure (Fig. 8a), comparable to that observed using
optical microscopy with a mixed acid etch. The dark vertical lines
on the image indicate an EBSD scan area, which was executed
using a fine step size of 0.035 µm. Figure 9 illustrates the EDS
data for the different elements collected from the scan of the area
shown in Fig. 8. During the EBSD scan, two iron-based phases
were scanned for: delta (δ) and gamma (γ). Figure 8(b) is a greyscale Fig. 6. Optical micrographs. (a) A region near the weld surface, and (b) a
image quality map overlaid with the colour phase map showing dark-field view of the same region.
red (γ) as the only phase present, indicating the area to be 100%
austenite. The inverse pole figure map was also a single colour
(Fig. 8c), again signifying that the mapped area comprised a single
crystallographic orientation. The EDS maps in Fig. 9 show that
there are inclusions present, comprising predominantly oxygen,
manganese and silicon, and a reduced iron content. It is clearly
observed that levels of chromium, molybdenum and sulphur
all increase around the edges of the cellular structure, which
also shows a drop in iron level to compensate for the increases
in other elements. The nickel and phosphorus, do not display
significant fluctuations throughout the maps.
Welds were also examined produced using a diffractive
optical element to shape the laser beam. Large EBSD scans over
the fusion zone have shown the grain size to be similar to that
of conventional laser welding, i.e. approximately 200 µm. The
observed microstructure was not cellular as seen when using
a Gaussian beam with conventional optics, and so an EBSD/
EDS scan over an area of higher magnification was carried Fig. 7. Inverse pole map of an interfacial region of the weld. The upper left
out. Figure 10(a) shows a image quality map of the scan area, corner shows the weld surface.

© 2005 The Royal Microscopical Society, Journal of Microscopy, 217, 167– 173
 M I C RO S T RU C T U R A L C H A R AC T E R I Z AT I O N O F AU TO G E N O U S L A S E R W E L D S 171

Fig. 8. (a) SEM image showing EBSD scan area in

enclosed rectangle, (b) the phase map and (c) the
inverse pole figure map.

and Fig. 10(b) the phase map of the same area. The green areas a diffractive optical element, a partially austenitic structure
in this map show where δ-ferrite is present, which matches the is produced in the fusion zone. The grain size is still ∼200
dark areas in Fig. 10(a). Figure 10(c) provides more evidence µm, but small regions of δ-ferrite are also found. This is likely
for ferrite inclusions by showing an inverse pole map exhibiting to be due to the increased quench rate seen when using a
areas of different crystallographic orientation within grains. square uniform irradiance distribution.
Further evidence for δ-ferrite can be seen in Fig. 11, showing
EDS maps of the scanned area shown in Fig. 10(a). Segregation
References
of chromium, molybdenum, and sulphur, all ferrite forming
elements, is seen in Fig. 11 in the form of increased concentra- David, S.A., Goodwin, G.M. & Braski, D.N. (1979) Solidification behaviour
tions at the areas which have been previously identified as of austenitic stainless steel filler metals. Weld J. 58, 330s–336s.
ferrite. Similarly, nickel, an austenite former, concentrated David, S.A., Vitek, J.M. & Hebble, T.L. (1987) Effect of rapid Solidification
on stainless steel weld metal microstructures ands its implications on
heavily in areas away from ferrite. This shows that by using
the Schaeffler diagram. Weld J. 66, 289s–300s.
both EBSD and EDS, a very detailed analysis of a laser weld in
Davies, R.H., Dinsdale, A.T., Chart, T.G., Barry, T.I. & Rand, M.H. (1989)
stainless steel can be produced.
Application of MTDATA to the modelling of multicomponent equilibria.
High Temp. Sci. 26, 251–262.
Conclusions Guo, W. & Kar, A. (1999) Prediction of microstructures in laser welding of
stainless steel AISI 304. J. Laser Apps. 11, 185 –189.
• EBSD and EDS have successfully been utilized to characterize
the metallurgy of laser welded samples of 316L stainless steels.
Kell, J., Tyrer, J.R., Higginson, R.L. & Thomson, R.C. (2004) Computer
generated holographic elements used in laser powder fusion. ICALEO
The unique combination of simultaneous determination 2004 Techn. Digest, 23, 65.
of chemistry and crystallography provides a rapid tool for Kotecki, D.J. & Siewert, T.A. (1992) WRC-1992 Constitution diagram for
conclusive identification of the complex solidification structures stainless-steel weld metals – a modification of the WRC-1988 diagram.
forming from different laser welding processes. Upon solidi- Weld J. 71, 171s–178s.
fication, the weld metal forms significantly larger grains (∼200 Lippold, J.C. (1994) Solidification behaviour and cracking suscepti-
µm), than those found in the substrate (∼20 µm). The larger bility of pulsed laser welds in austenitic stainless-steels. Weld J. 73,
129s–139s.
grains exhibit a cellular appearance within the grain, with
Locke, E., Hoang, E. & Hella, R. (1972) Deep penetration welding with
each cell being approximately 2–4 µm in diameter.
high power CO2 lasers. Weld J. 51, 245s–249s.
• In this work when using a Gaussian beam with conventional
optics, solidification produces a microstructure comprising
Lyman, C.E. (1979) Analytical electron-microscopy of stainless-steel weld
metal. Weld J. 58, 189s–194s.
100% austenite, as evidenced by the EBSD scans. However, Schaeffler, A.L. (1974) Constitution diagram for stainless steel weld metal
microsegregation of chromium, molybdenum and sulphur 2. Shaeffler diagram. Metal Prog. 106, 227.
is visible at the outer edges of the cells within the structure. Sundman, B. (2001) TCFE database supplied by TCAB for use with MTDATA.
• When using a uniform irradiance distribution, provided by Available from Thermo-Calc Software AB, Stockholm.

© 2005 The Royal Microscopical Society, Journal of Microscopy, 217, 167– 173
172 J. K E L L E T A L .

Fig. 9. EDS data from all detected elements from the

scan area shown in Fig. 8.

© 2005 The Royal Microscopical Society, Journal of Microscopy, 217, 167– 173
 M I C RO S T RU C T U R A L C H A R AC T E R I Z AT I O N O F AU TO G E N O U S L A S E R W E L D S 173

Fig. 10. (a) Image quality map showing EBSD scan

area, (b) the phase map and (c) the inverse pole figure
map. This weld has been produced using an alternate
method of beam shaping. Using diffractive optical
elements as an alternative to conventional lenses
provides a significantly improved method of custom
beam shaping (Kell et al., 2004).

Fig. 11. EDS data from detected elements from the scan
area shown in Fig. 10(a). Oxygen, manganese and silicon
have been omitted as they show the same as in Fig. 9.

© 2005 The Royal Microscopical Society, Journal of Microscopy, 217, 167– 173