Title: Lecture 25: The Liquid State Nucleation and Metallic Glasses
1Lecture 25 The Liquid State Nucleation and
Metallic Glasses
- PHYS 430/603 material
- Laszlo Takacs
- UMBC Department of Physics
2The states of matter
- Gas - relatively simple
- There is a reference state, the ideal gas, that
can be understood in significant detail. Real
gases are usually described as deviations from
the ideal gas state due to interaction between
molecules. - Crystalline solid - relatively simple
- There is a reference state, the ideally periodic
infinite crystal, that can be understood by
focusing on the unit cell. Real crystals
deviate from the ideal state due to finite
crystallite size and defects. - Liquid - inherently complicated
- There is no good reference state. Liquids are
dense like solids with strong interaction between
the molecules, but their structure is random, at
least it has neither long range order nor unit
cell. There is some short range order, both
correlations in position and specific local
geometries. The structure is dynamic, the
molecules are in constant motion. - Intermediate cases
- Amorphous materials - dense, random, but static
- Quasicrystals - orientational order, but no
periodicity - Partially ordered materials, e.g. liquid
crystals
3Structure from diffraction experiments
The intensity of the diffracted beam depends on
the phase difference between elementary rays
scattered by individual atoms.
- Gas The relative atomic positions are random,
constructive and destructive interference average
out. Same as scattering by individual molecules. - Crystal If there is constructive interference
between corresponding atoms in two neighboring
unit cells, the interference is constructive
between atoms from every unit cell, diffraction
peak is observed. - Notice that XRD of crystals does not derive the
structure from scratch, but only selects the
structure from a list of possible structures
allowed by symmetries. A 1-d (or possibly 2-d)
measurement may not be able to reconstruct a 3-d
structure. - Liquids and glasses There is some short range
order and consequently an interference pattern,
but only statistical information can be obtained
from diffraction, the position of every atom
cannot be determined.
4Characterization of a liquid or amorphous
structure
- Radial distribution function the probability of
finding an atom at a distance r from an average
atom.
For a completely random structure (gas) or for
large distances where there is no more short
range order, RDF approaches a parabola
corresponding to the average atomic density.
- Notice that there is clear short-range order
- The nearest neighbors are quite clearly defined,
both their distance and their number (the area
under the first peak.) - The second and third coordination shell are
identifiable. - The curve approaches the 4?r2?0 parabola.
- The bars indicate the distance and number of
neighbors in crystalline hcp Zn. The short range
order in the liquid is quite different - RDF
cannot be obtained by broadening the crystalline
distances.
5What do we measure in diffraction?
- Scattered intensity corrected for instrumental
effects is the interference function, S(Q),
where Q is the scattering vector (the change of
the wave vector from the initial to the scattered
state, Q 4?/? sin?.) From S(Q) the radial
distribution function can be calculated
- This seems to be a straightforward procedure, but
- The interference function is known only to a
finite Q, the integral must be truncated. - Instrumental effects must be considered very
carefully, as the pattern consists of broad
features, not sharp lines.
6XRD results on amorphous FeP electrodeposited and
melt quenched alloys
- Interference functions. Notice that the top three
curves contain a small crystalline fraction. The
lowest two define the fully amorphous
concentration range, 16.7 to 22.9. - Hiltunen, Lehto, Takacs, 1986
RDF, P concentration increases down. Notice that
the short range order is similar but stronger
than in liquids. There is little variation
between samples, it is difficult to identify
trends.
7Partial correlation functions for Ni60Nb40
- The measured interference function is an average
of the partial interference functions
corresponding to the different atom pairs,
weighted with the atomic scattering amplitudes.
If three independent measurements with different
scattering amplitudes are performed, the partial
correlation functions can be separated. - The three measurements here were made using XRD,
neutron diffraction with natural Ni and neutron
diffraction on a sample made with 58Ni isotope. - Forgács, Hajdu, Sváb, J. Takács (1980)
8Ni60Nb40 partial RDFs
- There are clear differences between the curves,
e.g. in the first neighbor distance and the shape
of the split second peak. Unfortunately,
decomposing similar curves from independent
measurements, even different samples, results in
large experimental uncertainties. - Another method of obtaining similar results is
anomalous scattering of X-rays.
9Crystallization
- It is simple Over the melting point a material
is liquid, below it is crystalline. - No! Starting to form a new phase also requires
formation of a new interface!
10- Crystallization from the liquid state must always
start from a nucleus. - Homogeneous nucleation from a cluster within the
liquid. - Heterogeneous nucleation from a pre-existing
surface, e.g. wall of the container or surface of
impurity particle.
11The free energy balance of homogeneous nucleation
Volume free energy Surface energy ?, assumed
to be direction independent For a spherical
nucleus of radius r
If r is small, the positive surface term
dominates, the formation of a small nucleus is
not favorable. But if a large nucleus forms
against the odds, for large r the negative volume
term dominates, the nucleus is stable. The two
regions are separated by the radius of the
critical nucleus
The volume energy gain increases with increasing
undercooling, consequently the critical radius
decreases, that is random fluctuations can more
easily - and more frequently -create a nucleus
larger than the critical size. At Tm the volume
energy gain becomes zero, the critical size
approaches infinity, there is no nucleation.
12Heterogeneous nucleation
- Nucleation can start at pre-existing surfaces,
with substantial gain in surface energy. The most
typical places for heterogeneous nucleation are
the surface of the container and high-melting
particles present in the melt (seeds added
intentionally or impurities, e.g. oxides.) - Heterogeneous nucleation dominates close to the
melting point, but homogeneous nucleation becomes
important at low temperatures, as the number of
heterogeneous nucleation sites is constant, but
the number of homogeneous nuclei increases with
decreasing temperature. - In order to achieve large undercooling,
heterogeneous nucleation has to be avoided. The
best is a small droplet with no room for a seed
particle, levitated freely in a rf magnetic
field. In industrial settings only a few degrees
of undercooling take place, but 15 of Tm is
possible in the laboratory.
13The nucleation rate
- The nucleation rate is proportional to the
probability of forming a nucleus larger than
critical size
Close to Tm the size and therefore the free
energy of the critical nucleus is large, the
nucleation rate is very slow, substantial
nucleation takes a very long time. At low
temperature, T in the denominator is small,
nucleation is slow because of the slow
dynamics. In between there is a degree of
undercooling where homogeneous nucleation is the
fastest. Growth is a diffusion-driven process
that has a rate according to the Fulcher equation
of
Finish Start
The fastest nucleation happens close to the
temperature a the tip of the curve.
Crystallization - can be avoided, by fast enough
cooling to avoid the nucleation line. This is how
metallic glasses are made.
The temperature variation of the nucleation and
growth rates explain the shape of the TTT -
time-temperature-transformation diagram.
14What materials can form a glass?
- Some materials - like mixtures of oxides,
chalcogenides, complex organic materials - are
easy glass formers. In those cases nucleation is
very difficult because a large cluster of
molecules must fit together by random motion to
acct as a crystalline seed. Many of those
materials do crystallize if cooling is very slow
or seeds for heterogeneous nucleation are
provided. - The structure of metals is simpler, thus
nucleation is easier. Until 1959 it was believed
that metals were always crystalline. This is not
the case. Crystallization can be avoided also in
metallic melts, if the difference between Tm and
T0 is small. In a typical alloy system Tm can be
strongly temperature dependent - deep eutectic
points are advantegous - while T0 varies
relatively little. - 1960 P. Duwez, Au-Si, later Pd-Si
- Late 1960s The first ferromagnetic metallic
glasses Fe80P12C8 - 1975 The METGLAS 2605 family (Fe80B20 and
related compositions) - Late 1970s Ni-Nb and other early-late transition
metal systems - 1985-1995 a decade of decreasing interest
- 1996 Inoue (Tohoku, Sendai) and W. Johnson
(CalTech, Pasadena), bulk metallic glasses, such
as Al-Y-Ni, La-Al-Ni Zr-Ni-Al-Cu, Zr-Ti-Cu-Ni-Be
and Mg-Cu-Y
15Some typical metallic glass forming systems
- Au-Si the first one Fe-B ferromagnetic
Nb-Ni metal-metal
16Forming traditional metallic glasses requires
106 C/s cooling rate - possible only for thin
ribbons or sheets.Melt spinning is the most
frequently used method to reach such cooling
rates.
17STM images of a crystalline and glassy Zr alloy
18Mechanical property comparison for a bulk
metallic glass
19Other subjects related to metallic glasses
- Stability and crystallization
- E.g. Fe(80)B(20) melt ? Fe(80)B(20) glass ? Fe
Fe4B ? Fe Fe3B ? Fe Fe2B - Magnetism - no crystal structure, no
magnetocrystalline anisotropy, stress sensitivity
can be minimized by varying the composition - E.g. (Fe1-xCox)75Si15B10 at about x 0.9
- Partially crystallized magnetically hard-soft
alloys. -
- Other materials produced by rapid quenching
Quasicrystals, metastable compounds