Exploration and mitigation of protrusion behavior in Ga-ion doped h-BN memristors

Mucun Li, Enxiu Wu,,2 Linyan Xu,,a) Xiaodong Hu, and Xiaopu Miao

AFFILIATIONS 1 State Key Laboratory of Precision Measurement Technology and Instruments,School of Precision Instruments and Opto-electronics Engineering,Tianjin University,No.92 Weijin Road,Tianjin 300072,People’s Republic of China

2State Key Laboratory of Transducer Technology,Institute of Electronics,Chinese Academy of Sciences,Beijing 100190,People’s Republic of China

ABSTRACT Using hexagonal boron nitride(h-BN)to prepare resistive switching devices is a promising strategy.Various doping methods have aroused great interest in the semiconductor fiel in recent years,but many researchers have overlooked the various repetitive anomalies that occur during the testing process.In this study,the basic electrical properties and additive protrusion behavior of Ga-ion-doped h-BN memristors at micro–nanoscale during the voltage scanning process are investigated via atomic force microscopy(AFM)and energy dispersive spectroscopy.The additive protrusion behavior is subjected to exploratory research,and it is concluded that it is caused by anodic oxidation.An approach is proposed that involves fillin the AFM chamber with nitrogen gas to improve the stability of memristor testing,and this method provides a solution for enhanced testing stability of memristors.

KEYWORDS Hexagonal boron nitride,Memristor,AFM,Anodic oxidation

I.INTRODUCTION

The von Neumann architecture is used widely in computers,but it limits significantl the information exchange speed between the central processing unit (CPU) and memory because of factors such as CPU access speed,storage element response speed,information transfer speed,and memory structure and performance.1–3Instead,researchers have proposed an integrated storage–computing architecture that can fundamentally solve the problem of information storage and processing,breaking through the von Neumann bottleneck to some extent and improving the system performance of storage chips.4

As a fourth-generation basic circuit element,the memristor was firs proposed by Chua in 1971.5The device has a metal–insulator–metal (MIM) sandwich structure,and the resistance state can be changed by adjusting the voltage between the metal electrodes.The device has resistance switching characteristics,and the resistance state is nonvolatile.In the MIM structure,the insulating layer is the most important part,for which more and more 2D materials are being used in MIM devices.6–9In the study reported herein,hexagonal boron nitride (h-BN)—which is a material akin to graphene—was selected for the dielectric layer;this material has a bandgap of 5.5 eV and a dielectric constant of 3,making it a promising dielectric layer for 2D electronic devices.10,11

Traditional mechanical exfoliation produces h-BN dielectric layers with almost no memristive effect in MIM memristor devices,but rather a long-term high-resistance state.To improve the performance of memristors,Wuet al.12proposed an Al doping process to modify HfO2memristors,Yuet al.13proposed a nitrogen doping process to modify TiO2memristors,and Genget al.14proposed a Ga+doping strategy to modify h-BN memristors.Also,it has been reported that during micro–nanoscale testing of memristors,bumps may occur via dielectric breakdown-induced epitaxial growth.15

In the study reported herein,we used Ga+injection to modify h-BN by doping,and we constructed a structure with an atomic force microscopy (AFM) probe as the top electrode,doped h-BN as the dielectric layer,and Au as the bottom electrode for electrical characteristic testing.More importantly,we investigated the bumps that occurred during the voltage testing period,and we performed elemental analysis using energy dispersive spectroscopy(EDS).We designed an experiment involving nitrogen filling which offered improvement by avoiding bumps,and we conclude that reducing the content of oxygen and water vapor in the measurement environment can avoid device failure due to bumps.

II.EXPERIMENTAL WORK

A.Fabrication of MIM structure

In the MIM sandwich structure shown in Fig.1,the conductive probe of the AFM serves as the top electrode,and the dielectric layer and bottom electrode were fabricated as follows.First,a 10-nm-thick layer of Au was deposited as the bottom electrode on a surface of SiO2(285 nm)/Si using electron beam evaporation at a deposition rate of 0.1 Å/s.Next,the h-BN flake (purchased from Shenzhen Six Carbon Technology Co.,Ltd.,China) were mechanically exfoliated and transferred onto a polydimethylsiloxane (PDMS) stamp.With the assistance of an optical microscope(Eclipse LV150N;Nikon),the material was aligned and transferred onto the bottom Au electrode via the PDMS stamp.During testing,the bottom Au electrode was connected to the sample stage using conductive silver paste,forming a loop with the AFM conductive probe.This completed the construction of the top electrode,dielectric layer,and bottom electrode in the MIM device.

B.Electrical measurement via AFM

The Kelvin probe force microscopy (KPFM) mode in a commercial AFM (Dimension Icon;Bruker) was used to measure the surface potential changes of the material before and after Ga+beam doping.The KPFM mode can be used to measure surface potential at the nanoscale with high resolution and is a quantitative measurement method.The contact potential difference signal at the position of h-BN before and after Ga+doping reflect the work function before and after processing,and based on that difference,the built-in potential of the ion-doped region was obtained.

Meanwhile,PeakForce Tunneling AFM (PF-TUNA) was used to characterize the electrical properties of the material at micro–nanoscale,allowing the conduction mechanism of the memristor to be discussed.The working principle of PF-TUNA is shown in Fig.1;it is based on PeakForce Tapping,with the PF-TUNA module incorporated into the AFM measurement system.The conductive AFM probe used was MESP-V2 (with an elastic modulus ofk=3 N/m and a resonance frequency off0=75 kHz),which was coated with CoCr alloy on both sides for electrical conduction.While the probe was in contact with the sample,a certain voltage could be applied to the conductive probe to form a circuit among the sample stage,the sample,the conductive probe,the probe holder,and the measurement system to obtain the surface electrical information of the sample during morphology measurement.The electrical characteristics were derived from two aspects:(i)the current in the entire doped region was measured with a certain voltage;(ii)theI–Vcharacteristics of the device were measured via point-by-point scanning in the doped region.

C.Element analysis via EDS

The sample was placed in the chamber of the scanning electron microscopy(SEM)/EDS instrument,and the latter was evacuated to a pressure of 10-3mbar.The working distance was set to 10 mm,with a voltage of 1 kV and a beam current of 0.2 nA.The sample was then brought into focus,and areas with and without protrusions were identifie for EDS analysis,after which the elemental changes between those areas were observed.

III.RESULTS AND DISCUSSION

A.Generation of build-in potential

The surface morphology changes of the sample after Ga+doping are shown in Fig.2(a).Cross-sectional analysis revealed that a 1.4-nm depression appeared on the sample surface after Ga+doping,and a protrusion was produced at the edge of the treatment area.Simultaneously,surface potential measurement was performed as shown in Fig.2(b),where the doping area is in red and the intrinsic area is in green.By comparison,the built-in potential induced by ion beam doping can be estimated to be～-434.6 mV.

B.Measurement of electrical characteristics

After completing the surface potential measurements,to avoid the variables that may have arisen during the measurement process,we selected a new area for Ga+doping.The focused-ionbeam doping parameters remained unchanged,but the doping area was reduced to 5 × 5 μm2.Figure 3(a) shows an AFM morphology image of the doped area;the center point of the image was selected as the testing point,and 25 single-point scanning measurements were performed there.As shown in Figs.3(c)–3(e),the current changed suddenly at 6 V during the firs scan,indicating that the device switched from high to low resistance.During the second scan,the point of sudden change moved to 3.5 V,making it easier for the device to switch from high to low resistance.In the third and subsequent scans,the points of sudden change were all at 3–4 V.Figure 3(b)shows an AFM morphology image of the device after the 25 single-point scanning measurements.There was clearly a protrusion at the scanning point,the cause of which remains to be investigated,which we do in the subsequent sections.

FIG.1.Electrical testing system based on PeakForce Tunneling AFM(PF-TUNA).

FIG.2.(a)Surface morphology and(b)surface potential of hexagonal boron nitride(h-BN)after Ga+doping.

FIG.3.Results of PF-TUNA measurements in Ga+doped area:(a)AFM topography before single-point ramp measurement;(b)AFM topography after 25 single-point ramp measurements;(c)-(e)I-V curves obtained during first three ramp measurements.

Meanwhile,a device with the same h-BN/Au/SiO2/Si substrate was also fabricated,an ion beam was implanted on a 3 × 3 μm2area of the h-BN with the same doping parameters,and a global PF-TUNA constant-voltage scan was performed with a scanning voltage of 10 V.Figure 4(a)shows the AFM surface morphology of the doped region,Fig.4(b)shows the results of the global PF-TUNA constant-voltage scan,and Fig.4(c) shows the voltage scanning results in three dimensions.The scanning results show many current points in the doped area,which confirm the conduction mechanism of a memristor,namely the theory of conductive filaments In the high-current signal area,several conductive filament were formed that participated in conduction on the device,and we conclude that many conductive nano-filament constituted the conduction mechanism of the memristor at the device level.As shown in Fig.4(c),the conductive nano-filament were formed under the electric fiel generated by the voltage in the insulator layer of the MIM device,and the conductive state of the device was changed by the formation of these conductive nano-filaments

FIG.4.(a)AFM morphology,(b)constant-voltage scanning test results in ion-beam-doped area via PF-TUNA mode,and(c)3D current map of(b).

FIG.5.(a)Variation of surface height of material additive protrusion and(b)AFM morphology of protrusion generation process with increasing number of ramp scans.

C.Discovery and exploration of protrusion behavior

In Sec.III B,we noted that the Ga+-doped device underwent surface morphology changes after single-point ramp voltage scanning,which manifested as a protrusion at the scanning point.To investigate this phenomenon,we conducted several experiments and tested various hypotheses to fin a method to eliminate the effect of the protrusion.As shown in Figs.5(a)and 5(b),AFM surface morphology characterization revealed that there was a certain height of protrusion on the surface of the material after 25 single-point ramp scans in the Ga+doping region.We speculated whether the height of the protrusion was related to the number of single-point ramp scans,and we conducted the following exploratory experiments.

TABLE I.Change of height of material surface with each ramp sweep.

As shown in Fig.5,20 single-point ramp scans were performed in the center of the target area with a scanning voltage of 10 V,and based on those AFM results,the height of the additive protrusion did increase with the number of scan voltage cycles.Table I gives the results for the protrusion height and its increment with the number of ramp scans.As indicated,the height of the material generally increased with more ramp scans,and after 20 single-point ramp scans,the protrusion height of the material reached 7.9 nm.Based on these data,we conclude that there is an additive and protrusive effect on the surface of the material after Ga+doping when applying a voltage scan.According to the literature,the protrusion may be due to two factors:(i) the induced epitaxial growth of the material under the electric field (ii) the anodic oxidation processing of the AFM probe.

To investigate the underlying cause of the surface protrusions,EDS material analysis was performed with a 5 × 5 μm2square on the h-BN/Au/SiO2/Si substrate chosen as the Ga+injection region.Figure 6(a)shows the AFM surface morphology after Ga+injection.To generate new protrusions on a larger scale,we selected four small areas in the ion injection region and performed constant-pressure scans of 800×800 nm2on each of them.As shown in Fig.6(b),these four small areas exhibited protrusions of varying heights,making them suitable test samples for subsequent elemental analysis,with the small region in the upper left corner selected as the object for EDS analysis.In the SEM/EDS instrument,as shown in Fig.7,the testing area was selected inside and outside the protrusion for EDS elemental analysis,and the analysis results are shown in Fig.8.

Based on the test results shown in Fig.8,it was found that the proportion of oxygen increased from 22.06%to 24.86%and the proportion of nitrogen decreased from 9.79% to 8.33% after formation of the protrusion.According to the chemical reaction for high-temperature oxidation of BN,i.e.,

FIG.6.AFM surface topography of h-BN material after ion-beam doping injection:(a)before and(b)after constant-voltage scanning in divided region.

FIG.7.EDS measurement:selected area is(a)inside and(b)outside the protrusion.

FIG.8.EDS test results: (a) inside protrusion area;(b) outside protrusion area.

it is possible to analogize the oxidation–reduction that occurs when the probe contacts the sample at the micro–nanoscale.Regarding the products in Eq.(1),two nitrogen atoms escape from BN in the form of NO2gas,and six oxygen atoms remain in the product B2O3of BN;therefore,the proportion of nitrogen decreases,and the proportion of oxygen increases.This is consistent with our EDS results,from which we conclude that the protrusions generated during the scanning process of Ga+-doped h-BN result from the anodic oxidation processing of h-BN.To avoid this surface morphology change,we designed an experiment using nitrogen gas as introduced in Sec.III D.

D.Analysis and elimination of protrusion behavior

In Sec.III C,we noted that the reason for the formation of protrusions on the material surface is anodic oxidation processing;when the voltage between the probe and the sample stage is applied to the sample surface,an oxidation reaction occurs.Therefore,to avoid the device’s inability to reset due to these protrusions,we conducted an experiment in the AFM chamber fille with nitrogen,in which there was a built-in thermometer and hygrometer connected by a rubber hose.

The chosen sample was still Ga+-doped h-BN/Au/SiO2/Si,with the Au connected to the sample stage through conductive silver glue.Regarding the environmental conditions,the room temperature was 24°C and the humidity inside the chamber was 51%,and four single-point ramp scanning experiments were conducted on the Ga+-doped modifie region under these conditions.Figure 9 shows the AFM morphology after scanning,with several obvious protrusions in the center.

FIG.9.AFM topography image after single-point ramp scanning test before nitrogen injection.

FIG.10.AFM topography images after multiple single-point ramp scans.

Next,the chamber door was closed,and the nitrogen valve was opened to adjust the nitrogen pressure to 40 MPa.The nitrogen fillin lasted for 15 min,but because of experimental limitations,we could only judge the nitrogen content in the chamber based on the humidity change therein,with the humidity inside the chamber observed to drop from 50% to 10%.The nitrogen valve was then closed to avoid vibration interference during AFM measurement.Under these conditions,multiple single-point ramp experiments were conducted on other areas of Ga+-doped samples,and the experimental results are shown in Fig.10.In the center of the scanning area,after 23 single-point ramp scans,the surface morphology remained unchanged,and after 19 additional scans,the surface morphology remained the same.In an environment fille with nitrogen,the probability of anodic oxidation behavior is minimal because of the reduced oxygen content.Therefore,fillin the AFM environment with nitrogen not only avoids the formation of protrusions during scanning but also confirm that the protrusion behavior is due to the anodic oxidation between the probe and the material.

IV.CONCLUSIONS

Herein,we presented a Ga+doping method for h-BN memristors.We constructed a MIM sandwich structure using an AFM conductive probe as the top electrode,h-BN as the dielectric layer,and Au as the bottom electrode,and we tested its basic resistanceswitching characteristics.Also,we observed the protrusion behavior of the resistive switching device during the ramp voltage scanning process,and we performed elemental analysis on the protrusions caused by this behavior.The results showed that this protrusion behavior is due to an anodic oxidation process and can be eliminated in an environment with a high concentration of nitrogen gas and low humidity,thereby providing a stable electrical measurement environment for resistive switching devices during the electrical measurement process.

ACKNOWLEDGMENTS

The authors thank engineers at Bruker for helpful discussions about test methods related to this work.The research was supported by the Youth Fund of the National Natural Science Foundation of China(Grant No.622041701004267).

AUTHOR DECLARATIONS

Conflict of Interest

The authors have no conflict to disclose.

DATA AVAILABILITY

The data that support the finding of this study are available from the corresponding author upon reasonable request.