This course aims to provide a succinct overview of the emerging discipline of Materials Informatics at the intersection of materials science, computational science, and information science. Attention is drawn to specific opportunities afforded by this new field in accelerating materials development and deployment efforts. A particular emphasis is placed on materials exhibiting hierarchical internal structures spanning multiple length/structure scales and the impediments involved in establishing invertible process-structure-property (PSP) linkages for these materials. More specifically, it is argued that modern data sciences (including advanced statistics, dimensionality reduction, and formulation of metamodels) and innovative cyberinfrastructure tools (including integration platforms, databases, and customized tools for enhancement of collaborations among cross-disciplinary team members) are likely to play a critical and pivotal role in addressing the above challenges.
Material Property, Material Structure, and Manufacturing Processes
Loading...
来自 Georgia Institute of Technology 的课程
Materials Data Sciences and Informatics
61 个评分
从本节课中
Materials Knowledge and Materials Data Science
• Understand property, structure and process spaces
• Learn about Process-Structure-Property Linkages
• Learn what does Materials Knowledge mean
• Learn about a role of Data Science in Materials Knowledge System
• Overview approaches and main components of Data Science
• Learn about a new discipline - Materials Data Sciences
与讲师见面
Dr. Surya KalidindiProfessor
The George W. Woodruff School of Mechanical Engineering
[MUSIC]
Hello, everyone.
Welcome to the Materials Data Science and Informatics class.
I'm Surya Kalidindi, the instructor for this class.
The title for today's lesson is Material Property, Material Structure, and
Manufacturing Processes.
The expected outcomes for today's lessons are to understand the meanings of this
comes in the title in the context of materials development efforts.
And then in the process we also learned that materials design and
process design are in worst problems that are very difficult to solve.
Let's start. With the word property.
Properties are usually defined and measured.
A typical standard test for problem measuring, mechanical properties,
is the tension test and this tension test is described in the ASTM standard E8.
In the tension test, one typically starts with a cylindrical sample and
I place a load along the axis of the sample.
As you apply the load, the cylindrical sample stretches, and when it stretches,
for a while it actually undergoes uniform deformation in the sense that the cross
section remains constant throughout the length of the sample.
And after a while you get what is called as nicking, or localized deformation.
Eventually the sample breaks.
One recourse the load displacement or load extension.
And extension is defined as L minus L nought, the change in length.
And then define the stress and strain, where the stress is defined as load
divided by the initial cross-sectional area.
And strain is defined as Extension divided by initial cross section area.
And the extension is L minus L not.
So once one has the load and the extension you normalize
it by the geometric parameters of the sample and define the stress and strain.
Once you have the stress/strain curve,
you can now define properties on the stress/strain curve.
So the first property of interest might be yield strength.
Yield strength is defined as the stress at which elastic deformation or
inelastic deformation sets in the sample.
So in design it's an important property because you want to design your
You're component [INAUDIBLE] that there's no permanent defamation or
inelastic defamation.
Another property of [INAUDIBLE] is the ultimate tensile strength.
Ultimate tensile strength is also very useful property, because it tells
you how much stress you can apply on the sample without breaking the sample.
A third property from that is apparent here from this curve
is called the total elongation.
It's percentage strain at which the sample breaks.
That falls in extremely useful property because you want to design
your conference.
Such a way that they don't break to give you enough strain before they break,
they give you enough warning before they break.
So, these are three of the properties you can define on the stress strain
from a tension test.
There are more,
but this hopefully will give you some idea of examples of properties.
Let's now define property in a little bit more general way.
The property refers to a physical property of interest.
which means that you are imposing some sort of a physical
loading condition on the sample and you're measuring it's response.
It is defined as a characteristic of the materials' response to the applied load
and it is always normalized, suitably normalized.
And what we mean by that is exactly what I described in the previous slide,
where we took the load and the extension, and normalize it by area and
length to get stress and strain.
Without normalization, if you define a property it is
likely to be sensitive to small changes in the geometry of the sample.
That you use for the test.
So the normalization gets rid of that sensitivity and
makes the property more useful.
In structural applications, there are many properties of interest.
These include things such as elastic stiffness,
elastic modulus, yield strength, fracture toughness, fatigue strength.
And so on so forth.
In reality, we're often interested in property combinations.
We seldom are interested in individual properties but
interested in property combinations
as an example as a specific example of a property combination of interest.
Let's look at this combination, Elongation versus Tensile strength.
In the previous light we have defined these properties again to remind you.
Tensile strength tells you something about how much load you can apply on the sample
before the sample.
Under those plastic deformation or sometimes breaks.
And the elongation you see an indicator of
how much we can stretch the sample before it breaks.
These are both very useful and ideally you would want a combination that has
high tensile strength and high elongation.
In a plot like this, high tensile strength and high elongation are in these areas.
In fact if you look at all the property combinations,
the values of property combinations that you can get in a very broad
range of steels I think that at about a few thousand steels the percentage.
Even in this plot.
You see that, that particular combination the high strength and
high elongation is not that easily achievable.
At least it's not achievable in the known materials so far.
It doesn't mean it cannot be attained in any material, but it just means that in
the steals that have produced and tested so far, that combination has been elusive.
That's one of the observation you can make.
The second observation that becomes clear from a plot like this
Is that even in a single steel, if you look at one of these ellipses,
that ellipses represents a single grade of steel.
That means that the chemical composition is reasonably well controlled.
So even when you control the chemical composition of the particular steel,
you still get a substantial variance.
In the properties and
if you look at the ranges of these properties these are substantial.
And therefore, the challenge and
design because you don't know whether you can design for the property here or
can you design for the property there or somewhere in between.
Ideally what we would like to do is to do two things.
One is we want to control the property variance such whether.
Properties are more tightly controlled.
And we would like to get access to properties,
combinations that are of value in engineering design.
So we would like to produce new materials that would exhibit properties in
this area.
So if we want to do that, the next logical question is, what do properties depend on?
And if you think about it the properties depend on material internal structure.
That should be rather obvious because the material internal structure has to
control the properties.
However, the difficulty is the material internal structure is
extremely complex and in fact it spans many many land scales.
If you look at the smallest land scales and
in particular pay attention to the hale bars here.
So if you look at the scale of the nanometer below,
you'll see that the atomic structure is fairly well organized,
at least in metals, which are polycrystalline.
So in the crystalline solids or polycrystalline solids,
the atomic structure is rather well organized.
However, if you start looking at this area indicated as A and
B, you'll start seeing that there are defects.
Although it is crystal in and although it is ordered with a large extent,
it's by no means perfect and these defects are extremely important.
Here in this particular graph, you see another defect between the two arrows and
that's another kind of defect and that is a denominator length scale.
Now if you actually go to the micron lens care, which is
three orders of magnitude larger than a nanometer, you get other kinds of effects.
What you see here as lines are essentially what are called dislocations.
These dislocations represent defects that are missing half planes
within the crystal structure.
Within the three dimensional crystal structure.
And you can see that these locations
are arranged in very complicated ways at a much higher length scale.
Now we are talking about a 100 micron length scale which is still much,
much smaller than a macro scale material sample that you can hold in your hand.
At this length scale that is still yet
detail of the material Polycrystalline structure.
What you are seeing as individual creatures here, anyone of this creatures,
or you're seeing as those creatures, those are called grains or individual crystals.
So basically in anyone of this grains you have a crystalline structure
that may look like that.
But when you go to another grain the crystal structure,
the basic crystal structure might be the same.
But the orientation of the crystal structure might be different.
So in one of these grains the crystal structure might look like that.
And in another grain the crystal structure might look different.
So there is a rich heterogeneity and
a rich hierarchy of details in the material internal structure.
And one might initially think that these are a small number of defects and
the defects may not be that important.
and that would be a wrong assumption because material science has shown us for
many years that the defects structures and
how they get distributed in the internal structure of the material.
How they are spaced are all extremely important in controlling
the mechanical properties of the material.
In fact, defects rule the mechanical properties of the material.
So once we understand that, we come to the realization
that when we do it in [INAUDIBLE] with one material structure.
Again, although I'm only showing the molecular structure at one lens scale,
I want you to continuously think through the hierarchical structure of the material
at different land scales.
So one structure will produce one stress strain response, and if you now change
the structure and do a test on a sample that has a different structure.
You will get a different stress strain level and therefore the properties change.
So once we know how to control the structure and
again the structure is hierarchical across many length scales.
Then we can control the properties.
In order to simplify our terminology, we will simply refer to the materials
internal structure as microstructure or even simply structure.
And again I would like you to keep in mind that the word structure or
microstructure refers to the entire hierarchical stretch out the material
all the way from atomistic or subatomistic to the micro scale.
So there is a lot of detail in the word structure.
If we understand that the structure plays an important role in controlling
the properties, the next logical question to ask is,
how big is this structure space?
The word space here represents the set of all possible structures.
So the structure space is a very large and
unimaginably large space that are many structures one can think of.
If you think of the structure space as an ellipse
the elements of this space are structures.
So for example a point in that space might be one structure, another point in that
space might be another structure and they're different structures.
And the third point might be another structure.
So if you can start imagining the range of all possible structures you might be
interested in that space is a gigantic space especially when you
remember that structure here is hierarchical structure.
It's not structure just at one land scale.
Given that, how do we actually control the structure?
The answer to that comes from the middles processes..
The middles processes is actually hybrid processes.
Because it's not a single unit manufacturing step,
it in fact is an ordered sequence of unit manufacturing steps.
For example, in metals, when you make an ornament from a metallic alloy,
you typically go through a large number of steps.
That'll start all the way from refinement, to casting, shaping, machining,
joining and finishing.
And some of the steps are shown in the schematic below.
Process steps can be very complex, can involve many,
many steps before you get to the final product.
The space, the corresponding space of Hybrid Process Space,
as you can imagine again.
Is going to be very large.
Again, to get to that point, let's just introduce a notation where we say
hybrid process space is made of several unit manufacturing routes.
So one hybrid process space produces one structure.
And that structure has a property set.
If we go another hybrid process space, you'll get another structure and
another property set.
Now notice that all I did in going from HP1 to HP2 is
I simply changed the sequence of.
The unit process steps, P2 and P5.
The point I'm trying to make here is that,
it change in the sequence of the unit processes that are used in
the hybrid process can result in different microstructures with different properties.
It doesn't actually means that the hybrid process, this can be extremely large
because its the combinatorial combination of many unit processes.
So one can define hybrid process simply as a sequence of processes and
even if we have a limited set of unit processes.
One can make an unimaginably large number of combinations.
And therefore, the hybrid process space is extremely large.
Now we have seen the three spaces of interest.
We are interested in hybrid process space because that's the way we can
control the structures and once we control the structures.
We know or we can expect certain properties.
So what are we really interested in in a materials development?
In a materials development we want to first understand
how hybrid processes control structure.
We want to get these linkages.
We actually want to define these linkages, and
then we want to understand how they structure the property of interstate.
And again we want to be able to quantify and understand these linkages.
Of course if we start with a different process, we get a different structure and
a different combination.
And we might sometimes even be invested in structures for
which we don't know how to make them.
But we want to know what properties that produce because we want to know what
are the optimum properties that might be feasible theoretically feasible.
So one can ask and answer a lot of these questions if
one focuses on establishing structure property processing linkages.
So this is the focus of Matims development of words in that
we are constantly trying to establish that this Process-Structure-Property linkages.
In particular, we want lower dimensional representations of the PSP linkages.
We want them to be low-dimensional because these spaces are extremely large.
So if it takes awhile to evaluate each linkage, so
if you give me a process and it's going to take several weeks or
months to predict or estimate what the microstructure is going to look like.
Then searching through this space and
exploring this space is going to cause a huge problem.
We really want to focus our efforts on getting this low
dimensional representation of PSP linkages that's what we think
is going to get us to product design and process design in very efficient ways.
Once we establish this PSP linkages then we come to the real engineering problems.
In real engineering problems,
we are really interested in moving in the opposite direction.
In the previous slide we were actually going in this direction and
now we realized that for a real engineering application
We want to drive the information in the opposite direction.
What that means is for a real application, we are interested in properties,
of particular properties of interest.
And this property of interest then drives the micro-structure of interest.
And that then tells me what process I should be using.
We really want to go in the opposite direction in using these linkages.
And these endless problems are significantly challenging because as we
said before the spaces are extremely large and
this is opportunity that data science presents.
Data science and the things you learn in this class are going to help you
understand how to formulate this link.
I just particularly in an efficient ways that we can solve in most problems.
So in summary, what did we learn in this lesson?
We learned that the term property, structure, and
process have very specific meaning in the context of materials development efforts.
And these terms are heavily connected to each other.
And these connections are extremely important if we have to make better
products.
And these connections can be very complex but then the data science has a particular
opportunity to help us achieve these connections in very efficient ways.
Thank you.
