WEBVTT

1
00:04:18.130 --> 00:04:19.159
Oliva Primer-PNNL: Hi Graham!

2
00:04:23.030 --> 00:04:23.860
Grant Johnson: Hey, Oliver.

3
00:04:24.740 --> 00:04:28.417
Oliva Primer-PNNL: Yeah, I'm in the hospital. But I say, oh, I can join waiting for

4
00:04:28.820 --> 00:04:34.480
Oliva Primer-PNNL: to finish the surgery. So I so I'd say I can connect. So you want to present? Yes.

5
00:04:34.480 --> 00:04:37.173
Grant Johnson: I'm gonna start and then chun long. And

6
00:04:37.660 --> 00:04:41.139
Grant Johnson: Zhao, gunner, gonna can you tell me if you can see my slides.

7
00:04:41.300 --> 00:04:43.930
Oliva Primer-PNNL: Of course. Yeah, yes, I can see your slides.

8
00:04:43.930 --> 00:04:45.550
Grant Johnson: It's presentation, mode.

9
00:04:45.550 --> 00:04:47.390
Oliva Primer-PNNL: Yeah, it is in presentation. One.

10
00:04:47.700 --> 00:04:48.400
Grant Johnson: Okay.

11
00:04:49.790 --> 00:04:50.940
Chun-Long Chen (PNNL): Yeah. Now it's good.

12
00:04:51.570 --> 00:04:53.930
Chun-Long Chen (PNNL): Why, I had a another link. I just.

13
00:04:53.930 --> 00:05:00.080
Grant Johnson: No, there were 2 links, so I logged into the one from last week. If no one shows up, I'll I'll switch to the other one.

14
00:05:00.650 --> 00:05:04.289
Chun-Long Chen (PNNL): Okay, yeah, I I clicked the other one there. I was the only one there.

15
00:05:04.290 --> 00:05:06.380
Grant Johnson: Was there? Was there anybody in the other one?

16
00:05:06.380 --> 00:05:09.210
Grant Johnson: No, currently nobody. Is there. Okay?

17
00:05:14.980 --> 00:05:20.319
Chun-Long Chen (PNNL): Yeah, I don't know why the other one is weekly, right, or both weekly, or

18
00:05:21.360 --> 00:05:23.349
Chun-Long Chen (PNNL): actually, I don't know which one joined now.

19
00:05:23.350 --> 00:05:27.670
Grant Johnson: I I did the one that we use last week. So I think that's all right.

20
00:05:27.670 --> 00:05:29.210
Chun-Long Chen (PNNL): Maybe that's the right one.

21
00:05:30.590 --> 00:05:32.640
Chun-Long Chen (PNNL): I think the other one could be

22
00:05:32.790 --> 00:05:37.619
Chun-Long Chen (PNNL): just that used to be a master meeting right for the leadership.

23
00:05:39.210 --> 00:05:40.590
Chun-Long Chen (PNNL): Shall Garcia.

24
00:05:42.260 --> 00:05:42.860
Xiaogan Liang: Hi Eric.

25
00:06:14.760 --> 00:06:15.540
Christian Mailhiot - Sandia National Laboratories: See.

26
00:06:40.440 --> 00:06:42.500
Christian Mailhiot - Sandia National Laboratories: Hi, Jim, how are you doing.

27
00:06:42.500 --> 00:06:44.740
James Ang: Hi, Christian! Well, I'm doing well.

28
00:06:45.530 --> 00:06:46.100
Christian Mailhiot - Sandia National Laboratories: Great.

29
00:06:53.340 --> 00:06:56.020
James Ang: Wow! I feel like I know everyone on the call right now.

30
00:06:56.924 --> 00:06:57.459
Grant Johnson: Right.

31
00:06:58.250 --> 00:07:01.200
Chun-Long Chen (PNNL): I think we usually need to wait for a few minutes right.

32
00:07:01.200 --> 00:07:04.590
James Ang: Oh, yeah, yeah, we'll wait. We'll wait. We'll wait till we get a bigger quorum.

33
00:07:04.890 --> 00:07:05.570
Chun-Long Chen (PNNL): Yeah.

34
00:07:05.790 --> 00:07:09.779
Chun-Long Chen (PNNL): So, Jeff, from your email seems that you still work on your subcontract.

35
00:07:10.840 --> 00:07:18.070
James Ang: Yeah, we don't have any of our subcontracts in place. So I'm I'm I'm requesting a deferral from next week, which.

36
00:07:18.070 --> 00:07:18.690
Chun-Long Chen (PNNL): Oh!

37
00:07:18.690 --> 00:07:19.999
James Ang: Hopefully. That'll be fine.

38
00:07:20.320 --> 00:07:23.700
Chun-Long Chen (PNNL): We had our sub all subcontract, placed like.

39
00:07:23.830 --> 00:07:25.839
Chun-Long Chen (PNNL): I think, at least one month ago.

40
00:07:26.700 --> 00:07:27.920
James Ang: Oh, that's good!

41
00:07:27.920 --> 00:07:28.560
Chun-Long Chen (PNNL): Yeah.

42
00:07:28.820 --> 00:07:31.479
Grant Johnson: University of Washington was actually the slowest.

43
00:07:32.970 --> 00:07:33.350
James Ang: Okay.

44
00:07:33.350 --> 00:07:34.150
Grant Johnson: Sweet enough.

45
00:07:35.660 --> 00:07:37.619
Ting Cao: There was a connection issue. Sorry.

46
00:07:58.780 --> 00:07:59.780
Grant Johnson: 19.

47
00:08:01.380 --> 00:08:02.220
Chun-Long Chen (PNNL): Thank you.

48
00:08:03.150 --> 00:08:04.280
Ting Cao: Hello! How are you?

49
00:08:06.060 --> 00:08:10.509
Chun-Long Chen (PNNL): So who is the organizer? Paul? Is Paul available today?

50
00:08:13.200 --> 00:08:17.969
Grant Johnson: I didn't see an email from him saying he wasn't. He usually did the introductions, if I remember.

51
00:08:17.970 --> 00:08:18.650
Chun-Long Chen (PNNL): Yeah.

52
00:08:27.850 --> 00:08:29.500
Chun-Long Chen (PNNL): wait 5 min. Maybe.

53
00:08:30.930 --> 00:08:35.359
Chun-Long Chen (PNNL): I mean, usually it will be great. It's like like,

54
00:08:36.740 --> 00:08:40.799
Chun-Long Chen (PNNL): like before the presentation. If everybody can see the title.

55
00:08:42.840 --> 00:08:44.699
Grant Johnson: I'm still sharing right. Shanlong.

56
00:08:44.700 --> 00:08:56.299
Chun-Long Chen (PNNL): Yeah, yeah, we can see your 1st line. What I mean is like, the email, right? Kinda like, Oh, we by weekly meeting. We can send the email. Let everybody know who will present, and then we'll

57
00:08:56.460 --> 00:08:58.240
Chun-Long Chen (PNNL): what the topic will be.

58
00:08:59.450 --> 00:09:03.889
Grant Johnson: Yeah, there is a schedule, but it's been revised a few times.

59
00:09:04.680 --> 00:09:07.869
Chun-Long Chen (PNNL): But I don't think that schedule it. Go it go

60
00:09:08.190 --> 00:09:11.700
Chun-Long Chen (PNNL): goes to everyone. At least, I have never seen that schedule.

61
00:09:22.430 --> 00:09:23.989
Chun-Long Chen (PNNL): Oh, no poise here.

62
00:09:27.060 --> 00:09:29.599
James Ang: Let's see, I see Paul is now on.

63
00:09:29.830 --> 00:09:31.619
Chun-Long Chen (PNNL): Yeah, no point is the host.

64
00:09:31.940 --> 00:09:34.349
Chun-Long Chen (PNNL): I think he's yeah. Yeah. Hey, Paul.

65
00:09:37.550 --> 00:09:38.589
Christian Mailhiot - Sandia National Laboratories: Hi Paul.

66
00:09:39.880 --> 00:09:44.590
jsnelso: Hey? Real quick. I think there's some other folks that are on another.

67
00:09:45.090 --> 00:09:48.009
Chun-Long Chen (PNNL): There were 2 link. I think someone need to go there.

68
00:09:48.010 --> 00:09:48.690
Christian Mailhiot - Sandia National Laboratories: This week.

69
00:09:48.690 --> 00:09:50.570
Chun-Long Chen (PNNL): And tell them to join this one.

70
00:09:50.860 --> 00:09:54.180
jsnelso: Yeah, I'll I'll tell them, and I'll be right back.

71
00:09:54.180 --> 00:09:55.300
Chun-Long Chen (PNNL): Okay. Thanks. Yeah.

72
00:09:55.300 --> 00:09:55.920
Grant Johnson: Yeah.

73
00:09:57.230 --> 00:09:59.670
James Ang: I think the reason there are 2 links is, one is

74
00:09:59.940 --> 00:10:06.390
James Ang: for just like the pis and deputies for the meerkat projects.

75
00:10:06.560 --> 00:10:11.789
James Ang: and the other is this larger one which goes to all of the team members.

76
00:10:12.240 --> 00:10:12.910
Chun-Long Chen (PNNL): Okay.

77
00:10:12.910 --> 00:10:18.100
Grant Johnson: So this was the same link from last week, which is why I used it hopefully. This is the right one.

78
00:10:18.550 --> 00:10:22.560
James Ang: I think it's the right one, Grant. It's it's the one with many more attendees.

79
00:10:22.560 --> 00:10:23.280
Grant Johnson: Okay.

80
00:10:23.670 --> 00:10:25.800
Chun-Long Chen (PNNL): Yeah, I think it just happened that one.

81
00:10:25.800 --> 00:10:26.480
James Ang: Good.

82
00:10:26.480 --> 00:10:28.440
Chun-Long Chen (PNNL): Overlap right for this week.

83
00:11:18.190 --> 00:11:18.865
jsnelso: Okay.

84
00:11:55.390 --> 00:11:56.260
Chun-Long Chen (PNNL): At 10.

85
00:11:56.410 --> 00:11:59.640
Chun-Long Chen (PNNL): Do you currently have a student on this project?

86
00:12:00.710 --> 00:12:03.959
Ting Cao: Yes, I'm in the process of locating one.

87
00:12:04.320 --> 00:12:09.469
Chun-Long Chen (PNNL): Okay, yeah. Later, send us the student information.

88
00:12:09.750 --> 00:12:28.299
Ting Cao: Right? So we still need to figure out how much effort the student need to invest because he's currently working also on another project. So I think, since since the grant itself doesn't cover a full student. So then talk to him to understand research, expectation, and the deliverable.

89
00:12:28.630 --> 00:12:33.759
Chun-Long Chen (PNNL): Okay, that's fine. Yeah. I mean, for our bi-weekly meeting, we need a like.

90
00:12:34.400 --> 00:12:38.040
Chun-Long Chen (PNNL): we are still organize this like email list. Yeah.

91
00:12:38.040 --> 00:12:39.830
Ting Cao: Yeah. Great. Thanks.

92
00:12:49.830 --> 00:12:54.449
Grant Johnson: All usually does some introductions or business first, st right before we start.

93
00:12:54.920 --> 00:13:00.420
Chun-Long Chen (PNNL): Yeah, we can wait. I think it probably went to other link.

94
00:13:49.411 --> 00:13:54.800
Angelo Dragone (SLAC): Not sure all is connected. I doesn't seem he was.

95
00:13:54.800 --> 00:13:56.730
Chun-Long Chen (PNNL): Here like a few minutes ago.

96
00:13:56.730 --> 00:13:58.860
Angelo Dragone (SLAC): Oh, okay, maybe to step on.

97
00:13:58.860 --> 00:14:04.389
Chun-Long Chen (PNNL): I think Christian, like Christian, went to check the other link. They will.

98
00:14:04.510 --> 00:14:06.270
Chun-Long Chen (PNNL): Now some people.

99
00:14:08.020 --> 00:14:08.630
Angelo Dragone (SLAC): Oh!

100
00:14:08.630 --> 00:14:10.170
Chun-Long Chen (PNNL): Oh, yeah, crazy. You're back.

101
00:14:10.390 --> 00:14:12.630
Christian Mailhiot - Sandia National Laboratories: Yeah, Jeff. Jeff Nelson went out to

102
00:14:14.850 --> 00:14:18.150
Christian Mailhiot - Sandia National Laboratories: tell the other attendees he should be right back.

103
00:14:18.380 --> 00:14:19.160
Chun-Long Chen (PNNL): Okay.

104
00:14:19.650 --> 00:14:21.580
Chun-Long Chen (PNNL): It's Paul over there.

105
00:14:24.190 --> 00:14:25.740
Angelo Dragone (SLAC): 3, 3 months.

106
00:14:38.900 --> 00:14:44.570
Angelo Dragone (SLAC): Oh, and that is bizarre. We have 2 week, 2 meetings.

107
00:14:48.600 --> 00:14:55.170
Grant Johnson: Yeah, this one's the same link from last week in the previous meetings. And then there's another. There's a new one for today. For some reason.

108
00:14:57.950 --> 00:14:58.640
Angelo Dragone (SLAC): Oh.

109
00:15:01.660 --> 00:15:02.470
Angelo Dragone (SLAC): that's true!

110
00:15:25.450 --> 00:15:27.850
Chun-Long Chen (PNNL): Maybe we should just start.

111
00:15:30.890 --> 00:15:33.169
Grant Johnson: I was gonna wait to see if Jeff came back.

112
00:15:33.750 --> 00:15:34.490
Paul McIntyre’s phone: Yeah.

113
00:15:34.490 --> 00:15:35.300
Angelo Dragone (SLAC): All 10 seconds.

114
00:15:39.000 --> 00:15:41.420
Christian Mailhiot - Sandia National Laboratories: Yeah, what? You go ahead and start.

115
00:15:43.090 --> 00:15:45.449
Grant Johnson: Okay, that's fine. Thank you.

116
00:15:46.285 --> 00:15:53.040
Grant Johnson: So yeah, good afternoon. Everyone. My name is Grant Johnson. I'm a scientist at Pacific, Northwest National Lab.

117
00:15:53.190 --> 00:15:57.789
Grant Johnson: We're still getting some noise there. So if you could mute yourself, please, that'd be great.

118
00:15:57.970 --> 00:16:10.500
Grant Johnson: So I'm pleased today to give you an overview of our microelectronics Science Research Center Project, which is titled Assembly of Molecular Memristors for energy, efficient computing.

119
00:16:10.630 --> 00:16:33.169
Grant Johnson: and the sort of theme of the overall project is captured by this graphics that we just recently had made up where we have molecular memristors in this case, which are actually molecular metal oxides or polyoxymethylates. And then we use biological scaffolds in this case, peptoids to arrange those into device level structures for testing. So

120
00:16:33.230 --> 00:16:55.759
Grant Johnson: I'll give kind of an overview of the project, and I'll talk about my thrust, which is focused on sort of local interactions. I'll then hand it off to Chun Long Chen, who's the project manager for this project here at Pnnl, and he'll tell you more about the Macromolecular assembly. And then Zhaogen Lang will tell you about the device testing that we're doing with these materials.

121
00:16:57.220 --> 00:16:57.900
Grant Johnson: Okay.

122
00:17:01.430 --> 00:17:26.600
Grant Johnson: so I think this audience knows pretty well the goals of neuromorphic computing and the desire to build biologically inspired systems that mimic the brain's parallel event driven and energy efficient information processing. So this is inherently different to sort of Von Neumann computing architectures where you're shuffling information back and forth between computers and memory.

123
00:17:26.950 --> 00:17:39.389
Grant Johnson: And we need materials that basically function as biological synapses. So there's a lot of work that's been done on neuromorphic chips and architectures like these memoristic crossbar arrays.

124
00:17:39.500 --> 00:17:52.150
Grant Johnson: But there are still some material challenges associated with these, and we think that our molecular approach to memristors might have some inherent advantages to address some of these issues. So

125
00:17:52.450 --> 00:18:22.410
Grant Johnson: Memristers play a key role in neuromorphic computing. This is due to their nonlinear dynamics, their multilevel or analog conductance, and the fact that they can be very low power use in certain implementations. They also have the advantages of being non-volatile fast, and again overcoming that von Beneumann bottleneck. So the traditional architecture usually looks something like this. You have your memristive material with a top and bottom electrode.

126
00:18:22.440 --> 00:18:35.189
Grant Johnson: And then the characteristics of this, the voltage current response looks something like this for a digital response. And like this for an analog response, which is more, what we're interested in for these neuromorphic computing applications

127
00:18:35.840 --> 00:18:47.140
Grant Johnson: and the market from enristers is predicted to grow dramatically over the coming years. As these things are increasingly used in it, telecoms and in other industries.

128
00:18:48.690 --> 00:19:14.469
Grant Johnson: So, as I mentioned. Despite, you know, all their advantages, there are also some challenges that need to be overcome. One of these clearly is the variability, the device, device and cycle to cycle variability in addition, degradation. And both of these basically relate to the stochastic switching mechanisms that impede reproducible memrist or performance.

129
00:19:14.610 --> 00:19:38.230
Grant Johnson: So on the right hand side. There are just the various mechanisms by which these materials can function. In our case we're concerned with metal oxide. So the 2 most important phenomena, of course, are oxygen vacancy, migration, and also conductive filament formation, and we'll explain a little bit about how our molecular memristors can be used to modulate both of those effects.

130
00:19:40.120 --> 00:19:46.789
Grant Johnson: So this is kind of a overview figure that summarizes the objectives of our project.

131
00:19:46.920 --> 00:19:54.740
Grant Johnson: We aim to self-assemble molecular memristors with long range order for resilient and energy efficient neuromorphic computing.

132
00:19:55.080 --> 00:20:17.089
Grant Johnson: So you can see there on the left hand side. Hopefully, you can see my laser pointer. We want to take these molecular metal oxides, which are sort of nanometer scale molecular metal oxides. These have multiple electronic charge states or redox states, each of which gives a discrete current voltage response.

133
00:20:17.660 --> 00:20:43.680
Grant Johnson: But we need a way to put those into a material with long range order that's well organized on a 2D electrode interface. So to do that we're leveraging peptoids which are biomolecular polymers which will allow us to assemble these polyoxomethylates into more ordered structures, creating individual synapses basically for a single polyoxomethylate and then into larger synaptic networks.

134
00:20:44.050 --> 00:21:03.709
Grant Johnson: So the key games here are controlling that local environment around this molecular metal oxide to give it the current resistance performance that we want, and then arranging it into longer range order on electrode materials that are compatible with conventional Cmos technologies.

135
00:21:03.850 --> 00:21:21.659
Grant Johnson: So the general workflow on the right looks something like this where we design and control these memoristic states. Locally, we tune this at a larger scale, and then we start going up to signal and memory functionality and then evaluating these devices for energy efficiency and neuromorphic computing applications.

136
00:21:23.800 --> 00:21:43.929
Grant Johnson: Our project addresses 3 scientific knowledge gaps through 3 integrated research thrusts. The 1st one that's led by myself is focused on the local interactions between these polyoxomethylates and sequence-defined peptoids. And what effect that has on their memoristic states and absorption behavior on 2D interfaces.

137
00:21:44.140 --> 00:22:02.709
Grant Johnson: The second knowledge gap is centered around the effects of peptoid side chain, chemistry, structure, and composition, and how that can be tuned to control the long range order of these assemblies, and also the formation of these Nano confined Ion transport channels through which these conductive filaments will form.

138
00:22:03.150 --> 00:22:16.499
Grant Johnson: and the 3rd thrust, led by Zhaogen Lang from University of Michigan, is focused on assembling these materials into devices and actually testing their energy efficiency for computing operations.

139
00:22:18.680 --> 00:22:41.130
Grant Johnson: So here are some images that capture the thrust that I described on the previous slide, we actually take a slightly different approach to many of the people in this center and that we're using solution phase synthesis to make these Polyoxomethylates which are shown on the left hand side here. So we can tune what sort of transition metals are in these, the size and charge states of these metal oxides.

140
00:22:41.130 --> 00:22:52.189
Grant Johnson: and then we also connect them through different covalent linkages to peptoids which serve as the host matrix to create an ordered array.

141
00:22:52.560 --> 00:23:08.759
Grant Johnson: So we can do deposition experiments where we create these in very clean environments on 2D interfaces, and we interrogate the current voltage response of individual clusters, using scanning tunneling microscopy and infrared spectroscopy.

142
00:23:09.180 --> 00:23:21.639
Grant Johnson: But then we start to assemble these with peptoids into larger scale arrays which we can then characterize, using larger scale techniques, such as conductive afm and simulation, such as molecular dynamics.

143
00:23:22.510 --> 00:23:30.570
Grant Johnson: The 3rd thrust then involves taking these materials and putting them into both lateral and lateral and vertical meristic configurations

144
00:23:30.750 --> 00:23:39.369
Grant Johnson: and analyzing their current voltage characteristics, the memoristic states and evaluating them for computing applications.

145
00:23:41.540 --> 00:23:48.440
Grant Johnson: So this is just a picture of the team. So this project is led out of Pacific Northwest National Laboratory.

146
00:23:48.490 --> 00:24:12.419
Grant Johnson: I serve as the lead for thrust. One Chan Long Chen leaves the peptoid task, and then the folks shown on the bottom here bring a whole range of expertise in material synthesis, controlled deposition, multimodal characterization, and modeling. And then we have some postdocs which are going to be joining us shortly at the end of the July and August to push forward this effort in the laboratory.

147
00:24:13.780 --> 00:24:35.970
Grant Johnson: We've partnered with several other institutes shown here to bring in some additional expertise in theoretical modeling machine learning and device fabrication testing. So we have ting Cao from University of Washington is helping us with our surface simulations. Zhaogen, who I mentioned, is leading our device assembly and testing thrust.

148
00:24:35.970 --> 00:24:53.609
Grant Johnson: and then go away. Who brings expertise in machine learning models to help us explore the broad parameter space of these materials and select those that are most promising for our applications. And then Jan at Berkeley is also helping us with the deposition and device testing.

149
00:24:55.480 --> 00:25:06.039
Grant Johnson: So our 1st thrust again that I lead is focused on the local interactions in the immediate region around these molecular metal oxides.

150
00:25:06.180 --> 00:25:34.440
Grant Johnson: So as you can see here, these clusters are on the order of a half nanometer. In size they have different heteroatoms at the center, things like phosphorus, silicon, and germanium, which can be used to tune the energy levels of the electronic orbitals. You can also exert a lot of control over those orbitals through the substitution of the transition metals things like vanadium, molybdenum, and tungsten. So we will be exploring that through selective substitution reactions.

151
00:25:34.780 --> 00:25:51.400
Grant Johnson: the next step, of course, is then covalently attaching these molecular metal oxides to the polymer, the biopolymer, the peptoid in this case. And we're exploring both Thiol click chemistry and condensation reactions as ways to achieve that.

152
00:25:51.950 --> 00:26:12.180
Grant Johnson: We have unique capabilities which I'll tell you a little bit about in a minute, where we can deposit these sort of complex molecules from the gas phase, using a technique known as ion soft landing. This allows us to create very well defined interfaces that are well suited to characterization with spatially resolved techniques like scanning tunneling microscopy and spectroscopy.

153
00:26:12.550 --> 00:26:26.519
Grant Johnson: So this allows us to get the sort of individual behavior current voltage response. The memorative states of these polyoxymethylates on surfaces and understand how that's tuned by the parameters that I just described.

154
00:26:28.800 --> 00:26:58.029
Grant Johnson: So here's some evidence of what I was just speaking about. So if you have one of these molecular metal oxides on a surface they can have up to 4 different conductive States or 4 logic states. This is shown here on the right hand side again, in these plots. So each one of those has the potential to be used to store a certain bit or to actually provide a certain synaptic weight in a neural network.

155
00:26:59.050 --> 00:27:19.080
Grant Johnson: In addition to controlling their electronic conductivity, these polyoxymethylates can also regulate cation, migration and conductive filament formation. So as you know, one of the big problems with memristors is this stochastic filament formation, which is often not reproducible, cycle to cycle or device to device

156
00:27:19.160 --> 00:27:32.109
Grant Johnson: and polyoxomethylates which have these localized unoccupied molecular orbitals can basically serve as transit corridors for these ions, making those processes much more reproducible.

157
00:27:34.230 --> 00:27:51.169
Grant Johnson: So one technique that we've developed here at our lab is this technique known as Ion soft landing, and the key advantage here is it allows us to take a mixture containing many complex things. So this is a mass spectrum of these molecular metal oxides.

158
00:27:51.670 --> 00:28:12.250
Grant Johnson: and if you were to drop, cast these or try to solution, cast them, you would get everything present in solution. But with this mass spectrometry approach, we can individually select each of these ions and isolate it on a surface and examine the effect of stoichiometry, heteroatom substitution and covalent functionalization.

159
00:28:12.490 --> 00:28:25.679
Grant Johnson: and then we have some in situ capabilities to actually examine the structure of these things on the interface, and how they change when we apply a voltage and change the redox or conductive state of these metal oxides.

160
00:28:25.800 --> 00:28:34.540
Grant Johnson: and we have some capabilities as well for uhv sample transfer, so these can be moved over to Stm. And Sts. And characterized without exposure to air.

161
00:28:36.230 --> 00:28:52.250
Grant Johnson: So one of the things we think is going to be quite important, for these molecular memoristers are the local interactions between the polyoxymethylate and the peptoid with the support, and then also with adjacent molecular metal oxides.

162
00:28:52.320 --> 00:29:13.359
Grant Johnson: And one way we can get information about. This is by studying the spectroscopy of these things as a function of their surface coverage. And this is an example here, from a paper from a couple years ago, where we basically imaged these materials on a surface as a function of surface coverage, and showed how the vibrational modes of these things evolve as a function of almost 40 wave numbers

163
00:29:13.490 --> 00:29:21.919
Grant Johnson: from an environment where they're interacting mostly with the electrode surface to an environment where they're interacting with each other in a higher coverage regime.

164
00:29:24.830 --> 00:29:54.140
Grant Johnson: As I mentioned, we are going to study the individual mineristic properties of these epoxy methylates and their assemblies on surfaces, and this is done using Pnnl's Stm. And Sts capabilities shown here. We have several instruments both for low temperature and variable temperature measurements, and this is also coupled with X-ray photoelectron spectroscopy and the ability to dose reactive gases and examine the effect of those environmental conditions.

165
00:29:56.680 --> 00:30:13.030
Grant Johnson: So the project just started this year. So we do have some exciting results to share. These are some results of these molecular metal oxides deposited onto graphite. These are conductive Afm images shown here at 2 different coverages.

166
00:30:13.030 --> 00:30:26.910
Grant Johnson: And then we also have current voltage and Didv curves here, revealing that, in fact, these materials do show multiple discrete conductance states which we believe will be useful for memrist or applications.

167
00:30:28.610 --> 00:30:45.120
Grant Johnson: So I think this is the last slide for my thrust. I just want to emphasize again that the Polyoxomethylates themselves are one memoristic element. But then we need a way to arrange those into macromolecular assemblies with long range order.

168
00:30:45.200 --> 00:31:04.669
Grant Johnson: and to do that we have to covalently attach them to the peptoid. So we're exploring various tethering schemes, using this Tris alkoxyl chemistry shown here, where we can form both amide bonds and also do click chemistry to create strong covalent bonds to the host scaffold.

169
00:31:05.520 --> 00:31:11.035
Grant Johnson: So again, I'm just gonna bring you back to our high level diagram so that

170
00:31:11.560 --> 00:31:23.439
Grant Johnson: summarizes the sort of work that I'm doing on the local interactions of these memristors. I'm going to hand it off now to my colleague, Chun Long Chen, who's going to tell you more about the macromolecular assembly and peptoid work.

171
00:31:25.650 --> 00:31:26.980
Chun-Long Chen (PNNL): Okay, thanks. Guys.

172
00:31:27.670 --> 00:31:33.124
Chun-Long Chen (PNNL): So as Greg mentioned also in from this overall

173
00:31:34.140 --> 00:31:42.349
Chun-Long Chen (PNNL): picture, right? So this project, what we are really doing is like, we want to take this bow, inspire concept.

174
00:31:42.570 --> 00:31:51.848
Chun-Long Chen (PNNL): as you know, like, from the top of the figure, right? So the the iron transport in the brain. So it's very important for this

175
00:31:52.350 --> 00:32:18.820
Chun-Long Chen (PNNL): energy, efficient neuromorphic computing. So what we try to do for stress 2, right is how we can also create this kind of long range ordering right to tune the Ion transport electron transport. And then we can use a lot of information learned from thrust one. For example, this tunerable memorized state right? How we can organize them into a

176
00:32:18.880 --> 00:32:34.689
Chun-Long Chen (PNNL): long range order. So which kind of show in this cartoon? Right? If we develop this framework structure. And then how we can use this different charge state. And then the ordering of the peptoid to regulate the

177
00:32:34.970 --> 00:32:39.660
Chun-Long Chen (PNNL): iron transport electron transport for building energy efficient device.

178
00:32:41.440 --> 00:32:44.309
Chun-Long Chen (PNNL): Grant, can you? Are you controlling the slide?

179
00:32:47.540 --> 00:32:52.220
Chun-Long Chen (PNNL): Oh, okay, yes. Thanks. So, yeah. Thanks. God.

180
00:32:52.771 --> 00:33:06.828
Chun-Long Chen (PNNL): Like, I mentioned right? This, the thrust. 2. What we really want to do is now say, okay, how we can use the self assembly or the peptide into crystalline material to organize this

181
00:33:07.840 --> 00:33:24.630
Chun-Long Chen (PNNL): memoriesta state material right into this long range ordering. And then to do that, we need to put this memoristic material onto a substrate right which could be the the memoristic electrode, or this semiconductive material.

182
00:33:24.630 --> 00:33:38.110
Chun-Long Chen (PNNL): So we originally put a 4 different way to make this kind of material. One is, for example, if we make this sofa assembled crystal material, we can do this surface agnostic coating.

183
00:33:38.340 --> 00:34:01.539
Chun-Long Chen (PNNL): By doing that we will have layer by layer. This ordered material onto the substrate surface to regulate the iron or electron transport. For example, if we have the solar electrolyzed, this solar ion diffusion through the filament. So then we can use Peptide and pump to control that stage.

184
00:34:01.750 --> 00:34:21.090
Chun-Long Chen (PNNL): And we also, we used to put a like a Pcvd. To do this Pcvd glows of the peptoid material crystal material on the surface. But then, due to the re scope of the work, we kind of like remove the Pcvd from our

185
00:34:21.409 --> 00:34:23.190
Chun-Long Chen (PNNL): task temporary.

186
00:34:23.840 --> 00:34:25.740
Chun-Long Chen (PNNL): So grant, next next slide.

187
00:34:28.020 --> 00:34:52.579
Chun-Long Chen (PNNL): Yeah. So again. So this is kind of repeat what I said. Right? Like, you know, the source 2, it's really say, okay for source one. Once we understand a lot of this electronic structure, and how the local interaction which could be the covalent attachment of the peptoid with the palm or peptoid sequence, right inference, the palm charge state.

188
00:34:52.650 --> 00:35:08.139
Chun-Long Chen (PNNL): Now the question is, how we can use those molecular level understanding. Now build into this crystalline material right to tune the electron and ion transport. So then we can deposit this material onto the

189
00:35:08.663 --> 00:35:21.510
Chun-Long Chen (PNNL): this memory. So like to do to develop this organic inorganic interface interface. Right? So to understand, for example, some of the key question we want to

190
00:35:22.160 --> 00:35:25.389
Chun-Long Chen (PNNL): understand down the road is like how the

191
00:35:25.570 --> 00:35:40.770
Chun-Long Chen (PNNL): peptoid sequence, or the interaction in inter with this inorganic interface, and how we can actually use the inorganic with specific lattice to inference. The ordering of this peptoid pond structure.

192
00:35:40.880 --> 00:35:51.899
Chun-Long Chen (PNNL): And then by doing that, we can, you know, get a lot of fundamental understanding how the ion or the electron flow during the memorized

193
00:35:52.130 --> 00:35:57.440
Chun-Long Chen (PNNL): device and the performance process. And next slide grant.

194
00:35:58.910 --> 00:36:02.849
Chun-Long Chen (PNNL): So this is basically one slice to show.

195
00:36:02.850 --> 00:36:03.540
Niri Govind: What kind of.

196
00:36:03.540 --> 00:36:19.720
Chun-Long Chen (PNNL): How what we had done before. We can actually create a crystalline material right? So the left figure showing, if we keep the same hydrophobic domain. So we can add a lot of different functional group

197
00:36:19.720 --> 00:36:35.989
Chun-Long Chen (PNNL): to do this basically kind of like cold crystallization process. So then, we are able to not only get this crystalline nanomaterial right? So in for this project, we want to develop like palm containing crystalline material, but because of this

198
00:36:36.550 --> 00:36:44.699
Chun-Long Chen (PNNL): high tune ability coming from the peptoid and also coming from this coal crystallization process. So we are able to

199
00:36:45.090 --> 00:36:46.290
Chun-Long Chen (PNNL): tune the

200
00:36:46.320 --> 00:37:11.529
Chun-Long Chen (PNNL): palm local environment which in this last one right? We Grant mentioned, like the local interaction, it's very important to tune in that memorous state. So there you can imagine for this thrust. Once we align those palm in an order, the way. And then we can also tune in the local environment, basically to achieve some kind of like a protein like.

201
00:37:11.560 --> 00:37:17.539
Chun-Long Chen (PNNL): you know, nature system, right? So the local environment is important for iron electron transport.

202
00:37:17.600 --> 00:37:24.559
Chun-Long Chen (PNNL): So the middle figure showing this is actually the Pepto Assembly did on the MOS. 2.

203
00:37:24.810 --> 00:37:54.470
Chun-Long Chen (PNNL): The reason we want to use this is because MOS 2 is also one of the very interesting memoristic material which can, you know, tune the band gap right for building the memoristic device here. What we want to do is we want to use this MOS, 2 kind of temporary, the peptoid assembly. Right? So then, we are able to develop this peptoid palm and the MOS 2 hybrid material. Use this as a

204
00:37:54.960 --> 00:37:58.929
Chun-Long Chen (PNNL): tunable memoristic material for device, and the

205
00:37:59.420 --> 00:38:22.100
Chun-Long Chen (PNNL): right figure is to show. Actually, if we understand the lattice of the inorganic material, we are able to use this inorganic lattice right? Which could be Hopg, like conducting material, or in this middle one, like MOS, 2 semiconductor material. So then, we are able to use this inorganic material

206
00:38:22.100 --> 00:38:34.530
Chun-Long Chen (PNNL): which could be used as memory electrolyte or the buffer material. Right? So now say, okay, use those inorganic lattice to temperate the ordering of this

207
00:38:34.540 --> 00:38:40.380
Chun-Long Chen (PNNL): peptoid palm material which obtained from source one next one.

208
00:38:42.300 --> 00:39:10.810
Chun-Long Chen (PNNL): Yeah. So this is the basically the result we had before to show through the self assembly process. It is, you know, totally doable we can align like we can align this cluster right? In this case we use a palm cluster to show if we have the hydrophobic domain for the tube formation which is on the left figure to show.

209
00:39:11.270 --> 00:39:39.850
Chun-Long Chen (PNNL): and then we can put the palm over there and then to align this into nanotube structure right which we have well aligned this Nano cluster. So in this project we basically use the same approach to align the different palm cluster. So the middle one to show if we change hydrophobic domain. And then we can get this nano sheet. And then the right image basically also show like, we can also put this

210
00:39:39.880 --> 00:39:56.960
Chun-Long Chen (PNNL): cluster in the middle to get this highly crystal material. So this is basically the slides to show through our previous work. We have demonstrated. Align the palm into order. The way to achieve the long range order is totally doable

211
00:39:57.220 --> 00:39:58.340
Chun-Long Chen (PNNL): next slide.

212
00:40:01.210 --> 00:40:27.289
Chun-Long Chen (PNNL): Yeah. So in this slide, like I mentioned, like, we actually originally put 4 different way to make this peptoid pump material onto the inorganic substrate. So the 1st one is doing this solution crystallization which basically we started with amorpho. Right, then let the peptide and the palm to crystallize and then use hydrophobic interaction. Control the crystallization

213
00:40:27.840 --> 00:40:33.950
Chun-Long Chen (PNNL): in the in the solution. So then we can get a lot of this freestanding crystalline a nano sheet.

214
00:40:33.980 --> 00:40:44.139
Chun-Long Chen (PNNL): The second approach is this temporary approach which we can use inorganic substrate as the temporary. The advantage for this second approach is

215
00:40:44.569 --> 00:41:04.750
Chun-Long Chen (PNNL): since later, going to present stress 3 to show they already built some device with MOS 2. And then you can imagine right? Once we have this inorganic memory, and then we can actually use this inorganic surface as the temporary. Further, to grow this

216
00:41:05.680 --> 00:41:21.669
Chun-Long Chen (PNNL): Peptide palm and then use a Peptide palm to further control the iron and the electron transport. And the 3rd one is about building a crystal framework structure. So here we kind of remove the the Pcvd.

217
00:41:21.790 --> 00:41:50.051
Chun-Long Chen (PNNL): And then we will use molecular dynamic simulation to guide us the material synthesis and also go away and don't change from Michigan State. They are using machine learning to kind of like screen the kind of existing like palm crystal and see how we can understand more what kind of interaction, what kind of ordering will be important to build the

218
00:41:50.700 --> 00:42:02.920
Chun-Long Chen (PNNL): the regulated ion and electron transport. And then we also have the in situ Afm, some other characterization to look at. Say, okay, once we build a memoristic device.

219
00:42:03.250 --> 00:42:16.725
Chun-Long Chen (PNNL): And then what kind of surface charge? And then, like, you know, surface potential. Or you know, the bank gap change of the inorganic material will be useful for us to get a certain

220
00:42:17.510 --> 00:42:33.629
Chun-Long Chen (PNNL): performance from the memorista, for example, Xiao Gampo will mention, like, you know, long-term memory, short-term memory, how we can understand more about the you know, material property related to those performance of the memorester

221
00:42:34.100 --> 00:42:35.810
Chun-Long Chen (PNNL): next slides, please.

222
00:42:36.770 --> 00:42:52.049
Chun-Long Chen (PNNL): Yeah. So this is one size to show. For example, right? Once we can build a peptid Nano sheet. So we have this cartoon, which we had a paper last year to show. Like once we have this highly crisp Nano sheets.

223
00:42:52.080 --> 00:43:15.699
Chun-Long Chen (PNNL): Now we want to build the functional right into this nano sheet. One way we can do is you can imagine those kind of ball like right if those are the palm. And now we can use a lot of different chemistry, for example, like those quick chemistry on the right to cover, attach the palm into this sheets forming peptoid. Next slide, please.

224
00:43:16.760 --> 00:43:25.460
Chun-Long Chen (PNNL): Yeah. So this is the result to show. Once we started with, I think here we have 3 different

225
00:43:25.970 --> 00:43:35.129
Chun-Long Chen (PNNL): palm cluster. Right? So that's because they have different chemistry we can use, for, you know, quick chemistry. Or for this.

226
00:43:35.240 --> 00:43:46.039
Chun-Long Chen (PNNL): like am a reaction. So then, basically, once we get this pump containing peptoid, and then we are able to use this for self assembly. And

227
00:43:48.890 --> 00:43:52.740
Chun-Long Chen (PNNL): just me, I think. Can you guys see my slides.

228
00:43:53.900 --> 00:43:55.039
Zbynek Novotny’s iPad: I can't see it.

229
00:43:55.040 --> 00:44:00.665
Chun-Long Chen (PNNL): Oh, okay, I think then maybe it just you know, you cannot see the slides.

230
00:44:01.250 --> 00:44:19.370
Chun-Long Chen (PNNL): Okay, so now we'll just continue. Yeah. So basically, this is to show like, okay, once we have successful covalent attachment right to this as sheets forming peptoid. So then we use Afm Sem to show, indeed, that we have very nice sheets here, and then next slide, please.

231
00:44:20.580 --> 00:44:22.599
Chun-Long Chen (PNNL): and then we also have a

232
00:44:22.740 --> 00:44:31.880
Chun-Long Chen (PNNL): the Tm image to show. Indeed, right? We can see individual like palm cluster around these Nano sheets

233
00:44:32.120 --> 00:44:33.349
Chun-Long Chen (PNNL): next slides.

234
00:44:35.440 --> 00:44:45.319
Chun-Long Chen (PNNL): Yeah, I think now, like I was transfer this to shall gong to tell us how you know we are doing for the device. 4. Star 3.

235
00:44:46.050 --> 00:45:11.290
Xiaogan Liang: Thank and grant. And now talk about the Thrus 3. And so in this direction we push on the device physics. And it's about the memories to device fabrication and the characterization for neuromorphic computing. So as grant and trend long and mentioned advantages using the peptoi and based material to build the memory stores. So here I want to further emphasize.

236
00:45:11.290 --> 00:45:16.570
Xiaogan Liang: and such a device, oriented effort is driven by this hypothesis that

237
00:45:16.570 --> 00:45:43.739
Xiaogan Liang: the long range order in the peptoid nanostructures. And also there's a wider range of tunability of such macromolecular assemblies, so can provide the new method for us to create. We call this a biorealistic synaptic devices. So, for example, and using the different peptide nanostructures with different the molecular structure, we could build the memory structures

238
00:45:43.740 --> 00:45:50.319
Xiaogan Liang: with lateral or vertical channel configurations, so they they can be programmed

239
00:45:50.350 --> 00:45:58.980
Xiaogan Liang: to have a short term or long-term memory behaviors, or the or the different dynamic and response characteristics.

240
00:45:58.980 --> 00:46:22.720
Xiaogan Liang: And so they can meet the different neuromorphic compute computing scenarios. Also, in this project, we hope to build the neuromorphic computing network system using such peptoid based memory store. So one of the important feature we want to try is in such a network the interaction among the different synaptic nodes.

241
00:46:22.820 --> 00:46:38.180
Xiaogan Liang: It's driven by the Ionic coupling, and instead of the traditional charge control processes. So this will be very different, traditional. This is Cmos based memorative integrate circuits.

242
00:46:38.230 --> 00:47:02.599
Xiaogan Liang: So, but to achieve such goal, we need to address the important knowledge gap. The key question is, what kind of peptide nanostructures or assemblies that can be really used to make a memory stores that can result in the low energy consumption for operation? Or how do the different peptoid structure

243
00:47:02.600 --> 00:47:20.930
Xiaogan Liang: influence the transport behaviors of the carriers and the ions, and the result in the low energy consumption. So to address such question, we have a bunch of the research activity proposed. I'll grant next next slides, please.

244
00:47:22.110 --> 00:47:41.239
Xiaogan Liang: So first, st we're going to assemble the materials developed in the Stras, one Thrus, 2 into the memory store device structures and explore how the different macromolecular assembly structure affect the memorative switching behaviors of such device.

245
00:47:41.240 --> 00:48:04.710
Xiaogan Liang: And it's and we're gonna using the voltage program and the pulse program method to study the characteristics of such device and evaluate the device consumption. And also we will. We will study the dynamic characteristics of the such devices in response to the the spiking signal. So there are a bunch of the

246
00:48:04.710 --> 00:48:33.640
Xiaogan Liang: benchmark testing we're going to perform like a paired pulse of facilitation, long-term potential Asian depression. And more specifically, in this project, we're going to use a peptoid materials to build, for example, the non-volatile memory stores so that memory stores with a long-term memory capability. So we're going to test its performance at a synaptic components in the hardware based

247
00:48:33.650 --> 00:48:50.889
Xiaogan Liang: spiking neural network to see how good this kind of device can handle the spatial temporal data. And also we're gonna make the memory stores with a shorter memory or or the memory stores with very dynamic

248
00:48:50.920 --> 00:49:17.310
Xiaogan Liang: and response characteristics in response to like a spiking signal and some house based the real time signal, and it will give will be tested for processing the the spatial temporal information carried by the real time. Analog signal. So this is important for the edge computing and robotic control application grand, next place next.

249
00:49:17.430 --> 00:49:18.780
Xiaogan Liang: But yeah, thank you.

250
00:49:18.910 --> 00:49:37.990
Xiaogan Liang: And so we're going to use a method like nanofabrication nanodithography to make a nano scale and the memory stores involving the different peptoid nanostructure. And we're going to perform the electronic Ionic transport characterizations in combination with the materials, characterization.

251
00:49:37.990 --> 00:49:40.210
Dong Chen: It is so small.

252
00:49:41.280 --> 00:50:05.189
Xiaogan Liang: We will also build the test band the neural network systems and incorporate the Peptoi memory stores and either testing at the synaptic components or testing at the dynamic memory store to study their response to the real-time signal or spectrum based signal next page, please.

253
00:50:07.810 --> 00:50:29.070
Xiaogan Liang: So in my lab at the University of Michigan, and recently my Phd student established. These are called memory stir, based reservoir computing networks. And right now we are using the 2D layered materials. And in this project we're going to incorporate the Peptide based memory

254
00:50:29.070 --> 00:50:42.319
Xiaogan Liang: and test the performance in the in the neuromorphic computing scenarios. So here the relevant computing is a mathematical framework for building the recurrent neural network

255
00:50:42.330 --> 00:51:05.310
Xiaogan Liang: and and used for parallel computing and control applications. So they take the dynamic signal and without the any the digital electronic process or involvement involvement of the software. And such in such a network can directly generate the the anticipated control signals or some other output

256
00:51:05.350 --> 00:51:09.160
Xiaogan Liang: for the parallel computing applications.

257
00:51:09.180 --> 00:51:30.169
Xiaogan Liang: So traditionally, this kind of the reservoir computing is implemented in the form of the software or the firmware. And but here we, using the memory store to build all hardware based reservoir computing network. So compared with the software

258
00:51:30.170 --> 00:51:44.570
Xiaogan Liang: based rather computing network, such hardware devices can enable much lower energy consumption and a much smaller footprint. So it's a very good for the the system integration.

259
00:51:44.610 --> 00:51:48.630
Xiaogan Liang: And next next slides, please, yeah.

260
00:51:49.110 --> 00:52:04.349
Xiaogan Liang: And Grant, can can you play the left video? So here, I'd like to show you using our current rather computing. And we can demonstrate several different robotic control processes.

261
00:52:04.550 --> 00:52:24.510
Xiaogan Liang: For example, if you have the left video here. And yeah, if you click the, I think, yeah, here. So this is a target tracking and navigation of a robotic rover. So during this process, you see, the position signal of the target is directly send it to the right of our computing network

262
00:52:24.570 --> 00:52:38.729
Xiaogan Liang: and almost no involvement of the software or digital electronics. So the reservoir computing layer directly generate the control signal to control the movement of the Rover

263
00:52:39.000 --> 00:52:45.360
Xiaogan Liang: and and grant. But please play the yeah. Another one is, my students want to use such

264
00:52:45.470 --> 00:52:55.994
Xiaogan Liang: rather computing layer to control a drone. But it's too challenging. So right now, they decided, just control one arm of the drone. So here you can see

265
00:52:57.240 --> 00:53:15.779
Xiaogan Liang: Here's like a balancing bar using a propeller to control the the position. So during this control process, the position signal, the real tight signal is direct, physically applied to the right of our computing process. The layer and the red of our computing layer, and you

266
00:53:16.020 --> 00:53:22.849
Xiaogan Liang: completely using the physical process to generate the control signal to keep the system in in the balance.

267
00:53:23.020 --> 00:53:27.799
Xiaogan Liang: Grant, can you go back to the previous slides? I want to emphasize one thing.

268
00:53:31.370 --> 00:53:55.000
Xiaogan Liang: Yes, thank you. So so in this project we we the reason we choose. A rather computing network is, if you look at this schematic wheel in, there's a blue color part. There's a rather layer. So it's a very dynamic process to take the input signal. And so here we need a short term memory stores with nonlinear switching behaviors

269
00:53:55.030 --> 00:54:15.629
Xiaogan Liang: and the readout layer the orange part to generate the final output signal. So we need a long-term memory stress with linear switching behavior. So it's a very good scenario. We can test the Peptide based memory stores, because it's this kind of versatile chemistry of the Peptide materials

270
00:54:15.630 --> 00:54:26.490
Xiaogan Liang: give us the opportunity. We can build the short term and the long term memory stores to meet the different and the neuromorphic computing the scenarios

271
00:54:27.070 --> 00:54:29.569
Xiaogan Liang: I saw. There's question from Ryan.

272
00:54:30.550 --> 00:54:39.785
Ryan Coffee’s iPhone (2): Yes, thank you. So my experience before with reservoir compute is for control system and tokamax. And the key was that they used like

273
00:54:40.390 --> 00:54:51.000
Ryan Coffee’s iPhone (2): training like a lot of training to try and get the weights right and the connections and training ended up being like a genetic algorithm for the random connectivity.

274
00:54:51.410 --> 00:54:54.230
Ryan Coffee’s iPhone (2): But is this continuously training itself.

275
00:54:55.274 --> 00:55:06.129
Xiaogan Liang: For our rather computing, we don't need to train the rather layer, so it's a completely rely on the a little bit of randomness of the the devices and.

276
00:55:06.130 --> 00:55:09.820
Ryan Coffee’s iPhone (2): Golf resistances start out random, and then they Brendan

277
00:55:09.820 --> 00:55:11.329
Ryan Coffee’s iPhone (2): in over time. Is that right?

278
00:55:11.670 --> 00:55:25.430
Xiaogan Liang: Yeah, they just needed a nonlinear response behavior that can can map the kinetic information carried by the input signal to a high dimensional data space. That's good. Yeah. So if

279
00:55:25.430 --> 00:55:41.999
Xiaogan Liang: the the reverse state vector is a diversified enough, so the readout layer can map to what we want for the anticipated control signal. So we need to train the readout layer so that a long-term memory stirs with the linear switching behaviors.

280
00:55:42.511 --> 00:55:47.809
Ryan Coffee’s iPhone (2): Cool. So the the did. You say that the the reservoir is the short term Memristers.

281
00:55:48.640 --> 00:55:52.859
Xiaogan Liang: Okay, good. Cool. No, that's brilliant. It's an autonomous system.

282
00:55:53.080 --> 00:55:59.829
Xiaogan Liang: Yeah, thank you. So we need this a very fast update rate. So we can use for the real time application.

283
00:55:59.830 --> 00:56:03.270
Ryan Coffee’s iPhone (2): Yes, yeah. Okay. I got it. That's wonderful. Thank you.

284
00:56:03.430 --> 00:56:07.359
Xiaogan Liang: Thank you. Yeah. And grant the next slide, please.

285
00:56:08.980 --> 00:56:10.240
Xiaogan Liang: Yeah. Next one.

286
00:56:13.930 --> 00:56:31.849
Xiaogan Liang: Yeah. Right now. You know, we work closely with Pnl, we already got the the Peptide nano nano structure samples from Pnl collaborators like Peptide Nano sheets Nano Rod nanotube. So we

287
00:56:32.090 --> 00:57:00.109
Xiaogan Liang: right now we have been making the memory stores using such material. So here I show you the just example of the memory stores using the multi nanotube channel. And also here, you can see there's a memory store. We we try to use a single nanotube. But right now our fabrication process is quite a by chance. It's quite random. But it's good enough for individual device fabrication right now. But later, we we're going to propose

288
00:57:00.110 --> 00:57:06.160
Xiaogan Liang: the the research to develop methods. We can generate array of such memory stores.

289
00:57:06.180 --> 00:57:07.889
Xiaogan Liang: Our next slides, please.

290
00:57:09.870 --> 00:57:30.449
Xiaogan Liang: and my students already performed some preliminary characterization, and especially for the nanotube memory stores, and we already see this hysteresis Iv. Characteristics that indicates there is the memorative switching behaviors between the high resistance and low resistance states.

291
00:57:30.450 --> 00:57:49.019
Xiaogan Liang: So here, we said, there's a set compliant current. Limit. So you can see there's abrupt change. We haven't optimized this Iv the voltages, this programming process. We hope we can further make a make it more reliable switching cycles.

292
00:57:49.020 --> 00:58:14.169
Xiaogan Liang: and we also perform this pulse. Programmed characterization. Try to use the the voltage pulses to tune the conductance States. We see the change of the conductance States. But we haven't got very neat this set reset courses. So we're still working on that hopefully. Next time I can show you a much better this. Go up, go down this this kind of pulse program courses

293
00:58:14.320 --> 00:58:16.470
Xiaogan Liang: next slides, please.

294
00:58:17.730 --> 00:58:27.900
Xiaogan Liang: Okay, there's also result from the Lawrence Berkey National Lab. I don't know if the doctor or Dr. John is here like to pre.

295
00:58:27.900 --> 00:58:33.969
Chun-Long Chen (PNNL): No, I think both of them are not here. I mean, I I can also briefly mention this one. It's fine. Yeah.

296
00:58:34.160 --> 00:58:35.399
Xiaogan Liang: Thank you.

297
00:58:35.400 --> 00:58:50.799
Chun-Long Chen (PNNL): Basically, they are making this vertical device. So the result for this slides to show, even though, like on the left right, it's a silk, I mean for this project. We don't really use silk, but this is a proof of concept to show.

298
00:58:50.800 --> 00:59:15.940
Chun-Long Chen (PNNL): because silk have some ordering, and then, when they compared to this amorphous Peptide material, even though both have memory right, the ordering actually can make the performance much better, which is as we expected, because now we are going to use this capability to start to measure more, this self-assembled crystalline peptide material, and including those

299
00:59:15.980 --> 00:59:22.140
Chun-Long Chen (PNNL): peptor, like the palm containing Peptor sheets, I showed in the source, too. Yeah.

300
00:59:22.970 --> 00:59:25.140
Chun-Long Chen (PNNL): I think next the next slide.

301
00:59:26.740 --> 00:59:56.420
Xiaogan Liang: So in the near future. And we're going to do several things. First, st we further improve the our programming and setting the processes to figure out the the device physics for operating the peptide based memory stores, and we will do more the material characterization. And especially we want to visualize the kinetic, the behaviors of ions, and or the formation of the Ionic filaments in the Peptide nanotube.

302
00:59:56.450 --> 01:00:18.969
Xiaogan Liang: And also we want to. I've just mentioned. We want to develop the upscalable approaches that can produce the the larger arrays of a Peptide based memory store, not just random devices on the substrate, and also, like the trend loan and Grant mentioned, and we're going to study the the interface between the peptwise nanostructure and the

303
01:00:18.970 --> 01:00:29.630
Xiaogan Liang: the inorganic two-dimensional layer materials. There could be a new opportunity to generate the memory stores with high uniformity. Next slides, please.

304
01:00:30.870 --> 01:00:55.250
Xiaogan Liang: So this is the approach I mentioned. We propose this Nano and Macro printing process. We hope to produce the the array and of pipe toy based the memory store. So we start with Pdms. Template the protrusive features, define the device array, then, using the shear direction printing approach, we want to selectively deposit

305
01:00:55.250 --> 01:01:12.750
Xiaogan Liang: type 2 nanotube on the protrusive surfaces on the template. So after printing, we can stand out the arrays of such nanotube structure. Finally, after metalization, we can have the larger array for such devices.

306
01:01:13.480 --> 01:01:15.619
Xiaogan Liang: Grant the next page, please.

307
01:01:16.460 --> 01:01:46.029
Xiaogan Liang: and also we will mention that in addition to the memory structures based on the pure peptoy nanostructure, we also want to study the hybrid structure right integrate, and 2D layered materials and the Pipetoi assemblies. So here's 1 example. Right now, we already start to work on that. So recently my students developed this this vertically arranged Bismu cyanide.

308
01:01:46.040 --> 01:01:50.189
Xiaogan Liang: And also we have a modern disulfide based memory stores.

309
01:01:50.370 --> 01:01:55.989
Xiaogan Liang: And such memory store has quite a unique switching behavior. Grant the next page, please.

310
01:01:58.940 --> 01:02:06.889
Xiaogan Liang: So here you can see this is a set reset behavior of our business, this vertically arranged device.

311
01:02:06.990 --> 01:02:30.810
Xiaogan Liang: and using the voltage pulses, we can realize analog conductance tuning. But more important. You can see here, once we stop the application of the voltage pulses, stop the sighting the device going to hold the conductance days for really long time. So that indicates really stable, non-volatile retention of the conductance days.

312
01:02:30.980 --> 01:02:52.600
Xiaogan Liang: And also you. From this result you can see there's almost no post setting relaxation, because many reported analog memory stores after setting, there is a significant relaxation of the the conductance, and so sometimes people need to integrate the such memory structure with a current regulator.

313
01:02:52.600 --> 01:03:01.730
Xiaogan Liang: And, for example, a current limiting transistor to precisely set the conductance days. We want to set the synaptic

314
01:03:01.730 --> 01:03:13.759
Xiaogan Liang: notes value. So for such device, we have a have a potential. We could build the memory, stir network without such current regulator.

315
01:03:13.820 --> 01:03:26.149
Xiaogan Liang: And but right now the challenge is, and such devices still exhibits significant device to what device the the consistency issue. And next page, please.

316
01:03:27.360 --> 01:03:34.929
Xiaogan Liang: So right now, we're doing this. So in up in my lab we deposit the button electrodes

317
01:03:35.020 --> 01:03:52.910
Xiaogan Liang: and deposit business dynamite. We call it half device sample, and we sent such a sample to Pnl, and at the Pnl the the collaborator will deposit a peptwine nanostructures nanotube, nano rod.

318
01:03:52.910 --> 01:04:10.019
Xiaogan Liang: also different palm structures like Grant mentioned, soft landing, the iron soft landing and and Chenlo mentioned the different morphology, and we want to incorporate the long range order in such a structure. So we expect.

319
01:04:10.030 --> 01:04:39.850
Xiaogan Liang: and after fabrication. So when a sample. Go back to my lab. We're going to finish the device structure. So and such device. We expect the incorporation of the pipeline nanostructure can regulate the formation side of the filaments. So we expect to see improve the device to device uniformity consistency. So that's good for our future and the system into the integration. Yeah, so I will stop here.

320
01:04:39.850 --> 01:04:46.839
Chun-Long Chen (PNNL): And how about? Let's just end up here. Let's see if others have question. I know, like we kind of started late.

321
01:04:52.590 --> 01:04:55.220
Chun-Long Chen (PNNL): like any any question for us.

322
01:05:02.500 --> 01:05:03.830
MCINTYRE PAUL: I had a question. Can you hear me.

323
01:05:03.830 --> 01:05:04.399
Chun-Long Chen (PNNL): Hang on!

324
01:05:04.530 --> 01:05:05.370
Chun-Long Chen (PNNL): Yes.

325
01:05:06.580 --> 01:05:19.634
MCINTYRE PAUL: Hi, this is Paul. Sorry I had some audio trouble. So I'm using my phone. Hope you can hear the question. Basically I I was wondering, to what extent are you planning to?

326
01:05:20.540 --> 01:05:42.569
MCINTYRE PAUL: look at the heterogeneity of properties of these structures? I'm reminded for some of the later slides on the electrical characterization and trying to develop more ordered arrays of of wire like

327
01:05:43.030 --> 01:06:01.819
MCINTYRE PAUL: on structures. I'm reminded of carbon nanotube research where there's a pretty lengthy history of characterizing the variability from, you know, properties from one wire to another depending on its structure, or

328
01:06:01.950 --> 01:06:13.329
MCINTYRE PAUL: the way that it's bundled with other wires is. There is that kind of activity planned to sort of sort through these structures, and see how variable their properties are.

329
01:06:16.190 --> 01:06:16.680
Xiaogan Liang: You're

330
01:06:16.680 --> 01:06:42.749
Xiaogan Liang: once we get this array, we want to evaluate the the variation from the device to device. But my understanding, there's a peptoid like the nanotube maybe it's different from the I know the previous. There's a research about assemble the like the carbon nanotube. But but I expect. Maybe the train. Logan? Answer question, how to control the maybe the uniformity of the.

331
01:06:42.750 --> 01:06:59.390
Chun-Long Chen (PNNL): Yeah, the nice thing for this system is like, when we started this peptide nanotube or the nano sheet. We really started with this molecular self assembly. Right? We can precisely control the structure, the crystalline material.

332
01:06:59.580 --> 01:07:05.730
Chun-Long Chen (PNNL): And then we can keep actually the same methodology right? But then tune in the chemistry.

333
01:07:05.920 --> 01:07:10.790
Chun-Long Chen (PNNL): So now you can imagine right if we build a device with with exactly same

334
01:07:10.900 --> 01:07:13.500
Chun-Long Chen (PNNL): kind of tube, but different chemistry.

335
01:07:13.750 --> 01:07:18.472
Chun-Long Chen (PNNL): I believe you know, for from that aspect. Right? We can kind of like

336
01:07:19.090 --> 01:07:25.450
Chun-Long Chen (PNNL): adjust the challenge coming from the carbon nanotube right? Because in the carbon nanotube synthesis there might be some edge

337
01:07:25.570 --> 01:07:51.709
Chun-Long Chen (PNNL): generity over there. But here, like, since we are doing molecular self assembly approach, we know our crystal material system pretty well, and then, once we know the performance from the device. We can actually narrow down to specific tube right? And then just keep the same methodology and change the chemistry. See how the chemistry, as we kind of proposed, you know, to tune in the

338
01:07:51.780 --> 01:07:55.890
Chun-Long Chen (PNNL): like the regulation of the Ion transport. Yeah.

339
01:07:57.920 --> 01:08:00.500
Chun-Long Chen (PNNL): And, Paul, did we answer your question?

340
01:08:01.300 --> 01:08:02.350
MCINTYRE PAUL: Yeah, thank, you.

341
01:08:02.520 --> 01:08:03.900
Chun-Long Chen (PNNL): Okay, thanks. Yeah.

342
01:08:04.870 --> 01:08:12.139
Chun-Long Chen (PNNL): I know some of you might need to go right. But if you like to ask other question, you know. Happy to.

343
01:08:13.220 --> 01:08:14.380
Grant Johnson: Honestly don't know.

344
01:08:14.380 --> 01:08:19.869
Grant Johnson: I have to drop off at 4, but you're you and Jogan are welcome to stay and and take further questions.

345
01:08:19.870 --> 01:08:25.969
Chun-Long Chen (PNNL): Okay, and see if there there are any more question. And now we kind of start late. Yeah, so.

346
01:08:27.510 --> 01:08:29.800
Ryan Coffee’s iPhone (2): I have to jump. But thank you very much.

347
01:08:30.010 --> 01:08:31.920
Xiaogan Liang: Okay, thank, you, yeah.

348
01:08:32.380 --> 01:08:35.039
Chun-Long Chen (PNNL): I think if no more question, then we'll just end.

349
01:08:36.720 --> 01:08:37.660
Xiaogan Liang: Yeah.

350
01:08:38.740 --> 01:08:40.519
Chun-Long Chen (PNNL): Okay, thanks. Everyone.

351
01:08:42.939 --> 01:08:45.129
Chun-Long Chen (PNNL): See you next time. Yeah. Bye.

352
01:08:45.130 --> 01:08:45.760
Xiaogan Liang: A good night.