The self-assembling computer chips of the future | Karl Skjonnemand

103,690 views ・ 2019-03-13

TED


Please double-click on the English subtitles below to play the video.

00:13
Computers used to be as big as a room.
0
13246
3584
00:16
But now they fit in your pocket,
1
16854
1592
00:18
on your wrist
2
18470
1171
00:19
and can even be implanted inside of your body.
3
19665
3319
00:23
How cool is that?
4
23008
1281
00:24
And this has been enabled by the miniaturization of transistors,
5
24809
4337
00:29
which are the tiny switches in the circuits
6
29170
2492
00:31
at the heart of our computers.
7
31686
1776
00:34
And it's been achieved through decades of development
8
34051
3172
00:37
and breakthroughs in science and engineering
9
37247
2798
00:40
and of billions of dollars of investment.
10
40069
2672
00:43
But it's given us vast amounts of computing,
11
43352
2748
00:46
huge amounts of memory
12
46124
1805
00:47
and the digital revolution that we all experience and enjoy today.
13
47953
4942
00:53
But the bad news is,
14
53665
2768
00:56
we're about to hit a digital roadblock,
15
56457
3132
00:59
as the rate of miniaturization of transistors is slowing down.
16
59613
4350
01:04
And this is happening at exactly the same time
17
64471
2874
01:07
as our innovation in software is continuing relentlessly
18
67369
3998
01:11
with artificial intelligence and big data.
19
71391
3760
01:15
And our devices regularly perform facial recognition or augment our reality
20
75175
5040
01:20
or even drive cars down our treacherous, chaotic roads.
21
80239
4225
01:24
It's amazing.
22
84959
1207
01:26
But if we don't keep up with the appetite of our software,
23
86618
4667
01:31
we could reach a point in the development of our technology
24
91309
3787
01:35
where the things that we could do with software could, in fact, be limited
25
95120
4210
01:39
by our hardware.
26
99354
1271
01:41
We've all experienced the frustration of an old smartphone or tablet
27
101075
4508
01:45
grinding slowly to a halt over time
28
105607
3164
01:48
under the ever-increasing weight of software updates and new features.
29
108795
3975
01:52
And it worked just fine when we bought it not so long ago.
30
112794
3383
01:56
But the hungry software engineers have eaten up all the hardware capacity
31
116201
4510
02:00
over time.
32
120735
1306
02:03
The semiconductor industry is very well aware of this
33
123883
3612
02:07
and is working on all sorts of creative solutions,
34
127519
3884
02:11
such as going beyond transistors to quantum computing
35
131427
4311
02:15
or even working with transistors in alternative architectures
36
135762
4212
02:19
such as neural networks
37
139998
1603
02:21
to make more robust and efficient circuits.
38
141625
3013
02:25
But these approaches will take quite some time,
39
145270
3339
02:28
and we're really looking for a much more immediate solution to this problem.
40
148633
4627
02:34
The reason why the rate of miniaturization of transistors is slowing down
41
154899
4782
02:39
is due to the ever-increasing complexity of the manufacturing process.
42
159705
4686
02:45
The transistor used to be a big, bulky device,
43
165142
3250
02:48
until the invent of the integrated circuit
44
168416
3309
02:51
based on pure crystalline silicon wafers.
45
171749
2691
02:54
And after 50 years of continuous development,
46
174946
2779
02:57
we can now achieve transistor features dimensions
47
177749
3373
03:01
down to 10 nanometers.
48
181146
2529
03:04
You can fit more than a billion transistors
49
184361
2437
03:06
in a single square millimeter of silicon.
50
186822
2963
03:10
And to put this into perspective:
51
190273
2022
03:12
a human hair is 100 microns across.
52
192319
3826
03:16
A red blood cell, which is essentially invisible,
53
196169
2519
03:18
is eight microns across,
54
198712
1599
03:20
and you can place 12 across the width of a human hair.
55
200335
3400
03:24
But a transistor, in comparison, is much smaller,
56
204467
3100
03:27
at a tiny fraction of a micron across.
57
207591
3848
03:31
You could place more than 260 transistors
58
211463
3546
03:35
across a single red blood cell
59
215033
1978
03:37
or more than 3,000 across the width of a human hair.
60
217035
4464
03:41
It really is incredible nanotechnology in your pocket right now.
61
221523
4324
03:47
And besides the obvious benefit
62
227204
2188
03:49
of being able to place more, smaller transistors on a chip,
63
229416
3834
03:53
smaller transistors are faster switches,
64
233984
3492
03:58
and smaller transistors are also more efficient switches.
65
238166
4401
04:02
So this combination has given us
66
242591
2477
04:05
lower cost, higher performance and higher efficiency electronics
67
245092
4299
04:09
that we all enjoy today.
68
249415
2063
04:14
To manufacture these integrated circuits,
69
254415
2764
04:17
the transistors are built up layer by layer,
70
257203
3208
04:20
on a pure crystalline silicon wafer.
71
260435
2353
04:23
And in an oversimplified sense,
72
263332
2228
04:25
every tiny feature of the circuit is projected
73
265584
4281
04:29
onto the surface of the silicon wafer
74
269889
2332
04:32
and recorded in a light-sensitive material
75
272245
3679
04:35
and then etched through the light-sensitive material
76
275948
2939
04:38
to leave the pattern in the underlying layers.
77
278911
3021
04:42
And this process has been dramatically improved over the years
78
282612
4084
04:46
to give the electronics performance we have today.
79
286720
2773
04:50
But as the transistor features get smaller and smaller,
80
290279
3442
04:53
we're really approaching the physical limitations
81
293745
3037
04:56
of this manufacturing technique.
82
296806
1883
05:00
The latest systems for doing this patterning
83
300515
3105
05:03
have become so complex
84
303644
2303
05:05
that they reportedly cost more than 100 million dollars each.
85
305971
4730
05:10
And semiconductor factories contain dozens of these machines.
86
310725
4287
05:15
So people are seriously questioning: Is this approach long-term viable?
87
315036
4426
05:20
But we believe we can do this chip manufacturing
88
320441
3680
05:24
in a totally different and much more cost-effective way
89
324145
4023
05:28
using molecular engineering and mimicking nature
90
328966
3973
05:32
down at the nanoscale dimensions of our transistors.
91
332963
3613
05:37
As I said, the conventional manufacturing takes every tiny feature of the circuit
92
337267
4661
05:41
and projects it onto the silicon.
93
341952
2124
05:44
But if you look at the structure of an integrated circuit,
94
344818
2744
05:47
the transistor arrays,
95
347586
1974
05:49
many of the features are repeated millions of times.
96
349584
3629
05:53
It's a highly periodic structure.
97
353237
2608
05:56
So we want to take advantage of this periodicity
98
356331
3068
05:59
in our alternative manufacturing technique.
99
359423
2697
06:02
We want to use self-assembling materials
100
362144
3435
06:05
to naturally form the periodic structures
101
365603
2977
06:08
that we need for our transistors.
102
368604
2383
06:12
We do this with the materials,
103
372052
2142
06:14
then the materials do the hard work of the fine patterning,
104
374218
3437
06:17
rather than pushing the projection technology to its limits and beyond.
105
377679
4859
06:23
Self-assembly is seen in nature in many different places,
106
383909
3899
06:27
from lipid membranes to cell structures,
107
387832
3410
06:31
so we do know it can be a robust solution.
108
391266
3055
06:34
If it's good enough for nature, it should be good enough for us.
109
394345
3561
06:38
So we want to take this naturally occurring, robust self-assembly
110
398549
4800
06:43
and use it for the manufacturing of our semiconductor technology.
111
403373
3965
06:48
One type of self-assemble material --
112
408929
2615
06:52
it's called a block co-polymer --
113
412388
2247
06:54
consists of two polymer chains just a few tens of nanometers in length.
114
414659
4783
06:59
But these chains hate each other.
115
419466
2051
07:01
They repel each other,
116
421541
1484
07:03
very much like oil and water or my teenage son and daughter.
117
423049
3897
07:06
(Laughter)
118
426970
1357
07:08
But we cruelly bond them together,
119
428351
2774
07:11
creating an inbuilt frustration in the system,
120
431149
2695
07:13
as they try to separate from each other.
121
433868
2206
07:16
And in the bulk material, there are billions of these,
122
436716
3285
07:20
and the similar components try to stick together,
123
440025
3301
07:23
and the opposing components try to separate from each other
124
443350
2809
07:26
at the same time.
125
446183
1155
07:27
And this has a built-in frustration, a tension in the system.
126
447362
3754
07:31
So it moves around, it squirms until a shape is formed.
127
451140
4309
07:36
And the natural self-assembled shape that is formed is nanoscale,
128
456209
4048
07:40
it's regular, it's periodic, and it's long range,
129
460281
3727
07:44
which is exactly what we need for our transistor arrays.
130
464032
3858
07:49
So we can use molecular engineering
131
469347
2531
07:51
to design different shapes of different sizes
132
471902
3064
07:54
and of different periodicities.
133
474990
2063
07:57
So for example, if we take a symmetrical molecule,
134
477077
2731
07:59
where the two polymer chains are similar length,
135
479832
3075
08:02
the natural self-assembled structure that is formed
136
482931
2671
08:05
is a long, meandering line,
137
485626
2929
08:08
very much like a fingerprint.
138
488579
1810
08:10
And the width of the fingerprint lines
139
490951
2322
08:13
and the distance between them
140
493297
2010
08:15
is determined by the lengths of our polymer chains
141
495331
3911
08:19
but also the level of built-in frustration in the system.
142
499266
3294
08:23
And we can even create more elaborate structures
143
503320
2558
08:27
if we use unsymmetrical molecules,
144
507487
2439
08:30
where one polymer chain is significantly shorter than the other.
145
510839
4085
08:35
And the self-assembled structure that forms in this case
146
515749
2710
08:38
is with the shorter chains forming a tight ball in the middle,
147
518483
3800
08:42
and it's surrounded by the longer, opposing polymer chains,
148
522307
3841
08:46
forming a natural cylinder.
149
526172
2048
08:49
And the size of this cylinder
150
529089
2075
08:51
and the distance between the cylinders, the periodicity,
151
531188
3415
08:54
is again determined by how long we make the polymer chains
152
534627
3594
08:58
and the level of built-in frustration.
153
538245
2738
09:01
So in other words, we're using molecular engineering
154
541896
3878
09:05
to self-assemble nanoscale structures
155
545798
2825
09:08
that can be lines or cylinders the size and periodicity of our design.
156
548647
4910
09:14
We're using chemistry, chemical engineering,
157
554369
3297
09:17
to manufacture the nanoscale features that we need for our transistors.
158
557690
4789
09:25
But the ability to self-assemble these structures
159
565611
4049
09:29
only takes us half of the way,
160
569684
2437
09:32
because we still need to position these structures
161
572145
2809
09:34
where we want the transistors in the integrated circuit.
162
574978
3550
09:39
But we can do this relatively easily
163
579246
2738
09:42
using wide guide structures that pin down the self-assembled structures,
164
582008
6977
09:49
anchoring them in place
165
589009
1921
09:50
and forcing the rest of the self-assembled structures
166
590954
2847
09:53
to lie parallel,
167
593825
1350
09:55
aligned with our guide structure.
168
595199
2400
09:58
For example, if we want to make a fine, 40-nanometer line,
169
598510
4639
10:03
which is very difficult to manufacture with conventional projection technology,
170
603173
4138
10:08
we can manufacture a 120-nanometer guide structure
171
608274
4785
10:13
with normal projection technology,
172
613083
2504
10:15
and this structure will align three of the 40-nanometer lines in between.
173
615611
6591
10:22
So the materials are doing the most difficult fine patterning.
174
622226
4769
10:27
And we call this whole approach "directed self-assembly."
175
627790
3907
10:33
The challenge with directed self-assembly
176
633586
2754
10:36
is that the whole system needs to align almost perfectly,
177
636364
4476
10:40
because any tiny defect in the structure could cause a transistor failure.
178
640864
5281
10:46
And because there are billions of transistors in our circuit,
179
646169
2969
10:49
we need an almost molecularly perfect system.
180
649162
3228
10:52
But we're going to extraordinary measures
181
652977
2005
10:55
to achieve this,
182
655006
1167
10:56
from the cleanliness of our chemistry
183
656197
2992
10:59
to the careful processing of these materials
184
659213
2326
11:01
in the semiconductor factory
185
661563
1571
11:03
to remove even the smallest nanoscopic defects.
186
663158
4572
11:09
So directed self-assembly is an exciting new disruptive technology,
187
669311
5190
11:14
but it is still in the development stage.
188
674525
2569
11:17
But we're growing in confidence that we could, in fact, introduce it
189
677680
3861
11:21
to the semiconductor industry
190
681565
1687
11:23
as a revolutionary new manufacturing process
191
683276
2957
11:26
in just the next few years.
192
686257
2067
11:29
And if we can do this, if we're successful,
193
689014
3034
11:32
we'll be able to continue
194
692072
1531
11:33
with the cost-effective miniaturization of transistors,
195
693627
3258
11:36
continue with the spectacular expansion of computing
196
696909
3753
11:40
and the digital revolution.
197
700686
1882
11:42
And what's more, this could even be the dawn of a new era
198
702592
3545
11:46
of molecular manufacturing.
199
706161
2231
11:48
How cool is that?
200
708416
1531
11:50
Thank you.
201
710519
1158
11:51
(Applause)
202
711701
4209
About this website

This site will introduce you to YouTube videos that are useful for learning English. You will see English lessons taught by top-notch teachers from around the world. Double-click on the English subtitles displayed on each video page to play the video from there. The subtitles scroll in sync with the video playback. If you have any comments or requests, please contact us using this contact form.

https://forms.gle/WvT1wiN1qDtmnspy7