1
00:00:00,030 --> 00:00:03,430
Let's a let's get started again. Welcome back.

2
00:00:03,450 --> 00:00:06,540
This is now the 20th lecture of the condensed matter, of course.

3
00:00:06,900 --> 00:00:11,820
When we last left off, we were talking about magnetism and we thinking about magnetism one atom at a time.

4
00:00:12,150 --> 00:00:16,830
And in this lecture, we're going to upgrade and start thinking about collective magnetism.

5
00:00:17,190 --> 00:00:27,719
In other words, magnetism. When we have lots of atoms collecting magnetism, using the theme that more is different.

6
00:00:27,720 --> 00:00:31,140
When you put lots of things together, you can get new and different physics.

7
00:00:31,140 --> 00:00:35,580
Now, we did start a little bit of collective magnetism last time when we were talking about rules.

8
00:00:35,580 --> 00:00:43,379
There was the when we talked about hundreds of molecule, there was the driving force by horns roll to try to align spins between atoms in a molecule.

9
00:00:43,380 --> 00:00:49,709
But there is also the covalent bonding physics which tried to anti align spin so you can make a singlet or covalent bond.

10
00:00:49,710 --> 00:00:53,100
Those two things compete with each other and getting the details right to figure

11
00:00:53,100 --> 00:00:56,730
out whether the spins will align or anti align is sort of a complicated process.

12
00:00:57,000 --> 00:01:04,649
And we're going to, you know, brush that entirely under the rug and try to take a step back and look at a model that's going to sort

13
00:01:04,650 --> 00:01:09,810
of give us much of the physics that we're interested in without getting too bogged down in the details.

14
00:01:10,080 --> 00:01:17,460
So we have a couple of approximation in our model. So our toy model will have one know electron hopping.

15
00:01:18,540 --> 00:01:22,379
So we don't have to worry about any sort of band physics.

16
00:01:22,380 --> 00:01:26,370
We don't have to worry about tight binding model. You're not too worried about Pauli power magnetism.

17
00:01:26,370 --> 00:01:35,190
All of that has been thrown out. So with no electron hopping, we can assume that on each atom there is a spin, spin.

18
00:01:35,880 --> 00:01:39,000
S.I. On each atom. On each atom.

19
00:01:39,240 --> 00:01:46,920
Adam I and I put spin in quotes because this spin, what I really mean is a magnetic moment.

20
00:01:47,190 --> 00:01:54,419
That magnetic moment may be from an actual electron spin. It may be from many electron spins, it may be orbital angular momentum.

21
00:01:54,420 --> 00:01:57,899
It may be some combination of orbital and spin, angular momentum.

22
00:01:57,900 --> 00:02:03,930
But generally when you sort of make these so toI models, you just call it spin no matter what it's actually from.

23
00:02:04,920 --> 00:02:11,670
And then we'll write down the Hamiltonian. So here's a Hamiltonian three H equals and then it has two pieces.

24
00:02:11,670 --> 00:02:23,489
The first piece should look familiar g you b b dot s i that's just the coupling of the spin to the external magnetic field.

25
00:02:23,490 --> 00:02:29,890
This is the term that gave us power magnetism. Then we're going to add to it another term, some over.

26
00:02:30,090 --> 00:02:36,510
I'm going to put up minus sign and one half out front some I and j over atoms i

27
00:02:36,540 --> 00:02:47,489
comma j of j i j as I thought as j so this term here is an interaction term.

28
00:02:47,490 --> 00:03:06,540
So j j is the interaction between atom i and J between atoms i and J and we will set j j equal to j j i.

29
00:03:08,910 --> 00:03:16,470
And this is also sometimes called also equals known as exchange, the exchange constant.

30
00:03:18,240 --> 00:03:22,110
This is how the word exchange is used in condensed matter physics most of the time,

31
00:03:23,040 --> 00:03:27,180
and the factor of one half is put in there to prevent over counting.

32
00:03:27,330 --> 00:03:31,170
We want j12 to be the interaction between atom one and two.

33
00:03:31,380 --> 00:03:39,390
But in that sum there'll be a term S1 dot S2 and another term which is S2 S1 we don't want to over count them, so we just put a one half out front.

34
00:03:39,810 --> 00:03:49,139
Okay. So this model, model three, the Hamiltonian three three is known as the three meaning of the Hamiltonian three

35
00:03:49,140 --> 00:03:54,000
is known as the Heisenberg model after everyone's favourite uncertain physicist.

36
00:03:54,000 --> 00:04:04,120
Heisenberg model. Very frequently studied model in condensed matter physics and a very good representation of of many magnets.

37
00:04:04,360 --> 00:04:08,020
Now we're going to simplify this even further. Simplify.

38
00:04:11,050 --> 00:04:24,330
By setting j aj equal to j if I and J or neighbours i j neighbours, and we're going to set it equal to zero.

39
00:04:24,340 --> 00:04:33,880
Otherwise, in other ways we're just going to keep track of interaction between spins that happen to be nearest neighbours with each other.

40
00:04:33,940 --> 00:04:37,870
It's a fairly good approximation that spins on atoms that are close to each other,

41
00:04:37,870 --> 00:04:40,870
see each other, and if they're farther from each other, they don't see each other.

42
00:04:41,110 --> 00:04:45,999
And then we're also going to simplify. So this is simplification one and simplification two is four.

43
00:04:46,000 --> 00:04:51,010
Now we're going to set the external magnetic field to zero, and we may put it back in later.

44
00:04:51,430 --> 00:05:00,909
So our Hamiltonian now takes the form H is minus one half sum over i j and I'll write like this.

45
00:05:00,910 --> 00:05:04,270
So this notation means I and J our neighbours.

46
00:05:08,160 --> 00:05:12,580
Bracket notation saw a standard j.

47
00:05:12,610 --> 00:05:16,649
S I thought s. J. So we're only swimming over.

48
00:05:16,650 --> 00:05:20,379
I enjoy being neighbours and the constant is just a constant j here.

49
00:05:20,380 --> 00:05:24,420
And there's a couple of possibilities of what you can get out of this sort of Hamiltonian.

50
00:05:25,590 --> 00:05:28,620
So possibility one So let's write down case one.

51
00:05:30,210 --> 00:05:39,090
Case one is that J is greater than zero, in which case in order to make the energy of that Hamiltonian as small as possible,

52
00:05:39,270 --> 00:05:42,749
the two spins on adjacent sides will try to align with each other.

53
00:05:42,750 --> 00:05:57,750
So spins align. And when the spins align, what you get is for magnetism, ferromagnetic magnetism, magnetisation, which is not equal to zero.

54
00:05:58,020 --> 00:06:06,989
Now it doesn't matter in this Hamiltonian which direction the spins all align in, as long as the spins are all aligned with each other.

55
00:06:06,990 --> 00:06:11,850
So they can't be pointing up. They can all be pointing down They can all be pointing left I'll be pointing the right same.

56
00:06:11,850 --> 00:06:15,810
You'll end up with the same energy. What's important is all the spins align with each other.

57
00:06:15,840 --> 00:06:23,729
Okay, so this is the easy case to study. The more complicated case to study is the case.

58
00:06:23,730 --> 00:06:33,990
Two is j less than zero now with j less than zero the neighbours want to anti align.

59
00:06:37,780 --> 00:06:43,180
In order to make the Hamiltonian as small as possible or the energy as small as possible.

60
00:06:43,510 --> 00:06:47,080
So then what we have is if we have a let's see right here.

61
00:06:47,440 --> 00:06:55,510
So if we have like a square lattice like this, each with a spin on it, if the first spin is pointing up,

62
00:06:56,110 --> 00:07:00,489
then the second spin will want to point down in order to make the energy as small as possible.

63
00:07:00,490 --> 00:07:06,639
And then the third spin will point up this one will point down this one, point up this one point down so forth and so on.

64
00:07:06,640 --> 00:07:10,030
And we'll get these alternating spin structures.

65
00:07:10,270 --> 00:07:12,340
This is what's known as an antiferromagnetic.

66
00:07:17,800 --> 00:07:28,810
Or sometimes known as a nail state and e e l natural state after Louis Nile, who realised that these things actually exist in nature.

67
00:07:31,000 --> 00:07:35,980
And then what's interesting about this is that they have magnetic order, magnetic order,

68
00:07:37,840 --> 00:07:45,280
but they have no net magnetisation and equals zero y, they have no net magnetisation.

69
00:07:45,280 --> 00:07:48,729
Well, because you have the same number of up pointing spins as you have down pointing spins.

70
00:07:48,730 --> 00:07:53,140
Well, at least if I, if I drew them more aligned, it would look like there is an equal number of spin.

71
00:07:53,350 --> 00:07:56,830
Pointing spins is down. Pointing spins. My drawing isn't so good.

72
00:07:57,010 --> 00:08:03,460
So how is it that Louis now knew that this sort of situation could occur?

73
00:08:03,490 --> 00:08:10,150
Well, on one of the homework assignments, the last homework assignment, you are meant to calculate the magnetic susceptibility of a state like this.

74
00:08:10,390 --> 00:08:14,440
And there's actually some signature in the magnetic susceptibility that you can look for,

75
00:08:14,560 --> 00:08:21,880
which was which was the thing that Niall is looking at and what made him realise that these Antiferromagnetic states actually exist.

76
00:08:22,780 --> 00:08:32,020
But in the modern era there's actually a very direct way to see this sort of physics, which is to do neutron scattering on it.

77
00:08:32,030 --> 00:08:43,840
So let me actually you write that down. You can see with neutron scattering or neutron diffraction, I guess we should call it.

78
00:08:47,200 --> 00:08:57,729
So the elementary lattice constant here might be a, but the neutrons actually see a lattice constant, which is this big.

79
00:08:57,730 --> 00:09:06,129
This is the unit cell now because the neutrons, if a neutron is coming in with spin up, it sees an up spin as being a different atom from a down spin.

80
00:09:06,130 --> 00:09:09,640
And the reason that neutron seeing spin is being different from a down spin is the

81
00:09:09,640 --> 00:09:14,680
neutrons themselves have spin and they interact differently with the neutron.

82
00:09:14,680 --> 00:09:20,500
Coming in is a spin up atom. It will interact differently with this atom than with this atom because their spins are anti aligned.

83
00:09:20,710 --> 00:09:26,110
So it sees a unit cell which is this big a prime. So when you do your neutron diffraction experiment,

84
00:09:26,380 --> 00:09:33,130
you will see a different size of the unit cell when antiproton magnetism occurs, which will get a bigger unit cell.

85
00:09:34,690 --> 00:09:37,629
I should actually comment one more thing here, which is a little subtle,

86
00:09:37,630 --> 00:09:44,980
but there's a homework assignment on it as well that in this picture I've drawn or I should have drawn if I was a better artist,

87
00:09:45,280 --> 00:09:49,509
outspends down spins ups, spins ups and downs, spin. And that's a very classical way of thinking.

88
00:09:49,510 --> 00:09:54,250
I'm thinking of these spins as actually being, you know, vectors that you can point in any direction in space.

89
00:09:54,550 --> 00:09:59,650
If you have large spins in the sense of spin, you know, five halfs or spin as equals,

90
00:09:59,650 --> 00:10:03,280
two as equals, nine halves or as equals, three or four or something like that.

91
00:10:03,520 --> 00:10:11,770
Then having this picture in your head of these spins is just being vectors that can point any direction in space is a fairly good classical picture,

92
00:10:11,770 --> 00:10:16,389
which is fairly accurate, but if you're talking about spin one halves, then it's not so good anymore.

93
00:10:16,390 --> 00:10:21,430
You really have to think in terms of quantum mechanics and you can't draw these pictures of ups and downs, spin ups and downs,

94
00:10:21,430 --> 00:10:28,270
spin and spin one half Antiferromagnetic is a very complicated beast, and it's not really right to think of it this way.

95
00:10:28,270 --> 00:10:34,059
This is sort of more of a classical intuition that works for large, large spins, but not so well for small spins.

96
00:10:34,060 --> 00:10:37,930
And that will be presumably worked out in a homework assignment.

97
00:10:38,890 --> 00:10:42,070
All right. So one more comment about this.

98
00:10:42,070 --> 00:10:49,300
This case of anti from a magnetism is that you can have antiferromagnetic antiferromagnetic coupling

99
00:10:50,110 --> 00:10:57,130
from ag on a triangular lattice and then you'll get something really strange triangular lattice.

100
00:10:59,920 --> 00:11:03,910
So let's imagine we have a triangle here, three atoms in triangle.

101
00:11:04,720 --> 00:11:12,100
And if the first one happens to have his spin up, the second one up here wants to have a spin pointing in the opposite direction from this one.

102
00:11:12,100 --> 00:11:17,350
So he points spin down and then this one wants to have an opposite point in spin to each of his neighbours.

103
00:11:17,350 --> 00:11:23,319
But he's in trouble. He can point up and then make the opposite direction of this guy, but the same direction as this guy.

104
00:11:23,320 --> 00:11:26,710
Or he can point down which case he's opposite this guy, but same direction at this guy.

105
00:11:26,980 --> 00:11:30,130
So this guy's in trouble and he can't point happily either direction.

106
00:11:30,400 --> 00:11:36,400
So this is what you call a frustrated, frustrated antiferromagnetic.

107
00:11:41,240 --> 00:11:43,910
Everything causes frustration these days.

108
00:11:44,270 --> 00:11:52,820
Okay, so one more case of this, a.k.a. anti-terror magnetism or similar physics to anti fender magnetism that I want to mention,

109
00:11:53,180 --> 00:12:08,959
is the possibility that you have a basis or let's consider a two atom basis are two, two or more or more atom basis for our lattice for our crystal.

110
00:12:08,960 --> 00:12:18,890
So let's consider, for example, a big atom, a small atom, a big atom, small atom, big atom, big them small and big and small.

111
00:12:18,980 --> 00:12:22,490
Big. It's not like this. So here's a two atom basis.

112
00:12:24,920 --> 00:12:35,270
Okay, now imagine that we have antiferromagnetic coupling where the neighbouring atoms want to point their spins in opposite direction.

113
00:12:35,270 --> 00:12:41,780
So the big atom say points up and then the small atom points down big atom points up, small and points down.

114
00:12:42,140 --> 00:12:48,350
Big Adam points up. Okay, we can keep going here and you get the idea and.

115
00:12:48,710 --> 00:12:51,860
Okay, up, down, up, down, up.

116
00:12:53,510 --> 00:12:59,570
If the spin on the big atom is not equal to the spin on the small atom,

117
00:13:01,160 --> 00:13:08,630
then what we'll have is will have magnetisation not equal to zero because you'll have more magnetisation pointing up.

118
00:13:08,930 --> 00:13:12,350
Then you have magnetisation pointing down. Is that is that clear how that works?

119
00:13:12,560 --> 00:13:18,050
So even though you have antiferromagnetic coupling and each neighbouring spin wants to point in the opposite direction,

120
00:13:18,230 --> 00:13:23,870
you get a net magnetisation left over because some of the spins are bigger than the other spins.

121
00:13:23,870 --> 00:13:29,780
And this is what is known as very magnetism, very magnetism.

122
00:13:32,540 --> 00:13:37,340
And it may seem like it's kind of strange or unusual, but it's actually quite common.

123
00:13:37,610 --> 00:13:44,000
In fact, the world's most common magnet occurring in nature, its ion three oxide four,

124
00:13:44,210 --> 00:13:48,950
otherwise known as magnetite or lodestone that you can find in the earth sometimes.

125
00:13:48,950 --> 00:13:55,849
And it's magnetised when you pull it out of the earth and it will stick. Your refrigerator directly is actually a fairy magnet.

126
00:13:55,850 --> 00:14:01,100
It's made up of more than one type of atom, larger, larger moments on one type of atom,

127
00:14:01,100 --> 00:14:03,470
smaller moments on the other, and they aren't aligned with each other.

128
00:14:05,300 --> 00:14:12,050
Then there's an argument in the community as to whether you should call a very magnet, a subset of a magnet.

129
00:14:12,290 --> 00:14:15,410
And this is where, you know, people will argue with each other till they're blue in the face.

130
00:14:15,890 --> 00:14:19,730
And there's no actual good answer to this because people use it both ways.

131
00:14:19,970 --> 00:14:26,660
So some people say that a very magnet is a subset of a fellow magnet because it has non-zero magnetisation.

132
00:14:26,810 --> 00:14:31,580
It will stick to your refrigerator, you know, as the property that we think of as a magnet.

133
00:14:32,540 --> 00:14:38,810
On the other hand, some people think that the word foul magnet should be reserved for the case where all the spins point in the same direction.

134
00:14:38,960 --> 00:14:43,910
And we should call this a fairy magnet to distinguish it from the case where all the signs are pointing in the same direction.

135
00:14:43,910 --> 00:14:49,100
So anyway, you should just be aware that that's there's that little conflict of nomenclature in the community.

136
00:14:49,430 --> 00:14:57,830
Okay, so far so good. We're doing well still. All right. So let's move on to actually maybe write it again.

137
00:14:58,070 --> 00:15:01,940
We'll write down the Heisenberg Hamiltonian H Heisenberg again,

138
00:15:02,060 --> 00:15:14,480
minus one half some of our neighbours i j j as I talked SJ and as I mentioned and let's we're going to sort of focus mostly on J greater than zero.

139
00:15:14,480 --> 00:15:16,670
In other words, the ferromagnetic case.

140
00:15:18,800 --> 00:15:25,940
And as I mentioned this, a Hamiltonian doesn't care which direction the spins point as long as they're all pointing in the same direction.

141
00:15:26,180 --> 00:15:33,350
Now, unfortunately, most materials aren't like that. Most materials actually prefer the spins to point in some particular direction.

142
00:15:33,650 --> 00:15:38,960
And so this is in order to represent that, we need to add another term to the Hamiltonian.

143
00:15:39,110 --> 00:15:41,839
Actually, first, let's this is something I mentioned in the last lecture,

144
00:15:41,840 --> 00:15:48,530
but I'll mention again one thing that would make it obvious that the directions aren't all equivalent is if you imagine

145
00:15:48,530 --> 00:15:54,410
having a very long tetrad and total unit cell where you know the crystal is very long and thin in one direction,

146
00:15:54,680 --> 00:16:01,399
then you might imagine that maybe the spins want to point along, particular the long thin axes.

147
00:16:01,400 --> 00:16:09,980
Maybe so you might have as wanting to be up or wanting to be down, but not any direction in between.

148
00:16:10,040 --> 00:16:15,440
So this is sort of a case where you might expect this sort of anisotropy to exist.

149
00:16:15,710 --> 00:16:21,170
So in order to represent that, we write down an additional term of the Hamiltonian, D, H and I,

150
00:16:21,170 --> 00:16:30,860
Satrapi or H and I, Satrapi will call it D, I'm not sure which might take the form some overall atoms.

151
00:16:31,820 --> 00:16:37,970
Let's put a minus a Kappa out front. That's some constant RSI in the Z direction.

152
00:16:38,510 --> 00:16:44,060
Squared. So why is this going to prefer having the spin in some particular direction?

153
00:16:44,300 --> 00:16:47,540
Well, it wants to maximise the magnitude of Z.

154
00:16:47,780 --> 00:16:54,859
So if C is pointing in the in the direct up direction or the direct down direction, this Hamiltonian is as small as possible.

155
00:16:54,860 --> 00:17:03,320
It gets as negative as you possibly can as compared to having as point in the in on the as the equals zero plane where this gets less negative.

156
00:17:03,680 --> 00:17:09,440
So this kind of term, this anisotropy term is one we're going to consider a lot.

157
00:17:09,740 --> 00:17:18,680
So maybe I'll even write down likes to have s up or down, not as pointing sideways.

158
00:17:19,730 --> 00:17:26,510
Now for a cubic crystal. You might expect something d h cubic, which is a little more complicated.

159
00:17:26,510 --> 00:17:30,200
That would allow the spin to point along any of the crystal and axes,

160
00:17:31,400 --> 00:17:36,500
but not in between the crystal axes in a appropriate Hamiltonian for that might be

161
00:17:36,800 --> 00:17:46,430
something like this as i x the fourth plus s i y to the fourth plus s i z to the fourth.

162
00:17:46,790 --> 00:17:51,530
And the reason I wrote this to the fourth power is because if I wrote them to the squared power,

163
00:17:51,770 --> 00:17:56,000
then you just get a squared percent y squared, percent z squared, which is just s squared.

164
00:17:56,180 --> 00:18:00,500
And that doesn't care which direction the spin is pointing because it's just the magnitude of the spin.

165
00:18:00,740 --> 00:18:04,400
But if you write it as force as the fourth, that's why the fourth is Z the fourth.

166
00:18:04,610 --> 00:18:09,710
Then this prefers to have this been pointing along one of the crystal axes, but not in-between the crystal axes.

167
00:18:10,040 --> 00:18:14,150
So we might consider this one a little bit as well, but mainly we're going to focus on this one.

168
00:18:14,630 --> 00:18:19,160
Just one comment, comment again for small.

169
00:18:19,670 --> 00:18:23,810
What we've written here is sort of based on our intuition for spins.

170
00:18:24,020 --> 00:18:28,910
When we're thinking them as classical VEC, we're thinking of them as classical vectors for small spins.

171
00:18:29,300 --> 00:18:34,340
Small spins are these can be trivial.

172
00:18:34,580 --> 00:18:38,420
These are these terms can be trivial.

173
00:18:39,830 --> 00:18:46,730
What I mean by that. Well, suppose I tried to write this down for a spin one half.

174
00:18:47,180 --> 00:18:51,250
You remember this spin? One half poly operator Sigma Z.

175
00:18:51,980 --> 00:18:56,690
The Pauli Spin Matrix. If you square it for spin one half, spin one half.

176
00:18:57,290 --> 00:19:03,940
The Spin Matrix. Sigma Z squared is just one. It's a very uninteresting operator just gives you a constant.

177
00:19:04,090 --> 00:19:08,200
So if you try to write down a term like this for spin one half spin, it would do nothing.

178
00:19:08,230 --> 00:19:11,320
It wouldn't specify which direction the spin wants to point in.

179
00:19:11,710 --> 00:19:18,310
However, if you did it for a spin two or spin three, it would. That would be an analogous spin matrix, which does not square to one.

180
00:19:18,310 --> 00:19:25,030
It becomes non-trivial. So that's just something to be warned about. You can't write down anisotropic terms for small enough spins.

181
00:19:25,300 --> 00:19:28,990
Okay. All right. So we need to think about this.

182
00:19:29,000 --> 00:19:34,780
And I. Satrapi, which will prefer the spins to point in some directions over other directions.

183
00:19:34,780 --> 00:19:36,580
And we need to think about two possibilities.

184
00:19:36,910 --> 00:19:44,320
One possibility is that the anti Satrapi is strong and the other possibility is that the anisotropy is weak.

185
00:19:44,530 --> 00:19:53,769
So maybe I'll write down the weak case over here. We can I SATRAPI By which I mean that this kappa,

186
00:19:53,770 --> 00:20:02,620
this coefficient that occurs in the anisotropy term is much less than the j the exchange constant that occurs in the Heisenberg term.

187
00:20:02,950 --> 00:20:16,300
Now, in this case, what one has to do is one has to first or most importantly first use spins align with neighbours, spins align with neighbours.

188
00:20:18,280 --> 00:20:24,640
And when I say that first, I mean it's most important to align with your neighbours because J is the big energy scale in the system.

189
00:20:24,880 --> 00:20:35,470
Once you've aligned with your neighbours, second you align along a line with anisotropy, with anisotropy axis.

190
00:20:41,260 --> 00:20:47,920
So first you align with your neighbours and then everyone decides they're better off if they all align with the isotropic axis.

191
00:20:48,190 --> 00:20:52,480
Now if you compare that to the case over here, strong, anisotropy,

192
00:20:53,500 --> 00:21:03,880
strong and I Satrapi meaning this Kappa, this isotropic constant kappa is much bigger than J.

193
00:21:04,360 --> 00:21:11,860
Then first you want to align along and I Satrapi axis align along and I Satrapi.

194
00:21:15,620 --> 00:21:19,470
For example, all spins will point either up or down.

195
00:21:19,490 --> 00:21:24,260
If we're using that up and down and I SATRAPI that everyone has to point either up or down

196
00:21:24,290 --> 00:21:28,250
that's what you do first because that's the big energy scale in the second in the system.

197
00:21:28,550 --> 00:21:31,580
But then second, you align with neighbours.

198
00:21:31,970 --> 00:21:42,620
Align with neighbours. Now this may seem like it's a trivial distinction because at the end of the day the ground state is the same.

199
00:21:42,890 --> 00:21:47,530
The ground state is that everyone's aligned with their neighbours and everyone is aligned along the anisotropy axis.

200
00:21:47,540 --> 00:21:52,400
So why did I make a big deal about this? Well, it makes a difference. The two limits are different.

201
00:21:52,700 --> 00:21:55,460
If you think about what the low energy excitations are.

202
00:21:55,700 --> 00:22:08,330
So in this case, the low energy excitations, energy expectations are small, angular rotations of a whole bunch of spins.

203
00:22:08,420 --> 00:22:15,139
Everyone stays more or less aligned with their neighbours, but they rotate just a little bit from the anisotropy axis.

204
00:22:15,140 --> 00:22:23,000
So this is a small rotations. Small rotations staying aligned.

205
00:22:25,750 --> 00:22:38,080
Aligned with neighbours. Whereas over here the low energy excitations are completely different.

206
00:22:38,440 --> 00:22:47,700
Low energy excitations. Here.

207
00:22:47,710 --> 00:22:51,340
The thing that's important is that you have to stay aligned with the anisotropy axis.

208
00:22:51,550 --> 00:22:59,980
So the low energy excitation is a flip, some spin flips a spin, but keep it aligned with the anisotropy axis.

209
00:22:59,980 --> 00:23:03,010
But stay aligned but stay on.

210
00:23:03,970 --> 00:23:12,690
And I saw isotropic axis. Is it clear what the difference is between these two?

211
00:23:12,730 --> 00:23:19,139
Okay. So this is an interesting case here. And it's sufficiently interesting that people study in a lot and they write

212
00:23:19,140 --> 00:23:27,270
down an effective model for this one model which has the following properties.

213
00:23:27,270 --> 00:23:41,250
We write some sigma I, which is plus or minus s the magnitude of the spin, and then s I is as a vector is sigma I in the Z direction.

214
00:23:41,260 --> 00:23:48,960
So basically we're, we're only letting the spin have complete magnitude, pointing exactly up or full magnitude pointing exactly down.

215
00:23:49,410 --> 00:23:59,520
And then we can write out a Hamiltonian very similar to the Heisenberg form in Heisenberg form h equals.

216
00:23:59,520 --> 00:24:10,950
Okay, so that would be minus one half sum over i j neighbours j sigma i sigma j and then plus we'll couple it to the external field.

217
00:24:11,310 --> 00:24:14,610
G might be the sigma i.

218
00:24:15,740 --> 00:24:18,740
This Hamiltonian is known as the easing model.

219
00:24:20,360 --> 00:24:24,200
Easing model? Because it was invented by lens.

220
00:24:24,680 --> 00:24:25,370
Not easing.

221
00:24:25,790 --> 00:24:36,260
Easing was the graduate student who Lenz gave this to as a project in 1925, and he solved this for his Ph.D. project in one and one dimensional case.

222
00:24:37,400 --> 00:24:45,229
Now, I'm going to make a big deal about this easing model, because this easing model, it's extremely simple, and yet it's also extremely complex.

223
00:24:45,230 --> 00:24:48,650
It's simple because on each side there's only one degree of freedom.

224
00:24:48,650 --> 00:24:52,040
And either it points up or it points down, there's no other possibility.

225
00:24:52,040 --> 00:24:56,540
So it's very simple degrees of freedom. And yet and it couples only to its neighbour.

226
00:24:56,690 --> 00:25:03,559
And yet the physics of this model are extremely complex. And it's probably not an exaggeration to say the world has learned more about

227
00:25:03,560 --> 00:25:08,720
statistical mechanics from this one model than from anything else we possibly study.

228
00:25:09,560 --> 00:25:14,480
So using solve this model in 1925, in one dimension during World War Two,

229
00:25:14,480 --> 00:25:19,310
and Sager very famously managed to solve it without magnetic field in two dimensions.

230
00:25:19,580 --> 00:25:26,810
No one has solved the model, either in two dimensions with a magnetic field or in three dimensions or any higher dimension since then.

231
00:25:27,050 --> 00:25:31,910
So it's been a lot of study. Thousands and thousands of papers have been written about this model since then.

232
00:25:32,060 --> 00:25:38,930
And I should mention, just in the last few years, there's been a really valiant assault against solving this model in three dimensions,

233
00:25:38,930 --> 00:25:42,259
and maybe within a few years there will be a solution. Maybe not.

234
00:25:42,260 --> 00:25:50,420
We can be optimistic. But the reason that this is this model is so interesting is because it shows an awful lot of interesting physics,

235
00:25:50,720 --> 00:25:57,590
particularly about phase transitions, even in this this model that only has one degree of freedom on each on each site.

236
00:25:59,510 --> 00:26:07,730
Okay. So we'll come back to the easing model actually in the next lecture and then we will ask the question where we're going to put it here.

237
00:26:09,080 --> 00:26:13,250
Now let's put it over here. Here. Okay. For any model that we write down.

238
00:26:13,250 --> 00:26:22,490
So starting here, if we have a Hamiltonian, which is H of the Heisenberg form plus the anisotropy form.

239
00:26:25,220 --> 00:26:30,110
And if the anisotropy is strong, we can just replace the combination with the easing Hamiltonian.

240
00:26:30,640 --> 00:26:34,190
Does. Does the whole system align?

241
00:26:35,030 --> 00:26:40,550
The system align align at t equals zero.

242
00:26:43,340 --> 00:26:53,420
And in the model the answer is yes. In model yes, but in real magnets, no.

243
00:26:59,120 --> 00:27:03,889
So in real magic magnets, not all the spins in the system actually do align with each other.

244
00:27:03,890 --> 00:27:09,560
And to understand why this is, we have to realise that we actually left something important out of our Hamiltonian.

245
00:27:09,920 --> 00:27:16,459
So let's do a little bit of a docking experiment. Let's imagine you have a big chunk of some magnetic material.

246
00:27:16,460 --> 00:27:20,790
Another big chunk of some magnetic material. The this one has magnetisation here.

247
00:27:20,840 --> 00:27:24,980
This one has a magnetisation here. Then you bring these two chunks of materials together.

248
00:27:25,280 --> 00:27:29,090
Well, what happens when you've probably done this when you were a kid at some point,

249
00:27:29,300 --> 00:27:33,140
and when you bring them together, you realise that they want to sit like this.

250
00:27:33,680 --> 00:27:36,680
They want to sit head to tail because they have a dipolar interaction.

251
00:27:36,890 --> 00:27:42,530
And so the magnets want to come together, head to tail, and they really don't want to come together head to head like this.

252
00:27:42,800 --> 00:27:47,150
So this is telling us that we we left out some important piece of physics.

253
00:27:48,230 --> 00:27:53,030
And what we left out we left out magnetic energy.

254
00:27:54,110 --> 00:27:59,929
Magnetic energy. That's one way of thinking about it. Or dipole.

255
00:27:59,930 --> 00:28:00,950
Dipole interaction.

256
00:28:08,970 --> 00:28:18,060
In the last lecture, I mentioned that if you were thinking about two spins in a given atom and asking whether they want to align or not,

257
00:28:18,390 --> 00:28:23,970
the dipole dipole interaction between two spins in atom is insanely small, so you could ignore it.

258
00:28:24,390 --> 00:28:30,060
That's true. But if you have a big chunk of the material, you have an insanely large number of electrons there,

259
00:28:30,060 --> 00:28:37,050
each with an insanely small amount of dipole moment, and you put them all together, actually becomes a reasonably big dipole moment.

260
00:28:37,200 --> 00:28:40,440
And you do have to worry about this dipole dipole energy.

261
00:28:41,040 --> 00:28:48,150
Another way of thinking about it, exactly the same physics is to well, okay,

262
00:28:48,450 --> 00:28:55,829
if you think about the magnetic field lines in these two pictures here, magnetic field lines look kind of like this in this picture.

263
00:28:55,830 --> 00:28:59,280
Whereas in this picture, the magnetic field lines kind of look like this.

264
00:28:59,580 --> 00:29:08,280
And you remember with the magnetic energy, total metallic energy is integral d3r of B squared over two of you not.

265
00:29:08,670 --> 00:29:14,010
And there's the total B squared is a lot smaller if you align your dipoles head to tail.

266
00:29:14,010 --> 00:29:19,140
So you're reducing magnetic energy by putting your dipoles had to tail in this way.

267
00:29:20,190 --> 00:29:23,430
So what actually happens in a real physical system?

268
00:29:23,730 --> 00:29:28,980
So we're going to write H equals H, Heisenberg, Heisenberg plus H and I,

269
00:29:28,980 --> 00:29:35,940
Satrapi and Satrapi plus we're going to add H dipolar interaction or magnetic?

270
00:29:36,480 --> 00:29:45,060
Magnetic interaction, magnetic energy. Then what actually happens is you get here's your big system.

271
00:29:45,660 --> 00:29:51,450
The system will break up into what are known as domains, sometimes called vice domains,

272
00:29:52,410 --> 00:30:02,760
vice domains where some point up, some point down, some point up like this and break up into many little domains.

273
00:30:02,760 --> 00:30:07,290
Now, what's going on here at some point in this in this material?

274
00:30:08,540 --> 00:30:15,379
You'll have UPS bins right next to downs bins. And that makes the Heisenberg term in the Hamiltonian very unhappy because the

275
00:30:15,380 --> 00:30:19,430
Heisenberg term of the Hamiltonian wants the UPS spins to be next to up spins.

276
00:30:19,670 --> 00:30:26,470
So there's some energy cost along that, that transition along that, what's what's known as a domain wall.

277
00:30:26,480 --> 00:30:31,940
Maybe I'll even write that down. This is a the transition is called a domain wall.

278
00:30:35,850 --> 00:30:42,090
Between the two. So there's some Heisenberg Energy to have UPS spins next down spins on the domain wall.

279
00:30:42,300 --> 00:30:51,360
But you sacrifice that Heisenberg energy on the domain wall in order to save magnetic energy or dipolar energy for the whole system as a whole.

280
00:30:53,220 --> 00:31:04,860
If if you had a different type of anisotropy and I saw entropy, for example, cubic anisotropy, the domain wall structure might look different.

281
00:31:05,340 --> 00:31:15,120
You might have something that looks like like this or this points up this, points right, spins left down like that.

282
00:31:15,510 --> 00:31:22,380
And this is putting each dipole for a head to tail with the next dipole that improves the energy even better.

283
00:31:22,740 --> 00:31:31,680
Now what happens? In this case is that overall the total magnetisation of one of these magnets is now zero.

284
00:31:31,680 --> 00:31:38,460
Again, because you have just as many up domains as you have down domains and you get no magnetisation at all.

285
00:31:38,580 --> 00:31:42,149
And then you might scratch your head and say, well, wait a second, I know magnets exist.

286
00:31:42,150 --> 00:31:45,959
And, you know, so how is it we're going to get magnetisation out of our magnets?

287
00:31:45,960 --> 00:31:48,750
And so now we have to start thinking a little bit carefully.

288
00:31:48,960 --> 00:31:53,010
The first thing we can do and actually, I guess I have a slide better than drawing that pictures again.

289
00:31:54,000 --> 00:31:58,710
The first thing you can do is you can imagine if you add a magnetic field externally.

290
00:31:59,660 --> 00:32:07,160
To the system, the size of the domains where we re equilibrate and you know, the field that you apply is pointing up.

291
00:32:07,400 --> 00:32:12,260
You'll get more up domains than down domains and you'll get a net magnetisation again.

292
00:32:12,590 --> 00:32:21,080
But still, if you take away the magnetic field, the domains move back to their old equilibrium over here and you'll have no magnetisation.

293
00:32:21,290 --> 00:32:25,519
So this thing that we thought was supposed to be a ferromagnetic is in fact,

294
00:32:25,520 --> 00:32:31,340
a paramagnetic making can be a very good power magnet, which gives you a very high magnetisation for a very small magnetic field.

295
00:32:31,640 --> 00:32:41,520
But it's still, unfortunately, a power magnet. So how are we going to get this thing to be magnetic, even in the absence of an applied magnetic field?

296
00:32:41,790 --> 00:32:46,650
So the whole trick to making a real magnet, the kind of thing that sticks your refrigerator,

297
00:32:46,830 --> 00:32:52,650
is to find a way to prevent the domains from going back from this configuration to this configuration.

298
00:32:52,920 --> 00:32:58,920
So you train a magnet in a high magnetic field, so it gets to one of these configurations where there's more up spins,

299
00:32:59,190 --> 00:33:02,250
then down spins or up moments and down moments in this picture.

300
00:33:02,670 --> 00:33:06,630
But then you find a way to prevent the domains from moving back.

301
00:33:06,660 --> 00:33:11,760
This is what's called pinning. Pinning of domain walls.

302
00:33:11,760 --> 00:33:23,560
Pinning domain walls. And the way it's your typically painted domain wall is with disorder.

303
00:33:26,240 --> 00:33:29,770
And whether one is trying to design a magnet for, you know,

304
00:33:29,840 --> 00:33:34,159
for our use in electrical motors whether use trying to make a magnet that will stick to your refrigerator,

305
00:33:34,160 --> 00:33:41,150
whether you're trying to make a magnet for a magnetic disk drive, the whole art to making good magnets is to figure out how to make the disorder,

306
00:33:41,360 --> 00:33:47,360
prevent the magnetic domains from moving back to their original configuration, where the magnetisation goes back to zero.

307
00:33:47,640 --> 00:33:53,330
Okay, so why is it that disorder is going to help you? All right, so here's here's a picture that's supposed to show us here.

308
00:33:53,690 --> 00:33:58,160
So this is an easing dimension, the whole business about easing versus icing.

309
00:33:58,610 --> 00:34:01,730
I didn't say that. Did. Did I say that? Okay. So let me tell you.

310
00:34:01,940 --> 00:34:07,820
So easing was German. And apparently the way it's pronounced is is is easing in Germany, I believe.

311
00:34:08,870 --> 00:34:15,320
But then after he finished his dissertation, he went to the United States and people called him ising for the rest of his life.

312
00:34:15,650 --> 00:34:20,120
And so everyone the United States calls it the ising model. Everyone in Europe calls it the easing model.

313
00:34:20,390 --> 00:34:24,950
I'll probably never get it straight, but I think it is correct to call it to call it easing.

314
00:34:25,280 --> 00:34:30,260
Anyway, so this is an easing magnet. All the spins are pointing either up or down.

315
00:34:30,620 --> 00:34:37,010
And the higher the interaction, the interaction between spins is unhappy along the domain wall.

316
00:34:37,160 --> 00:34:42,530
And I've drawn all the unhappy spins in red here where the up spins, meet the down spins.

317
00:34:43,070 --> 00:34:50,870
And if you count them from top to bottom, there are 12 of these red bonds where there's a unhappy connection.

318
00:34:51,840 --> 00:34:58,980
Now in this picture, there's a little green atom there, which is supposed to be a defect someplace where there's no spin at all.

319
00:34:59,400 --> 00:35:05,850
If you happen to move over the domain wall to a different position over here, so it goes through the defect.

320
00:35:06,360 --> 00:35:09,540
There's only ten red bonds now, not 12.

321
00:35:09,810 --> 00:35:14,790
So the energy actually drops if the domain wall goes through the defect.

322
00:35:15,210 --> 00:35:18,300
It's a lower energy for domain walls to go through a defect.

323
00:35:18,480 --> 00:35:27,030
So they want to stick on the defect. And that's how you pin domain wall with a with some disorder by putting defects in the system.

324
00:35:27,270 --> 00:35:32,430
The once the the domain walls are in this configuration, you take away the magnetic field,

325
00:35:32,610 --> 00:35:35,700
but then they can't move back because they would have to give up some energy

326
00:35:35,880 --> 00:35:39,860
to move back to their original position because they're stuck on the defect.

327
00:35:40,440 --> 00:35:45,960
That's the general the general physics of how it is you make a real magnet which has m not equal to zero.

328
00:35:46,350 --> 00:35:48,270
So far so good. Okay.

329
00:35:48,600 --> 00:35:56,640
So since the, you know, the fundamental physics is really all wrapped up in the physics of the domain wall and how the domain wall moves,

330
00:35:56,760 --> 00:36:00,300
it's worth thinking about the domain while a little bit more carefully.

331
00:36:00,540 --> 00:36:06,150
And it turns out that most magnets in nature are not of the easing type where the anisotropy is strong,

332
00:36:06,360 --> 00:36:10,200
but there are this type and isotope is quite weak.

333
00:36:11,400 --> 00:36:17,580
So we have to sort of think about what happens in four domain walls and not in the easing limit.

334
00:36:17,580 --> 00:36:23,550
But in the weekend I saw a speed limit over here. There's no question of what a domain while looks like for the easing limit.

335
00:36:23,820 --> 00:36:29,070
You just have, you know, up spins and then they meet down spins and nothing interesting happens in between.

336
00:36:29,370 --> 00:36:36,300
But over here where we have weak and I saw entropy if you have up spins over here and you have down spins way over here,

337
00:36:37,470 --> 00:36:44,910
remember that the most important thing to do for this weekend is not to the limit is to keep everyone as aligned as possible with their neighbour.

338
00:36:45,210 --> 00:36:51,090
So the way you get from up into down spin is to just slightly rotate each spin just a little bit

339
00:36:51,330 --> 00:36:57,310
from his neighbour and eventually make it from pointing up to pointing down with a slow rotation.

340
00:36:57,930 --> 00:37:02,070
And this is what's known as, let's say this goes over and spins.

341
00:37:03,510 --> 00:37:07,410
Okay, I didn't draw that very well, but you get the idea and this is what's known as a block wall.

342
00:37:08,190 --> 00:37:18,719
BLOCK wall or nail wall. Now a wall now wall after Felix block the same blocks Theorem Guy and Louis.

343
00:37:18,720 --> 00:37:21,870
Now the guy who realised that we should have Andy for our magnets.

344
00:37:22,140 --> 00:37:28,020
So let's try to calculate the energy and the width of this on net block.

345
00:37:28,020 --> 00:37:31,229
Okay. There is a slight difference between a block wall and an AL Wall.

346
00:37:31,230 --> 00:37:40,020
It depends on whether the rotation of the spin is in the plane of of these atoms or it's transverse to the plane of the atoms.

347
00:37:40,650 --> 00:37:45,660
It's you know, it's a detail that you probably don't have to worry about, but there's a slight difference between the two.

348
00:37:45,930 --> 00:37:48,990
At any rate, I think what I've drawn here is actually a block wall.

349
00:37:52,230 --> 00:37:54,790
So let's try to calculate the energy of this.

350
00:37:54,790 --> 00:38:04,110
So the energy in this one dimensional picture, it's going to be some over I, I guess minus J, I see plus one.

351
00:38:05,580 --> 00:38:11,940
And here you might have think you might think that I've left out a factor of one half but but I haven't

352
00:38:11,940 --> 00:38:18,300
because here I since I'm summing over I only in one dimension as one as two only occurs in one term now.

353
00:38:18,510 --> 00:38:20,880
Okay. So I got the factor of a half right.

354
00:38:22,830 --> 00:38:33,180
And if I align all the spins aligned, I then get some over I, I guess I equals one two end of minus j as squared.

355
00:38:33,660 --> 00:38:46,170
Everyone is aligned with each other and if I twist it e twist minus e aligned, I'm then going to call it Delta e equals delta e.

356
00:38:46,980 --> 00:38:51,059
The amount of energy you pay to twist the spin from top to bottom that will be

357
00:38:51,060 --> 00:38:59,550
sum over i equals one to end of j s squared times one minus cosine of theta i.

358
00:38:59,910 --> 00:39:02,040
So for each time you twist the spin a little bit,

359
00:39:02,040 --> 00:39:13,830
you have to pay this sort of cosine energy theta is is okay so if one that's theta I the rotation of the spin from one step to the next.

360
00:39:14,190 --> 00:39:21,740
So let's assume that the theta is small and if we need to get from, from all the way to the left,

361
00:39:21,750 --> 00:39:28,980
all the way to write in and spins, let's just declare all the theta is to be pi over n so we have to make a pi rotation and step.

362
00:39:28,990 --> 00:39:38,970
So we'll say theta is pi over. And so cosine theta cosine theta is an approximately one minus one half pi over and squared.

363
00:39:39,270 --> 00:39:44,610
If I plug that into here I'll get a total of and spins.

364
00:39:44,610 --> 00:39:52,320
I'm summing over I have J s squared and one half pi over n squared.

365
00:39:52,650 --> 00:39:58,260
Multiply that out I then get pi squared over to J.

366
00:39:58,260 --> 00:40:03,980
S squared over and. And unsurprisingly, this tells me that the energy is lower.

367
00:40:04,190 --> 00:40:11,210
If I rotate this, the the spins more smoothly that each spin wants to be as aligned as possible with his neighbour.

368
00:40:11,630 --> 00:40:15,770
And so if we're only considering this part of the Hamiltonian, the spins would.

369
00:40:15,770 --> 00:40:19,520
If they rotated they will rotate infinitely slowly and they would be most happy that way.

370
00:40:19,820 --> 00:40:24,800
But they. But this has to compete with the anisotropy energy and isotropic energy.

371
00:40:26,610 --> 00:40:32,790
And Isotropic, which wants the spins to be either pointing directly up or directly down.

372
00:40:32,790 --> 00:40:40,620
And so we have to estimate how much energy you're going to pay to have em spins in this wall, which are neither up nor down.

373
00:40:40,920 --> 00:40:49,710
And okay, this is a rather crude approximation, but we'll just say that each spin that's not pointing up and down, you pay an energy X squared.

374
00:40:50,160 --> 00:40:54,300
Okay. I mean, this this number is probably off by a factor of two or pi or something like that,

375
00:40:54,600 --> 00:41:04,469
but at least it would give us the order of magnitude correctly. So the total energy of the block wall e total is the sum of the Heisenberg Energy pi

376
00:41:04,470 --> 00:41:11,670
squared over 2js squared over RN plus the anisotropy energy and Kappa s squared.

377
00:41:13,030 --> 00:41:16,540
And then we have to minimise this veto at all.

378
00:41:17,440 --> 00:41:29,500
Diane So that equal to zero and that gives us I just differentiate that we just get an equal square root of five squared over to J over Kappa.

379
00:41:29,950 --> 00:41:35,650
And so not surprisingly, what this is telling us is that if the anisotropy is extremely small,

380
00:41:35,980 --> 00:41:42,430
then and the, and the Heisenberg coupling is extremely strong, then the spins rotate extremely slowly.

381
00:41:42,610 --> 00:41:48,250
Whereas if you're in the other limit where the and I Satrapi is large and the Heisenberg coupling small

382
00:41:48,430 --> 00:41:55,510
then the spins rotate in a very small number and maybe even directly from up to down in one in one step.

383
00:41:55,870 --> 00:42:05,500
Now, in real magnets, it's very often the case. Often you can have a case where N is greater than or about equal to 100 atoms.

384
00:42:08,380 --> 00:42:14,920
So you can have these these block walls or these domain walls or now walls be actually quite, quite wide.

385
00:42:15,310 --> 00:42:20,260
Now, this is actually quite important if you start thinking about polycrystalline material.

386
00:42:20,800 --> 00:42:25,700
Polycrystalline kris. Crystal crystalline crystal in.

387
00:42:28,960 --> 00:42:38,860
Magnets. So now we're talking about lots of little crystal lights separated from each other of of magnetic material.

388
00:42:39,910 --> 00:42:47,770
And if the size of the crystal light in number of atoms going across is less, then this magic number the size of the domain wall.

389
00:42:48,100 --> 00:42:57,040
Then the whole atom, the whole crystal light has a point, have its magnetisation pointing in just one direction, has a point and only one direction.

390
00:42:57,900 --> 00:43:02,350
You know, each crystal I can point in a different direction, but each crystal light has to be pointed only in one direction.

391
00:43:02,470 --> 00:43:08,830
And the reason for this is because you don't have enough space within one crystal light to rotate the spin from one direction to another.

392
00:43:09,220 --> 00:43:16,950
It's too stiff. You can't. You know that the coupling between neighbours is too strong in order to get the spins to rotate within one crystal light.

393
00:43:17,140 --> 00:43:24,340
And you might think that this hundred atoms is rather small, you know, because each atom is only an angstrom or two angstroms or something.

394
00:43:24,580 --> 00:43:27,670
But if you have 100 arms in this direction, 100 times another direction,

395
00:43:27,670 --> 00:43:32,110
100 times another direction, all of a sudden you have a million atoms you're talking about.

396
00:43:32,140 --> 00:43:35,140
So these crystallised can actually have an awful lot of atoms in them.

397
00:43:35,140 --> 00:43:39,580
And you can still think of them as microscopic individual magnets pointing in just one direction.

398
00:43:40,080 --> 00:43:45,430
Okay, so, so, so far, so good. I will resist the urge to make jokes about one direction.

399
00:43:47,230 --> 00:43:55,770
All right. Okay, so let's imagine that we take one of these crystal lights and we are going to apply.

400
00:43:55,780 --> 00:44:04,569
So we have a crystal right here. And suppose it has an anisotropy axis in this ray z hat that it wants to point its magnetisation either up or down.

401
00:44:04,570 --> 00:44:09,880
And let's, for the sake of argument, we'll put the magnetic field in that direction as well.

402
00:44:10,240 --> 00:44:13,899
The energy and then the magnetic moment can be pointing in any direction.

403
00:44:13,900 --> 00:44:17,800
We like the energy of that. Chris Light is some insight.

404
00:44:18,310 --> 00:44:26,800
There's going to be minus B and then there's the anisotropy energy minus Kappa Z squared this.

405
00:44:27,750 --> 00:44:32,760
That wants the the magnetisation to point either directly up or directly down.

406
00:44:33,660 --> 00:44:45,420
So if we decide that, okay, so if the angle between M and B is theta, then we can rewrite this energy as E,

407
00:44:45,420 --> 00:44:54,150
not minus m b cosine theta minus kappa m squared, cosine squared theta.

408
00:44:54,510 --> 00:44:58,740
And in fact, we might even let x equals cosine theta.

409
00:44:59,580 --> 00:45:08,760
And then what we have is a quadratic inot minus maybe x minus, kappa, m squared, x squared.

410
00:45:10,020 --> 00:45:13,079
And we can plot this thing. Oops.

411
00:45:13,080 --> 00:45:18,060
Money out of room, aren't I? Okay. We can plot this thing as a function of theta or as a function of x.

412
00:45:18,540 --> 00:45:22,200
So this is the energy. And let's start in the case of B equals zero.

413
00:45:22,230 --> 00:45:28,800
So if you have your crystal light and zero magnetic field and your energy over here is x equals minus one,

414
00:45:29,340 --> 00:45:39,680
minus one, and over here is x equals plus one. What we have here is x equals minus one is theta equals pi, x equals plus one is theta equals zero.

415
00:45:39,950 --> 00:45:44,270
So this over here is the magnetisation pointing down.

416
00:45:45,380 --> 00:45:53,270
And over here we have the crystal light with the magnetisation pointing up and the energy is quadratic.

417
00:45:53,300 --> 00:45:56,510
Looks kind of like this. So there's two stable points.

418
00:45:56,810 --> 00:45:59,900
The stable point here is to have the magnetisation pointing down.

419
00:45:59,900 --> 00:46:02,719
And the stable point here is to have the magnetisation pointing up.

420
00:46:02,720 --> 00:46:05,810
And it's much less happy to have the magnetisation pointing in any other direction.

421
00:46:06,050 --> 00:46:13,700
This is not surprising because it's the anisotropy axis wants this, the magnetisation point either up or down.

422
00:46:14,030 --> 00:46:17,360
Okay. Now let's add a non-zero magnetic field.

423
00:46:19,550 --> 00:46:26,420
So now will it be not equal to zero or be small, but not equal to zero zero?

424
00:46:26,420 --> 00:46:29,570
So now as we plot again, we have a parabola.

425
00:46:29,870 --> 00:46:30,920
It's still quadratic,

426
00:46:31,310 --> 00:46:42,310
but the parabola is now shifted because of the linear term in that in the energy so that the peak of the parabola will not be in the centre anymore.

427
00:46:42,380 --> 00:46:45,800
The off centre. So we'll get a parabola. It looks like this.

428
00:46:45,800 --> 00:46:50,330
And again, over here, what we have is magnetisation pointing down,

429
00:46:50,330 --> 00:46:57,800
even though the magnetic field is pointing up and here we have the magnetisation pointing up and the magnet and the magnetic field pointing up.

430
00:46:58,100 --> 00:47:05,450
So this point here, having the magnetisation pointing in the same direction, the magnetic field, this is definitely the stable minimum.

431
00:47:09,130 --> 00:47:11,920
Whereas this point over here is metastable.

432
00:47:16,570 --> 00:47:22,150
In order to get from the metastable case of the magnetisation pointing in the opposite direction of the magnetic field.

433
00:47:22,300 --> 00:47:28,360
In order to get down to here where it would rather be, it has to get over this activation barrier.

434
00:47:28,780 --> 00:47:33,969
And this means you have to take all of your atoms, maybe a million of these atoms,

435
00:47:33,970 --> 00:47:38,740
and somehow flip them all over simultaneously, either thermally or by a quantum tunnelling.

436
00:47:39,730 --> 00:47:44,040
And that's a process that's so difficult that at low temperature it may never happen.

437
00:47:44,050 --> 00:47:51,820
So the that these you can have these little crystals of magnets have their magnetisation stuck in the wrong direction.

438
00:47:52,810 --> 00:48:04,540
Now let's turn up the magnetic field a little bit more. So now we have a large magnetic field, large B again, we have a parabola,

439
00:48:04,930 --> 00:48:09,700
but the but the maximum the parabola has been shifted all the way off to the left.

440
00:48:09,700 --> 00:48:16,870
The section is minus one. This is x equals plus one. And the maximum of the problem is way over here off the plot.

441
00:48:17,410 --> 00:48:26,530
And so this point here with the the magnetisation pointing in the opposite direction of the magnetic field is now unstable.

442
00:48:26,830 --> 00:48:33,610
And whereas this point over here is the only stable point with the magnetisation pointing the same direction as the magnetic field.

443
00:48:34,150 --> 00:48:37,180
Okay, so if the magnetic field is turned up enough,

444
00:48:37,480 --> 00:48:42,520
then you can flip over all of the magnetisation is to point in the direction of the magnetic field.

445
00:48:43,000 --> 00:48:49,090
It's not clear how that works. Okay, so this is. A picture you've seen before.

446
00:48:50,380 --> 00:48:56,650
It's a picture of a so-called hard foul magnet. Maybe even write that word down because it's used often hard feral magnets.

447
00:49:01,840 --> 00:49:07,330
Hard foul magnet is when you actually have em not equal to zero in b equals zero.

448
00:49:08,050 --> 00:49:14,710
And the way you get hard foul magnets and we explained that how you could get hard from magnets by pinning the domain walls.

449
00:49:14,980 --> 00:49:19,330
But you can also think about getting hard foul magnets from having lots of little crystal lights.

450
00:49:19,570 --> 00:49:26,020
And they you need to in order to push the crystal ice to flip over, you have to give them some coercive field.

451
00:49:26,020 --> 00:49:32,200
You have to give them a finite magnetic field in the wrong direction and in the opposite direction to get them to all flip over.

452
00:49:32,200 --> 00:49:35,589
If you just give out a small magnetic field in the other direction, they don't flip over.

453
00:49:35,590 --> 00:49:40,330
And this is what gives these magnets, hysteresis loops. Now, if you think about it for a second,

454
00:49:40,480 --> 00:49:45,970
you realise that this picture of pinning domain walls in a crystal and magnet, this picture, oops, where to go,

455
00:49:45,980 --> 00:49:51,490
this picture here where the domain wall is stuck on his position because it has an energy activation barrier

456
00:49:51,490 --> 00:49:58,900
to get off of that that position here where the atom gives it a lower energy if the domain wall goes to that,

457
00:49:59,260 --> 00:50:04,040
that point is extremely similar to what we have here with all these little crystal.

458
00:50:04,220 --> 00:50:10,570
The only difference here is that you can think with the crystal lights, the domain walls are always between the crystal lights.

459
00:50:10,570 --> 00:50:15,730
If you want to think of it that way, there's some crystal lights will point in one direction, some crystals will point in the other direction.

460
00:50:15,940 --> 00:50:19,210
And to move the domain wall, you have to flip over one more crystal light.

461
00:50:19,450 --> 00:50:21,010
So it's extremely similar physics.

462
00:50:21,640 --> 00:50:28,719
The moral of the story is that in order to get a real magnet, a hard foul magnet, somehow or other, you have to have disorder in your system,

463
00:50:28,720 --> 00:50:36,580
whether it's making the system into small, little crystallise or putting disorder in, it's going to pin the position of the domain walls.

464
00:50:36,880 --> 00:50:40,630
Okay. That's all for today. I will see you Thursday.

465
00:50:40,630 --> 00:50:43,660
Thursday? Yeah, Thursday. Okay. Have a good weekend.