Memristor Device for Security and Radiation Applications
The first physical demonstration of a non-volatile resistive-switching memory based on the nanostructured Pt/TiO2/Pt metal/insulator/metal stack from HP, has spurred the scientific community to develop memristive devices for a wide variety of applications. Owing to low-power and ultra-fast switching capabilities, memristors with nanoscale thickness geometry have been extensively investigated as potential replacements for flash memory technology in simple analog- and digital- computing applications. In Addition, both scalability and interconnectivity of memristors, through brain-inspired computing, have sparked a considerable move toward advancing of next-generation intelligent computing systems. On the horizon, other potential uses of the memristor have also emerged, particularly in sensing where attractive measurable changes in the I–V fingerprint of some device configurations have been demonstrated under certain types of extrinsic disturbances. Additionally, the unique and chaotic I–V response of some memristors opens the door for potential applications in hardware security. This chapter reports on novel approaches to utilize the electrical characteristics of the fabricated memristive devices for radiation sensing and security applications.
KeywordsMemristor Micro Nano Radiation Key Characteristics Security Switching IoT Communication
The first physical demonstration, in 2008, of a non-volatile resistive-switching memory based on the nanostructured Pt/TiO2/Pt metal/insulator/metal stack from HP , has spurred the scientific community to develop memristive devices for a wide variety of applications. Owing to low-power and ultra-fast switching capabilities, memristors with nanoscale thickness geometry have been extensively investigated as potential replacements for flash memory technology in simple analog- and digital- computing applications [2, 3, 4, 5, 6, 7]. In Addition, both scalability and interconnectivity of memristors, through brain-inspired computing, have sparked a considerable move toward advancing of next-generation intelligent computing systems [8, 9, 10].
On the horizon, other potential uses of the memristor have also emerged, particularly in sensing where attractive measurable changes in the I–V fingerprint of some device configurations have been demonstrated under certain types of extrinsic disturbances [11, 12, 13]. Additionally, the unique and chaotic I–V response of some memristors opens the door for potential applications in hardware security [14, 15]. This chapter reports on novel approaches to utilize the electrical characteristics of the fabricated devices presented in Chaps. 2 and 3 for sensing and security applications.
5.1.1 Memristor-Based Sensing
A growing number of fundamental studies on nano-sized memristors are emerging in sensing applications [16, 17, 18]. Although this field is yet to be developed, biological implementation holds the biggest share where implant- or portable-based memristor sensors are nowadays considered highly attractive for reducing the overall operational costs of vital prognostic tools [19, 20].
In bio-sensing, geometry and chemical adaptations of the nano-insulator component have both been reported essential to establish both the fluidic operation and the label-free recognition (specific or non-specific) of target species with the use of the memristive electrical fingerprint for sensing [18, 12]. Typically, the alteration of the I–V characteristics (e.g., voltage gap or ROFF/RON ratio [11, 13, 21]) in response to an external actuation (i.e., a change in physical or chemical environment) is considered to be the lead approach for exploiting the underlying operational features of memristor devices in sensing applications.
This chapter reports on a novel sensing area of using memristive devices for environmental health and safety applications. Ionizing electromagnetic radiation detection and dosimetry has been a worldwide challenge for general security purposes, particularly with regard to human exposure to x-rays and γ-rays. These radiations are extensively deployed for good purposes in research, medical imaging, radiation therapy, manufacturing, and environmental remediation [256–258]. They are also very frequently encountered in nuclear power plants and military industries, due to being primarily associated with the production, handling, and storage of hazardous materials and radioactive wastes [22, 23]. However, these radiations can be harmful due to their accumulative susceptibility and to their deleterious ionizing nature [24, 25]. Thus, frequent monitoring is mandatory to enhance public safety against undesired exposure due to accidental or hidden threats.
Classic radiation-protection dosimeters, based on (i) gas-ionization chambers , (ii) inorganic or organic scintillation (i.e., thermoluminescent) crystals [27, 28], (iii) radiographic or radiochromic dyes,  and (iv) semiconductor technologies [30, 31], have multiple shortcomings associated with either high power supply requirement, time consuming readout and calibration, lack of accuracy, and low spatial resolution or dynamic sensitivity with accumulated dose or temperature change . Continuous advancements of existing technologies are hence necessary to simultaneously tackle the portability, recyclability, and real-time monitoring faults.
The novel radiation sensing concept utilizing the capabilities of memristive devices can be explored in a close pathway to that established with semiconductor field-effect transistors (MOSFETs). Radiation detection with a MOSFET device is achieved via in situ trapping of charge carriers (e.g., electrons and holes) ejected from the insulator oxide layer as a result of a photoelectric effect. A shift in the threshold voltage (i.e., minimum gate-to-source voltage differential) is measured across a transistor and translated into a linear function of the accumulated absorbed dose [32, 33]. Similar to MOSFETs, radiation-induced photoelectric interactions can be emulated in memristive metal-oxide systems. The non-volatility and low power consumption of the memristor give the device the potential to replace the existing semiconductor-based radiation detectors. Sensing in this case would be achieved if measurable changes in the memristance I–Vcharacteristics are observed, provided that standard device operation is maintained. In recent space-related studies, metal-oxide memristors (e.g., Pt/TiO2–x(26 nm)/Pt and TiN/Ta/TaOx(10 nm)/TiN) were shown to have substantial hardness to ionizing electromagnetic radiations, which was mainly ascribed to radiation transparency due to nanoscale device thicknesses [34, 35, 36, 37, 38, 39, 40]. Predictive simulations with Pt/TiO2–x/Pt memristor, passively exposed to 1 meV γ-rays, projected that the probability of interaction becomes non-negligible beyond micron-scale oxide thicknesses .
5.1.2 Memristor-Based Security Applications
Ensuring that security aspects and key performance indicators (KPIs) are met is a challenge that adds value to any innovation in the technology domain. Hence, the evolution in electronic devices works hand in hand with security enhancement. The nonlinear I–V characteristic of memristors attracts the attention toward device deployment in security applications. Even though many researchers have highlighted the difficulties of producing similar memristor device behavior as a challenge for traditional memory and logic application , this randomness  and uniqueness of the memristor device behavior can be utilized for security applications.
Few secure communication systems are proposed in literature based on memristor devices. The work presented in the [14, 15] proposed building secures physical unclonable functions (PUFs) using the significant resistance variations of memristive crossbar. Encryption and decryption between the communicating devices using identical keys are considered an obstacle in this system, owing to the unpredictable behavior variations of the identical crossbars. Alternatively, the work presented in this chapter utilizes the uniqueness property of the fabricated nano/micro-thick memristor device to generate master and session keys. Moreover, a novel secure and efficient IoT conference communication system is proposed based on the unique memristive-based keys generated at the connected devices.
Many group secure communication systems are provided in literature. Burmester and Desmed (BD)  provided numerous conference key systems. Just and Vadenay  showed that the authentication feature is missing in BD system. Du et al.  proposed a modified synchronous counter-based scheme. Through such enhancements, Shim  showed that the verification of all the messages from all applicants is required to avoid the insider impersonation attack. The work in  proposed the practical enhancements of BD authenticated group key agreement systems that identify malicious insiders. This is achieved by using the trusted arbiter if the cheating has occurred in Mobile Ad hoc Network (MANET).
All the preceding systems are application layer-based algorithms using public key cryptography, which is not as efficient as symmetric key cryptography due to the required heavy computational cost. Alternatively, the work proposed in this chapter is the first to introduce secure IoT conference communication system based on memristor hardware, using symmetric key cryptography [48, 49].
5.2 Memristive-Based Radiation Sensing
The extensive fabrication and electrical testing mentioned in Chap. 2 have enabled the use of micro-thick TiO2memristors based on the optimal Al/Cu–D2 structure for radiation sensing experiments. Due to the great understanding of the basic electronic behavior of these devices without radiation; it is possible to ascribe novel observations to interactions developing with the radiation sources.
A cesium-137 γ-ray emitting source (type-D disk, Eckert & Ziegler, Germany), with an active diameter of 5 mm, a radioactivity of 18.1 μCi (0.67 MBq), and a primary emitted photon energy of 662 keV, was applied during the radiation sensing experiments. Radiation exposure tests were performed by placing the radioactive source on top of the hybrid Al/Cu memristor, facing the aluminum electrode. Radiation effects were monitored at room temperature, via real-time current measurement under non-switching −0.5 V pulse bias (pulse width 0.1 s, hold time 0.1 s).
After resetting the device to the original resistance value, the same experiment was repeated, during which the radiation source was put in contact with the operational device at 375 s from the beginning of the run. The initial nanoscale flowing current under −0.5 V bias gradually increased by 100-fold within 150 s window until it reached a value of 100 nA, which was fixed as the compliance value. When the radiation source was removed after 200 s of exposure, the current instantly dropped to 5 nA, and another turn-on was recorded 50 s earlier than that observed when the radiation was applied. Compared with the blank data, the first current jump reflects some probability of a radiation-induced conduction event. The second jump suggests persisting radiation-induced phenomena that would have exerted a synergy with the small voltage bias, both inducing faster switch-on, in agreement with associative effects, O’Kelly et al. recently explored with optical light and voltage pulses in neuromorphic nanowire devices.
Perceptible changes in the size of the negative hysteresis loop, in both Figs. 5.2a and b, provide further evidence of the sensing phenomenon from a different view that is based on the evolution of the hysteretic gap between the turn-off and turn-on currents. In this case, the ability of the memristor to detect γ-rays could also be traced from measurable changes seen in the ROFF/RON ratio of the device. Since the turn-on current was fixed by the compliance value, the changes in the size of the hysteresis gap reflected gradual modification of the ROFF value. In Fig. 5.2a, the ROFF value concluded from the turn-off current in the negative loop is found to be going further smaller under γ-irradiation. Since the ionizing radiations used are several orders of magnitude more energetic than the semiconductor’s bandgap, the deposited energy can create a photoelectric effect that generates a population of in situ secondary electrons within the switching material. The radiation sensing mechanism would hence be established on reading a lower device OFF resistance state (or higher OFF current) when enough charge trapping or tunneling leakage  is generated across the device. Both smaller OFF resistance state and shorter turn-on onset support the idea of a synergistic actuation pathway in active sensing mode; while γ-ray interactions could have participated in generating additional population of labile secondary electrons/holes pairs by photoelectric effect; the external electric field applied would have substantially minimized the recombination rate of these carriers by promoting charge transport. In Fig. 5.2b, seeing the gap gradually restoring back to its original magnitude in absence of the radiation source (as indicated by the tilted blue arrow), concludes that the memristor’s response time is not instantaneous and would require further material optimization for real-time monitoring applications .
5.3 Memristor-Based Secure Communication
5.3.1 I–V Characteristics and Key Generation
The memristive key generation scheme presented in this section depends mainly on the uniqueness property of the electrical characteristics of the fabricated memristor devices. This is realized and observed in both micro-thick and nano-thick memristor devices presented in Chaps. 2 and 3, respectively.
220.127.116.11 Micro-Thick TiO2 Devices
18.104.22.168 Nano-Thick HfO2 Device
The unipolar switching behavior presented in Chap. 3for the fabricated HfO2memristive crossbar follows the well-known filamentary-based switching mechanism. In such behavior, the creation and rupturing of the conductive filaments are considered probabilistic (rather than deterministic) processes. This can be mainly explained due to (i) (i) fabrication process variations  and (ii) the randomness in the number and the strength of the created filaments [42, 54]. As mentioned in Chap. 3, the thickness of the oxide layer is too small (10 nm) which leads to significant fabrication process variations due to the increased challenge in ions deposition. Moreover, the ions move randomly under the application of electric field. Thus, the number and the position of the created filaments vary within the same device and between the identical devices as well, under applying the same sweep voltages. This affects the strength of the formed conducting paths and consequently different ions profiles are obtained during reset operation.
5.3.2 Proposed IoT Conference Communication System
Proposed system acronyms and definitions
Unique secret key generated by TTP to share information among IoT device A, B, C
Unique secret key generated by A to share information among IoT device A, B, C
Secret Key between TTP and device A
Secret Key between TTP and device B
Secret Key between TTP and device C
Voltage applied across the device memristor
The width of the applied voltage pulse across the device memristor
Timestamp generated at device i
Nonce generated at device i
Encrypted with key K
Address of device I
- Step 1
To start the process, host A will contact the TTP informing it with the address of the nodes it wants to initiate a communication with. A sends TTP the addresses of B and C along with a nonce RA. The Message M1 is as follows:
After TTP receives the message from A and checks its content, TTP generates a session key KABC using its memristor and a timestamp TTP. TTP creates three messages encrypted with KAT, KBT, and KCT. These messages are sent back to A which is responsible for sharing the session key KABC with B and C. The message sent back to A is as follows:
- Step 2
key KABC. A generates tABC and VABC and uses its memristor to generate a new secure session key KABCnew and timestamp TA then forwards the messages from TTP. A also sends B and C another message to verify the used session key and share the newly generated KABCnew. A sends B the following:
And A sends C the following:
- Step 3
After B and C receive message from A, they verify the received messages and key and then send back another message containing their address and a fresh timestamp TB and TC at B and C, respectively; informing A that they are now aware of the communication session key that is used in future communications. B and C send messages M4 and M6, respectively. The messages are summarized as follows:
- Step 4
C sends B a message to inform it that it is aware of the new session key KABCnew and achieve authentication. C sends a message containing the address of B and its timestamp TC. The message is as follows:
- Step 5
Finally, B receives the message and replies with a message containing its timestamp TB and the address of C. The message is as follows:
5.3.3 Security Analysis
22.214.171.124 Mutual Authentication
Synchronization between A, B, C, and TTP was verified through claims (MEM,1A), (MEM,1B), (MEM,1C), and (MEM,1TTP). Passing synchronization test between the devices indicates that messages are sent and received by the intended parties in the right order and no manipulation has taken place. This ensures that the communicating entities can verify each other’s identity. Taking the communication between A device and TTP as an example, both platforms are sure of who sent the message because of the use of the secret key. Messages sent are fresh due to the existence of timestamps and both platforms can verify that the exchanged messages are actually intended to them, due to the concatenation of the address of the destination in the message.
The timestamps, nonce, and session keys exchange are verified for secrecy throughout the interaction by claims (MEM,2A to MEM,8A), (MEM,2B to MEM,7B), (MEM,12TTP to MEM,14TTP), and (MEM,2C to MEM,8C). The data remain confidential throughout the communication at A, B, C and TTP. At any time, only the owner platform can access the output of the request because of the encryption. In addition, the user request and timestamps are also encrypted and remain confidential. The essential information for decryption requires the knowledge of the secret keys.
Synchronization ensures that the exchanged information is not modified without being detected, claims (MEM,1A), (MEM,1B), (MEM,1C), and (MEM,1TTP) protect data integrity. The integrity of request, the output of the request, and all the timestamps are guaranteed at all times. Thus, any change or manipulation in messages is detected.
Claims proved that the exchanged information is kept secret from non-authorized parties. Only parties with proper decryption information are able to access the data. These keys remain confidential as proven by claims (MEM,7A, MEM,6B, and MEM,7C), therefore providing protection from unauthorized access attacks. When the user accesses the application and provides his/her credentials, this information is verified to decide whether to grant the user access to the application services or not. Moreover, by providing mutual authentication, platforms are able to verify each other, and any communication request or message sent by a non-trusted platform is ignored. The request, timestamps, and the output of the request are all confidential and can only be decrypted by authorized platforms.
126.96.36.199 Replay Attacks
The proposed system uses nonces and timestamps to provide a proof of freshness of messages. The timestamps and nonces remain secret and cannot be modified, and therefore, protection from replay attack is achieved.
- 8.L. Chua, Memristors: A new nanoscale CNN cell, in Cellular Nanoscale Sensory Wave Computing (Springer, Berlin, 2010), pp. 87–115Google Scholar
- 11.N.S.M. Hadis, A.A. Manaf, S.H. Herman, Comparison on TiO2 thin film deposition method for fluidic based glucose memristor sensor, in 2015 IEEE International Circuits and Systems Symposium (ICSyS) (2015), pp. 36–39Google Scholar
- 12.I. Tzouvadaki, N. Madaboosi, R. Soares, V. Chu, J. Conde, G. De Micheli et al., Bio-functionalization study of Memristive-Biosensors for early detection of prostate cancer, in Ph.D. Research in Microelectronics and Electronics (PRIME), 2015 11th Conference on (2015), pp. 17–20Google Scholar
- 14.Y. Gao, D.C. Ranasinghe, S.F. Al-Sarawi, O. Kavehei, D. Abbott, Memristive crypto primitive for building highly secure physical unclonable functions. Sci. Rep. 5, 12785 (2015)Google Scholar
- 19.E. Chen, J. Thornton, C. Mulchi Jr., Mapping circular current for a single brain cancer cell’s spatial-temporal orientations based on a memristor/memcapacitor. Sens. Transducers 183, 72 (2014)Google Scholar
- 21.N.S.M. Hadis, A.A. Manaf, S.H. Herman, S.H. Ngalim, ROFF/RON ratio of nano-well fluidic memristor sensor towards hydroxide based liquid detection, in Nanotechnology (IEEE-NANO), 2015 IEEE 15th International Conference on (2015), pp. 1078–1081Google Scholar
- 22.P.D. Wilson, The Nuclear Fuel Cycle: From Ore to Wastes (Oxford University Press, Oxford, New York, 1996), pp. 184–200; 252–271Google Scholar
- 23.C.S. Cook, initial and ionizing radiations from nuclear weapons, in Sources, Fields, Measurements, and Applications, 2nd edn, ed. By F.H.A. Tochilin (Academic Press, Cambridge, 1969), pp. 361–399Google Scholar
- 26.G.F. Knoll, Radiation Detection and Measurement, 4th edn (Wiley, New Jersey, 2010), p. 131Google Scholar
- 28.S. Pellegrin, C. Whitney, C.G. Wilson, A multichannel nanoparticle scintillation microdevice with integrated waveguides for alpha, beta, gamma, X-ray, and neutron detection, in Micro Electro Mechanical Systems, 2006. MEMS 2006 Istanbul. 19th IEEE International Conference on (2006), pp. 682–685Google Scholar
- 32.J. Izewska, G. Rajan, Radiation dosimeters. Radiat. Oncol. Phys. A Handb. Teachers Students 36, 71–99 (2005)Google Scholar
- 36.E. DeIonno, M.D. Looper, J.V. Osborn, H.J. Barnaby, W.M. Tong, Radiation effects studies on thin film TiO 2 memristor devices, in Aerospace Conference, 2013 IEEE (2013), pp. 1–8Google Scholar
- 38.E. DeIonno, A.L. White, Sensitivity of metal oxide memristors to radiation-induced displacement damage, in Aerospace Conference, 2014 IEEE (2014), pp. 1–5Google Scholar
- 42.S. Gaba, P. Knag, Z.Y. Zhang, W. Lu, Memristive devices for stochastic computing, 2014 IEEE International Symposium on Circuits and Systems (Iscas) (2014), pp. 2592–2595Google Scholar
- 43.M. Burmester, Y. Desmedt, A secure and efficient conference key distribution system, in Workshop on the Theory and Application of of Cryptographic Techniques (1994), pp. 275–286Google Scholar
- 44.M. Just, S. Vaudenay, Authenticated multi-party key agreement, in International Conference on the Theory and Application of Cryptology and Information Security (1996), pp. 36–49Google Scholar
- 45.X. Du, Y. Wang, J. Ge, Y. Wang, An improved ID-based authenticated group key agreement scheme. IACR Cryptol. ePrint Arch. 2003, 260 (2003)Google Scholar
- 46.S. Kyung-Ah, Further analysis of ID-based authenticated group key agreement protocol from bilinear maps. IEICE Trans. Fundam. Electron. Commun. Comput. Sci. 90, 295–298 (2007)Google Scholar
- 47.C.Y. Yeun, K. Han, D.L. Vo, K. Kim, Secure authenticated group key agreement protocol in the MANET environment. Inf. Secur. Tech. Rep. 13, 158–164 (2008)Google Scholar
- 48.H. Abunahla, D. Shehada, C.Y. Yeun, C.J. OKelly, M.A. Jaoude, B. Mohammad, Novel microscale memristor with uniqueness property for securing communications. In Circuits and Systems (MWSCAS), 2016 IEEE 59th International Midwest Symposium on (2016), pp. 1–4 Google Scholar
- 53.J.P. Strachan, D.B. Strukov, J. Borghetti, J.J. Yang, G. Medeiros-Ribeiro, R.S. Williams, The switching location of a bipolar memristor: chemical, thermal and structural mapping. Nanotechnology 22, 254015 (2011)Google Scholar
- 54.Q. Li, H. Xu, A. Khiat, Z. Sun, T. Prodromakis, Impact of active areas on electrical characteristics of TiO2 based solid-state memristors, in 2015 IEEE International Symposium on Circuits and Systems (ISCAS) (2015), pp. 185–188Google Scholar
- 55.C.J. Cremers, The scyther tool: verification, falsification, and analysis of security protocols, in International Conference on Computer Aided Verification (2008), pp. 414–418Google Scholar
- 56.C. Cremers, S. Mauw, Operational Semantics and Verification of Security Protocols (Springer, Berlin, 2012)Google Scholar