1 Bikas Das, Nikola Pekas, Rajesh Pillai, Bryan Szeto, Richard McCreery University of Alberta National Institute for Nanotechnology Edmonton, Alberta, Canada Yiliang Wu Xerox Research Centre of Canada Redox-gated Molecular Memory Devices based on Dynamic Doping of Polythiophene

2 Today’s solid state memory (>$100 billion annually): Dynamic Random Access Memory (DRAM) Static RAM Flash (cameras, USB stick) write/erase cell retention cycle life speed size power on 1 µsec- 10 msec 1T > 10 yrs No “universal” memory- we need different types for different needs T = transistor, C = capacitor *

3 Cell Phones IEEE Nanotech Magazine, December 2009 enabled mobile electronics (cell phones, MP3 players, etc) ~$65 billion/year (not counting disk drives) 16 billion units (i.e. chips) sold in 2009 Nonvolatile solid state Memory Devices: Scifinder hits: “nonvolatile memory”: 31,800 “resistance memory”: 9,900 “conductance switching”: 10,560 “Alternative” nonvolatile memory in ITRS roadmap: Phase Change RAM Magnetic RAM Polymer memory Fuse/antifuse Molecular memory Nanomechanical

4 ethyl viologen ClO 4 + polyethylene oxide read write/erase SiO 2 S D G PQT EV polythiophene PQT from Xerox Canada S S S S C 12 H 25 H 25 C 12 n S S S S C 12 H 25 H 25 C 12 n more to scale: 1000 nm 25 nm D S initial (neutral) σ ≈ S/cm eˉ PQT + “polaron” σ ≈ 10 S/cm + - “Redox gated” molecular memory:

V SG (V) PEO Start EV(ClO 4 ) 2 +PEO I SG, µA - G S D PQT + V SG PEO or PEO/EV NHE PQT + /PQT EV +2 /EV E o’, V V SG S G V SG drives a redox reaction: PQT + EV +2 → PQT + + EV + “write” “erase”

I S-D (nA) V S-D (V) Initial read write/erase SiO 2 S D G =-2V After V SG + - I SD (0.5 V) -3 “erase” pulse … “ON” state “OFF” state +3 “write” pulse … 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E time (s) ON/OFF > 10 4 PQT EV = +2V After V SG

7 + Write - Erase 100 W/R/E/R cycles ACS Appl. Mater. Interfaces 2013, 5, Note: resistance readout low energy W/E nondestructive “read” should scale well potentially longer cycle life than “flash”

8 σ ~ 10 S/cm σ ~ S/cm PQT (neutral) Raman shift, cm -1 (vs. 780 nm) 1460 cm -1 PQT + (polaron) 1405 cm -1 Raman permits monitoring of PQT + formation in working memory device Raman spectroelectrochemistry of PQT:

σ ~ 10 S/cm σ ~ S/cm PQT PQT + Drain, initial Raman shift (cm ) Raman intensity (a.u.) Source, V SG = +2 * * Source, V SG = -2 Drain, V SG = +2 Drain, V SG = cm cm cm cm -1 S D - PQT + EV(ClO 4 ) 2 +PEO G Substrate V S-G Raman laser V SG pulses are “switching” polymer between high and low conductance states.. top view Source, initial 1460 Solid state spectroelectrochemistry:

10 V S-G S D G D S + - (a) Initial neutral Drain Source (under) Gate V SG = +2V polaron Drain Source (under) Gate G a t e V SD =-2V neutral Raman shift (cm -1 ) Drain Source (under) Gate polaron generation mediates conductance and “memory” polaron is produced in entire source/channel/drain region J. Am. Chem Soc. 2012, 134, 14869

11 S D V SG - + N P+P+ P+P+ P+P+ A¯ N N P+P+ N N N N N EV + A‾ EV +2 S D V SG 25 nm 1000 nm - + P+P+ P+P+ P+P+ P+P+ EV + A‾ A¯ N e¯ Polythiophene is not only a redox polymer but also a conducting polymer, so electron transport “laterally” is quite efficient. e¯ A¯ N N N N N N N EVA 2 S D V SG - + P+P+ P+P+ P+P+ A¯ P+P+ P+P+ P+P+ P+P+ P+P+ P+P+ EV + A‾ A‾ = ClO 4 ‾

12 Read W/E Now, monitor S-D current during +3 V S-G “write” pulse 0 V -3 V +0.5 V ~1 µm PEO-viologen PQT S D 30 nm A closer look at dynamics, using a circuit equivalent to a bi-potentiostat:

13 I SD (mA) I SD (right axis) I SG (µA) Time (s) (b) I SG (left axis) G S D Read W/E 0 V -3 V +0.5 V RC “charging” current generating conducting polaron SD conduction (“readout”) note 1000x “gain” P+P+ P+P+ P+P+ P + generation P+P+ P + propagation charging (RC) Which one(s) control speed?

14 Time (s) I SG µ(A) (a) I SG (µA) An interesting effect of atmosphere: vacuum (right scale) ACN vapor (left scale) air (right scale) I SG, “write” RC charging? ionic mobility? propagation speed? RC (msec) Ambient0.18 Vacuum0.037 ACN vapor0.081 all < 1 msec, can’t be the problem G S D R W/E 0 V -3 V +0.5 V

15 G Read W/E 0 V -3 V +0.5 V PEO-viologen S D Propagation of polarons into channel: Propagation is essential, could it be the slow step?

16 “Propagation” experiment: S D G i SG i SD i SG “write” i SD “read”, 10 µm gap propagation delay i SD “read”, 1 µm gap i SD “read”, 10 µm gap +3 V

17 Canadian ECS- 23

18 Canadian ECS- 24

19 Canadian ECS- 25

20 Canadian ECS- 26

21 Canadian ECS- 27

22 Canadian ECS- 28

23 Canadian ECS- 29

24 Canadian ECS- 30

25 PEO + ClO 4 ˉ close-up of S electrode: - +3 V + + “W” pulse generates polarons anions move toward polarons, compensate + charge polarons “fill” the Source region polarons propagate throughout PQT layer, anions continue to migrate from PEO the bad news: iR losses in PEO layer reduce “write” current, thus requiring ~ 500 msec to “fill” source region. the good news: there are solid electrolytes with conductivity >1000x higher than PEO-ClO 4, which should reduce “write” time to μsec. We can also make the electrolyte at least 10x thinner

26 Some progress: G S D drop cast PEO-EV 1-2 μm G S D spin coated PEO-EV ~0.3 μm PEO-EV Drop Cast PEO-EV Spin Coated “write” pulse PEO-EV Drop Cast PEO-EV Spin Coated “write” pulse

t, msec 26 o C 57 o C 47 o C I SG, “write” ~100 X higher “write” current than drop cast, ambient acetronitrile vapour: Note: effective “write” speed of ~2 msec instead of ~500 msec t, msec 26 o C 57 o C 47 o C I SD, “read”

28 XRCC polymer CMOS with support electronics Thin film transistor substrate Target: high “value added” to CMOS by integrating molecular nonvolatile memory

29 Low density prototype tester: