1
00:00:00,050 --> 00:00:02,500
The following content is
provided under a Creative

2
00:00:02,500 --> 00:00:04,010
Commons license.

3
00:00:04,010 --> 00:00:06,350
Your support will help
MIT OpenCourseWare

4
00:00:06,350 --> 00:00:10,720
continue to offer high-quality
educational resources for free.

5
00:00:10,720 --> 00:00:13,340
To make a donation or
view additional materials

6
00:00:13,340 --> 00:00:17,226
from hundreds of MIT courses,
visit MIT OpenCourseWare

7
00:00:17,226 --> 00:00:17,851
at ocw.mit.edu.

8
00:00:26,090 --> 00:00:27,590
PROFESSOR: Last
class, we introduced

9
00:00:27,590 --> 00:00:28,730
thin-film technologies.

10
00:00:28,730 --> 00:00:30,710
We described some of
their generic advantages

11
00:00:30,710 --> 00:00:32,180
and disadvantages.

12
00:00:32,180 --> 00:00:34,980
I showed that particular book
that's going around in the back

13
00:00:34,980 --> 00:00:36,813
there, The Handbook of
Photovoltaics Science

14
00:00:36,813 --> 00:00:37,760
and Engineering.

15
00:00:37,760 --> 00:00:40,410
That describes several
of the technologies

16
00:00:40,410 --> 00:00:43,070
that we're discussing
today and last class.

17
00:00:43,070 --> 00:00:45,060
So last class we covered
general thin films

18
00:00:45,060 --> 00:00:47,610
and cad-tel-- cadmium telluride.

19
00:00:47,610 --> 00:00:50,250
Today we're going to be talking
about amorphous silicon, copper

20
00:00:50,250 --> 00:00:54,000
indium gallium diselenide, and
the last material system, which

21
00:00:54,000 --> 00:00:57,090
is actually more of a stack
or a composite of several.

22
00:00:57,090 --> 00:00:59,790
Then we'll enter a debate,
which will be kind of fun.

23
00:00:59,790 --> 00:01:03,140
So amorphous silicon-- we talked
about crystalline silicon.

24
00:01:03,140 --> 00:01:05,790
Crystalline silicon is
the crystalline form

25
00:01:05,790 --> 00:01:08,910
of the element known as
silicon, comprising a diamond

26
00:01:08,910 --> 00:01:10,820
cubic crystal structure.

27
00:01:10,820 --> 00:01:12,980
Amorphous silicon,
on the other hand,

28
00:01:12,980 --> 00:01:16,520
is a very broad term
catching a range of materials

29
00:01:16,520 --> 00:01:18,530
that lack long-range order.

30
00:01:18,530 --> 00:01:20,950
Let me be a little
bit more specific.

31
00:01:20,950 --> 00:01:23,560
Typically, there is some
semblance of short-range order

32
00:01:23,560 --> 00:01:25,050
in these materials.

33
00:01:25,050 --> 00:01:28,044
One can describe the
composition of amorphous silicon

34
00:01:28,044 --> 00:01:29,960
using, say, for example,
a radial distribution

35
00:01:29,960 --> 00:01:33,220
function-- the average distance
from one atom to the next.

36
00:01:33,220 --> 00:01:35,300
And you have these
distribution functions

37
00:01:35,300 --> 00:01:37,280
describing the material.

38
00:01:37,280 --> 00:01:38,960
It gets a little
bit more tricky when

39
00:01:38,960 --> 00:01:42,190
you try to define it on
a more detailed level,

40
00:01:42,190 --> 00:01:45,190
with average number
of nearest neighbors,

41
00:01:45,190 --> 00:01:48,630
the local configuration, the
number of stretched bonds

42
00:01:48,630 --> 00:01:49,470
locally.

43
00:01:49,470 --> 00:01:53,350
And the reality is that
with any amorphous material,

44
00:01:53,350 --> 00:01:55,490
you can deposit a
range of materials.

45
00:01:55,490 --> 00:01:57,740
It isn't just one
specific amorphous

46
00:01:57,740 --> 00:02:00,004
silicon material that
you get every time.

47
00:02:00,004 --> 00:02:01,670
Depending on your
deposition conditions,

48
00:02:01,670 --> 00:02:05,354
depending on the pressure, and
the temperature, and the power,

49
00:02:05,354 --> 00:02:07,520
during the PCVD-- the
plasma-enhanced chemical vapor

50
00:02:07,520 --> 00:02:11,130
deposition-- you obtain a range
of amorphous silicon materials.

51
00:02:11,130 --> 00:02:16,210
And within that
range, you can find

52
00:02:16,210 --> 00:02:19,070
that the mobility
of charge carriers

53
00:02:19,070 --> 00:02:22,460
can vary by several factors.

54
00:02:22,460 --> 00:02:25,490
The mobility, I believe,
in amorphous silicon

55
00:02:25,490 --> 00:02:30,201
can range easily between
0.001 and 0.1 centimeter

56
00:02:30,201 --> 00:02:31,200
squared per volt second.

57
00:02:31,200 --> 00:02:32,980
So that's two orders
of magnitude easily.

58
00:02:32,980 --> 00:02:35,146
And I have heard of amorphous
silicon materials that

59
00:02:35,146 --> 00:02:37,933
have gone as low as 10 to the
minus 4 centimeter squared

60
00:02:37,933 --> 00:02:41,440
per volt second, and as
high as approaching 1.

61
00:02:41,440 --> 00:02:44,500
So we have a range of material
qualities as a result,

62
00:02:44,500 --> 00:02:47,140
and also a range of performance.

63
00:02:47,140 --> 00:02:49,520
Likewise, there's a range
of stress in the film,

64
00:02:49,520 --> 00:02:51,940
depending on how you deposit
your amorphous silicon.

65
00:02:51,940 --> 00:02:56,750
So delamination is
of concern as well,

66
00:02:56,750 --> 00:02:59,100
depending on how you deposit it.

67
00:02:59,100 --> 00:03:01,530
And the challenges
from a material level

68
00:03:01,530 --> 00:03:05,630
is getting consistent quality
material, high-quality material

69
00:03:05,630 --> 00:03:08,990
every single deposition, and
overcoming certain degradation

70
00:03:08,990 --> 00:03:11,570
modes that are present
in amorphous silicon,

71
00:03:11,570 --> 00:03:14,200
not present in other material
systems-- certainly not present

72
00:03:14,200 --> 00:03:15,340
in crystalline silicon.

73
00:03:15,340 --> 00:03:18,050
For example, the
Staebler-Wronski defect

74
00:03:18,050 --> 00:03:20,960
that's still up for some
debate as to what precisely

75
00:03:20,960 --> 00:03:23,840
is the atomic nature
of that defect.

76
00:03:23,840 --> 00:03:26,490
But it is known that when
one exposes amorphous silicon

77
00:03:26,490 --> 00:03:29,270
to sunlight, that
the performance will

78
00:03:29,270 --> 00:03:31,310
degrade with a
relatively quick time

79
00:03:31,310 --> 00:03:33,970
constant over the period
of minutes and hours,

80
00:03:33,970 --> 00:03:35,700
and finally stabilize.

81
00:03:35,700 --> 00:03:38,050
So when you report
the cell performance

82
00:03:38,050 --> 00:03:41,130
of an amorphous silicon cell,
it's always very important

83
00:03:41,130 --> 00:03:43,730
to emphasize that this
is light-degraded or

84
00:03:43,730 --> 00:03:45,679
light-stabilized
efficiency, as opposed

85
00:03:45,679 --> 00:03:47,970
to the efficiency you get
when you take it straight out

86
00:03:47,970 --> 00:03:50,170
of the deposition
chamber, and it

87
00:03:50,170 --> 00:03:54,560
hasn't been exposed to high
amounts of sunlight yet.

88
00:03:54,560 --> 00:03:58,560
In general, amorphous silicon
has low hole mobility.

89
00:03:58,560 --> 00:04:01,760
Again, the precise root
cause on an atomic scale

90
00:04:01,760 --> 00:04:04,350
for the low hole mobility
is still up for some debate.

91
00:04:04,350 --> 00:04:06,800
There's a professor here
at MIT, Jeff Grossman,

92
00:04:06,800 --> 00:04:11,220
who has performed a combination
of molecular dynamics

93
00:04:11,220 --> 00:04:15,130
and density functional theory
to obtain amorphous silicon

94
00:04:15,130 --> 00:04:16,769
samples in DFT.

95
00:04:16,769 --> 00:04:21,760
And through those samples,
he and his colleagues

96
00:04:21,760 --> 00:04:24,460
were able to determine that
the low hole mobility might

97
00:04:24,460 --> 00:04:26,490
be caused by stretched
bonds, certain types

98
00:04:26,490 --> 00:04:28,787
of stretched bonds, within
the amorphous silicon.

99
00:04:28,787 --> 00:04:30,620
There are other hypotheses
out there as well

100
00:04:30,620 --> 00:04:32,320
in the literature.

101
00:04:32,320 --> 00:04:36,640
Finally, just getting a
uniform film deposition

102
00:04:36,640 --> 00:04:40,520
at high speed-- that's
really key-- a high speed,

103
00:04:40,520 --> 00:04:43,890
uniform deposition
of this layer has

104
00:04:43,890 --> 00:04:46,790
proved challenging in
commercial production so far.

105
00:04:46,790 --> 00:04:51,510
So all of this has combined
so that amorphous silicon

106
00:04:51,510 --> 00:04:54,880
has received a lot of
attention, but hasn't

107
00:04:54,880 --> 00:04:59,870
yet, in a massive way, been
economically successful.

108
00:04:59,870 --> 00:05:04,547
This is a cross-section SCM of
an amorphous silicon device,

109
00:05:04,547 --> 00:05:06,380
showing the TCO-- the
transparent conducting

110
00:05:06,380 --> 00:05:07,760
oxide-- on the top.

111
00:05:07,760 --> 00:05:09,610
The TCO was deposited
onto the glass,

112
00:05:09,610 --> 00:05:12,390
which appears at the very
top underneath the scale bar.

113
00:05:12,390 --> 00:05:15,000
The amorphous silicon layer
right here in the middle

114
00:05:15,000 --> 00:05:16,150
is shown.

115
00:05:16,150 --> 00:05:18,510
It is in a PIN structure.

116
00:05:18,510 --> 00:05:21,950
So we've learned
about P-N junctions.

117
00:05:21,950 --> 00:05:24,700
The I layer between
is an intrinsic layer,

118
00:05:24,700 --> 00:05:26,050
meaning lightly doped.

119
00:05:26,050 --> 00:05:30,350
And we'll see the band
diagram in the next slide.

120
00:05:30,350 --> 00:05:33,420
The silver contact in the back
here extracting the charge

121
00:05:33,420 --> 00:05:35,000
from the backside.

122
00:05:35,000 --> 00:05:39,520
So this is an example of the
amorphous silicon technology.

123
00:05:39,520 --> 00:05:42,150
Again, another cross-section
just for your reference,

124
00:05:42,150 --> 00:05:45,230
showing the stack there.

125
00:05:45,230 --> 00:05:50,170
OK, the PIN device
architecture-- right here

126
00:05:50,170 --> 00:05:53,242
we can see the P plus layer,
or the P layer, the N layer,

127
00:05:53,242 --> 00:05:55,200
and then the I here in
the middle-- the lightly

128
00:05:55,200 --> 00:05:57,500
doped, intrinsic layer.

129
00:05:57,500 --> 00:05:59,410
Note the band structure
or the band diagram,

130
00:05:59,410 --> 00:06:01,850
how it's arranged.

131
00:06:01,850 --> 00:06:04,740
And interestingly,
the layer thicknesses

132
00:06:04,740 --> 00:06:07,880
here are on the order
of 300 nanometers

133
00:06:07,880 --> 00:06:08,860
for the entire stack.

134
00:06:08,860 --> 00:06:12,870
It's rather thin, typically
under a micron in thickness.

135
00:06:12,870 --> 00:06:15,290
The reason it's very thin
is because amorphous silicon

136
00:06:15,290 --> 00:06:19,390
absorbs light much, much better
than crystalline silicon.

137
00:06:19,390 --> 00:06:21,580
For those who have a
solid-state physics background

138
00:06:21,580 --> 00:06:24,389
and want to know the root cause
of that, I'll explain briefly.

139
00:06:24,389 --> 00:06:26,680
For those who don't have the
background and can follow,

140
00:06:26,680 --> 00:06:27,190
don't worry.

141
00:06:27,190 --> 00:06:27,790
Plant a flag.

142
00:06:27,790 --> 00:06:29,100
We'll come right back to it.

143
00:06:29,100 --> 00:06:32,350
So in crystalline silicon,
we have crystal symmetry,

144
00:06:32,350 --> 00:06:35,310
and hence we have momentum
conservation rules

145
00:06:35,310 --> 00:06:38,150
when we excite charge carriers,
hence the indirect band gap

146
00:06:38,150 --> 00:06:39,890
nature of crystalline silicon.

147
00:06:39,890 --> 00:06:42,946
When we amorphize our
material, we no longer

148
00:06:42,946 --> 00:06:44,070
have that crystal symmetry.

149
00:06:44,070 --> 00:06:46,050
We no longer have
momentum conservation.

150
00:06:46,050 --> 00:06:48,340
We can excite directly
from your valence band

151
00:06:48,340 --> 00:06:49,540
into your conduction band.

152
00:06:49,540 --> 00:06:51,865
So that, in a very
quick, hand-wavy way

153
00:06:51,865 --> 00:06:54,460
is why amorphous silicon tends
to behave like a direct band

154
00:06:54,460 --> 00:06:56,090
gap semiconductor,
absorbing light

155
00:06:56,090 --> 00:06:59,100
extremely well with a very high
optical absorption coefficient.

156
00:06:59,100 --> 00:07:01,950
Hence, instead of
100 microns, we

157
00:07:01,950 --> 00:07:04,505
can get away with 300
nanometers of thickness.

158
00:07:04,505 --> 00:07:06,960
All right, back to our flag.

159
00:07:06,960 --> 00:07:08,489
We see here the PIN structure.

160
00:07:08,489 --> 00:07:10,780
Everybody should be able to
follow along and understand

161
00:07:10,780 --> 00:07:13,410
that if light excites a charge
carrier here in the middle,

162
00:07:13,410 --> 00:07:16,070
the electrons will be swept the
right, the holes to the left.

163
00:07:16,070 --> 00:07:19,450
And we'll have charge
separation occurring.

164
00:07:19,450 --> 00:07:25,840
Sometimes, we can have a stack
of materials like shown here,

165
00:07:25,840 --> 00:07:28,680
with varying band
gaps by, for example,

166
00:07:28,680 --> 00:07:31,600
blending germanium into
the amorphous silicon.

167
00:07:31,600 --> 00:07:33,860
This can modify the band
gap of the material.

168
00:07:33,860 --> 00:07:37,060
And we can wind up
with a stack of layers,

169
00:07:37,060 --> 00:07:42,564
or we could simply use two
layers of the same material.

170
00:07:42,564 --> 00:07:44,980
We wouldn't necessarily be
able to absorb more efficiently

171
00:07:44,980 --> 00:07:46,545
across the solar
spectrum, but we

172
00:07:46,545 --> 00:07:49,130
would be able to increase the
voltage output of the device.

173
00:07:49,130 --> 00:07:53,110
So there's a variety of
different amorphous silicon

174
00:07:53,110 --> 00:07:56,000
shall we say designs
out there in the market.

175
00:07:56,000 --> 00:07:58,850
Here's one example
of the triple stack.

176
00:07:58,850 --> 00:08:00,600
We have amorphous
silicon here at the top,

177
00:08:00,600 --> 00:08:01,650
with a large band gap.

178
00:08:01,650 --> 00:08:04,330
The band gap is somewhere
in the range of 1.7 eV.

179
00:08:04,330 --> 00:08:05,678
Yeah, question.

180
00:08:05,678 --> 00:08:08,670
AUDIENCE: I just have a
question of how that-- so

181
00:08:08,670 --> 00:08:10,169
does the [INAUDIBLE]
pair have to go

182
00:08:10,169 --> 00:08:13,670
to the interface between
the [INAUDIBLE] to separate?

183
00:08:13,670 --> 00:08:15,980
PROFESSOR: Yeah, yeah, so
in this particular case,

184
00:08:15,980 --> 00:08:17,790
I believe this is
a dual terminal

185
00:08:17,790 --> 00:08:22,700
device, which would be rather
tricky to perform an operation.

186
00:08:22,700 --> 00:08:25,390
What you have to imagine--
this is right here

187
00:08:25,390 --> 00:08:27,290
with very small perturbation.

188
00:08:27,290 --> 00:08:30,200
If you imagine it being
biased in sunlight,

189
00:08:30,200 --> 00:08:34,030
then you can easily see how a
larger voltage would result.

190
00:08:34,030 --> 00:08:36,734
Right here, you can imagine
a two-terminal device where

191
00:08:36,734 --> 00:08:39,520
you extract charge
from here and here,

192
00:08:39,520 --> 00:08:40,990
or perhaps a
four-terminal device--

193
00:08:40,990 --> 00:08:44,260
which is less common because it
requires more metal contacts.

194
00:08:44,260 --> 00:08:46,540
But that would contact
the device here, here,

195
00:08:46,540 --> 00:08:49,030
again here, and here.

196
00:08:49,030 --> 00:08:52,660
So you'd have the two
devices contacted.

197
00:08:52,660 --> 00:08:54,759
If you have a
two-terminal device,

198
00:08:54,759 --> 00:08:57,050
one can typically think about
it as adding the voltages

199
00:08:57,050 --> 00:08:58,530
together.

200
00:08:58,530 --> 00:09:00,520
Then the current would
be limited by the worst

201
00:09:00,520 --> 00:09:01,800
performer of those two.

202
00:09:01,800 --> 00:09:03,520
And that's why you
have this thinner

203
00:09:03,520 --> 00:09:05,089
than that one over
here, because we

204
00:09:05,089 --> 00:09:07,380
know that the light tends to
be absorbed preferentially

205
00:09:07,380 --> 00:09:08,342
near the front surface.

206
00:09:08,342 --> 00:09:10,300
We're absorbing an
equivalent number of photons

207
00:09:10,300 --> 00:09:13,380
in this thickness and
in that thickness.

208
00:09:13,380 --> 00:09:15,760
So the currents
tend to be matched.

209
00:09:15,760 --> 00:09:18,862
So that's why I would guess,
given the geometry of the fact

210
00:09:18,862 --> 00:09:21,070
that we have silver in the
back and TCO in the front,

211
00:09:21,070 --> 00:09:22,600
and no explicit contacts
here in the middle,

212
00:09:22,600 --> 00:09:24,933
that this is a two-terminal
device as opposed to a four.

213
00:09:27,960 --> 00:09:30,740
This stack right here
is representative

214
00:09:30,740 --> 00:09:34,150
of what is commercialized by Uni
Solar, a company in Michigan,

215
00:09:34,150 --> 00:09:35,770
where we have an
amorphous silicon

216
00:09:35,770 --> 00:09:37,890
layer on the top, wide
band gap semiconductor--

217
00:09:37,890 --> 00:09:40,810
somewhere in the range
of 1.7 to 1.9 eV,

218
00:09:40,810 --> 00:09:42,946
depending on the
deposition conditions.

219
00:09:42,946 --> 00:09:45,070
Again, I mentioned that
amorphous silicon is really

220
00:09:45,070 --> 00:09:46,580
a range of materials.

221
00:09:46,580 --> 00:09:49,080
There is no one
amorphous silicon.

222
00:09:49,080 --> 00:09:52,830
Depending on the
density of atoms

223
00:09:52,830 --> 00:09:54,620
inside of the amorphous
infrastructure

224
00:09:54,620 --> 00:09:58,160
and their local configuration,
you can vary the properties

225
00:09:58,160 --> 00:09:58,680
quite a bit.

226
00:09:58,680 --> 00:10:00,513
So the band gap can
vary, but typically it's

227
00:10:00,513 --> 00:10:02,980
in the range of 1.7 to 1.9 eV.

228
00:10:02,980 --> 00:10:05,570
Blending in germanium--
germanium being a bigger atom--

229
00:10:05,570 --> 00:10:08,460
tends to reduce the band
gap of the material.

230
00:10:08,460 --> 00:10:11,556
And you have successively
higher concentrations

231
00:10:11,556 --> 00:10:13,180
moving toward the
back, so the band gap

232
00:10:13,180 --> 00:10:15,080
shrinks from front to back.

233
00:10:15,080 --> 00:10:17,190
And you can wind up
with a stack like this.

234
00:10:17,190 --> 00:10:22,820
And this is an example here
of a three-cell stack, where

235
00:10:22,820 --> 00:10:25,180
you have the combined
quantum efficiency

236
00:10:25,180 --> 00:10:27,310
and the individual quantum
efficiencies of each

237
00:10:27,310 --> 00:10:31,510
of the sub-cells, showing you
how you can really fill up

238
00:10:31,510 --> 00:10:33,960
the entire solar spectrum
using a combination

239
00:10:33,960 --> 00:10:36,730
of these materials.

240
00:10:36,730 --> 00:10:39,140
Unfortunately, despite
all these efforts,

241
00:10:39,140 --> 00:10:41,880
the material is still
affected by what's

242
00:10:41,880 --> 00:10:44,500
called the Staebler-Wronski
effect in honor

243
00:10:44,500 --> 00:10:49,830
of the people who
initially determined it.

244
00:10:49,830 --> 00:10:52,264
The Staebler-Wronski effect
typically manifests itself

245
00:10:52,264 --> 00:10:53,930
as a degradation of
material performance

246
00:10:53,930 --> 00:10:55,330
as a function of time.

247
00:10:55,330 --> 00:10:59,600
It's also known that as the
layer thickness increases,

248
00:10:59,600 --> 00:11:02,150
the predominance of this
effect also increases,

249
00:11:02,150 --> 00:11:03,610
and its impact also increases.

250
00:11:03,610 --> 00:11:06,720
So it manifests itself as
a reduction of the fill

251
00:11:06,720 --> 00:11:08,420
factor, a reduction
of performance,

252
00:11:08,420 --> 00:11:10,470
and we see the trade-offs
during device design

253
00:11:10,470 --> 00:11:13,590
of amorphous silicon.

254
00:11:13,590 --> 00:11:19,220
So I would say, the essence
of amorphous silicon

255
00:11:19,220 --> 00:11:21,490
is that it's a very
promising material,

256
00:11:21,490 --> 00:11:23,050
has a wide degree of tunability.

257
00:11:23,050 --> 00:11:25,590
But the low hole mobility
limits its performance

258
00:11:25,590 --> 00:11:28,880
to somewhere around 10% for
just a single amorphous silicon

259
00:11:28,880 --> 00:11:31,900
layer, and in the low
teens for the stacks

260
00:11:31,900 --> 00:11:35,070
of multiple materials,
one on top of another.

261
00:11:35,070 --> 00:11:37,510
And this right here is
showing an interesting effect.

262
00:11:37,510 --> 00:11:39,330
When you begin heating
up your substrate,

263
00:11:39,330 --> 00:11:40,720
when you're depositing
amorphous silicon,

264
00:11:40,720 --> 00:11:42,060
you have amorphous material.

265
00:11:42,060 --> 00:11:44,850
If you heat up your substrate
to too high of a temperature

266
00:11:44,850 --> 00:11:46,890
during the deposition
process-- and this is also

267
00:11:46,890 --> 00:11:52,040
impacted by the concentration
of silane gas to hydrogen ratio

268
00:11:52,040 --> 00:11:53,830
during the deposition
process-- one

269
00:11:53,830 --> 00:11:56,620
can obtain a phase transition
from amorphous material

270
00:11:56,620 --> 00:11:59,660
into what's called
microcrystalline material.

271
00:11:59,660 --> 00:12:02,010
That's shown at the very top.
a-Si stands for amorphous

272
00:12:02,010 --> 00:12:05,200
silicon, mu-Si for
microcrystalline silicon.

273
00:12:05,200 --> 00:12:08,550
Microcrystalline silicon is
right at the phase transition

274
00:12:08,550 --> 00:12:11,900
between an amorphous silicon
and a crystalline variant

275
00:12:11,900 --> 00:12:14,140
polycrystalline
variant of silicon.

276
00:12:14,140 --> 00:12:16,400
And in some cases,
it can actually

277
00:12:16,400 --> 00:12:18,380
contain a mixture
of both amorphous

278
00:12:18,380 --> 00:12:20,520
and crystalline regions.

279
00:12:20,520 --> 00:12:23,550
And so it's very tricky to
nail to the position conditions

280
00:12:23,550 --> 00:12:24,720
just right.

281
00:12:24,720 --> 00:12:26,382
Somebody along the
way figured out--

282
00:12:26,382 --> 00:12:27,840
wait a second-- we
can actually use

283
00:12:27,840 --> 00:12:30,160
this microcrystalline
silicon to our advantage,

284
00:12:30,160 --> 00:12:32,050
and stack a layer of
microcrystalline silicon

285
00:12:32,050 --> 00:12:35,340
on the bottom, put a layer of
amorphous silicon on the top.

286
00:12:35,340 --> 00:12:38,671
And that was called a micromorph
technology as a result.

287
00:12:38,671 --> 00:12:41,170
And there are companies that
have commercialized micromorph.

288
00:12:46,100 --> 00:12:49,750
So the micromorph technology
commercialized by Oerlikon

289
00:12:49,750 --> 00:12:51,880
and previously by a company
in the United States

290
00:12:51,880 --> 00:12:53,090
called Applied Materials.

291
00:12:53,090 --> 00:12:55,131
And we'll get to that in
a couple slides as well.

292
00:12:57,810 --> 00:13:02,860
This right here is a brief
slide just addressing

293
00:13:02,860 --> 00:13:06,790
the issue of how
growth conditions can

294
00:13:06,790 --> 00:13:09,210
impact material quality.

295
00:13:09,210 --> 00:13:13,730
There are a number of growth
parameters, shall we say,

296
00:13:13,730 --> 00:13:17,610
that impact whether you
obtain an amorphous material

297
00:13:17,610 --> 00:13:19,920
with high mobility,
with low mobility,

298
00:13:19,920 --> 00:13:23,220
a microcrystalline material,
nanocrystalline material,

299
00:13:23,220 --> 00:13:24,360
and so forth.

300
00:13:24,360 --> 00:13:26,580
And if you're interested
in more details,

301
00:13:26,580 --> 00:13:31,010
these NREL reports from the
late 1990s and early 2000s,

302
00:13:31,010 --> 00:13:34,950
mid 2000s, are really a
treasure trove of information,

303
00:13:34,950 --> 00:13:37,650
if you really want to
get up to speed quickly

304
00:13:37,650 --> 00:13:41,780
on how to grow high-quality
amorphous silicon layers.

305
00:13:41,780 --> 00:13:43,340
OK, so these are
record amorphous

306
00:13:43,340 --> 00:13:45,250
silicon cell efficiencies.

307
00:13:45,250 --> 00:13:49,830
This, I believe, was
from the Martin Green

308
00:13:49,830 --> 00:13:52,610
tables that come out each
half year in Progress

309
00:13:52,610 --> 00:13:53,640
in Photovoltaics.

310
00:13:53,640 --> 00:13:56,940
So he and his colleagues
produce an article every half

311
00:13:56,940 --> 00:13:58,810
of a year in Progress
in Photovoltaics,

312
00:13:58,810 --> 00:14:01,735
describing the record
efficiencies of lab size

313
00:14:01,735 --> 00:14:03,860
cells-- maybe a centimeter
squared-- all the way up

314
00:14:03,860 --> 00:14:05,260
to full-size modules.

315
00:14:05,260 --> 00:14:07,390
So we have here
amorphous silicon

316
00:14:07,390 --> 00:14:09,950
and the blended amorphous
silicon germanium,

317
00:14:09,950 --> 00:14:11,450
or SiGe for short.

318
00:14:11,450 --> 00:14:13,770
People sometimes call
silicon germanium SiGe.

319
00:14:13,770 --> 00:14:17,240
So we have a variety of
different types shown here,

320
00:14:17,240 --> 00:14:20,770
and the areas of the
devices shown here.

321
00:14:20,770 --> 00:14:23,240
I would caution
over-interpretation

322
00:14:23,240 --> 00:14:24,940
of the very small area devices.

323
00:14:24,940 --> 00:14:26,970
If you're at 0.25
centimeters, you're

324
00:14:26,970 --> 00:14:32,140
at a very small linear dimension
along the side of the cell,

325
00:14:32,140 --> 00:14:35,650
and you can wind up being
impacted by edge effects.

326
00:14:35,650 --> 00:14:37,850
Sometimes this is
the active area

327
00:14:37,850 --> 00:14:39,590
of a much larger
device, where they've

328
00:14:39,590 --> 00:14:41,037
shrunken down the aperture.

329
00:14:41,037 --> 00:14:43,120
But depending on the
texturization of the surface,

330
00:14:43,120 --> 00:14:45,810
you still get light
trapping and, essentially,

331
00:14:45,810 --> 00:14:50,140
a larger active area than what
is actually being shaded off.

332
00:14:50,140 --> 00:14:54,780
So I would caution you against
over-interpreting these very

333
00:14:54,780 --> 00:14:56,720
small area measurements.

334
00:14:56,720 --> 00:15:00,980
At 1 centimeter squared larger,
you're typically in a regime

335
00:15:00,980 --> 00:15:03,460
where the results
are more believable.

336
00:15:03,460 --> 00:15:07,940
The Jse's shown here,
Voc's shown here.

337
00:15:07,940 --> 00:15:10,790
Note in certain
cases, the Voc you

338
00:15:10,790 --> 00:15:12,930
would expect from
amorphous silicon,

339
00:15:12,930 --> 00:15:15,270
if it's a band gap
of 1.7 eV, you'd

340
00:15:15,270 --> 00:15:21,080
expect a Voc somewhere in
the range of maybe 1.4 volts.

341
00:15:21,080 --> 00:15:23,970
And sometimes you get
much lower than that.

342
00:15:23,970 --> 00:15:25,860
Likewise, the short
circuit current

343
00:15:25,860 --> 00:15:28,210
is rather low, say, compared
to a crystalline silicon

344
00:15:28,210 --> 00:15:30,980
device because of
that higher band gap--

345
00:15:30,980 --> 00:15:34,990
the inability to capture some
of those lower energy photons.

346
00:15:34,990 --> 00:15:37,260
And so the record
single amorphous

347
00:15:37,260 --> 00:15:40,940
silicon device
performance somewhere

348
00:15:40,940 --> 00:15:43,420
buttressing up against
10%-- not quite.

349
00:15:43,420 --> 00:15:46,310
And then if you combine
different layers into a stack,

350
00:15:46,310 --> 00:15:50,720
for example here, you can
move into the low teens

351
00:15:50,720 --> 00:15:52,330
for your performance.

352
00:15:52,330 --> 00:15:54,750
Flag that-- these numbers here.

353
00:15:54,750 --> 00:15:56,550
We're going to get
back to this, and use

354
00:15:56,550 --> 00:15:58,520
that to understand why
some of the technologies

355
00:15:58,520 --> 00:16:01,750
didn't make it to
commercialization.

356
00:16:01,750 --> 00:16:05,440
This represents the largest
scale attempt to commercialize

357
00:16:05,440 --> 00:16:07,620
amorphous silicon technology.

358
00:16:07,620 --> 00:16:08,635
This was a turnkey line.

359
00:16:08,635 --> 00:16:09,760
Let's start from over here.

360
00:16:09,760 --> 00:16:12,770
This is a human being seated
at a computer-- two humans.

361
00:16:12,770 --> 00:16:15,400
This pod right
here-- this machine--

362
00:16:15,400 --> 00:16:17,930
represents one of these
little components right here

363
00:16:17,930 --> 00:16:20,050
that fits into a much
larger assembly line.

364
00:16:20,050 --> 00:16:22,254
This is the PCVD
reactor, actually.

365
00:16:22,254 --> 00:16:23,670
I believe there
were seven of them

366
00:16:23,670 --> 00:16:26,040
attached to each of these
little pods right here.

367
00:16:26,040 --> 00:16:28,920
So this the plasma-enhanced
chemical vapor deposition

368
00:16:28,920 --> 00:16:32,770
reactors were depositing
microcrystalline silicon

369
00:16:32,770 --> 00:16:35,290
and amorphous silicon
onto sheets of glass

370
00:16:35,290 --> 00:16:37,809
that were as large as you
could possibly transport.

371
00:16:37,809 --> 00:16:39,600
Does anybody just happen
to know the number

372
00:16:39,600 --> 00:16:40,400
off the top of their head?

373
00:16:40,400 --> 00:16:42,030
What is the largest
sheet of glass

374
00:16:42,030 --> 00:16:45,150
that you can physically
transport, say

375
00:16:45,150 --> 00:16:48,265
in commercial form, such
that you could buy it

376
00:16:48,265 --> 00:16:51,190
in bulk without custom design
of the Pilkington float glass

377
00:16:51,190 --> 00:16:51,704
process?

378
00:16:51,704 --> 00:16:54,120
AUDIENCE: It's as big as the
cross-sectional area of that.

379
00:16:54,120 --> 00:16:56,036
PROFESSOR: It's about,
yeah, yeah, definitely.

380
00:16:56,036 --> 00:16:58,090
That's a good place to start.

381
00:16:58,090 --> 00:16:59,959
It's actually
dictated by what can

382
00:16:59,959 --> 00:17:02,250
fit on some of the trucks
that transport the glass back

383
00:17:02,250 --> 00:17:02,749
and forth.

384
00:17:02,749 --> 00:17:05,269
You ever seen the glass
transport trucks with the sides

385
00:17:05,269 --> 00:17:06,810
where they put the
panes of glass in?

386
00:17:06,810 --> 00:17:09,329
And they have the chassis
of the truck like so,

387
00:17:09,329 --> 00:17:11,500
and the glass along the edge.

388
00:17:11,500 --> 00:17:14,230
It's around 3 by 3.3 meters.

389
00:17:14,230 --> 00:17:17,069
And so these deposition
systems could incorporate

390
00:17:17,069 --> 00:17:18,340
huge sheets of glass.

391
00:17:18,340 --> 00:17:20,089
And they would leave
it up to the customer

392
00:17:20,089 --> 00:17:21,569
whether they wanted
to quarter them

393
00:17:21,569 --> 00:17:24,040
and make reasonable-sized
modules, or whether they

394
00:17:24,040 --> 00:17:25,920
want the full-size
module and have

395
00:17:25,920 --> 00:17:28,099
to use cranes to
install them, which

396
00:17:28,099 --> 00:17:30,330
would reduce the labor
content of installation,

397
00:17:30,330 --> 00:17:33,190
but would shift the burden
onto the automation.

398
00:17:33,190 --> 00:17:36,410
So you see the panes
of glass coming along,

399
00:17:36,410 --> 00:17:39,300
pre-cleaning, and eventually
insertion to these devices.

400
00:17:39,300 --> 00:17:40,880
Why so many of those devices?

401
00:17:40,880 --> 00:17:45,170
Why so many PCVD reactors like
little piglets around a pod

402
00:17:45,170 --> 00:17:46,700
here?

403
00:17:46,700 --> 00:17:51,030
Each of these reactors
would have a cycle time

404
00:17:51,030 --> 00:17:53,310
somewhere on the order
of a few 10s of minutes--

405
00:17:53,310 --> 00:17:55,470
so my estimate would be
somewhere between 20 and 30

406
00:17:55,470 --> 00:17:57,020
minutes, maybe up to 40.

407
00:17:57,020 --> 00:17:58,640
Because the
microcrystalline silicon

408
00:17:58,640 --> 00:18:01,524
layer just took so
long to deposit.

409
00:18:01,524 --> 00:18:03,190
But without that
microcrystalline layer,

410
00:18:03,190 --> 00:18:05,780
it was difficult to
reach 10% efficiency.

411
00:18:05,780 --> 00:18:07,740
With just the amorphous
silicon layer,

412
00:18:07,740 --> 00:18:11,230
the efficiencies were in
the range of 6% to 7%.

413
00:18:11,230 --> 00:18:14,420
And if you start doing
calculations for a 6% or 7%

414
00:18:14,420 --> 00:18:18,080
module, you realize very quickly
that the cost of the glass,

415
00:18:18,080 --> 00:18:20,910
and the encapsulants, and
everything else adds up,

416
00:18:20,910 --> 00:18:23,654
because your device is
so low in efficiency

417
00:18:23,654 --> 00:18:26,070
that you need more encapsulant,
more gas, more labor, more

418
00:18:26,070 --> 00:18:28,380
installation to produce
a panel to produce

419
00:18:28,380 --> 00:18:30,040
the same amount of power.

420
00:18:30,040 --> 00:18:33,590
And as a result, the
system cost was very high.

421
00:18:33,590 --> 00:18:38,070
This technology couldn't compete
once the Chinese production

422
00:18:38,070 --> 00:18:40,130
started really ramping
up in the market,

423
00:18:40,130 --> 00:18:43,580
and decreasing the cost of
traditional crystalline silicon

424
00:18:43,580 --> 00:18:44,770
modules.

425
00:18:44,770 --> 00:18:50,300
And the SunFab Project,
which was Applied Materials

426
00:18:50,300 --> 00:18:54,860
name for this amorphous silicon
line-- or micromorph line,

427
00:18:54,860 --> 00:18:59,540
in reality-- was shut
down relatively recently,

428
00:18:59,540 --> 00:19:02,390
about a year ago,
when it I realized

429
00:19:02,390 --> 00:19:03,890
that this was no
longer commercially

430
00:19:03,890 --> 00:19:06,890
viable in the face of some
of the low-cost competition

431
00:19:06,890 --> 00:19:08,850
from crystalline silicon.

432
00:19:08,850 --> 00:19:13,110
All that could change very
quickly, if somehow one of you

433
00:19:13,110 --> 00:19:16,000
were to develop a way to, say,
double the record efficiency

434
00:19:16,000 --> 00:19:16,910
number.

435
00:19:16,910 --> 00:19:20,510
So if someone were to figure
out why exactly hole mobility is

436
00:19:20,510 --> 00:19:23,070
so low, impairing
transport inside

437
00:19:23,070 --> 00:19:26,080
of these materials,
in principle one

438
00:19:26,080 --> 00:19:29,080
could then envision rolling
it out into the SunFab line.

439
00:19:29,080 --> 00:19:31,030
There's already a turnkey
production equipment

440
00:19:31,030 --> 00:19:33,360
that's been built by Applied
Materials, ready to go.

441
00:19:33,360 --> 00:19:35,860
And you're off and running.

442
00:19:35,860 --> 00:19:38,000
So that's an
opportunity out there.

443
00:19:38,000 --> 00:19:40,850
This lists some of the
commercialization attempts.

444
00:19:40,850 --> 00:19:42,620
And the bottom line
at the end of the day

445
00:19:42,620 --> 00:19:46,550
was that 6% modules were too
inefficient to be profitable,

446
00:19:46,550 --> 00:19:48,479
unless you reach scales
of, say, gigawatts.

447
00:19:48,479 --> 00:19:51,020
And believe it or not, there
was one company out there called

448
00:19:51,020 --> 00:19:53,340
OptiSolar, the business
plan was literally

449
00:19:53,340 --> 00:19:55,890
to scale up to a gigawatt
faster than everybody else.

450
00:19:55,890 --> 00:19:59,530
And even though it
was lower performance,

451
00:19:59,530 --> 00:20:03,910
the sheer size and scale of
the manufacturing facility

452
00:20:03,910 --> 00:20:06,090
would reduce the cost
of the amorphous silicon

453
00:20:06,090 --> 00:20:09,210
relative to the crystalline
silicon competition.

454
00:20:09,210 --> 00:20:12,530
Unfortunately, OptiSolar
didn't make it, either.

455
00:20:12,530 --> 00:20:15,320
So the companies
that are left today

456
00:20:15,320 --> 00:20:17,820
producing amorphous
silicon include Oerlikon.

457
00:20:17,820 --> 00:20:21,840
This is a Swiss company that
produces turnkey equipment,

458
00:20:21,840 --> 00:20:25,766
and Uni Solar, which is
selling the actual modules

459
00:20:25,766 --> 00:20:26,640
of amorphous silicon.

460
00:20:26,640 --> 00:20:29,010
They have a nice roll to
roll deposition process

461
00:20:29,010 --> 00:20:33,210
on these stainless steel foils.

462
00:20:33,210 --> 00:20:35,220
Very nice factory--
believe it was

463
00:20:35,220 --> 00:20:39,160
toured by several members of
our current administration.

464
00:20:39,160 --> 00:20:41,747
And they're still going.

465
00:20:41,747 --> 00:20:42,542
AUDIENCE: Sorry--

466
00:20:42,542 --> 00:20:43,208
PROFESSOR: Yeah.

467
00:20:43,208 --> 00:20:44,874
AUDIENCE: Is the roll
that that person's

468
00:20:44,874 --> 00:20:48,570
holding-- what is he holding?

469
00:20:48,570 --> 00:20:50,650
PROFESSOR: I usually
think, based on the color,

470
00:20:50,650 --> 00:20:53,440
that this already has amorphous
silicon deposited onto it.

471
00:20:53,440 --> 00:20:56,500
But it could be at some
intermediate phase.

472
00:20:56,500 --> 00:20:59,830
I don't know for that specific
roll, but in principle,

473
00:20:59,830 --> 00:21:01,580
yes, at the end of the
manufacturing line,

474
00:21:01,580 --> 00:21:04,100
they'll have a full stack--
the triple amorphous

475
00:21:04,100 --> 00:21:07,550
silicon and then the two
layers of SiGe beneath it.

476
00:21:07,550 --> 00:21:10,340
AUDIENCE: What's--
the base is flexible,

477
00:21:10,340 --> 00:21:13,320
so what's the thing that the
film is deposited on, do you

478
00:21:13,320 --> 00:21:14,492
know?

479
00:21:14,492 --> 00:21:15,950
PROFESSOR: So what
they incorporate

480
00:21:15,950 --> 00:21:19,060
their flexible material
into typically,

481
00:21:19,060 --> 00:21:21,906
one of their products
is roof shingle.

482
00:21:21,906 --> 00:21:24,280
That looks very similar to a
standard, asphalt-based roof

483
00:21:24,280 --> 00:21:25,410
shingle.

484
00:21:25,410 --> 00:21:29,460
And it is very nice building
integrated technology.

485
00:21:29,460 --> 00:21:32,430
I happened to see one back
when I was a graduate student,

486
00:21:32,430 --> 00:21:37,370
and I was inspired by the
building integrated potential

487
00:21:37,370 --> 00:21:39,240
there.

488
00:21:39,240 --> 00:21:41,502
So there are attempts to
integrate amorphous silicon

489
00:21:41,502 --> 00:21:43,710
into regular crystalline
silicon technology, as well.

490
00:21:43,710 --> 00:21:45,835
Turns out amorphous silicon
passivates the surfaces

491
00:21:45,835 --> 00:21:46,490
very nicely.

492
00:21:46,490 --> 00:21:49,550
And you can create what's called
a HIT cell structure, which

493
00:21:49,550 --> 00:21:52,540
is a heterojunction with
a thin intrinsic layer,

494
00:21:52,540 --> 00:21:54,390
using a silicon substrate.

495
00:21:54,390 --> 00:21:57,880
Typically what's used
is an n-type base.

496
00:21:57,880 --> 00:22:00,300
And then the amorphous
silicon layers

497
00:22:00,300 --> 00:22:02,360
are deposited on either side.

498
00:22:02,360 --> 00:22:04,820
And that device
is currently being

499
00:22:04,820 --> 00:22:08,110
commercialized by a company
called Sanyo in Japan.

500
00:22:08,110 --> 00:22:11,490
The patents for the
base HIT cell structure

501
00:22:11,490 --> 00:22:14,980
expire very soon, if they
haven't expired already.

502
00:22:14,980 --> 00:22:19,540
And so there is a fair amount
of interest-- several companies

503
00:22:19,540 --> 00:22:22,100
out there right now attempting
to reproduce this process.

504
00:22:22,100 --> 00:22:23,356
But it is fairly tricky.

505
00:22:23,356 --> 00:22:24,730
And again, a lot
of it boils down

506
00:22:24,730 --> 00:22:26,940
to managing the
interfaces-- making

507
00:22:26,940 --> 00:22:30,110
sure the interface between the
crystal silicon and amorphous

508
00:22:30,110 --> 00:22:34,780
silicon are of high quality.

509
00:22:34,780 --> 00:22:37,570
So we have our HIT
cell structure here.

510
00:22:37,570 --> 00:22:40,670
Let me hop over to the
last thin-film materials

511
00:22:40,670 --> 00:22:42,880
so we have adequate
time for our debates.

512
00:22:42,880 --> 00:22:48,070
The copper indium gallium
diselenide, or SIGS for short--

513
00:22:48,070 --> 00:22:51,090
this is a quaternary
phase, sometimes comprised

514
00:22:51,090 --> 00:22:52,760
of even five elements.

515
00:22:52,760 --> 00:22:56,530
But let's show this is
the chalcopyrite structure

516
00:22:56,530 --> 00:23:01,320
right here, with copper shown in
black, indium or gallium shown

517
00:23:01,320 --> 00:23:05,430
in gray-- typically there's
some blend between the two--

518
00:23:05,430 --> 00:23:08,890
and the chalcogen-- the
selenium or sulfur--

519
00:23:08,890 --> 00:23:13,600
shown in the white right
here, typically selenium.

520
00:23:13,600 --> 00:23:18,190
So we have here an example
of the crystal structure.

521
00:23:18,190 --> 00:23:22,080
As I understand the situation,
this crystal structure

522
00:23:22,080 --> 00:23:26,000
came into being because copper
sulfide demonstrated potential

523
00:23:26,000 --> 00:23:30,440
as a photovoltaic material
early on, say in the late 1970s,

524
00:23:30,440 --> 00:23:31,480
early 1980s.

525
00:23:31,480 --> 00:23:34,320
There were attempts to grow
high-performance copper sulfide

526
00:23:34,320 --> 00:23:35,360
cells.

527
00:23:35,360 --> 00:23:38,520
They failed because
copper electromigrated.

528
00:23:38,520 --> 00:23:40,430
In the presence of an
electric field, copper,

529
00:23:40,430 --> 00:23:42,700
a very small ion
far to the right

530
00:23:42,700 --> 00:23:44,929
on the 3-d series
of elements-- so

531
00:23:44,929 --> 00:23:46,470
if you look at the
transition metals,

532
00:23:46,470 --> 00:23:48,140
copper is on the
far right, meaning

533
00:23:48,140 --> 00:23:51,010
you're adding more and more
and more protons to the nucleus

534
00:23:51,010 --> 00:23:52,700
as you go from left
to right, and you're

535
00:23:52,700 --> 00:23:54,297
pulling the outer shell in.

536
00:23:54,297 --> 00:23:56,630
So if you look at atomic
radius going from left to right

537
00:23:56,630 --> 00:23:58,020
in the periodic
table, you'll see

538
00:23:58,020 --> 00:23:59,920
that the atoms tend to shrink.

539
00:23:59,920 --> 00:24:01,750
So copper, being on
the far right-hand side

540
00:24:01,750 --> 00:24:05,200
of the 3-d elemental series,
has a smaller atomic radius

541
00:24:05,200 --> 00:24:08,520
compared to, say, iron,
or manganese, or titanium.

542
00:24:08,520 --> 00:24:10,950
And as a result, it
could move fairly easily

543
00:24:10,950 --> 00:24:14,480
throughout the lattice if it
is present in a charged state--

544
00:24:14,480 --> 00:24:17,340
if you have, say, copper plus--
and a large electric field

545
00:24:17,340 --> 00:24:20,190
distributed across a thin
film solar cell device.

546
00:24:20,190 --> 00:24:24,554
So copper sulfide devices ended
up degrading quickly over time.

547
00:24:24,554 --> 00:24:26,470
And it was realized that
some heavier elements

548
00:24:26,470 --> 00:24:28,750
needed to be included
to stabilize the crystal

549
00:24:28,750 --> 00:24:30,090
structure.

550
00:24:30,090 --> 00:24:33,530
So in the lab, the reason why
SIGS-- this particular material

551
00:24:33,530 --> 00:24:36,940
right here-- is so interesting
is because in the laboratory,

552
00:24:36,940 --> 00:24:40,250
efficiencies on the order
of 20% have been achieved.

553
00:24:40,250 --> 00:24:42,730
For large area-- you'll have
to correct this number right

554
00:24:42,730 --> 00:24:48,450
here-- you can increase that
to 15% plus-- 15.7% in fact,

555
00:24:48,450 --> 00:24:50,810
over large area devices.

556
00:24:50,810 --> 00:24:53,910
That's a very recent result
coming out of MiaSole,

557
00:24:53,910 --> 00:24:56,430
a company in California.

558
00:24:56,430 --> 00:24:59,580
There are dozens of
startup companies

559
00:24:59,580 --> 00:25:01,330
focused on SIGS development.

560
00:25:01,330 --> 00:25:03,780
And it is a very
challenging problem

561
00:25:03,780 --> 00:25:06,820
to get the stoichiometry
right, and to passivate

562
00:25:06,820 --> 00:25:09,890
the defects inside the
structure, including surfaces.

563
00:25:09,890 --> 00:25:11,480
Where many companies
fail or fall

564
00:25:11,480 --> 00:25:15,030
short is getting the
stoichiometry just right

565
00:25:15,030 --> 00:25:16,810
over a large area.

566
00:25:16,810 --> 00:25:20,530
It's as simple as a
manufacturing challenge.

567
00:25:20,530 --> 00:25:24,130
The idealized structure of
these thin-film devices--

568
00:25:24,130 --> 00:25:28,090
you have your SIGS layer
right here in the middle.

569
00:25:28,090 --> 00:25:30,530
In this case, it's
shown as CIS or cis,

570
00:25:30,530 --> 00:25:32,170
but you can add gallium as well.

571
00:25:32,170 --> 00:25:34,190
Typically, gallium
is blended in.

572
00:25:34,190 --> 00:25:37,240
Moly-- molybdenum-- back
contact on the glass--

573
00:25:37,240 --> 00:25:40,350
the glass in this case is just
the support for the structure.

574
00:25:40,350 --> 00:25:42,440
Molybdenum here would
be non-transparent.

575
00:25:42,440 --> 00:25:44,250
It'd be an opaque layer.

576
00:25:44,250 --> 00:25:46,570
The cadmium sulfide
on the front side

577
00:25:46,570 --> 00:25:53,350
here forms a very nice interface
with the SIGS layer underneath,

578
00:25:53,350 --> 00:25:57,320
preventing any interface
damage or any segregation,

579
00:25:57,320 --> 00:25:59,250
say, of gallium to
the surface here.

580
00:25:59,250 --> 00:26:01,930
And then the zinc
oxide layer on the top,

581
00:26:01,930 --> 00:26:05,940
which serves as the buffer
window layer, depending, is

582
00:26:05,940 --> 00:26:09,970
then deposited on the
cadmium sulfide layer.

583
00:26:09,970 --> 00:26:13,260
And eventually charge
is brought away.

584
00:26:13,260 --> 00:26:15,780
This right here is another
three-dimensional rendition

585
00:26:15,780 --> 00:26:16,960
of the same.

586
00:26:16,960 --> 00:26:19,890
What is unique about each
of those dozens of startup

587
00:26:19,890 --> 00:26:22,760
companies is the process in
which they deposit the SIGS

588
00:26:22,760 --> 00:26:24,324
layer, typically.

589
00:26:24,324 --> 00:26:25,740
So you have companies
like MiaSole

590
00:26:25,740 --> 00:26:28,470
that are invested in
sputtering of SIGS.

591
00:26:28,470 --> 00:26:30,780
You have companies
like Nanosolar

592
00:26:30,780 --> 00:26:33,930
that are invested in
ink-based printing of SIGS.

593
00:26:33,930 --> 00:26:35,750
So they print an ink
down, and then they

594
00:26:35,750 --> 00:26:37,810
sinter the ink to
form the layer.

595
00:26:37,810 --> 00:26:42,010
And like those two, there are
many other companies out there

596
00:26:42,010 --> 00:26:45,360
that are depositing SIGS in
some way, shape, or form.

597
00:26:45,360 --> 00:26:47,990
The folks at NREL
seem to be more

598
00:26:47,990 --> 00:26:53,250
in favor of thermal evaporation
based process for SIGS.

599
00:26:53,250 --> 00:26:57,960
Many of the more successful
industrial renditions of SIGS

600
00:26:57,960 --> 00:27:00,450
right now-- at least the ones
that appear most promising--

601
00:27:00,450 --> 00:27:03,290
are using other
deposition methods.

602
00:27:03,290 --> 00:27:06,050
So you'll see a
variety out there,

603
00:27:06,050 --> 00:27:08,840
of deposition methods
for SIGS technology.

604
00:27:08,840 --> 00:27:10,710
And one of the things
to keep track of,

605
00:27:10,710 --> 00:27:13,650
as you watch and monitor
all these startup companies,

606
00:27:13,650 --> 00:27:20,400
is to bin them by their process,
and to begin to discern trends.

607
00:27:20,400 --> 00:27:23,660
As we start to see some market
leaders evolve in the sector,

608
00:27:23,660 --> 00:27:26,150
begin to discern which
technologies are the most

609
00:27:26,150 --> 00:27:28,880
appropriate, perhaps for
large-scale deposition

610
00:27:28,880 --> 00:27:31,360
of this material system.

611
00:27:31,360 --> 00:27:33,740
Interfaces are critical.

612
00:27:33,740 --> 00:27:37,400
Here are some examples of
attempts at mapping out--

613
00:27:37,400 --> 00:27:41,710
and quite successful attempts at
mapping out the, shall we say,

614
00:27:41,710 --> 00:27:46,310
the band diagram of a
SIGS-based solar cell.

615
00:27:46,310 --> 00:27:49,100
One thing to keep in mind
is that, just like amorphous

616
00:27:49,100 --> 00:27:51,870
silicon represents a
broad class of materials,

617
00:27:51,870 --> 00:27:55,700
SIGS also represents a broad
class of stoichiometries,

618
00:27:55,700 --> 00:27:57,790
meaning the ratio of
indium to gallium,

619
00:27:57,790 --> 00:28:01,840
the ratio of-- if you include
it at all-- sulfur to selenium

620
00:28:01,840 --> 00:28:05,060
is critical in determining
the properties of both

621
00:28:05,060 --> 00:28:07,680
the bulk and the surfaces.

622
00:28:07,680 --> 00:28:09,870
Typically at the surfaces,
you will get termination

623
00:28:09,870 --> 00:28:12,350
by one atom, one atomic
species or another-- one

624
00:28:12,350 --> 00:28:13,650
elemental species or another.

625
00:28:13,650 --> 00:28:15,620
And that will determine,
to a large degree,

626
00:28:15,620 --> 00:28:18,170
the electronic quality
of that surface.

627
00:28:18,170 --> 00:28:19,780
So for a long time
in literature,

628
00:28:19,780 --> 00:28:25,850
you had debating world views of
what the band diagram of a SIGS

629
00:28:25,850 --> 00:28:27,140
device really looked like.

630
00:28:27,140 --> 00:28:29,720
And it turned out a lot
of people were right.

631
00:28:29,720 --> 00:28:31,670
They just had varying
starting materials.

632
00:28:31,670 --> 00:28:35,010
So they were comparing
apples to oranges.

633
00:28:35,010 --> 00:28:37,900
The thing that keeps
people going with SIGS

634
00:28:37,900 --> 00:28:40,150
is that in the
laboratory, efficiencies,

635
00:28:40,150 --> 00:28:43,170
I believe, now up to 21%
have been demonstrated.

636
00:28:43,170 --> 00:28:45,140
So that is really encouraging.

637
00:28:45,140 --> 00:28:47,140
People-- venture capitalists--
will look at this

638
00:28:47,140 --> 00:28:49,530
and say, wow, with
those high efficiencies,

639
00:28:49,530 --> 00:28:51,510
you could really
make solar cheap.

640
00:28:51,510 --> 00:28:53,320
The amount of, again,
glass, encapsulants,

641
00:28:53,320 --> 00:28:56,370
labor, and so forth that you
have to have for watt peak

642
00:28:56,370 --> 00:28:58,380
is lower, because
the cell can perform

643
00:28:58,380 --> 00:28:59,940
at a higher performance.

644
00:28:59,940 --> 00:29:03,350
But the downside is, there's a
gap between-- a substantial gap

645
00:29:03,350 --> 00:29:05,970
between-- what is commercially
manufacturable today

646
00:29:05,970 --> 00:29:08,000
and that record
efficiency device.

647
00:29:08,000 --> 00:29:12,880
Sometimes the practical
matters of commercialization

648
00:29:12,880 --> 00:29:14,590
are the ones that
impede a technology

649
00:29:14,590 --> 00:29:15,800
from getting to market.

650
00:29:15,800 --> 00:29:20,600
And I have a modicum of
faith that SIGS will make it.

651
00:29:20,600 --> 00:29:24,320
But if it fails to reach
its true market potential,

652
00:29:24,320 --> 00:29:27,690
if it fails to realize
its dream of becoming

653
00:29:27,690 --> 00:29:30,290
a significant fraction of
all solar panels produced,

654
00:29:30,290 --> 00:29:33,820
it will most likely be because
of these process engineering

655
00:29:33,820 --> 00:29:36,230
issues associated with
depositing over a large area,

656
00:29:36,230 --> 00:29:42,440
coupled to the scientific gap
in managing interfaces properly.

657
00:29:42,440 --> 00:29:47,010
So that's the story
of SIGS in a nutshell.

658
00:29:47,010 --> 00:29:48,820
There are some minor
issues, shall we say,

659
00:29:48,820 --> 00:29:51,520
associated with cadmium.

660
00:29:51,520 --> 00:29:58,730
There are attempts to go to
an all zinc oxide or maybe

661
00:29:58,730 --> 00:30:03,020
a graded zinc oxysulfide
buffering window layer here

662
00:30:03,020 --> 00:30:04,680
in the front to get
rid of the cadmium.

663
00:30:04,680 --> 00:30:08,230
We'll discuss that shortly.

664
00:30:08,230 --> 00:30:15,360
And I would say that's
probably the biggest

665
00:30:15,360 --> 00:30:16,490
near-term challenge.

666
00:30:16,490 --> 00:30:20,090
With the removal of cadmium,
you can access more markets.

667
00:30:20,090 --> 00:30:22,650
And secondly, SIGS
does contain indium,

668
00:30:22,650 --> 00:30:25,990
which is not in an infinite
supply in Earth's crust.

669
00:30:25,990 --> 00:30:29,320
It's debatable how much of
an issue that really is.

670
00:30:29,320 --> 00:30:31,370
And we'll get to that
in our discussions

671
00:30:31,370 --> 00:30:33,250
as well, in our debates.

672
00:30:33,250 --> 00:30:37,900
SIGS commercialization--
several startup companies

673
00:30:37,900 --> 00:30:41,780
and I would say a lot
of promising research

674
00:30:41,780 --> 00:30:45,291
and development
going on right now.

675
00:30:45,291 --> 00:30:49,370
This particular
technology right here--

676
00:30:49,370 --> 00:30:52,300
let's keep this company
nameless for now.

677
00:30:52,300 --> 00:30:54,920
This company represents
some of the difficulties

678
00:30:54,920 --> 00:30:57,006
in ramping up large area SIGS.

679
00:30:57,006 --> 00:30:59,450
They deposit it sometimes
over large areas,

680
00:30:59,450 --> 00:31:01,330
but would achieve
inhomogeneous results.

681
00:31:01,330 --> 00:31:05,000
And so the ultimate form
factor for the devices

682
00:31:05,000 --> 00:31:07,630
was to chop them
up into areas that

683
00:31:07,630 --> 00:31:10,510
are not too dissimilar from
an actual crystalline silicon

684
00:31:10,510 --> 00:31:11,230
device.

685
00:31:11,230 --> 00:31:14,360
So to chop up their layers into
individual, discrete cells,

686
00:31:14,360 --> 00:31:17,777
and then to connect them--
solder them together,

687
00:31:17,777 --> 00:31:19,360
much like a crystalline
silicon device

688
00:31:19,360 --> 00:31:21,730
would be put together
inside a module.

689
00:31:21,730 --> 00:31:25,960
So the dream of having this
inline, thin-film process

690
00:31:25,960 --> 00:31:30,760
roll to roll was reaching an
abrupt wall, shall we say,

691
00:31:30,760 --> 00:31:33,000
in the development process
because of the lack

692
00:31:33,000 --> 00:31:34,590
of large-scale uniformity.

693
00:31:34,590 --> 00:31:37,460
So you need to chop them up
into smaller units, which

694
00:31:37,460 --> 00:31:39,600
would add to the cost,
which would decrease

695
00:31:39,600 --> 00:31:42,435
the module performance as
well, and now you have gaps

696
00:31:42,435 --> 00:31:43,310
in between the cells.

697
00:31:43,310 --> 00:31:46,890
So it just goes to show
the difficulty of bringing

698
00:31:46,890 --> 00:31:48,980
a new technology to market.

699
00:31:48,980 --> 00:31:52,410
And hopefully, they're
doing better now.

700
00:31:52,410 --> 00:31:58,070
Lots of promising news for SIGS,
like this particular article--

701
00:31:58,070 --> 00:32:01,200
"The Rise of SIGS,
Finally?" question mark.

702
00:32:01,200 --> 00:32:05,570
There has always
been a lot of promise

703
00:32:05,570 --> 00:32:08,390
in the realm of SIGS because
of the high performance

704
00:32:08,390 --> 00:32:09,900
and high efficiencies reached.

705
00:32:09,900 --> 00:32:11,990
The big question is,
can you accomplish this

706
00:32:11,990 --> 00:32:15,650
at large scale for
large area modules?

707
00:32:15,650 --> 00:32:18,320
That's been the big challenge
so far-- and of course,

708
00:32:18,320 --> 00:32:19,890
materials availability.

709
00:32:19,890 --> 00:32:21,980
So let me pause right here.

710
00:32:21,980 --> 00:32:24,800
What I'm going to do-- we
are going to shift gears.

711
00:32:24,800 --> 00:32:27,110
And for the next 20
minutes-- unfortunately,

712
00:32:27,110 --> 00:32:29,160
not half an hour, but
for the next 20 minutes--

713
00:32:29,160 --> 00:32:31,780
we are going to
enter debate mode.

714
00:32:31,780 --> 00:32:34,930
What we're going to do is divide
the group into-- this group

715
00:32:34,930 --> 00:32:37,336
right here-- into
your project teams.

716
00:32:37,336 --> 00:32:39,210
And you'll cluster
together and work together

717
00:32:39,210 --> 00:32:40,557
on this next exercise.

718
00:32:40,557 --> 00:32:42,140
So for the very
beginning, we're going

719
00:32:42,140 --> 00:32:45,470
to have the cadmium
telluride debate.

720
00:32:45,470 --> 00:32:46,980
That will be debate number one.

721
00:32:46,980 --> 00:32:50,220
We'll be discussing the
pros and cons of cad-tel.

722
00:32:50,220 --> 00:32:52,470
Keep in mind that First
Solar, a gigawatt company,

723
00:32:52,470 --> 00:32:54,770
meaning one of the largest
companies in the world,

724
00:32:54,770 --> 00:32:58,940
and certainly the largest
company in the United States,

725
00:32:58,940 --> 00:33:02,080
is going full force producing
cadmium telluride modules.

726
00:33:02,080 --> 00:33:05,520
The debate question that
you'll be faced with is, is it

727
00:33:05,520 --> 00:33:08,060
a good investment for
the United States?

728
00:33:08,060 --> 00:33:10,980
Say, for example, you're
in charge of US government

729
00:33:10,980 --> 00:33:12,530
research and development funds.

730
00:33:12,530 --> 00:33:15,080
Should you be putting
your money into cad-tel?

731
00:33:15,080 --> 00:33:17,610
Or should you be putting your
money into something else?

732
00:33:17,610 --> 00:33:19,630
And we'll have two sides
of the debate-- one

733
00:33:19,630 --> 00:33:22,434
that will defend cad-tel
and say, no, it's

734
00:33:22,434 --> 00:33:23,600
a very promising technology.

735
00:33:23,600 --> 00:33:24,764
We should be going all in.

736
00:33:24,764 --> 00:33:26,430
And the other side
that will be adopting

737
00:33:26,430 --> 00:33:28,180
the counter argument
saying, well, cad-tel

738
00:33:28,180 --> 00:33:29,560
has these concerns.

739
00:33:29,560 --> 00:33:32,180
Perhaps we should be
looking elsewhere to invest.

740
00:33:32,180 --> 00:33:35,990
And that's the debate number
one-- cad-tel, no cad-tel.

741
00:33:35,990 --> 00:33:39,470
For the folks working on
earth-abundant materials,

742
00:33:39,470 --> 00:33:42,800
the second debate will be
focused on that question.

743
00:33:42,800 --> 00:33:44,580
Recently, there's
been a lot of interest

744
00:33:44,580 --> 00:33:47,203
in developing earth-abundant
alternatives to, say, SIGS,

745
00:33:47,203 --> 00:33:50,600
because of the indium
content, and cad-tel because

746
00:33:50,600 --> 00:33:52,460
of the tellurium content.

747
00:33:52,460 --> 00:33:55,909
And there's been a huge
amount of momentum behind it.

748
00:33:55,909 --> 00:33:57,450
But there are few
naysayers out there

749
00:33:57,450 --> 00:33:58,960
who say, well, wait a second.

750
00:33:58,960 --> 00:34:01,500
We've always come
up against shortages

751
00:34:01,500 --> 00:34:04,930
of one kind or another, since
the days of Malthus's warning

752
00:34:04,930 --> 00:34:08,530
about the limits of population
growth because of limited food

753
00:34:08,530 --> 00:34:09,830
supply.

754
00:34:09,830 --> 00:34:11,420
But then we developed
fertilizers.

755
00:34:11,420 --> 00:34:13,219
And that overcame that limit.

756
00:34:13,219 --> 00:34:15,570
Likewise, in these
resource abundance issues,

757
00:34:15,570 --> 00:34:17,909
I'm sure we're going to
find some new way to extract

758
00:34:17,909 --> 00:34:18,630
the metals.

759
00:34:18,630 --> 00:34:20,250
We'll find a way around it.

760
00:34:20,250 --> 00:34:23,110
We'll discover new
deposits and so forth.

761
00:34:23,110 --> 00:34:26,060
You hear similar arguments in
the peak oil debate, right?

762
00:34:26,060 --> 00:34:29,199
And so what I'd like to do
is to set up two teams there

763
00:34:29,199 --> 00:34:31,780
as well to debate this
particular question.

764
00:34:37,203 --> 00:34:39,179
[APPLAUSE]

765
00:34:39,179 --> 00:34:41,489
And we furthermore
recognize that several

766
00:34:41,489 --> 00:34:42,870
of these issues
that might appear

767
00:34:42,870 --> 00:34:46,482
black and white, in reality
have many shades of gray in PV.

768
00:34:46,482 --> 00:34:47,940
This cad-tel debate
is one of them.

769
00:34:47,940 --> 00:34:50,080
We'll hear the second
debate the next time.

770
00:34:50,080 --> 00:34:53,027
And anything from should we
invest in silicon technology

771
00:34:53,027 --> 00:34:55,610
or should we develop thin films
will have these shades of gray

772
00:34:55,610 --> 00:34:56,389
as well.

773
00:34:56,389 --> 00:34:58,980
So it's important to
recognize that simple fact

774
00:34:58,980 --> 00:35:01,757
as you gain maturity, and shades
of gray in your own ability

775
00:35:01,757 --> 00:35:03,090
to argument one side or another.

776
00:35:03,090 --> 00:35:06,010
So thank you, and we'll
see us on Thursday.