1
00:00:00,949 --> 00:00:02,773
I'm thrilled to be here tonight

2
00:00:02,773 --> 00:00:05,152
to share with you something
we've been working on

3
00:00:05,152 --> 00:00:07,242
for over two years,

4
00:00:07,242 --> 00:00:09,796
and it's in the area
of additive manufacturing,

5
00:00:09,796 --> 00:00:12,513
also known as 3D printing.

6
00:00:12,513 --> 00:00:14,231
You see this object here.

7
00:00:14,231 --> 00:00:18,039
It looks fairly simple,
but it's quite complex at the same time.

8
00:00:18,549 --> 00:00:21,800
It's a set of concentric
geodesic structures

9
00:00:21,800 --> 00:00:24,795
with linkages between each one.

10
00:00:24,795 --> 00:00:30,797
In its context, it is not manufacturable
by traditional manufacturing techniques.

11
00:00:31,343 --> 00:00:35,290
It has a symmetry such
that you can't injection mold it.

12
00:00:35,290 --> 00:00:38,879
You can't even manufacture it
through milling.

13
00:00:39,470 --> 00:00:42,117
This is a job for a 3D printer,

14
00:00:42,117 --> 00:00:46,598
but most 3D printers would take between
three and 10 hours to fabricate it,

15
00:00:46,598 --> 00:00:50,824
and we're going to take the risk tonight
to try to fabricate it onstage

16
00:00:50,824 --> 00:00:53,401
during this 10-minute talk.

17
00:00:53,401 --> 00:00:55,440
Wish us luck.

18
00:00:56,350 --> 00:00:59,624
Now, 3D printing is actually a misnomer.

19
00:00:59,624 --> 00:01:03,399
It's actually 2D printing
over and over again,

20
00:01:03,919 --> 00:01:07,761
and it in fact uses the technologies
associated with 2D printing.

21
00:01:08,401 --> 00:01:13,360
Think about inkjet printing where you
lay down ink on a page to make letters,

22
00:01:13,360 --> 00:01:18,346
and then do that over and over again
to build up a three-dimensional object.

23
00:01:18,346 --> 00:01:20,417
In microelectronics, they use something

24
00:01:20,417 --> 00:01:22,737
called lithography to do
the same sort of thing,

25
00:01:22,737 --> 00:01:24,945
to make the transistors
and integrated circuits

26
00:01:24,945 --> 00:01:26,997
and build up a structure several times.

27
00:01:26,997 --> 00:01:29,399
These are all 2D printing technologies.

28
00:01:30,099 --> 00:01:33,987
Now, I'm a chemist,
a material scientist too,

29
00:01:33,987 --> 00:01:36,711
and my co-inventors
are also material scientists,

30
00:01:36,711 --> 00:01:39,010
one a chemist, one a physicist,

31
00:01:39,010 --> 00:01:41,936
and we began to be
interested in 3D printing.

32
00:01:41,936 --> 00:01:47,531
And very often, as you know,
new ideas are often simple connections

33
00:01:47,531 --> 00:01:51,274
between people with different experiences
in different communities,

34
00:01:51,274 --> 00:01:52,751
and that's our story.

35
00:01:53,591 --> 00:01:56,122
Now, we were inspired

36
00:01:56,122 --> 00:02:00,893
by the "Terminator 2" scene for T-1000,

37
00:02:00,893 --> 00:02:05,836
and we thought, why couldn't a 3D printer
operate in this fashion,

38
00:02:06,426 --> 00:02:10,362
where you have an object
arise out of a puddle

39
00:02:11,052 --> 00:02:13,520
in essentially real time

40
00:02:13,520 --> 00:02:15,749
with essentially no waste

41
00:02:15,749 --> 00:02:18,071
to make a great object?

42
00:02:18,071 --> 00:02:19,488
Okay, just like the movies.

43
00:02:19,488 --> 00:02:22,877
And could we be inspired by Hollywood

44
00:02:22,877 --> 00:02:26,384
and come up with ways
to actually try to get this to work?

45
00:02:26,384 --> 00:02:28,450
And that was our challenge.

46
00:02:28,450 --> 00:02:31,817
And our approach would be,
if we could do this,

47
00:02:31,817 --> 00:02:35,671
then we could fundamentally address
the three issues holding back 3D printing

48
00:02:35,671 --> 00:02:38,086
from being a manufacturing process.

49
00:02:38,086 --> 00:02:40,617
One, 3D printing takes forever.

50
00:02:40,617 --> 00:02:45,841
There are mushrooms that grow faster
than 3D printed parts. (Laughter)

51
00:02:47,281 --> 00:02:49,417
The layer by layer process

52
00:02:49,417 --> 00:02:52,319
leads to defects
in mechanical properties,

53
00:02:52,319 --> 00:02:56,266
and if we could grow continuously,
we could eliminate those defects.

54
00:02:56,266 --> 00:03:01,398
And in fact, if we could grow really fast,
we could also start using materials

55
00:03:01,398 --> 00:03:06,042
that are self-curing,
and we could have amazing properties.

56
00:03:06,042 --> 00:03:10,151
So if we could pull this off,
imitate Hollywood,

57
00:03:10,151 --> 00:03:12,912
we could in fact address 3D manufacturing.

58
00:03:14,702 --> 00:03:17,953
Our approach is to use
some standard knowledge

59
00:03:17,953 --> 00:03:20,553
in polymer chemistry

60
00:03:20,553 --> 00:03:27,152
to harness light and oxygen
to grow parts continuously.

61
00:03:27,152 --> 00:03:30,099
Light and oxygen work in different ways.

62
00:03:30,099 --> 00:03:33,141
Light can take a resin
and convert it to a solid,

63
00:03:33,141 --> 00:03:35,295
can convert a liquid to a solid.

64
00:03:35,295 --> 00:03:38,829
Oxygen inhibits that process.

65
00:03:38,829 --> 00:03:42,080
So light and oxygen
are polar opposites from one another

66
00:03:42,080 --> 00:03:44,588
from a chemical point of view,

67
00:03:44,588 --> 00:03:48,001
and if we can control spatially
the light and oxygen,

68
00:03:48,001 --> 00:03:49,948
we could control this process.

69
00:03:50,288 --> 00:03:53,739
And we refer to this as CLIP.
[Continuous Liquid Interface Production.]

70
00:03:53,739 --> 00:03:55,615
It has three functional components.

71
00:03:56,465 --> 00:04:00,326
One, it has a reservoir
that holds the puddle,

72
00:04:00,326 --> 00:04:02,205
just like the T-1000.

73
00:04:02,205 --> 00:04:04,621
At the bottom of the reservoir
is a special window.

74
00:04:04,621 --> 00:04:06,112
I'll come back to that.

75
00:04:06,112 --> 00:04:09,892
In addition, it has a stage
that will lower into the puddle

76
00:04:09,892 --> 00:04:12,481
and pull the object out of the liquid.

77
00:04:12,481 --> 00:04:16,285
The third component
is a digital light projection system

78
00:04:16,285 --> 00:04:18,305
underneath the reservoir,

79
00:04:18,305 --> 00:04:21,578
illuminating with light
in the ultraviolet region.

80
00:04:22,048 --> 00:04:25,271
Now, the key is that this window
in the bottom of this reservoir,

81
00:04:25,271 --> 00:04:28,150
it's a composite,
it's a very special window.

82
00:04:28,150 --> 00:04:31,796
It's not only transparent to light
but it's permeable to oxygen.

83
00:04:31,796 --> 00:04:34,455
It's got characteristics
like a contact lens.

84
00:04:35,435 --> 00:04:37,716
So we can see how the process works.

85
00:04:37,716 --> 00:04:41,130
You can start to see that
as you lower a stage in there,

86
00:04:41,130 --> 00:04:45,309
in a traditional process,
with an oxygen-impermeable window,

87
00:04:45,309 --> 00:04:47,148
you make a two-dimensional pattern

88
00:04:48,008 --> 00:04:51,370
and you end up gluing that onto the window
with a traditional window,

89
00:04:51,370 --> 00:04:54,922
and so in order to introduce
the next layer, you have to separate it,

90
00:04:54,922 --> 00:04:58,451
introduce new resin, reposition it,

91
00:04:58,451 --> 00:05:00,910
and do this process over and over again.

92
00:05:01,400 --> 00:05:03,234
But with our very special window,

93
00:05:03,234 --> 00:05:06,563
what we're able to do is,
with oxygen coming through the bottom

94
00:05:06,563 --> 00:05:07,816
as light hits it,

95
00:05:09,256 --> 00:05:11,926
that oxygen inhibits the reaction,

96
00:05:11,926 --> 00:05:14,550
and we form a dead zone.

97
00:05:14,550 --> 00:05:18,869
This dead zone is on the order
of tens of microns thick,

98
00:05:18,869 --> 00:05:22,096
so that's two or three diameters
of a red blood cell,

99
00:05:22,096 --> 00:05:24,627
right at the window interface
that remains a liquid,

100
00:05:24,627 --> 00:05:26,577
and we pull this object up,

101
00:05:26,577 --> 00:05:28,969
and as we talked about in a Science paper,

102
00:05:28,969 --> 00:05:33,682
as we change the oxygen content,
we can change the dead zone thickness.

103
00:05:33,682 --> 00:05:37,374
And so we have a number of key variables
that we control: oxygen content,

104
00:05:37,374 --> 00:05:40,439
the light, the light intensity,
the dose to cure,

105
00:05:40,439 --> 00:05:42,401
the viscosity, the geometry,

106
00:05:42,401 --> 00:05:45,817
and we use very sophisticated software
to control this process.

107
00:05:46,697 --> 00:05:49,460
The result is pretty staggering.

108
00:05:49,460 --> 00:05:53,196
It's 25 to 100 times faster
than traditional 3D printers,

109
00:05:54,336 --> 00:05:56,170
which is game-changing.

110
00:05:56,170 --> 00:06:00,506
In addition, as our ability
to deliver liquid to that interface,

111
00:06:00,506 --> 00:06:04,246
we can go 1,000 times faster I believe,

112
00:06:04,246 --> 00:06:07,803
and that in fact opens up the opportunity
for generating a lot of heat,

113
00:06:07,803 --> 00:06:11,866
and as a chemical engineer,
I get very excited at heat transfer

114
00:06:11,866 --> 00:06:16,045
and the idea that we might one day
have water-cooled 3D printers,

115
00:06:16,045 --> 00:06:18,437
because they're going so fast.

116
00:06:18,437 --> 00:06:22,500
In addition, because we're growing things,
we eliminate the layers,

117
00:06:22,500 --> 00:06:24,474
and the parts are monolithic.

118
00:06:24,474 --> 00:06:26,564
You don't see the surface structure.

119
00:06:26,564 --> 00:06:29,057
You have molecularly smooth surfaces.

120
00:06:29,057 --> 00:06:33,297
And the mechanical properties
of most parts made in a 3D printer

121
00:06:33,297 --> 00:06:37,593
are notorious for having properties
that depend on the orientation

122
00:06:37,593 --> 00:06:41,354
with which how you printed it,
because of the layer-like structure.

123
00:06:41,354 --> 00:06:43,699
But when you grow objects like this,

124
00:06:43,699 --> 00:06:47,368
the properties are invariant
with the print direction.

125
00:06:47,368 --> 00:06:50,317
These look like injection-molded parts,

126
00:06:50,317 --> 00:06:53,729
which is very different
than traditional 3D manufacturing.

127
00:06:53,729 --> 00:06:57,259
In addition, we're able to throw

128
00:06:57,259 --> 00:07:00,835
the entire polymer
chemistry textbook at this,

129
00:07:00,835 --> 00:07:04,826
and we're able to design chemistries
that can give rise to the properties

130
00:07:04,826 --> 00:07:07,868
you really want in a 3D-printed object.

131
00:07:07,868 --> 00:07:09,205
(Applause)

132
00:07:09,205 --> 00:07:12,439
There it is. That's great.

133
00:07:14,049 --> 00:07:17,627
You always take the risk that something
like this won't work onstage, right?

134
00:07:18,177 --> 00:07:21,056
But we can have materials
with great mechanical properties.

135
00:07:21,056 --> 00:07:23,494
For the first time, we can have elastomers

136
00:07:23,494 --> 00:07:25,955
that are high elasticity
or high dampening.

137
00:07:25,955 --> 00:07:29,368
Think about vibration control
or great sneakers, for example.

138
00:07:29,368 --> 00:07:31,978
We can make materials
that have incredible strength,

139
00:07:32,828 --> 00:07:36,404
high strength-to-weight ratio,
really strong materials,

140
00:07:36,404 --> 00:07:38,517
really great elastomers,

141
00:07:38,517 --> 00:07:41,242
so throw that in the audience there.

142
00:07:41,242 --> 00:07:43,878
So great material properties.

143
00:07:43,878 --> 00:07:47,293
And so the opportunity now,
if you actually make a part

144
00:07:47,293 --> 00:07:50,973
that has the properties
to be a final part,

145
00:07:50,973 --> 00:07:54,073
and you do it in game-changing speeds,

146
00:07:54,073 --> 00:07:56,860
you can actually transform manufacturing.

147
00:07:56,860 --> 00:07:59,716
Right now, in manufacturing,
what happens is,

148
00:07:59,716 --> 00:08:02,678
the so-called digital thread
in digital manufacturing.

149
00:08:02,678 --> 00:08:07,717
We go from a CAD drawing, a design,
to a prototype to manufacturing.

150
00:08:07,717 --> 00:08:10,440
Often, the digital thread is broken
right at prototype,

151
00:08:10,440 --> 00:08:12,872
because you can't go
all the way to manufacturing

152
00:08:12,872 --> 00:08:16,587
because most parts don't have
the properties to be a final part.

153
00:08:16,587 --> 00:08:18,978
We now can connect the digital thread

154
00:08:18,978 --> 00:08:23,227
all the way from design
to prototyping to manufacturing,

155
00:08:23,227 --> 00:08:26,176
and that opportunity
really opens up all sorts of things,

156
00:08:26,176 --> 00:08:31,129
from better fuel-efficient cars
dealing with great lattice properties

157
00:08:31,129 --> 00:08:33,080
with high strength-to-weight ratio,

158
00:08:33,080 --> 00:08:36,508
new turbine blades,
all sorts of wonderful things.

159
00:08:37,468 --> 00:08:42,623
Think about if you need a stent
in an emergency situation,

160
00:08:42,623 --> 00:08:46,593
instead of the doctor pulling off
a stent out of the shelf

161
00:08:46,593 --> 00:08:48,822
that was just standard sizes,

162
00:08:48,822 --> 00:08:52,978
having a stent that's designed
for you, for your own anatomy

163
00:08:52,978 --> 00:08:54,789
with your own tributaries,

164
00:08:54,789 --> 00:08:58,038
printed in an emergency situation
in real time out of the properties

165
00:08:58,038 --> 00:09:01,477
such that the stent could go away
after 18 months: really-game changing.

166
00:09:01,477 --> 00:09:05,633
Or digital dentistry, and making
these kinds of structures

167
00:09:05,633 --> 00:09:08,814
even while you're in the dentist chair.

168
00:09:08,814 --> 00:09:11,530
And look at the structures
that my students are making

169
00:09:11,530 --> 00:09:13,504
at the University of North Carolina.

170
00:09:13,504 --> 00:09:16,313
These are amazing microscale structures.

171
00:09:16,313 --> 00:09:19,309
You know, the world is really good
at nano-fabrication.

172
00:09:19,309 --> 00:09:23,599
Moore's Law has driven things
from 10 microns and below.

173
00:09:23,599 --> 00:09:25,201
We're really good at that,

174
00:09:25,201 --> 00:09:29,241
but it's actually very hard to make things
from 10 microns to 1,000 microns,

175
00:09:29,241 --> 00:09:31,261
the mesoscale.

176
00:09:31,261 --> 00:09:34,094
And subtractive techniques
from the silicon industry

177
00:09:34,094 --> 00:09:35,510
can't do that very well.

178
00:09:35,510 --> 00:09:37,159
They can't etch wafers that well.

179
00:09:37,159 --> 00:09:39,109
But this process is so gentle,

180
00:09:39,109 --> 00:09:41,594
we can grow these objects
up from the bottom

181
00:09:41,594 --> 00:09:43,590
using additive manufacturing

182
00:09:43,590 --> 00:09:45,843
and make amazing things
in tens of seconds,

183
00:09:45,843 --> 00:09:47,932
opening up new sensor technologies,

184
00:09:47,932 --> 00:09:50,417
new drug delivery techniques,

185
00:09:50,417 --> 00:09:54,149
new lab-on-a-chip applications,
really game-changing stuff.

186
00:09:55,149 --> 00:09:59,983
So the opportunity of making
a part in real time

187
00:09:59,983 --> 00:10:02,816
that has the properties to be a final part

188
00:10:02,816 --> 00:10:05,792
really opens up 3D manufacturing,

189
00:10:05,792 --> 00:10:08,992
and for us, this is very exciting,
because this really is owning

190
00:10:08,992 --> 00:10:15,589
the intersection between hardware,
software and molecular science,

191
00:10:15,589 --> 00:10:19,755
and I can't wait to see what designers
and engineers around the world

192
00:10:19,755 --> 00:10:22,029
are going to be able to do
with this great tool.

193
00:10:22,499 --> 00:10:24,618
Thanks for listening.

194
00:10:24,618 --> 00:10:29,727
(Applause)