Introduction

Resistive Random Access Memory (RRAM)1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23 is the highly promising candidate for the next generation nonvolatile memory (NVM), because conventional charge-based memories, namely dynamic random access memory and flash memory, have too low capacitance after continuously downscaling into 1X-nm regimes. In addition, an RRAM array can be fabricated in the back end of line of a complementary metal-oxide-semiconductor circuit, which makes such device an excellent candidate for embedded NVM (eNVM) application. The typical write speed of RRAM device ranges from 100 ns to 1 μs, which is three-to-four orders of magnitude faster than flash memory. Such high-speed and process-compatible eNVM can enable hardware technologies such as artificial intelligence and neuromorphic computing1,6,7,8.

The charge transport mechanism of RRAM, however, is not fully understood, and it is generally attributed to charge transport in metallic filament because of its metal–insulator–metal (MIM) structure, where the insulator is usually formed by metal oxide–based dielectric. Previously we pioneered nonmetal GeOx RRAM, but the metal electrodes used might have contributed to the charge transport mechanism16,17,18,19,20,21. In the paper, we report the all nonmetal RRAM that does not contain any metal in both the electrodes and dielectric insulator. The purpose of all nonmetal RRAM device is to provide a different charge transport mechanism rather than the charge transport in normal metallic filament. Relatively small device variation and tight distribution can be reached in similar GeOx RRAM16 that are crucial for array design17. After forming the RRAM device under 6 V and 100 μA current compliance, a large resistance window of 1.9 × 104 was measured at room temperature (RT), which decreases slightly to 8.7 × 103 after 104 s data retention. The set/reset charge transport for low- and high-resistance states (LRS and HRS), deduced from the measured current–voltage (I–V) characteristics, is attributed to the charge transport mechanism by Shklovskii-Efros (S-E) percolation model.

Results

Figure 1 depicts the measured I-V characteristics of an N+-Si/SiOx/P+-Si RRAM device. During the forming step, the device was first subjected to a 6 V and 100 μA compliance current stress to attain the LRS. The same device was reset into HRS after a negative voltage bias. Then, the device was set to LRS again under a positive voltage bias. However, the positive set voltage was lower than the forming voltage once the RRAM switching function was established.

Figure 1
figure 1

I-V characteristics of N+-Si/SiOx/P+-Si RRAM device under forming, set and reset operations.

Full size image

The charge transport mechanism is crucial for RRAM devices. To understand the charge transport mechanism in this completely nonmetal RRAM, we further analyzed the measured I-V curves at different temperatures. Figure 2(a–c) depict the measured and modeled I-V curves in the virgin state (VS), HRS and LRS conditions, respectively. All state the HRS and LRS currents adhere to the Shklovskii-Efros (S-E) percolation model:

$$I={I}_{0}\,\exp (-\frac{{W}_{e}-{(Ce\frac{U}{d}a{V}_{0}^{\gamma })}^{\frac{1}{1+\gamma }}}{kT}),$$
(1)

where I0, We, a, V0, C and ɣ are the preexponential factor, percolation energy, space scale of fluctuations, energy fluctuation amplitude, numeric constant and it is equal to 0.25, critical index and it is equal to 0.9, respectively. The simulation by the S-E model gives reasonable model parameters to all resistance state (Fig. 2). The percolation energy decreases with decreasing resistance. Also, in the S-E model for LRS, the active contact area reduction of the charge involved in the transport is taken into account. The relation a × V00.52 = 1 × 10−7 cm·eV0.52 does not change from resistance to resistance. This is due to the fact that with decreasing resistance increases space scale of fluctuations a but decreases energy fluctuation amplitude V0. In addition, it can be said that the S-E percolation model is applicable to the LRS case, then it can be assumed that the conducting channel is not continuous. Hence, the results demonstrate that the charge transport of the N+-Si/SiOx/P+-Si RRAM in VS, HRS and LRS are described by the S-E percolation model. For more details on other models and their inapplicability to HRS, see the ref.24.

Figure 2
figure 2

I-V dependences of (a) VS, (b) HRS and (c) LRS currents of N+-Si/SiOx/P+-Si RRAM and fitting curves of S-E model.

Full size image

To further understand the device characteristics, material analyses were performed. Figure 3 displays the cross-sectional transmission electron microscope (TEM) image of this RRAM device. As depicted, the RRAM device was fabricated directly on a P+-Si substrate, followed by a 15-nm thick SiOx dielectric layer and a N+-Si top electrode. The SiOx layer was further analyzed using X-ray photoelectron spectroscopy (XPS). The sample surface was pre-sputtered to ensure that the native oxide did not influence the measurements. Figure 4 displays the XPS spectrum. From the peaks of O1s, Si2s, and Si2p, the mole fraction x in SiOx was determined to be 0.62. Because no metal or metallic ions were present in the whole RRAM device, metallic filaments were not formed13,14,15,16. In accordance with XPS experimental data certain fraction of vacancies exist in dielectric immediately after synthesis. The migration of oxygen vacancies plays an important role for current conduction.

Figure 3
figure 3

TEM image of N+-Si/SiOx/P+-Si RRAM devices.

Full size image
Figure 4
figure 4

XPS spectrum of SiOx layer.

Full size image

Figure 5 plots potential switching mechanisms. During the forming step, the current conducted through the initial Vo2+ inside the SiOx layer14. When the RRAM device was under sufficiently high positive voltage, soft breakdown in SiOx occurred and disrupted the covalent bonds25,26, generating unbonded Si ions, O2− and Vo2+. In refs.27,28, it was shown that Vo2+ and O2− anti-Frenkel pairs are not stable and should recombine immediately. We assume, that after generation of anti-Frenkel pairs, electrons are redistributed to maintain charge neutrality and new oxygen vacancies (Vo0) and interstitial oxygen atoms are formed29. Because the atomic size of O is significantly smaller than Si, the interstitial oxygen atoms and Vo0 could migrate inside SiOx under the applied electric field. At the end of the forming process, the interstitial oxygen atoms were attracted to the positive voltage and accumulated at the interface of top N+I junction. Once the conduction path was formed, electrons could transport through the Vo0 creating the LRS current pass in the SiOx layer. After application of a negative voltage, interstitial oxygen atoms moved away from the top N+I junction and recombined with Vo0 to rupture the conduction path- the reset process. After a positive voltage was applied again, the set process behaved as the forming process to form a conduction path, but under a lower positive voltage than the forming voltage due to not all generating in forming process Vo0 recombined in reset process.

Figure 5
figure 5

Potential microscopic conduction of N+-Si/SiOx/P+-Si RRAM device.

Full size image

Data retention is the necessary characteristics for NVM, and they are related to the nonvolatile behaviour and lifetime of an RRAM device. Figure 6 depicts the retention characteristics of the N+-Si/SiOx/P+-Si RRAM device. The completely nonmetal RRAM device could achieve favourable retention with a slight resistive window decay from 1.9 × 104 to 8.7 × 103 at RT and 3.6 × 103 to 1.2 × 103 at 85 °C after 104 s retention.

Figure 6
figure 6

Retention characteristics of N+-Si/SiOx/P+-Si RRAM devices at RT and 85 °C.

Full size image

Conclusion

A completely nonmetal RRAM device was demonstrated for the first time. A large resistance window of 1.9 × 104 at RT was measured. An excellent retention resistance window of 1.2 × 103 was obtained for 104 s retention at 85 °C. In addition, the charge transport of the N+-Si/SiOx/P+-Si RRAM in VS, HRS and LRS are described by the S-E percolation model. And the Vo0 migration played an important role in the set/reset functions.

Methods

The RRAM device was made on a highly doped P+-Si substrate with a resistance lower than 0.01 Ω per square, which was also used as a bottom electrode. After standard RCA clean, the native oxide on P+-Si wafer was removed by a dilute hydrofluoric (HF) acid (HF: H2O = 1:100) solution for 60 sec. Then, a 15-nm-thick SiOx was deposited by reactive sputtering. The composition ratio inside the SiOx was determined using XPS. Then, a 15-nm-thick amorphous N+-Si layer, was formed as the top junction electrode. The diameter of the fabricated device was 120 μm. The I-V characteristics was measured using an HP4155B parameter analyzer. The voltage was applied on the N+-Si (top electrode) side and P+-Si (bottom electrode) were grounded. The sweep rate is 0.5 V/s. A Thermo K-alpha system with an X-ray spot size of 400 μm was employed for XPS measurements. The cross-sectional image of the RRAM device was measured using a JEOL 2010F high-resolution TEM. The modeled data for HRS and LRS were fitted under positive and negative voltage bias, respectively.