Bipolar Resistive Memory with Functional Duality-Non Volatile Emerging Memory and Nano Biosensors

Abstract

Resistive Memory in recent year has emerged as a potential candidate in the field of Non Volatile Memory to solve the existing problems with FLASH. The operation novelty of ReRAM helps to evolve it from storage device to an effective ultra sensitive biomarker with a very simple structure and fabrication process steps. Basically as ReRAM is MIM capacitor like structure so for store data in terms of charge like DRAM is feasible effectively and at the same time capacitor as we know can be excellent for bioanalyte detection. So with same structure two purpose can be solved. Also we can see in this chapter that the biosensors with ReRAM will detect on Current -Voltage sampling method which is more efficient to detect with low sample volume. This chapter will give the readers a brief idea about the work done and ongoing research on Resistive memory as Non Volatile Memory as well as its potentiality as Biosensor.

1. Introduction

Modern era is the era of electronic gadgets and semiconductor memory is not only the integrated part of the modern electronics devices rather it acts as a heart of the device. From a longer period of time different research groups across the world in both industries and academics are working on various semiconductor memories and try to find the way out for better performances. The demand of development of the semiconductor memories are boosted due to enhancement usability of personal electronics appliances as well as amusement electronic gadgets such as laptops, palmtops, digital camera, tablet PCs, various kinds of smartphones, etc. [1]. This memory based electronic gadgets also play an important role in medical market. As the days passes the dramatic development in all these electronic gadgets put a demand of more powerful memory devices in front of researchers. On the basis of operation principle the semiconductor memory is classified into two major groups: volatile and non-volatile memory (NVM) [2]. Volatile memories are such kinds of memories which require an external power source to retain data. Static random access memory (SRAM) and dynamic random access memory (DRAM) are the two existing volatile memory in arket. On the other hand NVM are those kind of semiconductor memories which do not need any external power source for retaining their stored data. Some popular NVM leading in semiconductor memory market are phase change memory (PCM), FLASH memory, ferroelectric random access memory (FeRAM) and magnetic random access memory (MRAM) etc. Now all these memory devices have several advantages as well as disadvantages. For example DRAM is very high capacitive and higher density memory structure but it needs a refreshment operation in regular intervals of few milliseconds. On the other hand SRAMs have the ability of faster operation and it possesses a large numbers of memory cells but it has lower capacity than DRAMs. Among them the most leading and popular NVM that commercially rule the current semiconductor memory market is FLASH memory. The FLASH memory has two basic advantages for its rapid growth and applications [3]; one is its miniaturized structure and reliability property. Another one is its non volatility characteristics. But still the scaling issue and some operational shortcomings hinder FLASH memories to lead the market for longer periods of time. Additionally the demand of new NVM that can overcome the drawbacks of FLASH increases day by day. Some important key pitfalls of FLASH are limited endurance (∼10⁵ to 10⁶ cycles), small write-erase (W/E) time (1 ms to 1 μs) and higher operation voltage [4, 5]. But above all the serious bottleneck issue for FLASH is its scaling limit (cannot be scaled below 20 nm). So the requirement of emerging NVM which can capable to replace the FLASH is necessary. The market expectation is that this new kind of NVM can able to show longer endurance (>10⁶ cycles) with faster switching speed (few ns) and stable data retain ability property (∼10 years @ 85°C). Additionally this new kind of memory devices should have smaller feature size (<20 nm) and show lesser power consumption property and must be CMOS compatible. In this respect International Road Map for Semiconductors (ITRS) had proposed a guideline in 2015 where a brief comparison with emerging NVM technologies had a very good comparison with FLASH and showed which are the emerging NVM that have the potentiality to take over the lead in future NVM market from FLASH [6]. The comparisons are shown in Table 1.

		Base line technologies				Prototypical technologies			Emerging
		DRAM	SRAM	Flash		FeRAM	STT-MRAM	PCM	Redox FeRAM
				NOR	NAND
Cell elements		1 T1C	6 T	1 T		1 T1C	1(2)T1R	1 T(D)1R	1 T(D)1R
Feature size, nm	2013	36	45	90	22	180	65	45	D:9; B: <5
	2026	9	10	25	8	65	16	8
Cell area	2013	6 F^2	140 F^2	10 F^2	4 F^2	22 F^2	20 F^2	4F^2	4F^2
	2026	4 F^2	140 F^2	10 F2	4 F^2	12 F^2	8F^2	4F^2
Read time	2013	<10 ns	0.2 ns	15 ns	0.1 ms	40 ns	35 ns	12 ns	D: <50 ns
	2026	< 10 ns	70 ps	8 ns	0.1 ms	<20 ns	<10 ns	<10 ns	B: <10 ns
W/E time	2013	<10 ns	0.2 ns	1 μs/1 ms	1/0.1 ms	65 ns	35 ns	100 ns	0.3 ns
	2026	<10 ns	70 ps	1 μs/1 ms	1/0.1 m s	<10 ns	<1 ns	<50 ns
Retention time	2013	64 ms	—	10 y	10 y	10y	>10 y	>10y	10 y
	2026	64 ms	—	10 y	10 y	10y	>10 y	>10y
Write cycles	2013	>1E16	>1E16	1E5	1E14	1E14	>1E12	1E9	D: 1E12; B: 1E16
	2026	>1E16	>1E16	1E5	5E3	>1E15	>1E15	1E9
Write voltage (V)	2013	2.5	1	10	15	1.3–3.3	1.8	3	< 0.5
	2026	1.5	0.7	9	15	0.7–1.5	<1	<3
Read voltage (V)	2013	1.8	1	1.8	l.8	1.3–3.3	1.8	1.2	0.15
	2026	1.5	0.7	1	1	0.7–1.5	<1	<1
Write energy (J/bit)	2013	4E-15	5E-16	1E-10	>2E-16	3E-14	2.5E-12	6E-12	D: 1E-13,; B: 1E-17
	2023	2E-15	3E-17	IE-11	>2E-17	7E-15	1.5E-13	1E-15

Table 1.

Comparison of characteristics of DRAM,SRAM,FLASH with other emerging NVM published in international technology roadmap for semiconductor (ITRS), emerging research devices (ERD) chapter, 2015 [6].

Basically Resistive Memory devices are structure wise a simple metal-insulator-metal capacitor based devices, where a high-k dielectric insulator layer mostly made of metal oxide (MOs) or transitional metal oxides (TMOs) is sandwiched between two terminal metal electrodes. Depending on the types of material of the top metal electrodes the Resistive Memory are categorized into 2 types- (a) Conductive Bridging Random Access Memory (CBRAM) and (b) Resistive Random Access Memory (ReRAM).

On the other hand, resistive switching mechanism in MO/TMOs-based structure have been reported in various articles already [7, 8, 9, 10, 11]. But exploring the features of Resistive memory not as a storage device but as biosensor is very rare [12, 13, 14, 15, 16, 17]. Now along with memory characterization some bioanalyte detection tests also are carried out because these biological compounds are most important for maintaining a healthy body. As resistive switching is based on reduction-oxidation (redox) process, also many of the bioanalytes detection are based on redox process where the one of the common by product is H₂O₂, so the switching mechanism also can be justified through H2O2 sensing. Hence the interest on more work on RRAM as biosensor is hoped to come in upcoming time.

Hence, different bio-analyte, namely Creatinine, Urea, Sarcosine, LoXL2, Tributyrin, etc. detected by 2D Through Via Hole (TVH) and 3D cross-point (CP) structure is a completely new concept demonstrated in this chapter.

In this chapter we will do a basic review on ReRAM, CBRAM, 3D ReRAM or Cross-Point Memory and ReRAM as Biosensor in the following sections.

2. Resistive memory as a promising alternative of FLASH

Conventional semiconductor memories such as DRAM, FLASH have already reached to some limitations from technological point of view and so a replacement alternative is emerged as market demand. Now some emerging candidates in the field of NVM which are going to be considered as alternative of FLASH and considered as next generation NVM lead the semiconductor market are PCM, FeRAM, MRAM, and resistive random-access memory (RRAM). Now among all these memories RRAM has emerged as one best option due to several advantageous potentials such as it has a simple MIM structure, CMOS compatibility, excellent scalability potential, and long program/erase (P/E) endurance [18, 19, 20, 21]. From technological point of view RRAM is a promising candidate to replace FLASH according to ITRS report published in 2015 [6] as described in Table 1. According to this report, where FLASH cannot be scaled down below 20 nm there RRAM has demonstrated to be scaled down below 10 nm and are expected to be scaled down to below 5 nm by 2026. The processing cost is also very less in case of RRAM owing to its simple structure. There is no need of block erase like FLASH (block erase needed for every 3 ms interval). RRAM has excellent latency time (<30 ns) over NAND FLASH (∼75 μs) and superior write performance (140 MBPS for RRAM compare to 7 MBPS for NAND FLASH). Endurance for RRAM also very high (>10¹² [22] and > 10¹⁶ is expected to achieve). RRAM has ability of longer data retention in robust condition (∼10 years at 125°C temperature). Also from other properties like high memory density, low power consumption, low energy, better security are some key advantages of RRAM that make it’s a promising candidate for future leader in semiconductor memory field. Figure 1 shows the reported data that prove RRAMs’ superiority over conventional FLASH.

Figure 1.

Advantage of resistive memory over FLASH memory [22].

2.1 Resistive RAM basics

Normally the metal oxides (MOs) or transitional metal oxides (TMOs) are acting as insulator at normal state. But if an electric field is applied across them then at certain field the oxide breaks down and a conducting nano-filament is formed which decreases the resistance of the device. Almost more than 40 years ago such kind of fact was come to the picture when in 1962 Hickmott observed a resistive switching phenomena in a MIM structure [23]. Since that the research groups across the world became interested to evaluate the operation principle of RRAM and tried to come out with promising features of resistive switching that make it the suitable alternative of FLASH. The insulating material includes binary or multinary oxides, chalcogenides materials and also organic compounds. Now this insulator is sandwiched between two metal electrodes such as W, Pt, Cu, TiN, etc.

As per the different sources and research groups working globally Resistive random-access memory (ReRAM) is based on a simple three-layer structure of a top electrode, switching medium and bottom electrode as depicted in Figure 2. The resistance switching mechanism is based on the formation of a filament in the switching material when a voltage is applied between the two electrodes. There are different approaches to implementing ReRAM, based on different switching materials and memory cell organization. Those variables drive significant performance differences depending upon the switching materials being used. In addition to use as multiple-time programmable (MTP), few-time programmable (FTP) and on-time programmable (OTP) non-volatile memory applications.

Figure 2.

Simple 2D schematic of resistive memory structure.

ReRAM technology also can be used for security applications, where ReRAM is used for secure physical unclonable function (PUF) keys embedded in semiconductors.

Normally when a positive or negative bias is applied to the top electrode (TE) to create the filament and the bottom electrode (BE) is kept as grounded. The insulator or dielectric material between them is termed as switching material (SM).

Because the resistance switching mechanism is based on an electric field, the ReRAM cell is very stable, capable of withstanding temperature swings from −40°C to 125°C, 1 M+ write cycles and a retention of 10 years at 85°C.

2.2 Resistive switching mechanism

2.2.1 Modes of resistive switching: unipolar and bipolar

From mode of operation point of view resistive switching is classified into two sub category. One is called bipolar switching and another is called Unipolar switching. In a Unipolar switching the direction of current voltage (I-V) hysteresis curve do not depends on the polarity of the applied bias where as in case of bipolar resistive switching mode this phenomenon is just opposite. In bipolar switching the switching direction is highly depends on bias polarity. The clear view of both kinds of resistive switching is described in Figure 3.

Figure 3.

Typical unipolar and bi-polar resistive switching characteristics of RRAM [24].

In most of the publications based on ReRAM or CBRAM are describing bi-polar mechanism and in this article I am also describe about the bi-polar switching only.

2.2.2 Bipolar resistive switching mechanism

Usually the pristine device has a very high insulation property. An initial forming voltage [25] is required to change the initial resistance state (IRS) of the device. Some devices do not require any forming voltage. These are called forming-free devices [26, 27] and these are effective for low power low cost RRAM applications. Now different research groups have developed many physical models to describe the resistive switching phenomena but the exact mechanism occurs inside the device is yet to have a clear conception. In overall point of view the mechanism of resistive switching is described by two established models.

1. Filamentary type resistive switching [8] and

2. Interface mediated resistive switching [28]

These two models shown in Figure 4 are jointly used to explain the most of the insulating oxides based bipolar resistive switching phenomena [29, 30, 31].

Figure 4.

Typical switching mechanism models of RRAM (a) filamentary type resistive switching model, and (b) Interface mediated resistive switching model.

In CBRAM structure a thin layer of solid electrolyte or high-κ dielectric material is sandwiched between oxidized anode electrode made of active metals such as Cu, Ag and an inert cathode electrode made of inert metals such as W, Pt, and TiN etc. Figure 5 describes the CBRAM mechanism using a schematic diagram.

Figure 5.

Typical switching mechanism of cation based resistive switching memory or CBRAM.

When a positive bias is applied at oxidized electrode the metal ions are oxidized. Now the mobile cations then drifted towards cathode terminal through the dielectric material and form metallic path by connecting two terminal electrodes by a bridge like filamentary path. When the filament is formed the device start to conduct and device resistivity goes to LRS. When the reverse bias is applied again the metal ions drifted towards the active electrode under the influence of applied field and thus a filament dissolution or rupture occurs and the path is broken causing device to switch to HRS.

On the other hand in anion based switching under the influence of applied bias the insulator oxide break down occurs and the O₂-ions are drifted towards the electrode where positive bias is applied.

These results show the generation of oxygen vacancies (VO) inside the insulator material. When there are enough VO causing to form a filamentary path between two terminal electrodes the device starts to switch. Again when the bias with opposite polarity is applied the O2- ions moved in opposite side than the previous case causing the filling up of VO and by this process filament ruptures. This mechanism is described in Figure 6.

Figure 6.

Typical switching mechanism of anion based resistive switching memory or RRAM.

Although this mechanism is widely proposed by most of the research groups but still the exact mechanism is unfolded. The bias polarity is the main factor that controls the process of oxidation and reduction in this kind of devices [31, 32].

The most accepted CBRAM mechanism was first ever proposed by Prof. M.N. Kozicki [7] in solid electrolyte in 2004 and the same for RRAM was proposed by Prof. R. Waser and M. Aono in 2007 [24].

2.3 Resistive memory materials

To explore the switching mechanism in RRAM devices more elaborately different research groups have studied different switching characteristics by using different switching materials and changing the top and bottom electrodes. Although many materials have been studied but here we are citing few of them showing very promising characteristics and better performances. Table 2 shows some popular oxides [33, 34, 35, 36] used as RRAM dielectrics with their band gap (Eg), dielectric constant (κ) and Gibb’s free energy (ΔG).

Material	Band gap; (Eg) in eV	Dielectric constant (k)	Gibbs free energy (kJ/mole)
Al2O3	8.8	9	−1580
BaTiO3	3.42	100–600	−157.4
Gd2O3	5.3	17	−1760
a-C	2–5	2.5–6
GeO2	0.66	16.2	−518.845
HfO2	5.8	25	−1010.75
SiO2	9	3.9	−853.5
SrTiO3	3.2	2000	−1567.77
Ta2O5	4.4	22	−760.75
TiO2	3.2	SO	−923.375
WO3	3	4.S	−506.875
ZrO2	5.8	25	−1037.25

Table 2.

Band gap (Eg), dielectric constant (κ) and Gibbs free energy (ΔG) for different metal oxides [34, 35].

Among the oxides materials some of the promising candidates are TaOx [10, 18, 22, 37, 38], HfOx [11, 39, 40], and TiOx [21, 41].

Besides that some new oxides are also bring attention to the research groups such as GdOx [12, 42], WOx [43], BaTiO₃ [44]. SiOx also have a recent growing interest to the researchers [15, 45, 46, 47, 48] because of it has low Gibbs free energy (−853.5 kJ/mole) and high band gap (Eg ∼ 9) [34, 35]. Also SiOx is most CMOS compatible material with low cost.

On the other hand, after the proposed mechanism of CBRAM by Kozicki et al. [7], many groups as per reported by Jana et al. [49] giving interest in CBRAM structures. Among the popular materials with which the CBRAM works are carried out mostly are based on solid-electrolyte materials including GeSe [20, 50], GeS2 [51, 52], although there are few reports on high-κ oxides such as Ta₂O₅ [53, 54, 55, 56, 57], Al₂O₃ [58, 59, 60], TiO₂ [61], and SiOx [47, 62, 63, 64]. Some recent reports are based on 2D materials like MoS₂ based CBRAM structures [65, 66].

Among all the switching materials mentioned above Al₂O₃, HfO₂, SiO₂, Ta₂O₅,and TiO₂ are some of the most popular and widely used switching materials owing to their potentiality to generate higher ON/OFF ratio, Longer Program/Erase Endurance cycles, Faster switching speed, and Lower Energy Consumption.

Xiao Liang Hong et al. [67] from NTU Singapore had published a review article in Journal of Material Science in 2018 where they had shortlisted some basic criterion for choosing RRAM materials. Following 3 tables Tables 3–5 classified RRAM mechanism based on RRAM structure, RRAM top electrode and RRAM switching materials.

Type of RRAM Stack Structure	Switching Mechanism Type
a) Symmetric:	Mainly Unipolar Switching are observed.
Both Terminal electrodes are made with same metal. Like: Pt/HfOx/Pt; W/SiOx/W; TiN/TaOx/TiN etc.
b) Asymmetric:	Mainly Unipolar Switching are observed.
Both Terminal electrodes are made with different metal. Like: Pt/HfOx/TiN; W/SiOx/TiN; Ir/TaOx/W etc.

Table 3.

RRAM switching mechanism based on types of RRAM stack structure [67].

Type of RRAM Electrodes	Mechanism
a) Active Metal Electrodes	Mainly Bipolar Switching. Switching Types: ECM & VCM. ECM: Electro Chemical Metallization effect. In this case the filament is a metallic filament made of metal cations like Cu, Ag, Ni etc. (CBRAM). VCM: Valance Change Memory effect. Here the filament is fonned by migration of anions or oxygen vacancies. (RRAM).
Metals which can be get oxidized like Ti, Al, Ag, W, Cu, Ni, TiN, TaN, ITO.
b) Inert Metal Electrodes	Mainly Unipolar Switching. Switching Types: VCM.
Metals which ne\'er reacts with oxygen such as Pt, Ir, Ru, Au, Pd.
c) Novel Metal Electrodes	Mainly Bipolar Switching.
Graphene or CNT.

Table 4.

RRAM switching mechanism based on types of RRAM electrodes [67].

Type of Switching material stack	Mechanism	Advantages
a) Doped Switching layer	Dopants diffuse inside the switching materials and create more oxygen vacancies to improve switching and cause fanning free switching	Good data retention, potential of multi bit operation, highly scalability, faster switching speed.
If metal oxides doped by me active metals like Ti, Hf Ta, Al or semiconductor materials like Si, Ge
b) Bi-layer / Multilayer Switching material stack	Resistive switching occurs at higher resistivity layer and lower resistivity layer can improve the ON state resistance.	Better uniformity (diffusion of metal ions from the lower resistivity to the higher resistivity layer. This stabilizes the CFs to reduce the randomness of resistive switching (RS), Multi-Level Storage function, Low power operation, lower Reset current.
The switching layer will be combination of two metal oxides with one higher and one lower resistivity. Like HfOx/AlOx or SiOx/TaOx or GeOx/TiOx etc.

Table 5.

RRAM switching mechanism based on types of RRAM switching materials [67].

It is also observed that combination of Active & Inert electrode results higher endurance and longer data retention where as Graphene & CNT based electrode results ultra dense memory formation.

3. 3D resistive memory structure or cross-point memory

To compete the 3D NAND FLASH the RRAM should be upgraded next level for ultra-fast performances. Now for these one of the good options is to go for high density memory applications. For this we need to work on cross-point memory devices. The Cross Bar Architecture (CBA) or Cross-Point (CP) structure is a fundamental structure used in 3D Resistive Switching Memory applications. It consists of a grid-like arrangement of memory cells, with memory cells integrated between word-lines and bit-lines [68] as shown in Figure 7.

Figure 7.

Typical cross-sectional view of 3 × 3 cross Bar resistive memory device structure.

Seok et al. [68] had reported this cross-point structure allows for a high cell density, as it reduces the cell size to 4F^2, where F represents the feature size. This high-density ReRAM technology can be stacked in 3D, delivering multiple terabytes of storage on a single chip. Its simplicity, stackability below 10 nm and CMOS compatibility enable logic and memory to be integrated onto a single chip at the latest technology node. Designers can put logic, controllers and microprocessors next to memory in the same die, simplifying packaging and increasing performance. ReRAM’s simple structure and CMOS compatibility enable any foundry - CMOS or logic - to enter the ReRAM business by licensing CrossBar ReRAM technology for Systems-on-Chip (SoC) or standalone memory devices or embedded cryptographic keys.

However, despite its advantages in terms of cell density, the CBA or CP is not without its challenges. One of the main issues is the occurrence of cross-talk interference between memory cells which called as Sneak Path Current shown by red dotted line in Figure 7. This interference is caused by leakage current paths through neighboring cells with low resistances.

Many researchers are hence try to design a high-density memory which can operate at low current operation and deliver a high speed program and erase operation and also reduce the sneak path current. To achieve those requirements, different structures with different materials have been reported by several groups. In CP memory also different SM materials based reports are there among which some of the popular SM are TaOx [69, 70, 71], HfOx [72, 73, 74] AlOx [13, 75], and SiOx [47, 64].

CP memory can show excellent data storage capability with >3 × 10⁶ memory window [74] which make them suitable candidates for high density memory. The cross-point memory show its potentiality for low power VLSI applications with ultra-low operation current of 1 μA [69] and low Set/Reset voltages of ±0.5 V [47]. Also with Cross-bar architecture the longer P/E endurance >10¹¹ cycles at low program current ∼10 μA [74], excellent data retention >3 hour at robust condition of >125°C temperature [75], high speed operation of 50 ns with low energy per bit of 0.49 pJ [64] can be achievable.

According to the reported data some improvements have been obtained but operation current is still high in most of the cases. To reduce operating current of <1 μA, investigation is needed. In addition, it is also important to detect bioanalyte by using cross-point memory if we want to establish ReRAM has the potentiality to be work as ultra-sensitive biosensor. Although many terminal electrodes are used in CP structure but some recent study has shown if IrO_X will be used as top electrode CP-RRAM can be an excellent biosensor owing to the benefits of Ir for its porous nature shown by the FeSEM image in Figure 8, which can help it to act as an excellent sensing membrane for different bioanalytes and help the analytes to reach fast at IrO_X and SM which is the MO interface for redox reaction during switching of the memory device [12, 13].

Figure 8.

Field-emission scanning electron microscope (FESEM) image of 20 nm-thick IrO_X layer deposited on Si at 200 k magnification [12].

3.1 Crossbar memory operation

The basic operation of a crossbar-based resistive memory involves applying voltage pulses to the appropriate row and column lines to read or write data. To write data, a voltage pulse is applied across a selected row and column, causing the resistance of the memory cell at the intersection point to switch between high and low states. The resistance state of the memory cell can be sensed by applying a voltage across the row or column and measuring the resulting current.

3.1.1 Read operation

To read data from a memory cell in a crossbar architecture, a voltage is applied across a selected word line (row) and bit line (column). The resulting current through the memory cell is measured to determine its resistance state (HRS or LRS).

3.1.2 Write operation

To write data to a memory cell, a voltage or current pulse is applied across a selected word line and bit line. This pulse induces a physical or chemical change in the insulating layer, causing the resistance of the memory cell to switch between HRS and LRS.

3.2 Crossbar memory advantages and challenges

One of the advantages of the crossbar architecture is its high density. Since each intersection point in the crossbar represents a memory cell, a large number of memory cells can be packed into a small area. This high density makes crossbar-based resistive memory attractive for applications that require large storage capacities.

However, there are several challenges associated with the crossbar architecture. One major challenge is the sneak path problem. When voltage is applied to a selected memory cell, neighboring cells in the same row or column can also be affected due to parasitic currents. This can lead to unintended switching of multiple memory cells, resulting in data corruption. Various techniques such as voltage compliance, current sensing, and advanced algorithms are employed to mitigate the sneak path problem.

Another challenge is the scalability of the crossbar architecture. As the number of memory cells in a crossbar increases, the resistance of the interconnects and the parasitic capacitance also increase, which can degrade the performance and reliability of the memory array. Techniques such as hierarchical and three-dimensional (3D) crossbar architectures are being explored to overcome these scalability limitations.

3.2.1 Sneak path problem

One of the challenges in crossbar-based resistive memory is the sneak path problem. When voltage is applied to a selected memory cell, neighboring cells in the same row or column can be unintentionally affected due to parasitic currents. This can lead to unintended switching and data corruption. Various techniques, such as voltage compliance, current sensing, and advanced algorithms, are employed to mitigate the sneak path problem.

3.2.2 Scalability

The scalability of crossbar-based resistive memory is an important consideration. As the size of the crossbar array increases, there can be challenges in maintaining uniform switching characteristics and managing interconnect resistance and capacitance. Hierarchical and three-dimensional (3D) crossbar architectures are being explored to overcome these scalability limitations.

There are several techniques are employed to mitigate the sneak path problem in crossbar-based resistive memory. Here are some commonly used approaches:

1. Voltage Compliance: One approach is to apply voltage compliance limits during write operations. By setting an upper limit on the voltage applied to the selected lines, the parasitic currents in the neighboring cells can be restricted, reducing the likelihood of unintended switching. This technique helps in confining the switching operation to the intended memory cell.

2. Current Sensing: Current sensing techniques can be applied to detect and compensate for the sneak path currents. By measuring the currents flowing through the bit lines during read and write operations, it is possible to identify the contributions from the unintended sneak path currents. By subtracting or compensating for these currents, the desired current corresponding to the targeted memory cell can be accurately determined.

3. Advanced Algorithms: Sophisticated algorithms and error correction codes can be employed to overcome the sneak path problem. These algorithms take into account the behavior of the memory cells and the sneak path currents to reconstruct the intended data accurately. By employing error correction techniques, the impact of sneak path-induced errors can be mitigated.

4. Shielding Techniques: Physical or electrical shielding techniques can be employed to minimize the influence of parasitic currents. For example, placing additional metal layers or insulating materials between the rows and columns can help reduce the coupling between adjacent memory cells and mitigate the sneak path problem.

5. Selective Access Devices: Introducing selective access devices, such as transistors or diodes, at each intersection point in the crossbar can help in isolating the memory cell being accessed. These devices act as switches and allow the targeted memory cell to be electrically connected to the selected word and bit lines, while preventing the parasitic currents from affecting neighboring cells.

6. Process Optimization: The fabrication process of the resistive memory can be optimized to minimize the occurrence of sneak path currents. This can involve carefully engineering the materials, interfaces, and dimensions of the memory cells and interconnects to reduce the coupling between adjacent cells.

It’s important to note that while these techniques can mitigate the sneak path problem to a certain extent, eliminating it entirely remains a challenge. The choice and effectiveness of the mitigation techniques depend on the specific characteristics and requirements of the resistive memory technology being used. Ongoing research and development efforts are focused on further improving these techniques and developing new approaches to enhance the reliability and performance of crossbar-based resistive memory.

4. Resistive switching memory fabrication process

The fabrication process of resistive random-access memory (RRAM) involves several key steps to create the memory cells and the necessary components. Here are the general steps involved in RRAM fabrication:

1. Substrate Preparation: The process begins with the preparation of a suitable substrate, typically a silicon wafer. The substrate is cleaned and prepared to provide a clean and uniform surface for subsequent layers.

2. Bottom Electrode Deposition: A thin layer of a conductive material is deposited on the substrate to serve as the bottom electrode of the RRAM cell. Commonly used materials include metals such as titanium (Ti), tungsten (W), or tantalum (Ta).

3. Insulator Deposition: A layer of insulating material is deposited on top of the bottom electrode. This insulator layer plays a crucial role in the resistive switching mechanism of RRAM. Common insulator materials used in RRAM include metal oxides such as titanium dioxide (TiO₂), hafnium oxide (HfO₂), or tantalum oxide (TaOx).

4. Top Electrode Deposition: A second conductive layer, known as the top electrode, is deposited on top of the insulator layer. The top electrode completes the RRAM cell structure. Similar to the bottom electrode, it is often made of a conductive material like titanium, tungsten, or tantalum for RRAM and Copper or Silver for CBRAM.

5. Patterning: Photolithography and etching techniques are used to define the size and shape of the RRAM cells. Photolithography involves the use of photoresist materials and exposure to UV light through a mask to transfer the desired pattern onto the material layers. Subsequent etching processes selectively remove material from specific areas, creating the desired electrode and insulator patterns.

6. Passivation Layer: A passivation layer is deposited on top of the RRAM cells to protect them from external contaminants and provide electrical insulation. This layer helps ensure the long-term stability and reliability of the RRAM devices.

7. Interconnect Formation: Metal interconnects are formed to connect the RRAM cells to the peripheral circuitry. These interconnects provide the necessary electrical connections for read, write, and control operations. Multiple layers of interconnects may be formed, and dielectric materials are used to isolate and insulate the interconnect layers.

8. Front-End and Back-End Processes: Additional steps include the front-end and back-end processes, which involve the integration of transistors, addressing circuits, and other peripheral components. These processes ensure the RRAM devices can be accessed and controlled effectively.

9. Testing and Packaging: After fabrication, the RRAM devices undergo rigorous testing to verify their functionality and performance. Once the devices pass the testing phase, they are packaged to protect them from environmental factors and to enable integration into larger electronic systems.

It’s important to note that the specific fabrication steps and techniques can vary depending on the RRAM technology, research advancements, and manufacturing processes employed by different manufacturers. The steps mentioned above provide a general overview of the RRAM fabrication process.

A typical fabrication process flow of oxygen vacancy based resistive memory or RRAM having structure of W/SiO_x/TiN reported by Roy et al. [15] is described in Figure 9 and the same for metallic filament based resistive memory of CBRAM having structure of Cu/Cu-Al alloy/Ta₂O₅/TiN reported by Roy et al. [56] is described in Figure 10 and the fabrication process of a typical 3 × 3 crossbar RRAM of the author [unpublished] having structure of Ir/SiO₂/W is shown below in Figure 11 respectively in below.

Figure 9.

Typical fabrication process flow step by step of a W/SiO_x/TiN 2D RRAM structure.

Figure 10.

Typical fabrication process flow step by step of a Cu/Cu-Al alloy/Ta₂O₅/TiN 2D CBRAM structure.

Figure 11.

Typical fabrication process flow step by step of a Ir/SiO₂/W 3D RRAM 3 × 3 Crosspoint structure.

The typical OEM image of a 2D RRAM/CBRAM devices is look like Figure 12 and a typical SEM image of a single cross point of 3D RRAM is look like Figure 13 below.

Figure 12.

Typical planer view of OEM image with top (left hand side) and bottom (right hand side) electrode pads of a 2D or through via hole (TVH) resistive memory device.

Figure 13.

Typical planer view of SEM image with top (vertical orientation) and bottom (horizontal orientation) electrode pads of a 3D or Crosspoint (single) resistive memory device [13].

5. Typical resistive memory electrical characterization and memory test

As described in Section 2.2 about Resistive Switching mechanism we already know that there are two types of resistive switching mechanism-a) Filamentary type and b) Interface type. Figures 14 and 15 showed the typical filamentary and interface type switching through pictorial illustration.

Figure 14.

Typical process diagram of filament formation and rupture of filamentary resistive switching mechanism [8].

Figure 15.

Typical process diagram of interface type resistive switching mechanism [28].

It is observed from Figure 14 that in Filamentary type switching when there is no bias is applied the initial state there will be no filamentary path formed inside the insulator layer sandwiched between the terminal electrodes. So the resistance of the insulator will be very high and we called this phase as Initial Resistance State or IRS. Now when first a bias is applied at top electrode a filamentary path will form which may be metallic path for CBRAM devices or may be oxygen vacancy mediated non-metallic filamentary path for RRAM devices. This process is called Formation and the voltage at which the filament forms is called as Formation Voltage. As the filamentary path form so there will be a flow of current through that percolating path inside the insulator switching material and the resistivity of the switching material decreases and that resistance state is known as Low Resistance State or LRS. Now at this point if the Top Electrode bias will be reversed then the filament from the electrode/switching material interfaces start to dissolve or rupture. And immediately the switching material’s resistance state again go high and we called that state as High Resistance State or HRS. As this process the complete filament not dissolve and part of it will remain back inside the switching material so the resistance state will not be as high as that during the initial state so HRS is lower than IRS. Now again when the polarity of bias applied at top electrode change then again filament form but this time to form filament lower voltage is required to apply. This voltage is called SET voltage and the voltage required to rupture the filament is called RESET voltage.

The ratio of HRS to LRS is known as Memory Window or Resistance Ratio or ON/OFF Ratio. This ratio is considered to be >10 for an ideal case but as high this ratio will be achieved the data storage capability will be more.

On the other hand for Interface type switching mechanism the transition between HRS to LRS will be mediated by oxygen vacancy concentration at Electrode-Switching material interfaces. If the concentration is high it is denoted as Higher Resistance State or HRS and if the concentration is low then it is called Lower Resistance State or LRS as shown in next page.

It is observed that LRS for filamentary type switching is independent of the device area where as HRS increases as the area decreases. On the other hand for interface type switching both LRS and HRS increases as the device area decreases.

Now in this chapter for memory characterization filamentary effects will be described mainly whereas for biosensor characterization interface type switching is described mainly. For memory characterization usually in general three types of electrical measurements are performed.

5.1 Dual-sweep IV characteristics

In this method a sweep voltage with a particular steps is applied across one of the terminal electrodes where keeping other electrode as grounded, and the current inside the switching material due to formation and rupture of filament are measured. A typical IV characteristics with formation cycle, set operation, reset operation, HRS, LRS, IRS is described in Figure 16. The current during formation cycle or the current reach after the device is SET is known as Compliance Current (CC). Whereas current at SET voltage is called Set or Program Current and the same for RESET voltage is called Reset or Erase Current, respectively.

Figure 16.

Typical current–voltage (IV) characteristics of filamentary type of resistive switching mechanism.

From the figure we can see a current is initially gradually increasing with voltage then at SET voltage it jumps suddenly to a high magnitude indicated by compliance current and then after increasing of voltage current got saturated. Now if voltage is reducing to zero the current initially for a long time remain in saturated level and then suddenly drops to zero. The region in IV curve between 0 V to Set voltage indicated by “1” is called HRS region where as the region between Set voltage to 0 V indicated by “2” is called LRS region.

Now again when the bias polarity is reversed the current will increase from 0 and now the direction of current is reversed. Now at Reset voltage this current suddenly drops to lower value same as HRS state current. This phase the filament got ruptured. As we find both under positive and negative bias polarity there are an exchange of current level with applied voltage between HRS and LRS so this IV sweep also called Bipolar Double Sweep Current Voltage resistive switching characteristics.

5.2 Data retention characteristics

Data retention means under the stress condition how long a memory device can hold the data. To measure this characteristics the device is initially kept in either LRS or HRS region. Then one voltage between 0 V and SET volt has been chosen as Read Voltage and for a longer period of time like 2 or 3 hour or several hours the device resistance state is measured with a time sample having a particular time gap like 2 min or 30 s etc. Then again the state of the device make changes and repeat the same thing for the new state. Thus if it is found that the device can hold the particular value of respective HRS and LRS state at read voltage then it is considered the device has good data retention capability. This measurement can be done at room temperature as well as higher temperature. One the typical data retention author had measured for a Pt/HfO₂/TiN based devices at 100°C is shown in next page (Figure 17).

Figure 17.

Typical data retention characteristics of filamentary type of resistive switching mechanism at 100°C temperature [11].

5.3 Endurance characteristics

Endurance characteristics measurement show that how the device can hold the data if a stress is applied continuously to the device inform of AC or DC voltage. Based on the type of stress signal applied across the device the Endurance characteristics is of two types.

5.3.1 DC endurance

In this process a representative DC IV sweep has been carried out and it is checked how long the memory window or ON/OFF ratio can hold the value >10. A typical 3000 repetitive cycles at 300 μA of Cu/Ir/TiNxOy/TiN CBRAM device reported by Dutta et al. [17] has been shown below (Figure 18).

Figure 18.

Typical 1000 representative DC cycles of a Cross-Point array at 50 mA compliance current [17].

5.3.2 Program-erase endurance/pulse endurance

This process instead of DC sweep a typical AC pulse voltage train as shown in Figure 19 is applied. During SET pulse the device is bring to ON state or programmed or data is written while during RESET pulse the device is bring to OFF set or data got erased. During read pulse data are read.

Figure 19.

Typical pulse format used as AC pulse for P/E endurance characteristics.

Now using similar kind of AC pulse a typical Pulse or Program Erase Endurance characteristics of RRAM device measured by author is (Figure 20).

Figure 20.

Typical pulse endurance measured for 3 different program current as 1 μA, 10 μA, and 50 μA by using -2V amplitude and 100 ns pulse width AC pulse for programming and +1 V amplitude and 100 ns pulse width AC pulse for erase [15].

6. Resistive memory potentiality as ultra sensitive biomarker devices

Resistive switching memory (ReRAM) is a type of non-volatile memory that stores information in the resistance state of a solid-state device. It has been primarily explored for its potential applications in data storage and computing. However, researchers have also investigated its use in biosensing applications, including as a biosensor [14, 15, 16, 17]. A biosensor is a device that detects and measures biological substances, such as proteins or DNA, by converting a biological response into an electrical signal. Resistive memory works by changing its resistance in response to an applied voltage or current. This change in resistance can be used to detect the presence of biological substances.

Using ReRAM as a biosensor involves leveraging its ability to change resistance in response to certain stimuli, such as the presence of specific molecules or biological substances. This property can be harnessed to detect and analyze various biological targets, including proteins, DNA, and other biomolecules.

In this section we explore elaborately the new potentiality of RRAM devices parallel to storage device as biomarker or biosensor devices. The basic principle behind using ReRAM as a biosensor is to functionalize the surface of the ReRAM device with a specific recognition element, such as antibodies, enzymes, or aptamers. These recognition elements are designed to selectively bind to the target molecules of interest. When the target molecules interact with the recognition elements on the ReRAM surface, it leads to a change in the resistance state of the device. This change in resistance can be measured and correlated with the concentration or presence of the target molecules.

One advantage of using ReRAM as a biosensor is its compatibility with integrated circuit technology. It can be fabricated using standard semiconductor fabrication processes, allowing for the integration of sensing and signal processing functionalities on a single chip. Additionally, ReRAM-based biosensors can offer label-free and real-time detection of biomolecules, making them potentially useful in various applications, such as medical diagnostics, environmental monitoring, and drug discovery.

However, it’s important to note that ReRAM-based biosensors are still an active area of research, and several challenges need to be addressed before their widespread adoption. These challenges include improving the sensitivity, selectivity, and stability of the devices, as well as optimizing the functionalization techniques for immobilizing recognition elements on the ReRAM surface.

In summary, ReRAM-based biosensors have the potential to provide label-free and real-time detection of biomolecules. While there is ongoing research in this area, further advancements are needed to fully exploit their capabilities and address existing challenges.

In the following sections I describe about some of the recent research works of using RRAM as biosensor to detect various bioanalytes and pH/H₂O₂ by applying the idea of oxidation-reduction or profanation-deprotonation [14].

The recent advanced research on various potentiality of RRAM has establish the idea that owing to its high sensitivity and specificity, resistive memory biosensors have the potential to revolutionize the way we monitor and diagnose diseases, detect pollutants, and ensure food safety.

One example of a biosensor is the glucose sensor used by diabetics to monitor their blood sugar levels. The sensor contains an enzyme that reacts with glucose in the blood, producing a signal that is converted into a reading of the glucose level. Another example is the biosensor used to detect E. coli bacteria in food samples. The biosensor contains antibodies that bind to the bacteria, producing a signal that indicates the presence of the pathogen.

Resistive memory biosensors have been used in various fields, including healthcare and environmental monitoring. In healthcare, they have been used to detect diseases such as cancer and Alzheimer’s by detecting specific biomarkers in blood samples. In environmental monitoring, they have been used to detect pollutants in air and water samples. The potential applications of resistive memory biosensors are vast and exciting, with the possibility of detecting a wide range of biological molecules in real-time.

6.1 Advantages of resistive memory as a biosensor

• One of the main advantages of using resistive memory as a biosensor is its high sensitivity. Resistive memory devices can detect very small changes in resistance, which allows them to detect even low concentrations of analytes. This makes them ideal for applications where high sensitivity is required, such as disease diagnosis and environmental monitoring.

• Another advantage of resistive memory biosensors is their low cost and ease of fabrication. Unlike some other types of biosensors, resistive memory devices can be fabricated using standard semiconductor processing techniques, which makes them relatively inexpensive to produce. Additionally, they can be integrated with other electronic components, which makes them easy to use and integrate into existing systems.

6.2 Applications of resistive memory biosensor

6.2.1 Healthcare sector

1. Resistive memory biosensors have the potential to revolutionize healthcare by providing fast, accurate, and non-invasive methods for disease detection and patient monitoring. One example of this is the use of resistive memory biosensors to detect glucose levels in diabetic patients. By measuring changes in resistance caused by glucose binding to the sensor surface, these biosensors can provide real-time monitoring of blood sugar levels without the need for painful finger pricks.

2. Another example is the use of resistive memory biosensors to detect biomarkers in cancer patients. These biosensors can detect minute concentrations of cancer-specific proteins in bodily fluids, allowing for earlier and more accurate diagnosis of the disease. They can also be used to monitor the effectiveness of cancer treatments by tracking changes in biomarker levels over time.

6.2.2 Environmental monitoring

Resistive memory biosensors are becoming increasingly popular in environmental monitoring due to their high sensitivity and specificity. These biosensors can detect pollutants such as heavy metals, pesticides, and organic compounds in air and water with great accuracy. One example of how resistive memory biosensors have been used in environmental monitoring is the detection of nitrogen dioxide (NO₂) in air. By using a resistive memory sensor coated with a thin layer of metal oxide, researchers were able to detect NO₂ levels as low as five parts per billion. This technology has the potential to greatly improve air quality monitoring in urban areas.

6.2.3 Agriculture sector

Resistive memory biosensors can revolutionize agriculture by detecting plant diseases and monitoring soil quality with high accuracy and precision. By measuring the electrical resistance of a conductive filament, these biosensors can detect changes in the chemical composition of soil or plant tissue that indicate the presence of disease or nutrient deficiencies. For example, resistive memory biosensors have been used to detect fungal infections in tomato plants by measuring the levels of specific volatile organic compounds emitted by the plant. This allows farmers to identify and treat the infection before it spreads to other plants. Additionally, resistive memory biosensors can be used to monitor soil moisture levels and nutrient content, providing farmers with real-time data on crop health and growth.

6.2.4 Food safety

Resistive memory biosensors have the potential to revolutionize food safety by detecting pathogens and monitoring food quality. By using resistive memory technology, biosensors can quickly and accurately detect harmful bacteria such as E. coli and Salmonella in food products, preventing outbreaks and ensuring consumer safety. In addition to pathogen detection, resistive memory biosensors can also monitor food quality by detecting changes in pH, temperature, and other factors that can affect food spoilage. This allows for early detection of spoilage and can help reduce food waste by ensuring that only safe and high-quality products are sold to consumers.

6.2.5 Security sector

Resistive memory biosensors have the potential to revolutionize security measures by detecting explosives and monitoring air quality in public spaces. These sensors can detect trace amounts of explosive materials, making them an invaluable tool for security personnel. They can also monitor air quality in real-time, alerting authorities to potential hazards before they become a threat. In addition to their use in traditional security settings, resistive memory biosensors can also be used in transportation systems such as airports and subway stations. By monitoring air quality and detecting explosives, these sensors can help prevent terrorist attacks and ensure the safety of travelers.

6.3 Measurement of sensitivity by using RRAM biosensors

In this section first I describe here that how the analytes can be detected in RRAM biosensor. The electrical pulse based sensitivity analysis in RRAM devices can be measured in two methods, one by Capacitance-Voltage measurement (CV) and another is Current-Voltage (IV) measurement. As IV-measurement require less sample volume and perform a fast response, also by this process RRAM filament formation by redox operation also can be more justified so in this section I discuss about the measurement process as well as the reports only based on IV measurement process.

Before the IV measurement first of all a Buffer solution (Tris Buffer) is prepared and then the enzyme and the bioanalyte solution are added to the solution to change the molecular concentration of the solution. Now all the cases there two kinds of solutions are prepared - normal buffer solution with pH 7.4 and the buffer-enzyme-analyte solution with modified pH. Usually the pH enhanced but sometimes it is also reduced depending on the nature of the analytes.

6.3.1 Solution preparation

First a Tris Base is needed to purchase then the TRIS base is dissolved in DI water according to its molecular weight. Then the buffer solution was prepared by adjusting pH value at 7.4 using an HCl solution. Afterward, enzyme followed by bioanalyte are dissolved in the TRIS-HCl (pH 7.4) buffer solution to make a stock solution using the chemical formula. Now addition of bioanalytes will change the concentration of buffer solution. Usually the pH of the analytes mixed buffer solution pH will vary from pH 2 to pH 12 depending on the nature of the analytes is acidic or basic.

6.3.2 IV-measurement

For sensitivity measurement initially the RRAM device is made to be in LRS state by SET the device. Then using micro-pipette put a droplet of buffer solution from stock at the junction of via point or cross-point for RRAM where top and bottom electrode pads merge as shown by device area with arrow in Figures 12 and 13.

Now in case of 2D RRAM it is find a better response with either Cu electrode for CBRAM [16, 17] and W [15] or Ir [15] or IrO_x electrode [12, 13, 14] for RRAM to act as a Sensing membrane. As Ir or IrO_x is porous in nature and Cu/W is oxidizing metal in nature so the sensing will be better. As in almost majority bioanalyte with buffer solution during application of voltage under the electrolysis operation decomposes and generates H₂O₂ which help to change the oxidation state of sensing membrane electrode and due to oxidation/reduction or protonation/deprotonation mechanism the shotkey barrier of metal/oxide interface changes and the current also changes. Now then with a short duration time a I-V-t sampling operation is undergone which show the shift of current due to change of pH concentration from buffer to bioanalyte mixed buffer solution which gives a value of normalized current and from normalized current the sensitivity can be calculated. Figure 21(a-f) below show the typical measurement process of I-Vt sampling of RRAM biosensor for both 2D and 3D resistive memory devices. Figure 21(a) describes real time image of the probe-station with the device placed on chuck and probes connected with SMUs dedicated for bias and ground connection during measurement. Figure 21(b) shows a typical OM image of a 2×2 μm² via-hole memory device , while Figure 21(c) illustrates the schematic diagram of RRAM memory structure also used for bio-sensing measurement. Figure 21(d) describes the images of the measurement system composed of B1500 and 16440A PGU, and Figure 21 demonstrates a typical real-time measurement screen shot of long P/E endurance of >10⁹ cycles with a small P/E pulse width of 100 ns. Finally, Figure 21(f) represents the preparation method of bio-sensing measurement composed of 1μL micropipette and three containers containing stocks of buffer solution, enzyme solution and bioanalyte solution (here creatinine). This setup had already elaborately described in my previous publication [15].

Figure 21.

Experimental setup for the measurement of memory and bio-sensor characteristics. (a) Real time image of the probe-station with the device placed on chuck and probes connected with SMUs dedicated for bias and ground connection during measurement is shown. (b) Typical OEM image of a 2×2 mm² via-hole RRAM memory device. (c) A schematic diagram of RRAM memory structure also used for bio-sensing measurement is described. (d) Images of measurement system composed of B1500 and 16440A PGU are given. (e) A typical real time measurement screen shot of long P/E endurance of >10⁹ cycles with a small P/E pulse width of 100 ns is achieved. (f) A preparation method of biosensing measurement composed of 1 mL micropipette and three containers containing stocks of buffer, buffer and 10 unit creatinine diminase enzyme solution, and 100 nM creatinine solutions are shown [15].

Figure 22 below describes the typical biosensor measurement for crossbar device. Now, bioanalytes reduces first the TE from higher to lower oxidation states (M^z+ → M⁰). For example, the Ir⁴⁺ changes to Ir3⁺/Ir⁰ oxidation state for Ir or IrO_x TE. At a higher concentration of bioanalyte so i.e.Ir⁰ state changes to Ir³⁺/Ir⁴⁺ (Ir⁰ ↔ Ir³⁺ + 3e⁻; Ir⁰ ↔ Ir⁴⁺ + 4e⁻). The oxidation-state increment leads to increased work function (Φ_SB) of TE metal. This oxidizing current is detected, which decreases with elevated bioanalytes concentration. Therefore, the Φ_SB value at the TE/SM interface increases gradually as well as the sensing current decreases with increasing bioanalytes concentration.

Figure 22.

A 3D illustration of typical experimental process of urea sensing through our fabricated cross-point structure which is connected to B1500 measurement unit to perform the electrical characteristics.

During this IV-t sampling the typical IV and change in shotkey barrier height and the oxidizing current changes due to pH variation from different concentration of bioanalytes with reference to buffer solution are shown in Figure 23.

Figure 23.

(a) Typical IV characteristics with pH7 (buffer) and pH10 (some bioanalyte mixed buffer solution) of RRAM biosensor (b) & (c) typical shotkey barrier height under positive and negative bias, respectively, and (d) IV-t sampling curve measured by keeping electrolyte solution with pH7 and pH10 at sensing membrane for 5s and measuring IV characteristics.

Now normalized current can be calculated by using the following formula

$$\mathit{\text{normalized current}}=\frac{I_c-I_0}{I_0}\begin{array}{l}{\mathbf{I}}_{\mathbf{0}}=\mathit{\text{Current\hspace{0.17em}change\hspace{0.17em}for\hspace{0.17em}buffer}}+\mathit{\text{enzyme}}\\ {\mathbf{I}}_{\mathbf{c}}=\mathit{\text{Current\hspace{0.17em}change\hspace{0.17em}for\hspace{0.17em}bioanalytes}}\end{array}$$E1

A typical normalized current plot can be drawn from the typical IV-t sampling curve as shown in Figure 24 by measuring the changes of final current bioanalyte with varying concentration and the typical curve will be looking as Figure 25.

Figure 24.

IV-t sampling curve shift from buffer solution reference with varying concentration of bioanalyte solution.

Figure 25.

A typical normalized current plot with varying creatinine concentration as reported in [15].

As most of the bioanalytes after electrolysis decomposed to generate H₂O₂, finally sensitivity can be calculated by the relation between H₂O₂ concentration (α) and normalized current shift (ΔI_N) which is purely exponential which can be obtained by linearly fitted Figure 23. And the obtained relation is

$$\alpha =exp\left[\frac{∆I}{a}-\frac{b}{a}\right]$$E2

where a is the intercept and b is the slope of linearly fitted curve at shown in Figure 25.

Now there are multiple reports on various bioanalyte sensing [12, 13, 14, 15, 16, 17, 76, 77, 78].

Table 6 shows the different RRAM structures with different analytes detection with lowest concentration.

Structure	Sensing Membrane	Bioanalyte	Detection Method	Lower Detection Level	Usefulness
IrOx/GdOx/W -CP RRAM [12]	IrOx	Urea	CV	1 mM	High level of Urea can cause Ulcers, Indigestion and several Renal Diseases, so early detection of Urea is required
IrOx/Al2O3/W -CP RRAM [13]	IrOx	Glucose	CV	10pM	Glucose detection is an important to prevent hyperglucemia and hypoglucemia
W/GeOx/W-2D RRAM [14]	W	Sarcosine	CV	50 pM	Sarcosine is indicator of Prostrate Cancer so low concentration detection can help for early prevention and treatment.
Ir/SiOx/TiN -2 [15]	Ir	Creatinine	IV-t sampling	100 nM	High level of Creatinine can cause High blood pressure, diabetes and Kidney failure and several other Renal Diseases, so early detection of Creatinine can help for early treatment.
Cu/AlOx/a-COx/TiNxOy/TiN-2D CBRAM [16]	Cu	Saliva	IV-t sampling	1 pM	Non invasive way Glucose detection from Saliva by using lower sample volume.
Cu/Ir/TiNxOy/TiN-2D CBRAM [17]	Cu & Ir	Tributyrin	IV-t sampling	1 pM	High concentration of Tributyrin may cause gastrointestinal disturbances, such as diarrhea, bloating, and stomach cramps etc. So early detection can prevent these.
IrOx/Al2O3/TaOx/MoS2/TiN-2D RRAM [76]	IrOx	Ascorbic Acid	IV-t sampling	1 pM	High concentration of Ascorbic Acid might cause side effects such as stomach cramps, nausea, heartburn, and headache. So early detection can prevent these.
Cu/MoS2/TiN-2D CBRAM [77]	Cu	Dopamine	IV-t sampling	10 pM	Nausea, vomiting, orthostatic hypotension, headache, dizziness, and cardiac arrhythmia are the most common side effect of dopamine, so early detection can prevent these.
W/Al2O3/TaOx/TiN-2D RRAM [78]	W	Dopamine	IV-t sampling	1 pM	Nausea, vomiting, orthostatic hypotension, headache, dizziness, and cardiac arrhythmia are the most common side effect of dopamine, so early detection can prevent these.

Table 6.

Resistive memory-based biosensors with different analyte detection with lower detected concentration.

Now based on the reports below show some of the equations of bioanalyte decomposition like urea, creatinine, Sarcosine and LOXL2. In these Creatinine and urea are very important for kidney disease identification where as Sarcosine and LOXL2 are important for prostrate and breast cancer detection.

$$N-\mathit{\text{Lysine}}+O_2+H_{2}O\stackrel{\mathit{\text{LOXL}}2}{\leftrightarrow }\mathit{\text{Allysine}}+{\mathit{NH}}_3+H_2{O}_2$$E3

$$\mathit{Sar}\mathit{cos}\mathit{ine}+O_2+H_{2}O\stackrel{{\mathit{SO}}_x}{\leftrightarrow }\mathit{\text{Formaldehyde}}+\mathit{\text{Glycin}}+H_2{O}_2$$E4

$$\mathit{\text{Urea}}+H_{2}O\stackrel{\mathit{\text{Urease}}}{\leftrightarrow }2{\mathit{NH}}_4^{+}+{\mathit{CO}}_2+2{\mathit{OH}}^{-}$$E5

$$\mathit{\text{creatinime}}+H_{2}O\stackrel{\begin{array}{l}\mathit{\text{creatimine}}\\ \mathit{\text{deaminase}}\end{array}}{\leftrightarrow }N-\mathit{\text{methylthydantion}}+{\mathit{NH}}_4^{+}+{\mathit{OH}}^{-}$$E6

The typical sensitivity plot of urea sensing by using the formula in (2) is shown below (Figure 26).

Figure 26.

A typical normalized current plot with varying urea concentration by linearly fitting the curve as per shown in Figure 25 to calculate the sensitivity.

The lowest sensitivity will be denoted by the corresponding bioanalyte concentration values for which the normalized current of the bioanalyte will be minimum from the curve.

6.4 Challenges of resistive memory as a biosensor

One of the main challenges of using resistive memory as a biosensor is achieving high sensitivity and selectivity. While resistive memory biosensors have shown promising results in detecting biomolecules, there is still a need to improve their performance in terms of accuracy and reliability. Researchers are currently exploring various approaches to enhance the sensing capabilities of resistive memory biosensors, such as incorporating functionalized nanomaterials or optimizing the device architecture.

Another challenge is the integration of resistive memory biosensors into practical applications. The fabrication process of resistive memory devices can be complex and costly, which limits their scalability and commercialization. To address this issue, researchers are investigating new materials and manufacturing techniques that can simplify the production process and reduce the cost of resistive memory biosensors.

6.5 Future of resistive memory as a biosensor

The future of resistive memory biosensors is exciting and full of potential. As technology advances, we can expect to see these biosensors being used in a variety of fields, including agriculture, food safety, and security. In agriculture, resistive memory biosensors can be used to detect plant diseases and monitor soil quality. This can lead to more efficient and sustainable farming practices, ultimately increasing crop yields and reducing the use of harmful pesticides.

In the field of food safety, resistive memory biosensors can be used to detect pathogens and monitor food quality. This can help prevent food borne illnesses and ensure that consumers are getting safe and high-quality products. In terms of security, resistive memory biosensors have the potential to be used for detecting explosives and monitoring air quality in public spaces. This can help keep people safe and prevent dangerous situations from occurring.

Overall, the future of resistive memory biosensors is bright and full of possibilities. With continued research and development, we can expect to see these biosensors being used in even more innovative ways in the years to come.

7. Conclusion

In conclusion, resistive memory can be found to have a completely new are of application than its conventional high density memory utilization is acting as biosensors. RRAM biosensors have the potential to revolutionize various fields, including healthcare, environmental monitoring, agriculture, food safety, security, energy, and biomedical research. By using conductive filaments to detect changes in electrical resistance, these biosensors offer high sensitivity, specificity, and selectivity compared to other types of biosensors. They also have the advantage of being low-cost, label-free, and easy to fabricate.

However, there are still challenges that need to be addressed, such as improving the stability and reproducibility of the conductive filaments, optimizing the sensing performance, and reducing the interference from external factors. Nevertheless, with ongoing research and development, resistive memory biosensors hold great promise for enhancing our ability to detect and monitor various biological and environmental parameters.