Laser Ultrasonics Sensing of Metallurgical
Processes
Contributors: Thomas Garcin
Centre for
Metallurgical Process Engineering, The University of British Columbia, 309-6350
Stores Rd., Vancouver, BC, Canada V6T 1Z4
Contact: thomas.garcin@ubc.ca
Last modified: August 4th 2016
The document summarises important
applications of laser ultrasonics dedicated to the real time sensing of
metallurgical processes at high temperature. Emphasis is brought to the
assessment of the range of process parameters where the technology LUMet is
applicable. For some of the applications, a summary of feasibility is provided
in this documents, for some other, the interested user must refer to literature
available upon request.
Contents
Evaluation
of austenite grain size in metal
Monitoring
of phase transformation in steel
Monitoring
of ferrite recrystallization in steel
Monitoring
of Austenite recrystallization in steel
Monitoring
of recovery process in steel
Evaluation
of grain size in other metals
Monitoring
of recrystallization in Aluminum alloys
Evaluation
of residual stresses in Aluminum alloys
Monitoring
of phase transformation in Titanium
Ageing
study in beta stabilized Titanium alloys
Evaluation
of texture in hexagonal material
Monitoring
of sintering process
Evaluation
of plastic strain ratio in metals
Elastic
Moduli measurements in other compounds
Metals and there elaboration process are
constantly improved to meet the desired properties for novel technology in aeronautic,
civil engineering, automotive and biomedical industries. These efforts aim to
find an optimum balance between weight, strength, ductility, fracture
toughness, weldability and reactivity to the environment. New investigations
are therefore routinely conducted in universities, research institutes as well
as in the industries using numerous characterization techniques. For more than
two decades, laser ultrasonics has grown and now finds a place alongside other
in-situ methods to provide quantitative information on microstructure evolution
during industrial processing routes.
The technology laser ultrasonics was
developed in the bases of generating and detecting ultrasound remotely [1].
Among others, two important benefits of using lasers to generate and detect
ultrasound are i) the measurement on parts with complex shapes and ii) the
investigations at elevated temperature during processes. The focus is brought here
only to the later. Numerous scientific papers, patents as well as “proof of
concepts” have demonstrated the potential of such technology for the in-situ
microstructure control during thermo-mechanical processes of metals. This
report summarises the metallurgical applications for which laser ultrasonics is
well adapted to measurements in real time and identify areas where, on the
other hand, it has some limitations.
Ultrasound properties in a polycrystalline
sample, i.e. velocity and attenuation, are function of the material properties.
For an ultrasound pulse propagating in a polycrystalline material, the
ultrasound velocity is directly related to the temperature, elastic constants
and sample texture. It can be used to evaluate phase transformation,
recrystallization and other phenomena affecting the density and bulk elastic
constant of the sample. The attenuation has three sources that are the
diffraction (associated to the sample geometry), scattering (caused by elastic
mismatch between grains) and internal friction or absorption (generated by
numerous mechanisms such as dislocations movement, magnetic domains, interface
migration, solute diffusion, etc …).
The principle to measure velocity and
attenuation consists on generating a broadband ultrasound pulse in the sample
and detecting several echoes after propagation through the microstructure. The
time require for the pulse to travel in a given section of the sample is the
delay and the amplitude variation between two different echoes is the amplitude
decay. It is in addition necessary to estimate the propagation distance between
the two echoes which is related to the sample dimension (known a priori). Phase
fraction, mean grain size, recrystallized fraction and other metallurgical properties
such as elastic constants, plastic strain ratio, and texture component can then
be estimated in a certain range of process parameters.
It is not intended to go further into the
description of the principle of the technology but rather to focus on the
factors that allow and limit its applicability to industrial process control in
metallurgy. Number of references can be found in literature where extensive
details are provided with example of applications to various materials.
The range of applicability and methodologies
are defined based on numerous studies that have been conducted in various
materials for the last 30 years. The following table contains a non-exhaustive
list of important process variables allowing measurements with this technology.
There are correlations between the different process variables which directly
affects the range of applicability indicated in each section.
For sake of clarity, the comments listed
below are based on the typical capabilities of the Laser Ultrasonic for
Metallurgy (LUMet) sensor (TECNAR Ltee) attached to the Gleeble 3500
thermomechanical simulator (Dynamic System Inc. Poestenkill, NY). In this
sensor, a frequency-doubled Q-switched Nd:YAG laser with a wavelength of 532 nm
is used for the generation of a wide band compressive ultrasound pulse. The
duration of the laser pulse is approximately 6 ns, it has a maximum energy of
72 mJ and up to 50 pulses can be generated per second. The laser pulse produces
a broadband ultrasound pulse by vaporizing a small quantity of material at the
surface (of the order of a micrometer per hundred laser pulses). Successive
arrivals of the ultrasound pulse at the generation surface are detected with a
frequency-stabilized Nd:YAG pulsed laser which illuminates the surface 50 times
per second, with an infrared radiation at a wave length of 1064 μm and a
pulse duration of 90 μs. The infrared detection laser light reflected on
the specimen surface is demodulated inside a photo-refractive crystal using an
active interferometer approach. The ultrasound properties measured in this
technique are representative of the average properties of the material over a
volume created by the surface of the laser spot (about 2 mm) multiplied by the
sample diameter.
Sample
geometry: Flat
sheet with thickness between 1 to 15 mm Cylindrical
sample with diameter of 10 to 12 mm and with ratio between length and
diameter between 1.5 and 1. Round
tube Square
bar Hexagonal
bar |
· The thickness
range provided here usually allows for the measurement of at least one back
wall echo which is required for both attenuation and velocity measurement. · The
measurements on thick samples are more challenging due to the decrease on the
signal to noise ratio at elevated temperature and in sample with large grain
size. · The two
others dimensions of the sample (width and length) must be large compare to
the thickness. At least 2 to 3 times the thickness is recommended to avoid
reflection of the pulse on other surface. (For instance, measurement in the
plan RD,ND or TD,ND of a thin plate are usually not possible) · The
surface roughness as well as oxide layer also play a role. Measurements are
becoming more challenging in surface with roughness larger than 50 to 100
μm. The effect of various type of oxide layers remains to be fully
investigated. It is known to depend strongly on the bonding of the layer,
i.e. a homogeneous oxide layer well bonded to the sample surface is less
dramatic than a flaky oxide layer with poor adhesion developing in the course
of a treatments. · Measurement
on thin specimen can be problematic as a smaller number of grains are sampled
through thickness during the measurement creating additional uncertainty in
the measured data. · The signals
measured on thin sample (1mm and less) contain a large amplitude low
frequency wave which can be considered almost as a steady wave. It is
developing when the ultrasound wavelength is roughly the size of the sample
thickness. This wave must be filtered out reducing the effective bandwidth
available for the analysis of material parameter. · Measurement
on cylinder are possible as long as the alignment between the sample and the
D-laser is sufficiently good to collect some signal. Faceting both side of
the cylinder enhanced the signal to noise and reduces the number of echoes
originated from reflection on round surface. |
|
From
5 to 200 μm |
· Grain
size measurements can be conducted in materials which are sufficiently highly
anisotropic to cause measurable scattering. · The
measurement of grain size was currently validated in the following materials:
Austenite in low alloyed steels, Cobalt and nickel superalloys. Grain size
measurement in aluminum, magnesium, are difficult due to the small scattering
in these materials. Measurements of grain size in phase with magnetic
properties like ferrite in steel has not been validated yet. · Grain
size lower than a couple of micrometers usually cause very weak scattering
even in highly anisotropic materials making measurement challenging. · According
to the established methodology for grain size measurement, the smallest grain
size that can be measured is usually dictated by the mean grain size in the
reference sample used for the signal processing. · The
mean grain size estimated from ultrasound attenuation corresponds to the
average grain size through the sample thickness. For instance, the
measurement conducted in a partially recrystallized sample corresponds to the
mean grain size considering both the family of recrystallized grains and the
recovered grains. · The
calibration or analytical approach are usually constructing using homogeneous
grain size distribution, i.e. maximum grain size is 2 to 3 time larger than
the mean value. Large deviations from this criteria will lead to deviation
from metallographic data. · Calibrations
as well as analytical model are usually constructed for materials with
equiaxed or spherical grains. Strong variation in grain morphology need to be
integrated in the methodology to avoid deviation from metallographic data. It
was observed that the attenuation becomes lower [1] (thus predicted grain
size is lowered) when the ultrasound pulse propagate in the long axis of the
grain. · When
the grain size becomes large, the number of grain through thickness can
decreases dramatically, leading to dispersion in the measurements, i.e.
smaller number of grain sampled through thickness leads to bad statistics. As
a rule of thumb, a good enough statistic is obtained when the ratio of the
sample thickness to the grain diameter is larger than 50. · For
large grain size (> 200 um) or wide grain size distribution, conventional
established methodology using single scattering approaches are no longer
reliable and must be improved. |
|
From
room temperature to 1300 C |
· Temperature
play a role in the development of factors that can affect the quality of the
measured ultrasound waveform. High temperature can leads to the formation of
an oxide layer which won’t usually stick to the surface very well and cause
challenges in the generation as well as detection of the ultrasound. The
surface damage caused by the ablation during generation is also a function of
the temperature (This can be problematic in application where the same
position on the sample is used for a large number of measurement, i.e. to
follow continuously a process for instance) It is usually recommended not to
conduct more than 600 pulses at the same position. · Temperature
also affect the difference between the properties of different phases and
therefore the ability to measure a process. The difference between the bulk
elastic constant of the parent and product phase can in some material be
measurable only in a given temperature range. |
|
|
· The
ability to measure a certain metallurgical process varies from metals to
metals. As an analogy, the volume change between parent and product phase has
to be sufficiently different to measure the phase transformation by dilatometry.
In the same way, the difference between the bulk elastic properties of the
parent and product phase should be sufficiently different to be measured by
laser ultrasound. · The
difference between elastic constants of two phases varies with temperature. For
instance, almost no difference exists between the ferrite and austenite phase
in steel around 900°C where the austenite formation occurs making this
process rather complicated to monitor in-situ by use of velocity. However,
the difference between the bulk elastic constant of ferrite and austenite is
quite large below 700°C, temperature at which the austenite decomposition
occurs during cooling. This phase transformation is therefore totally
measurable in a certain temperature range only. · In the
same idea, the monitoring of recrystallization with laser ultrasonics rely on
the variation of the bulk elastic constant during recrystallization. This
elastic constant varies due to the texture change associated with
recrystallization. Depending on the intensity of the texture variation during
a particular process or a given material initial state, the change in elastic
constant can be large enough or contrary to weak for it to be measurable by
velocity measurements. |
A number of applications have proved to be
reliable and are detailed in the table below.
·
Range of measurements: o
Mean grain size range from 5μm to 200
μm, o
Temperature range from 900°C to 1300°C o
Grain aspect ratio (long axis divided by short
axis) range from 1 to 0.7 o
Grain size distribution (Maximum grain
size divided by mean grain size) range from 1 to 3 o
Minimum number of grain through thickness
must be larger than 50 o
Calibration already available for the
aforementioned range of parameter [2]. o
Measurement can be conducted during
isothermal holding and continuous heating ·
Steel compositions: o
Low carbon and low alloy steel (HSLA,
TRIP, DP, CP) o
Not yet validated for austenitic stainless
steel o
Difficult in Interstitial Free steel ·
Note: o
The measurement require a reference sample
with the same geometry than that of the measured sample and fine grain size
(< 5 μm) o
Measurement on thick samples affects the
measurable maximum grain size o
Measurement at very high temperature
affect the measurable maximum grain size o
Measurement on thin sheet cause
statistical problem due to limited number of grains through thickness o
Sample thickness must be known with a
precision of about 0.1 mm o
Temperature must be known with a precision
of about 10 °C o
Error estimated to be 10 to 15 % of the
absolute mean grain size ·
References: [3]–[10] |
·
Range of measurements: o
Austenite decomposition when occurring
below 750°C o
Austenite formation when associated with
texture change, i.e. when starting from a cold rolled state for instance. o
Measurement can be conducted during
isothermal holding and continuous heating o
Can be conducted during cooling rate up to
150°C/s o
Measurement is not affected by gas blowing
on the sample o
Cannot be conducted during water quenching
·
Steel compositions: o
Low, medium and high carbon steel ·
Note: o
The measurement of the fraction
transformed require the continuous evaluation of velocity from the fully
austenite region to the lower temperature range o
For a single shot measurement of the fraction
transformed, a calibration is necessary and the accurate knowledge of the
sample temperature is required o
As with dilatometry, the measurements does
not provide the fraction of individual phase such as ferrite, bainite and
martensite. o
Evaluation of phase transformation above
the Curie temperature (about 750°C) are challenging due to the fact that the
velocity of the parent and product phase are similar o
When using a calibration, sample thickness
must be known with a precision of about 0.1 mm o
When using a calibration, temperature must
be known with a precision of about 10 °C ·
Reference [11], [12], [13] |
·
Range of measurements: o
In the ferrite temperature range o
Measurement can be conducted during
isothermal holding and continuous heating ·
Steel compositions: o
Low, medium and high carbon steel o
Ferritic stainless steel o
Other iron based alloys with a ferritic
structure ·
Note: o
The measurement of the recrystallized
fraction require the continuous evaluation of velocity from the fully
deformed state to the fully recrystallized state o
For a single shot measurement of the
recrystallized fraction, a calibration is necessary and the accurate
knowledge of the sample temperature is required o
The measurement sensitivity depends on the
intensity of texture change during the recrystallization process, i.e. the
larger the texture evolution the better the sensitivity. o
When using a calibration, sample thickness
must be known with a precision of about 0.1 mm o
When using a calibration, Temperature must
be known with a precision of about 10 °C ·
Reference
[14],[15][3] |
·
Range of measurements: o
Typically conducted in Gleeble during
holding at high temperature after uniaxial compression testing o
Challenging for large grain austenite
grain (>150 μm) with thick sample to conserve acceptable signal to
noise ratio ·
Sample geometry: o
Cylindrical sample with diameter of 10 to
12 mm and with ratio between length and diameter between 1.5 and 1. o
Can be adapted to measurement on plate ·
Steel compositions: o
Low, medium and high carbon steel ·
Note: o
The measurement of the recrystallized
fraction require the continuous evaluation of attenuation and velocity from
the fully deformed state to the fully recrystallized state o
For a single shot measurement of the
recrystallized fraction, a calibration is necessary and the accurate
knowledge of the sample temperature is required o
The measurement sensitivity depends on the
intensity of texture change during the recrystallization process, i.e. the
larger the texture evolution the better the sensitivity. o
When using a calibration, sample thickness
must be known with a precision of about 0.1 mm o
When using a calibration, temperature must
be known with a precision of about 10 °C ·
Reference [16] |
·
Few studies shows a variation of velocity
during recovery which was associated to the variation in dislocation damping.
·
The attenuation is both a function of the
grain scattering and evolution in dislocation density. ·
Model are needed to estimate qualitatively
the change in dislocation density ·
Limited study where conducted overall ·
Measurement are challenging and usually
require simultaneous stress relaxation measurement. ·
The user must refer to available
literature ·
Reference [17],[18], [19] |
·
Range of measurements: o
Mean grain size range from 5 μm to
200 μm, o
Temperature range from room temperature
1300°C o
Grain aspect ratio (long axis divided by
short axis) range from 1 to 0.7 o
Grain size distribution (Maximum grain
size divided by mean grain size) range from 1 to 3 o
Minimum number of grain through thickness
must be larger than 50 ·
Materials: o
Any metal with anisotropy factor larger
than 2.5 o
Validated for Cobalt based super alloys o
Validated for Nickel based super alloys ·
Note: o
The measurement require a reference sample
with the same geometry than that of the measured sample. o
Measurement on thick samples affects the
measurable maximum grain size o
Measurement at very high temperature
affect the measurable maximum grain size o
Measurement on thin sheet cause
statistical problem due to limited number of grains through thickness o
Sample thickness must be known with a
precision of about 0.1 mm o
Temperature must be known with a precision
of about 10 °C o
Error estimated to be 10 to 15 % of the
absolute mean grain size ·
References: [20],[21] |
The
user must refer to available literature to evaluate the feasibility Reference
[22], [23], [24], [25] [26] |
The
user must refer to available literature to evaluate the feasibility Reference
[27] |
·
Range of measurements: o
Measurement can be conducted during
isothermal holding and continuous heating o
Can be conducted during cooling rate up to
150°C/s o
Measurement is not affected by gas blowing
on the sample o
Cannot be conducted during water quenching
·
compositions: o
Pure Titanium and Titanium alloys ·
Note: o
The measurement of the fraction
transformed require the continuous evaluation of velocity during beta to
alpha or alpha to beta transition o
For a single shot measurement of the
fraction transformed, a calibration is necessary and the accurate knowledge
of the sample temperature is required o
Extensive grain growth in the beta phase
can lead to lack or repeatability in the measurement. o
When using a calibration, sample thickness
must be known with a precision of about 0.1 mm o
When using a calibration, temperature must
be known with a precision of about 10 °C Reference
[28] |
·
Range of measurements: o
Measurement can be conducted during
isothermal holding ·
Steel compositions: o
Beta stabilized Titanium alloys ·
Note: o
The measurement of the fraction
transformed may require preliminary calibration o
When using a calibration, sample thickness
must be known with a precision of about 0.1 mm o
When using a calibration, temperature must
be known with a precision of about 10 °C o
Reference
[29] |
The
user must refer to available literature to evaluate the feasibility Reference
[30], [31], |
The
laser ultrasonic technology can be used to monitor sintering processes. [32] [34] The
rational is relatively simple. In a two phase materials, the elastic
properties as well as the density of the compound are a function of the
property of the individual phases weighted by the volume fraction of each
phase. In compacted metallic powder, the two phases are the void and the
powder which have obviously very different elastic property and density. The
simple manner to quantify sintering process therefore is to follow the
variation of the ultrasonics velocity during the process. As the fraction of
void decreases, the velocity v = sqrt(Elastic modulus/density) will increase. Attenuation
is more complicated to explain but likely provide also a way to estimate the
stage of sintering as void will act as strong scattering centre and disperse
the wave energy. The paper from Liu is more fundamental and explains this
concept of attenuation measurement applied to the sintering processes. [33] |
The
user must refer to available literature to evaluate the feasibility Reference
[35] |
The
user must refer to available literature to evaluate the feasibility Reference
[36] |
[2] “Kruger, S.E., US7353709,
2008.pdf.” .
[27] A. Moreau and C.
Man, “Laser-Ultrasonic Measurements of Residual Stresses on Aluminum 7075
Surface-Treated by Low Plasticity Burnishing,” pp. 97–108, 2005.