Draft proposal for concepts for opto memristor and proposed microscope design for testing the said optical memristor materials

Proposal:

Microscope Design for Ultrafast Pump and Probe Polarimetric

Characterization of Proposed Novel Materials and Structures Intended

For Use As Optical Memristors Working In Nonthermal Opto-Magnetic

Regime

Faisal Halim, City College of New York (CUNY)

Posted: Monday, 31st May, 2010

Course Professor: Gilchrist, Chemical Engineering, City College of New York (CUNY)

NOTE: This is a draft – I have not yet completed the microscopy aspect of the project,

and I may have to make a few refinements to the optomemristor concept. The

optomemristor concept is my own, and relies on prior work from [5,6,7].

As I have no way to test my ideas they remain what I would describe as “crazy ideas,”

but if anyone find my work useful, then I would appreciate being cited. Thanks.

Abstract: In order to realize an optical memristor the Inverse-Faraday Effect has

been investigated (through literature search) as a means for dynamically altering the

magnetization vector of materials, thus altering their Magneto-Optic Kerr Rotation

response and their Faraday Effect response faster than any other measurable change in

response that can be engineered and measured, and microscopy methods have been

investigated (through literature search) that can be used to characterize these responses

from microscopic samples (since practical devices made from these materials will need to

have microscopic components) at the speeds at which the optical memristor devices are

intended to operate.

Aim: The purpose of this project was to devise methods to make an optical memristor,
which will be an optical equivalent of HP’s electrical memristor [1], originally proposed

by Leon Chua [2], and to devise methods to characterize the materials and devices, once

they have been fabricated, in the microscopic size and ultrafast time regimes in which

they are intended to be used.

Introduction: Light propagates faster than electrons and holes and so the optical
responses of materials occur at much shorter time scales than they occur for electrical

responses. As a result there has been a push in the industry, for many years, to make

optical equivalents of electrical devices. Out of the four fundamental passive electrical

components possible [2] the memristor is one that brings neural networking closer to

reality, thus enabling a whole host of applications, like simulating the complexity of a

small animal’s brain, as was done by Jo, et al. [3]. At this junction lies the optical

memristor: a (possible) enabling technology that will also have a speed advantage from

being all-optical.

For a device to be all optical, and thus have all the advantages of using light in place of

electrons or holes, the signals that it receives have to be optical, the signals that it sends

have to be optical, and the processing (which may be enhancement, attenuation, or

change of polarization of the incoming light, or the addition or subtraction of two or more

signals, etc.) has to occur due to the interaction of the electric or magnetic field of the

light wave(s) with the material that the device is made of. The result of the interaction

may be steered (or influenced) by the presence of a constant field (electric or magnetic)

that does not need to be manipulated by non-optical means (which would negate the

speed advantages of a fully optical device).

To realize a practical all-optical memristor, or optical memristor, one would need a

device such that:

• Exhibits the following behavior:

o It should be possible to modify one aspect of its optical response quickly and

dynamically (i.e., the same material/device will have one of its optical

properties changed as and when needed, under optical excitations, and at time

scales that will not create speed disadvantages). To make an analogy, think of

a memristor as a variable resistor where one is using current to very quickly

vary the resistance, rather than using a hand operated knob (which is relatively

slow). An optical memristor, following this line of reasoning, is then

something whose optical response can be changed (very quickly) using light.

So, for example, a high intensity light pulse can be used to set the optical

memristor’s “resistance” value, and a low intensity light pulse can then be

used to take the desired reading.

o The change should be thermally irreversible, and should occur only when

exposed to light of sufficient intensity (lower intensity light pulses can then be

used to take the reading that was intended). As a consequence of this the

device’s setting is non-volatile – it does not change if one turns the device off.

o The changes made to the device’s behavior should be reversible.

o The device’s setting should not be affected by taking a reading.

o The contrast between the two extremes of the device’s output should be

consistently unambiguous when measured.

o The optical response of the device should not vary very sharply at either

extreme of its parameters, but should rather have a smooth gradient so that its

intermediate responses are easily accessible/measurable. This will make it

possible to subdivide the range of parameters that the device can be operated

with into smaller steps. Ideally, the device’s response should be a straight line

through the origin (the gradient would have to be small enough so that the

output readings can be taken unambiguously), where the abscissa denotes the

state that the device has been set to and the ordinate denotes the resulting

response to a signal that comes in to get processed/modified.

o The device should be durable (i.e., it should have high fatigue resistance).

Organic photochromophores, for examples, have the desired reversibility, but

they break down after a few thousand transitions [4] – that is not desirable in a

device which might go through those transitions within a fraction of a second.

A practical optical memristor should not degrade with use from having its

setting rewritten.

• Fits the following requirements:

o The physical dimensions of the device should be small, so that a large number

of devices can be put on a small chip.

o The response time of the device should be small enough, as compared the

existing electrical memristor, to justify the expense of development and

implementation.

• Does not depend too much on technology that has not yet been invented:

o The materials and procedures for fabricating the device should ideally rely on

mature technologies, so as to allow widespread use.

o The methods required for characterization of the materials and devices should

ideally not require too many innovations in characterization technology or

paradigm so as to reduce the risk of creating a characterization method that

may not work for the items being tested.

It should be noted that while the memristors created by HP had variable resistances

(though they did not exactly fit all of the mathematical criteria that Chua put forth [2])

one does not need to dogmatically stick to Chua’s definitions to make a practical device.

In fact, one can choose any particular parameter one wants to (i.e., any one that one can)

modify dynamically in order to make a passive device whose behavior can be modified to

suit a dynamically changing requirement. Such a device would not be an optical

memristor in the literal sense of the word, but it would certainly fit the bill in the

figurative sense. So, one can make an optical memristor with a dynamically variable

absorption coefficient, or a dynamically variable refractive index, or a dynamically

variable polarization, etc.

Optical Memristor: The most promising route towards the realization of optical
memristance (after discarding photochromic switching materials, for fatigue problems,

and after discarding optically induced hysteresis in liquid crystals, for their lack of

thermal stability) led towards optically induced changes in a material’s dipole moment

and optical hysteresis: since magnetic materials have hysteresis curves and since light is

an electromagnetic wave it should be possible to alter the magnetic state of a magnetic

material using photonic excitation, and since magnetic states (in tapes, discs, etc.) are

thermally stable the optically induced magnetization changes should also be thermally

stable. It was found that Hansteen [5] and Stanciu [6] had already thought of this idea and

successfully tested it, with Rasing, Kimel, Kirilyuk, Hunderi, et al [7-8], for the purpose

of improving magneto-optical disc drives, which currently use a laser to irreversibly erase

data (through demagnetization) and use an externally applied magnetic field to write new

data. The primary difference between the kind of materials that [5-8] used and what an

optical memristor would best use would be that an optical memristor’s memristive

material (“memresponsive” would perhaps be more appropriate) would need to have the

kind of shallow hysteresis curves (which do not change rapidly towards either extreme of

memristance) that HP’s memristors had (the looping is not very important):

Figure 1: Shallow hysteresis curve for memristor, that does not change abruptly at either extreme [1]

A material meant for recording binary information, on the other hand, requires a

hysteresis curve that is steep, and sharp at the edges [9]:

Figure 2: Steep hysteresis curve for recording binary bits – changes very abruptly [9]

Optical memristance will also need to be very fast (to compete with the electrical

memristor). Since it will have to be fast now, and it will need to be competitive in the

future it cannot use an externally applied magnetic field to generate a magnetization

vector within the optical memristor’s material (after it has been erased by an ultrafast

laser that heats the material to its Curie temperature [10]) because despite

demagnetization being an ultrafast process “writing” a new magnetization vector into a

material can be done faster than is possible by the application of a magnetic field to a hot

magnetic material (the re-write frequency is hampered by the cooling time, and this

method requires the material to dissipate heat quickly for the next re-write, thus

introducing thermal management problems [11]). A faster process would be a nonthermal

optical control of magnetism [7]:

Figure 3: Time Scales for Magnetic and Optical Processes [7]

Other advantages of nonthermal control are that deterministic magnetization cannot be

achieved if an applied magnetic field pulse (no matter how strong) is shorter than 2 ps

[12], and that spontaneous magnetization reversal does not occur under a nonthermal

regime, which happens with magnetization of a material after thermalization to the Curie

temperature, as was found during single-pulse magneto-optic microscopy experiments

[13].

Mechanism of Nonthermal Optical Control of Magnetism: The magneto-optical

Faraday Effect [14] says that the magnetization vector of a material will affect light

propagating through that material (for example, linearly polarized light passing through a

magnetized material will act as if it went through a polarizer). The (opto-magnetic)

Inverse-Faraday Effect, predicted by Pitaevskii [15], predicted that nonthermal optical

control of magnetism would be possible (i.e., light could change the magnetization of a

magnetic material that it was passing through, if it was intense enough), but there used to

be doubts about that since producing the effect experimentally was challenging [16-18].

With the advent of ultrafast laser pulses, however, one can generate extremely strong

fields, and this has now enabled the experimental observation of the Inverse Faraday

Effect [7-8]. The Inverse Faraday Effect is essentially a photon-direction and spin

preserving stimulated Raman Effect. Essentially, an electron in a non-degenerate state

absorbs a higher energy photon – hω1 – (which has less energy than the material’s band

gap, thus eliminating the possibility of electronic transitions and thus, large thermal

effects) from a spectrally broad ultrashort pulse and goes through a spin flip in the ground

state (which takes up energy hω1- hω2) before going up into a virtual state (so it is as if

the electron was excited by a photon of lesser energy than the energy of the photon that it

absorbed – hω2); then this electron undergoes stimulated emission upon being hit by

another photon of energy hω2 and returns to the ground state. So, the number of photons

is conserved, but one photon has given up part of its energy in flipping an electron’s spin.

The hω1- hω2 value depends on the material and its temperature conditions.

Figure 4: Ultrafast spin-flip via the process of the stimulated Raman scattering [7].

Mechanism for Maximizing Coherent Quantum Control of Spins: A system will be

set up to shape the beam that will be incident on the opto-memristor material so that the

beam will have two peaks in the Fourier domain: one at hω1 and one at hω2. This will

reduce the number of photons with unwanted energy values that compete for interactions

with electrons.

Figure 5: Pulse shaping for reduce competition from photons of unwanted energy [7]

Controlling Magnetization Vector Precession – the Double Pulse Method: A single

pulse of circularly polarized light, in the absence of an external magnetic field will start

the magnetization vector of the optomemristor material precessing, and it will keep

precessing (with a speed dependent on the material properties, excitation parameters, and

the beam propagation direction relative to the crystallographic axes) until the electrons go

back to the ground state.

Figure 6: Taken from [7]

If the electrons are hit with a second pulse while they are precessing then there will either

be constructive or destructive interference of the spin precessions (if there is constructive

interference then the precession still eventually dies, but the second pulse just adds its

own amplitude and lifetime to the existing precession).

Figure 8: Taken from [7]

Optomemristor materials can be chosen for their precession speed, their spin-orbit

coupling resonant frequency, or their bandgap. Considering that the Radboud University

Nijmegen group [7-8] were able to use materials with different properties for their pulsed

experiments it should be possible to fabricate optomemristors with various sets of

advantages.

Optomemristance Measurements: Optomemristance, as described above,
can be measured using the Faraday Effect [14], i.e., by measuring the rotation of linearly

polarized light that has just passed through the sample, or by measuring the MOKE

(Magneto-Optic Kerr Effect) [19], which is the rotation of the polarization of linearly

polarized light that has just been reflected from the sample.

Optomemristor Material Schemes: For devices that will utilize MOKE the material can

be a ferromagnetic or an antiferromagnetic, or a maybe a ferrimagnetic garnet, or

crystalline thin film.

Figure 9: Possible configuration for optomemristor device utilizing MOKE for probing

For devices that will incorporate the Faraday Effect for taking readings garnets or

crystalline films can be used, as well as films with quantum dots (metallic, as well as

semiconductors, such as CdSe [20]) held in position by a polymer matrix, perhaps with

carbon nanotubes in the matrix, so as to enhance the Faraday rotation, since the carbon

nanotubes will take on the surrounding magnetic field [21].

Figure 10: Possible configuration for optomemristor device utilizing Faraday Rotation for probing

Optomemristor Detection Schemes: The probe readings that will be taken from the

optomemristor/optomagnetic memristor proposed here will be in the form of MOKE

rotation (Fig. 11) Faraday Effect (Fig. 12) rotations. Therefore, the detector system will

involve an analyzer that will block out scattered and transmitted light that has not

undergone rotation, so that the rotated light (linearly polarized probe beam light whose

polarization got rotated upon interaction with the sample) can excite photomultiplier

tubes (PMTs) [22] or APDs.

Figure 11: Typical MOKE Setup [13]

Figure 12: Typical Setup for Measuring Faraday Rotation [7]

During development stages for a new process (i.e., when the method is still in its early

stages) it may be necessary to probe not just the overall polarization rotation result, but

also rotation resulting from various depths of the sample. While MOKE systems will not

allow probing very deep into the material they can still be probed layer by layer, so to

speak. Such a system will be extremely useful for the development of an optical

memristor, especially since such a device has never really been tried, and the uniformity

of performance of material at the different depths may be important for ensuring that no

part of an optomemristor material undergoes undue stress (which would degrade device

longevity). A confocal MOKE microscope that could be used for such a purpose can be

found in [23]:

Figure 13: Confocal MOKE Microscope [23]

Microscopy Requirements: Given that the each mem

References:

[1] Dmitri B. Strukov, Gregory S. Snider, Duncan R. Stewart & R. Stanley

Williams, The missing memristor found, Nature 453, 80-83 (1 May 2008),

doi:10.1038/nature06932

[2] Chua, L. O. Memristor - the missing circuit element. IEEE Trans. Circuit Theory

18, 507–519 (1971)

[3] Sung Hyun Jo, Ting Chang, Idongesit Ebong, Bhavitavya B. Bhadviya, Pinaki

Mazumder and Wei Lu, Nanoscale Memristor Device as Synapse in

Neuromorphic Systems, Nano Lett., 2010, 10 (4), pp 1297–1301, DOI:

10.1021/nl904092h

[4] He Tian and Yanli Feng, Next step of photochromic switches?, J. Mater. Chem.,

2008, 18, 1617 - 1622, DOI: 10.1039/b713216f

[5] Fredrik Hansteen, Ultrafast Optical Control of Magnetization in Ferrimagnetic

Garnets, PhD Thesis, Radboud University Nijmegen, The Netherlands - 2005

[6] Claudiu Daniel Stanciu, Laser-Induced Femtosecond Magnetic Recording, PhD

Thesis, Radboud University Nijmegen, The Netherlands – 2008

[7] Alexey V. Kimel, Andrei Kirilyuk, and Theo Rasing, Femtosecond opto-

magnetism: ultrafast laser manipulation of magnetic materials, Laser & Photon.

Rev. 1, No. 3, 275–287 (2007) / DOI 10.1002/lpor.200710022

[8] Alexey V Kimel, Andrei Kirilyuk, Fredrik Hansteen, Roman V Pisarev and Theo

Rasing, Nonthermal optical control of magnetism and ultrafast laser-induced spin

dynamics in solids, J. Phys.: Condens. Matter 19 (2007) 043201 (24pp)

doi:10.1088/0953-8984/19/4/043201

[9] J.-Y. Bigot , M. Vomir, L.H.F. Andrade, E. Beaurepaire, Ultrafast magnetization

dynamics in ferromagnetic cobalt: The role of the anisotropy, Chemical Physics,

Volume 318, Issues 1-2, 15 November 2005, Pages 137-146,

doi:10.1016/j.chemphys.2005.06.016

[10]Jean-Yves BIGOT, Femtosecond magneto-optical processes in metals, TRENDS

IN FEMTOSECOND LASERS AND SPECTROSCOPY, C. R. Acad. Sci. Paris,

t. 2, Série IV, p. 1483–1504, 2001

[11]J. Hohlfeld, Th. Gerrits, M. Bilderbeek, and Th. Rasing, H. Awano and N. Ohta,

Fast magnetization reversal of GdFeCo induced by femtosecond laser pulses,

PHYSICAL REVIEW B, VOLUME 65, 012413, DOI:

10.1103/PhysRevB.65.012413

[12]I. Tudosa, C. Stamm, A. B. Kashuba, F. King, H. C. Siegmann, J. St¨ohr, G. Ju,

B. Lu, and D. Weller, Nature 428, 831 (2004)

[13]M. Elazar, M. Sahaf, L. Szapiro, D. Cheskis, and S. Bar-Ad, Single-pulse

magneto-optic microscopy: a new tool for studying optically induced

magnetization reversals, OPTICS LETTERS / Vol. 33, No. 23 / December 1,

2008

[14]M. Faraday, Philos. Trans. R. Soc. Lond. 136(1), 104–123 (1846)

[15]L. P. Pitaevskii, Sov. Phys.-JETP (USA) 12, 1008 (1961)

[16]G. Ju, A. Vertikov, A.V. Nurmikko, C. Canady, G. Xiao, R. F. C. Farrow, and A.

Cebollada, Phys. Rev. B 57, R700 (1998)

[17]P. J. Bennet, V. Albanis, Y. P. Svirko, and N. I. Zheludev, Opt. Lett. 24, 1373

(1999)

[18]R. Wilks, R. J. Kicken, M. Ali, B. J. Hickey, J.D. R. Buchanan, A. T. G. Pym,

and B. K. Tanner, J. Appl. Phys. 95, 7441 (2004)

[19]M. Elazar, M. Sahaf, L. Szapiro, D. Cheskis, and S. Bar-Ad, Single-pulse

magneto-optic microscopy: a new tool for studying optically induced

magnetization reversals, OPTICS LETTERS / Vol. 33, No. 23 / December 1,

2008

[20]Shashi B. Singha, Mukta V. Limayea, Sadgopal K. Datea and Sulabha K.

Kulkarni, Room temperature ferromagnetism in thiol-capped CdSe and CdSe:Cu

nanoparticles, Chemical Physics Letters, Volume 464, Issues 4-6, 23 October

2008, Pages 208-210, doi:10.1016/j.cplett.2008.09.020

[21]Christophe Galland and Ataç Imamoğlu, All-Optical Manipulation of Electron

Spins in Carbon-Nanotube Quantum Dots, Phys. Rev. Lett. 101, 157404 (2008),

DOI: 10.1103/PhysRevLett.101.157404

[22]A. Laraoui, M. Albrecht, and J.-Y. Bigot, Femtosecond magneto-optical Kerr

microscopy, Optics Letters, Vol. 32, Issue 8, pp. 936-938 (2007),

doi:10.1364/OL.32.000936

[23]A. Laraoui, V. Halté, M. Vomir, J. Vénuat, M. Albrecht, E. Beaurepaire and J.-Y.

Bigot, Ultrafast spin dynamics of an individual CoPt3 ferromagnetic dot, The

European Physical Journal D - Atomic, Molecular, Optical and Plasma Physics,

Volume 43, Numbers 1-3, July 2007, DOI 10.1140/epjd/e2007-00120-y

Draft proposal for concepts for opto memristor and proposed microscope design for testing the said optical memristor materials

More Related Content

What's hot (20)

Similar to Draft proposal for concepts for opto memristor and proposed microscope design for testing the said optical memristor materials (20)

More from M. Faisal Halim (15)

Recently uploaded (20)

Draft proposal for concepts for opto memristor and proposed microscope design for testing the said optical memristor materials