Detection of bovine serum albumin using hybrid TiO2 + graphene oxide based Bio – resistive random access memory device


Bio – molecules detection and their quantification with a high precision is essential in modern era of medical diagnostics. In this context, the memristor device which can change its resistance state is a promising technique to sense the bio - molecules. In this work, detection of the Bovine Serum Albumin (BSA) protein using resistive switching memristors based on TiO2 and TiO2 + graphene oxide (GO) is explored. The sensitivity of BSA detection is found to be 4 mg/mL. Both the devices show an excellent bipolar resistive switching with an on/off ratio of 73 and 100 respectively, which essentially demonstrates that the device with GO, distinguishes the resistance states with a high precision. The enhanced performance in the GO inserted device (~ 650 cycles) is attributed to the prevention of multi-dimensional and random growth of conductive paths.


Indeed, resistive switching (RS) memristive devices and bio – resistive random access memory (RRAM) devices1,2 have the potential to replace the landscape of existing electronic devices pertinent to memory3, logic4 and sensing5 applications. Essentially, the functionality of memristive device can be attributed to a state of alteration and dynamics of the same can be engineered to a target specific application. Particularly when it comes for sensing, there has been a great interest in nano bio – sensing due to the possibility for minimal invasive, therapy personalization and real time monitoring.

Among various memristive devices6,7,8, RS memory device has gained much interest owing to its simple structure, easy fabrication, high density, excellent stability and low power consumption9,10,11. The responsible physical phenomenon that governs in memristive RS devices has been believed as migration of oxygen ions which modifies the resistive properties of the devices in a non – volatile manner. In general RS device consist two-terminals and the active material (typically transition metal oxide) would be sandwiched between two metal electrodes. Resistance of such a device can be switched between high resistance state (HRS) and low resistance state (LRS) by applying an electric field across the electrodes. If positive and negative polarities are mandatory to have SET (HRS to LRS) and RESET (LRS to HRS) switching, such RS can be termed as bipolar resistive switching (BRS). On the other hand, in unipolar RS (URS), one polarity is sufficient for SET and RESET switching. These memristive devices have many advanced applications such as neuromorphic systems12,13, memory logic14,15 and analogue computing16. Bio molecules such as proteins, DNA, enzymes, bioactive small molecules, dopamine and uric acid have been detected using nanomaterials like gold nanoparticles17,18, graphene19,20, carbon nanotubes21,22, nanowires23 and quantum dots24.

Among all the available proteins, bovine serum albumin (BSA) has been identified as multifunctional one and plays an important role in delivering the fatty acid/amino acid etc. Essentially, BSA can interact with many organic, inorganic molecules, drugs, ionic metals and radicals due to multiple binding sites on exposed surface of molecule apart from its biomedical application such as targeted drug delivery25,26. On top of that it consists of ample biochemical applications including enzyme – linked immunosorbent assay, immunoblots, and immunohistochemistry. It has been believed that reduction of BSA may cause various diseases in cattle families. In this respect, the quantification of BSA is very much crucial in order to understand the severity of particular disease. Some methods have been explored to detect the BSA such as triangular silver nanoplates using spectrophotometric method27, optical fiber based on a Mach-Zehnder interferometer etc28. However, these state-of-the-art technologies of detecting BSA protein by optical methods are cumbersome, labor intensive, which demands an alternative approach with environmental friendly, non-toxic and easy detection method. Hence, it is important to explore detection of BSA using TiO2 based RS memristor due to its low operational voltage and with high sensitivity29. Upon adding BSA protein to TiO2 RS memristor device, multiple binding sites of BSA (R-COO− and R-NH+3) may interact with predominant surface groups of TiO2 (Ti2 = O−, Ti-OH, Ti2 = OH), which may lead to drastic change in transport/optical properties of the TiO2 memristor device. On top of that it has been proved that graphene oxide (GO) consists of oxygen containing functional groups30. As a result of this, the conductivity/optical properties of the device may enhance by introducing GO as a sandwich layer between BSA and TiO2. Salient features of present work are (a) detection and confirmation of BSA using TiO2 and TiO2 + GO based RS memristor devices (b) increased sensitivity in a device with GO (c) excellent on/off ratio ~100 (d) achieving multi-bit data storage by controlling the compliance current (e) confirmation of filamentary kind of switching using conductive surface atomic force microscope (C – AFM).

Results and Discussion

Figure 1a,b depict schematic of the devices that we used to perform I–V characteristics. Figure 1a consists the configuration as Ag/BSA/TiO2/FTO (TB). Here, fluorine doped tin oxide (FTO) was used as bottom electrode and Ag as top electrode. Initially TiO2 was deposited on top of FTO substrate by drop casting method and subsequently BSA was added on TiO2 for the possible detection of it. The concentration of BSA was varied from 0–15 mg/mL (4 mg/mL, 6 mg/mL, 8 mg/mL and 15 mg/mL respectively). I–V characteristics were carried out using two probe method by Keithley 2400 source/sense meter. In all the devices that we discuss in this paper, the current was sent perpendicular to the plane of the film (CPP) configuration. Figure 1b depicts the device with configuration Ag/BSA/GO/TiO2/FTO (TGB). The importance of graphene oxide layer (GO) is to enhance the detection sensitivity of the BSA with lower concentration. Results those demonstrated in this manuscript are reproducible and also tested with various devices.

Figure 1

Schematic of the device on which I–V measurements were performed using two probe method with a Keithley 2400 Source meter. (a) Consists of the configuration as Ag/BSA/TiO2/FTO (TB). Here, fluorine doped tin oxide is used as bottom electrode and Ag as top electrode. Initially TiO2 is deposited on top of FTO and then BSA protein was added to it. Current perpendicular to the plane configuration was used to perform I–V characteristics by keeping electrical probes one on FTO and other one on Ag pad. (b) To enhance the detection sensitivity of low concentration of BSA, a GO layer was introduced between TiO2 and BSA due to which device configuration was changed to Ag/BSA/GO/TiO2/FTO (TGB). (c) TEM image of TiO2 + GO nanoparticles measuring about 120 nm. Inset shows the selected area diffraction pattern indicating anatase phase of TiO2 and presence of GO (002) plane in the complex. (d) The morphological features of BSA molecules after encapsulation on TiO2 + GO surface were characterized by Field Electron Scanning Electron Microscopy (FESEM) at a voltage of 20 keV which reveals the spherical particle like morphology distributed on the surface of graphene oxide.

Figure 1c shows the TEM image of polycrystalline TiO2 nanoparticles + GO with an average particle size of 120 nm. Polycrystalline nature of TiO2 was confirmed by selected area diffraction pattern (SAED) as shown in the inset of Fig. 1c. The diffraction rings were indexed to the anatase TiO2 structure (yellow). The presence of (002) plane (blue) indicates the presence of GO in the composite. Figure 1d shows the Field emission scanning electron microscope (FESEM) image of BSA/GO/TiO2/FTO device. The voltage applied was 20 keV. It is clear from the graph that BSA molecules are present on the top surface of the device. The preliminary characterization of the BSA protein and the presence of protein in TGB compound were confirmed by UV-Vis spectroscopy and FTIR measurement (Fig. S1 in the Supporting Information).

For albumin protein detection various analytical methods have been developed, however, spectrophotometric methods are fast and highly sensitive. Essentially, upon binding the protein with oxide surface, there would indeed be a change in the optical properties of resultant composite (TiO2 + BSA) and the same can be confirmed with the spectroscopic methods. In the similar direction, spectroscopic methods like absorbance and fluorescence methods have been envisaged to detect DNA concentration, DNA damage, glucose, detection of rabbit IgG using TiO2 nanoparticles31,32,33,34. Zhang et al. has demonstrated a simple and highly sensitive method to detect BSA protein using triangular silver nanoplates using UV-Vis spectrophotometry27. Yet in another study, gold nanorods have been used to detect the concentration of albumin using both the absorbance and fluorescence methods35. In the present case, we used both the methods to confirm the presence of BSA. The spectroscopic method of BSA detection is shown in the supporting information.

Figures 2a,b represents the evidenced I–V characteristics of Ag/TiO2/FTO and Ag/BSA/TiO2/FTO devices respectively. Initially the device Ag/TiO2/FTO is at high resistance state and upon sweeping voltage from 0–2 V, indeed there is a sharp jump in current at 0.9 V, which indicates that device changes its resistance state from HRS to LRS and the corresponding voltage can be termed as VSET. Above VSET, as there is a compliance limit of 10 μA (to avoid dielectric breakdown), we do see a saturation for the current that we measured. In the reverse run from 2 V–−2 V, the device is in LRS until −1.9 V (VRESET). Above this again it reaches to HRS and continued to be in the same state until 0 V. Such a behavior (having VSET and VRESET for different polarities of the voltage) is typical for a bipolar resistive switching (BRS). In the present investigation, we have observed BRS on all the devices that we investigated. Bipolar resistive switching has well been established in TiO2 based resistive random access memory devices36,37.

Figure 2

I–V characteristics of (a) Ag/BSA/TiO2/FTO and (b) Ag/BSA/GO/TiO2/FTO devices in voltage sweeping mode (with a compliance current of 10 µA) at room temperature showing bipolar switching. The detection of BSA protein is manifested as a shift in values of VSET from 0.9 V (for T) to 0.77 V (for TB). (c) Effect of protein on the switching voltages were studied on TiO2 by varying the concentration of protein which indicates a systematic shift of VSET for different concentration of BSA. Inset shows variation of VSET with BSA concentration. (d) Enhancement in the switching performance has been observed by inserting a graphene oxide layer between TiO2 and BSA protein with VSET changes from 0.77 V (TB) to 0.32 V (TGB). The systematic shift of VSET can be seen for all three devices (T, TB, TGB). (e) Multi-bit data storage of TGB device under different compliance currents of 10 µA, 100 µA, 500 µA. (f) Endurance characteristics of TGB device in LRS state for different compliance current at 10 mV read voltage.

Now we discuss about detection of BSA protein with TiO2 based Bio – RRAM through electrical measurements. In order to demonstrate our efforts, initially we introduced BSA protein on top of TiO2 layer and investigated the resistive switching properties. The detection of BSA protein is manifested as a shift in values of VSET. Figure 2b manifests such a shift in VSET upon adding BSA. VSET is shifted from 0.9 V (for T) to 0.77 V (for TB) and this variation of VSET with BSA addition motivated us to perform a systematic study. Pertinent to this study, we added different concentrations of BSA (4, 6, 8 and 15 mg/mL) on independent device and looked into resistive switching properties. Figure 2c shows the systematic shift of VSET for different concentration of BSA. For instance, the value of VSET changes from 0.9 V (for 0 mg/mL) to 0.4 V (for 15 mg/mL), hinting that the present device is highly sensitive for the detection of BSA protein. In order to make it more clear, we also showed the variation of VSET with BSA concentration as an inset of Fig. 2c. Previous studies on transport properties of single layer BSA protein in molecular transistors has exhibited a low operational voltage38. In addition, a two terminal transport experiment at nano-scale has shown Ohmic behavior, which indicates the metallic nature of the protein39 and leads to a decrease in switching voltage with an increase in the BSA concentration. It could be seen that the HRS current increases by 2 orders of magnitude when protein concentration enhances to 15 mg/mL. The lower value of resistance in HRS could be due to electrons transfer between protein molecules, which may be due to decrease in average separation between protein molecule upon increasing the concentration. We also compared the sensitivity of our device with previous literature reports27,35 where they could able to detect as low as 0.5 ng/mL and 0.2 mg/mL (UV region), 0.05 mg/mL (Vis-IR region), 0.0013 mg/mL (by Fluorescence spectra) respectively. However, in our case we could able to detect 4 mg/mL by electrical measurements.

In order to improve the sensitivity of BSA protein detection, we introduced a GO layer between TiO2 and BSA protein. Corresponding results are shown in Fig. 2d. It is evident that for 4 mg/mL of BSA protein, the VSET changes from 0.77 V (TB) to 0.32 V (TGB), which suggests that the GO indeed enhances the transport properties by pushing VSET to a lower value. This is an important observation which may have many practical implications. It could be that GO layer provides better environment for conductive filament growth inside the switching layer, which offer lower switching voltages and good endurance characteristics. As a result of above, the randomness of filament evolution is prevented by inserting a GO layer and only a fewer conductive paths are stabilized for the switching. During set process, we believe that oxygen groups are removed from GO, which may migrate towards bottom electrode. Due to the removal of oxygen group, the C–C bonds on GO may shorten, which leads to shorter hopping distance for the carriers. Essentially, shorter hopping distance makes easy electron migration and can accompany lower resistance and lower values of voltages for switching the device40.

Multibit data storage capability of the device is checked with different values of compliance current say 10 µA, 100 µA and 500 µA. It is evident from Fig. 2e that three different LRS states are distinguishable for different compliance limits. In addition, higher the compliance limit, better is the LRS current and stronger the filament. Figure 2f depicts the evidenced LRS state endurance behavior of TGB device with different compliance current limit of 10 µA, 100 µA and 500 µA and at a readout voltage of 10 mV. It is clear from both the experiments that we could tune the LRS states with different compliance limits. On top of that, reading of various resistance states are successfully achieved at 10 mV, which demonstrates that compliance current is a key parameter and can allow one to access multilevel LRS states.

Now we discuss about the conceivable mechanism for the BSA protein adsorption on TiO2 surface. It has widely been investigated by many researchers and reported a conformational change in the protein molecule due to adsorptions on TiO2 surface41,42. This change not only influences the characteristics of adsorbed protein but also directs the behavior of surface. In our experiment as we used DI water, the pH of it is around 7.4. In this pH region the BSA protein and TiO2 surfaces can carry negative sign and can have electrostatic repulsion, which leads to less adsorption onto oxide surface43. Whatever may be the case, the absorption of protein on oxide surfaces can primarily be happen through (a) electrostatic attraction and (b) ligand exchange between BSA and oxide surfaces. The adsorption of protein on oxide surfaces can happen by van der Waals forces, the sorbate conformational stability and hydrophobic interactions. Surface hydrophilicity, surface charge and aggregation of oxide particles can also provide the possible spaces for protein adsorption on oxide surfaces44,45. As electrostatic interactions is one of the main driving forces in protein adsorption, although the overall charges of both TiO2 and protein surfaces have same sign, adsorption can still happen through electrostatic attraction. The conceivable mechanism might be that different regions of protein surface consists different signs of charge (positive or negative), which can attract positively charged particle surfaces46. As there is less adsorption, the excess negative charges may participate in the switching process which causes the devices to form filaments with a lower set voltage.

If we consider an experiment with condition of pH 7.4, the predominant surface groups of TiO2 are Ti2 = O− and Ti-OH, with few Ti2 = OH. On the other hand, the important functional group of BSA protein are R-COO− and R-NH3+43,47. We believe that an interaction between these functional groups lead to surface modification and predominant change in I–V curves with a shift in switching voltages. The kind of electrostatic interaction that occurs between aforementioned groups are given below

$${\rm{Ti}}-{\rm{OH}}+{\rm{R}}-{{\rm{NH}}}_{3}^{+}+\to {\rm{Ti}}-{{\rm{OH}}}_{2}^{+}:{{\rm{NH}}}_{2}-{\rm{R}}\,({\rm{electrostatic}}\,{\rm{interactions}})$$
$${\rm{Ti}}-{\rm{OH}}+{\rm{R}}-{{\rm{COO}}}^{-}+\to {\rm{Ti}}-{\rm{OH}}\ldots {}^{-}{\rm{O}}{\rm{OC}}-{\rm{R}}\,({\rm{electrostatic}}\,{\rm{interactions}})$$
$${{\rm{Ti}}}_{2}={{\rm{O}}}^{-}+{\rm{R}}-{{\rm{NH}}}_{3}^{+}\to {{\rm{Ti}}}_{2}={{\rm{O}}}^{-}{}^{+}{\rm{N}}{{\rm{H}}}_{3}-{\rm{R}}\,({\rm{hydrogen}}\,{\rm{bonding}}\,{\rm{interaction}})$$

In order to differentiate real change in signal to noise for these devices, we have established the noise range for I-V detection of BSA with different concentrations. Figure 3(a–f) shows such noise measurements, which demonstrates range of voltages (noise) at which SET switching can happen for different concentrations of BSA protein. Upon repeated voltage cycling, we do not see a predominant variation in switching voltages, which demonstrates the stability of the filaments. For instance, the noise for change in voltage for parent device is ~0.6 V. However, the range of noise for voltage diminishes to 0.06 V for BSA with concentration (15 mg/mL). This indeed, demonstrates the enhanced stability of filament in all the devices with BSA. It is apparent from the figure that each device with specific concentration has its own noise range for voltage. Inset of each graph shows that there is no significant change in switching voltage for the devices despite repeated cycling.

Figure 3

Resistive switching of TiO2 based RRAM devices with different concentrations of BSA. These graphs differentiate real signal change to noise range for I-V detection of the BSA with different concentrations. Inset of each graph shows that there is no significant change in switching voltage of the devices with number of cycles.

Now we discuss about the actual mechanism for the detection of BSA through our XPS measurements. The actual thought to prove our claim is that the amount of voltage that one need to apply to get switching would essentially depends on the number of oxygen vacancies. For instance if a film has more oxygen vacancies, less would be the voltage required for the filament formation and SET switching. On the other hand, if a film consists less oxygen vacancies, more would be the voltage required for the filament formation and SET switching. Hence, we believe that number of oxygen vacancies would decide whether the switching must happen at lower voltages or higher voltages.

Particularly in TiO2 based RRAM devices, oxygen vacancies are the primary constituents for the filament formation. Upon adding, BSA to TiO2 device, we could able to see that there is decrease in the switching voltage, which may be due to the enhanced number of oxygen vacancies. We believe that the chemical reaction between predominant surface groups of TiO2 (such as Ti2 = O AND Ti-OH, Ti2=OH) and functional groups of BSA protein (R-COO- and R – NH3+) might be playing an important role in controlling number of oxygen defects in the device. In order to confirm our claims, we have performed X-ray photoelectron spectroscopy (XPS) measurement on the parent device TiO2 (Fig. 4(a)) and TiO2 + BSA (15 mg/mL) (Fig. 4(b). It is evident from Fig. 4(a) that the Ti4+ peak exists in doublet state and corresponding peaks are related to Ti 2p3/2 (binding energy of 458.1 eV) and Ti 2p1/2 (binding energy of 463.9 eV), which is consistent with the earlier reports. In addition to this, there is a broad peak at binding energy of 459.1 eV corresponding to Ti3+ 2p1/2 in Ti2O348,49. Increased number of oxygen vacancies are identified directly from the area under the peak of Ti4+. Essentially, the area under the curve of Ti4+ peak diminishes upon adding BSA concentration of 15 mg/mL (Fig. 4(b)). The reduction of Ti4+ peak area in the 15 mg/mL added BSA protein (TiO2 + BSA) compared to parent TiO2 indicates the generation of oxygen vacancies to compensate and balance the system electrostatically. Corresponding chemical equation can be written as follows50.

$$4{{\rm{Ti}}}^{4+}+{{\rm{O}}}^{2-}\to 4{{\rm{Ti}}}^{4+}+2{{\rm{e}}}^{-}/\square \,+\,{}_{2}{}^{1}{\rm{O}}_{2}\to 2{{\rm{Ti}}}^{4+}+2{{\rm{Ti}}}^{3+}+\,\square +\,{}_{2}{}^{1}{\rm{O}}_{2}$$

where □ is an empty site originated due to the removal of O2− in the lattice.

Figure 4

(a) XPS spectra of TiO2 indicating the two main peaks of Ti4+; Ti4+ 2p3/2 (binding energy of 458.1 eV), Ti4+2p1/2 (binding energy of 463.9 eV) and a shoulder peak at binding energy of 459.1 eV corresponding to Ti3+ 2p1/2 in Ti2O3. (b) XPS data of 15 mg/mL added BSA protein (TiO2 + BSA).

Thus, it is expected that the concentration of oxygen vacancies might be increasing with increasing the protein concentrations in the TiO2 based RRAM devices, which actually helps in detecting the BSA protein by indicating a reduced switching voltage in the I–V characteristics of the device. S. carrara et al. has also proposed detection of rabbit antibodies through I–V characteristics with silicon nanowire memristive bio-sensors where they have pointed out a voltage gap in the bio-functionalized device and the same has been attributed as a detection parameter1. We also believe that the BSA protein might be encapsulating the TiO2 particles. This might be the reason why we do see lot of noise in the XPS data with BSA (Fig. 4(b)).

Now we discuss about endurance characteristics of the device. Cycling endurance of RRAM device is one of the important features to be studied from the application point of view. Memory margin of all RRAM devices are calculated using the formula (ROFF  −  RON)/RON ≈ ROFF/RON. In general, a memory window of 10 is sufficient enough to distinguish information between HRS and LRS. As shown in the Fig. 5a, TiO2 based RRAM device was repeatable for 30 cycles with resistance ratio of 12. However, with addition of BSA protein, the ratio increased to 73 and the device can be switched for 100 cycles (Fig. 5b). It is evident from the Fig. 5c that TGB device shows more durability and up to 650 cycles we could switch between ON/OFF with a ratio of 100 after which the device fails. In oxide based RRAM devices, the most common defects are oxygen vacancies. Thus, when an external bias is applied to the active material, oxygen ions are pulled out of the equilibrium position, which results in the generation of vacancies and current conduction. This is believed to happen in the devices through oxygen vacancies based on conductive filaments  between top and bottom electrodes. Similarly, the filaments rupture due to drifting of oxygen ions from top metal electrode and recombining with the vacancies. However, as the number of switching cycles increase, additional defects are generated in the vicinity of filaments which results in an increase of the radius of filaments51. Moreover, the different electrode material in the device structure is very likely to form a wider filament on one side and a narrower in other side. Due to continuous accumulation of vacancies, the recombination rate of oxygen ion and vacancy pairs are also reduced, which eventually leads to device failure52. Therefore, the possible reason for the device failure after 650 cycles could be due to oxidation of metal electrode or an abrupt increase in the local density of oxygen vacancies and formation of additional conductive channels in the active layer which have resulted in lowering the resistance of the device and remains in an irreversible LRS state, which indicates the endurance failure.

Figure 5

Endurance performance of the three devices. (a) Ag/TiO2/FTO based RRAM device was repeatable for 30 cycles with a on/off ratio of 12 (b) On/Off ratio increases to 73 for the Ag/BSA/TiO2/FTO device (c) graphene oxide inserted device (TGB) shows enhanced endurance characterists and increase in the resistance ratio up to 102 at 0.2 V read voltage. (d) Current 3D conducting atomic force microscope (C-AFM) images of the Ag/BSA/GO/TiO2/FTO device scanned in an area of 1 × 1 µm2 by applying a voltage bias of 9 V and −9 V respectively. (i) Formation of conductive channels on applying a bias voltage of 9 V (ii) disappearance of conductive paths in the reverse bias of −9 V indicating the role of nano-channels in the switching phenomena. (e) Schematic representation of the switching mechanism in the Ag/BSA/GO/TiO2/FTO device, where the hollow circles represent empty charge trapping sites and the solid circles depict the filled charge sites. (i) Pristine state of the device without external bias. (ii) When a positive potential is applied to the top electrode, charge carriers injected from electrode fill few of the empty charge sites (iii) As the bias voltage reaches set voltage, most of the charge trapping centers are filled and hopping of charges from one filled sites to other is initiated, resulting in the formation of conductive paths which corresponds to the low resistance state. (iv) When reset voltage is reached, dissolution of conductive filaments occurs in the weakest part of the filament and the device switches to HRS.

Conductive Atomic Force Microscopy (C-AFM) is a powerful technique  and is used  to map conductive channels. This has the ability of spatially resolving local heterogeneities of electrical response and simultaneously probing the current distribution on the surface of the material. In this study, the metallic AFM tip behaves as a nano-sized electrode in the switching of the device. Images were taken using Bruker SCANASYST-AIR Probes (Silicon Tip on Nitride Lever) having 12 nm, 70 kHz and 0.4 N/m as curvature radius, oscillation frequency and cantilever spring constant respectively. The experiment was carried out at room temperature and at atmospheric pressure without applying any current compliance. Figure 5d illustrates the current distribution C-AFM image of TGB based memory device. We applied a +/−9 V bias on an area of 1.0 × 1.0 µm2 at 1.01 Hz in contact mode. The current mapping images were obtained by moving the cantilever across the film surface. When a positive potential was applied, a large number of conductive filaments are seen on the surface and is shown in Fig. 5d(i). On the other hand, filaments are not evident for the opposite bias voltage and are shown in Fig. 5d(ii). Hence, the presence of conductive paths are due to widely accepted filament model resistive switching.

The variation of HRS and LRS is explained on the basis of Schottky emission mechanism (ln I vs V1/2 variation is linear) and trap controlled space charge limited conduction mechanism (consists of ohmic, trap assisted region and space charge region). More details about the behavior are given in supporting information. Conceivable physical mechanism for both the mechanisms can be explained as follows using Fig. 5e. In TiO2, the most common defects are oxygen vacancies and these oxygen vacancies are believed to form trap sites for electrons53,54. When no external bias is applied, the device would be in HRS state as shown in Fig. 5e(i). At low voltages, some trap centers would be filled with charge carriers and at high voltage, hopping of charge carriers would takes place through charge trapping centers to Ag electrode, which leads to low resistance state for the device (Fig. 5e(ii,iii)). HRS state can be achieved by applying reset voltage, due to the fact that injected charge carriers would break the weakly linked conduction in the device (Fig. 5e(iv))55,56.

Experimental Section

Material preparation

Synthesis of graphene oxide

For the preparation of GO, modified Hummer’s method57 was adapted. First, the precursor solution was prepared by mixing 1 g of graphite powder and 0.5 g of sodium nitrate to 23 ml of concentrated sulfuric acid. Then, 3 g of potassium permanganate was added slowly to the prepared solution after 1 hr. To prevent thermal explosion, the solution beaker was kept in ice bath and stirred constantly for 12 hours at 35 °C. Further, this solution was diluted by adding 500 ml of deionized water under vigorous stirring and then treated with 30% H2O2 (5 ml). Finally, the mixture was washed with hydrochloric acid and deionized (DI) water and GO was obtained by filtering and drying the resultant material.

On top of that TiO2 (powder form) and BSA proteins were purchased from Sigma-Aldrich and Himedia respectively. Subsequently, solution of BSA was prepared by dissolving 50 mg of BSA to 10 ml of deionized water. In addition, both the TiO2 and GO solutions were prepared by adding them to DI water with 1 mg/ml respectively.

Device fabrication

Using the above prepared solutions, we have fabricated three resistive random access memory (RRAM) devices, Ag/TiO2/FTO (T), Ag/BSA/TiO2/FTO (TB), Ag/BSA/GO/TiO2/FTO (TGB) respectively. Prior to device fabrication, ultrasonication was used to clean FTO substrates. Cleaning of FTO substrate was done at least 10 min with deionized water, acetone and 2 – propanol individually. For all devices, FTO was used as bottom electrode and silver as top electrode. The resistance of FTO is approximately 15 Ω. The devices were prepared by drop casting the solution onto FTO substrate and dried at a temperature of 75 °C (for T device). A relatively low temperature of 35 °C was supplied to TB and TGB devices for the drying purpose in order to prevent the damage for BSA proteins if any. Silver conducting epoxy was used as top electrode. The UV-Vis absorbance spectra of BSA protein, all three devices (T, TB, TGB) were recorded using a Perkin Elmer LAMBDA spectrophotometer between 200 to 700 nm. Chemical bonding within the protein was studied by Fourier transform infrared spectroscopy (FTIR), using a FTIR spectrometer Bruker Tensor 37 by ATR method in the range of 4000 to 500 cm−1 with a spectral resolution of 4 cm−1. The fluorescence (FL) spectra were collected by Edinburgh FLS 1000 fluorescence spectrometer with an excitation wavelength of 340 nm. Microstructural studies were done using the field emission scanning electron microscope (Zeiss ultra 55 FE-SEM). The transmission electron microscopy (TEM) investigations were performed using JEOL JEM 2100 instrument at an accelerating voltage of 200 kV. In order to prepare sample for TEM, the TiO2 and GO were dissolved in DI water and subsequently ultrasonicated for 10 minutes. The suspensions were dropped on to a standard carbon grid. The electrical measurements were performed at room temperature using a Keithley 2400 source meter with a compliance limit of 10 µA. In order to gain information on local electrical conduction and to observe filament formation, we have used a conductive atomic force microscope (C-AFM). In this method, nano-sized AFM tip itself works as top electrode. The measurement was performed by applying a voltage between the C-AFM tip which acts as top electrode and FTO as bottom electrode. We applied two different bias voltages, 9 V and −9 V, on an area of 1.0 × 1.0 µm2 at 1.01 Hz in contact mode.


In summary, we have investigated the detection of BSA protein using TiO2 and TiO2 + GO based RRAM devices. Lowering of switching voltages, enhanced endurance characteristics (up to 650 cycles) and enhanced ON/OFF resistance ratios (~102) were observed in TiO2 + GO RRAM device in comparison with bare TiO2 RRAM device. Enhanced characteristics in TiO2 + GO device are attributed to prevention of the multi-dimensional and random growth of conductive paths. Our XPS measurements infer that decrease in switching voltage upon adding BSA to enhanced oxygen vacancies. Detection of current ~nA in C-AFM investigation confirm the presence of filamentary switching. Present results will be helpful for future bio- RRAM devices, where detection of bio- molecules would be through memristor devices.


  1. 1.

    Carrara, S. et al. Memristive-biosensors: A new detection method by using nanofabricated memristors. Sensors and Actuators B: Chemical 171, 449–457 (2012).

  2. 2.

    Tzouvadaki, I., Parrozzani, C., Gallotta, A., De Micheli, G. & Carrara, S. Memristive biosensors for PSA-IgM detection. BioNanoScience 5, 189–195 (2015).

  3. 3.

    Pavan, P., Bez, R., Olivo, P. & Zanoni, E. Flash memory cells-an overview. Proceedings of the IEEE 85, 1248–1271 (1997).

  4. 4.

    Duan, X., Huang, Y. & Lieber, C. M. Nonvolatile memory and programmable logic from molecule-gated nanowires. Nano letters 2, 487–490 (2002).

  5. 5.

    Hu, J. & Liu, S. Responsive polymers for detection and sensing applications: current status and future developments. Macromolecules 43, 8315–8330 (2010).

  6. 6.

    Scott, J. F. & De Araujo, C. A. P. Ferroelectric memories. Science 246, 1400–1405 (1989).

  7. 7.

    Wuttig, M. Phase-change materials: Towards a universal memory? Nature materials 4, 265–266 (2005).

  8. 8.

    Ney, A., Pampuch, C., Koch, R. & Ploog, K. H. Programmable computing with a single magneto resistive element. Nature 425, 485–487 (2003).

  9. 9.

    Waser, R., Dittmann, R., Staikov, G. & Szot, K. Redox‐based resistive switching memories–nanoionic mechanisms, prospects, and challenges. Advanced materials 21, 2632–2663 (2009).

  10. 10.

    Kwon, D. H. et al. Atomic structure of conducting nanofilaments in TiO2 resistive switching memory. Nature nanotechnology 5, 148–153 (2010).

  11. 11.

    Lee, M. J. et al. A fast, high-endurance and scalable non-volatile memory device made from asymmetric Ta2O5− x/TaO2− x bilayer structures. Nature materials 10, 625–630 (2011).

  12. 12.

    Jo, S. H. et al. Nanoscale memristor device as synapse in neuromorphic systems. Nano letters 10, 1297–1301 (2010).

  13. 13.

    Prezioso, M. et al. Training and operation of an integrated neuromorphic network based on metal-oxide memristors. Nature 521, 61–64 (2015).

  14. 14.

    Zidan, M. A., Strachan, J. P. & Lu, W. D. The future of electronics based on memristive systems. Nature Electronics 1, 22–29 (2018).

  15. 15.

    Borghetti, J. et al. Memristive’switches enable ‘stateful’ logic operations via material implication. Nature 464, 873–876 (2010).

  16. 16.

    Li, C. et al. Analogue signal and image processing with large memristor crossbars. Nature Electronics 1, 52–59 (2018).

  17. 17.

    Xia, F. et al. Colorimetric detection of DNA, small molecules, proteins, and ions using unmodified gold nanoparticles and conjugated polyelectrolytes. Proceedings of the National Academy of Sciences 107, 10837–10841 (2010).

  18. 18.

    Zhang, J., Song, S., Wang, L., Pan, D. & Fan, C. A gold nanoparticle-based chronocoulometric DNA sensor for amplified detection of DNA. Nature protocols 2, 2888–2895 (2007).

  19. 19.

    Zheng, Q. et al. Graphene-based biosensors for biomolecules detection. Current Nanoscience 10, 627–637 (2014).

  20. 20.

    Lu, C. H., Yang, H. H., Zhu, C. L., Chen, X. & Chen, G. N. A graphene platform for sensing biomolecules. Angewandte Chemie 121, 4879–4881 (2009).

  21. 21.

    Kim, S. N., Rusling, J. F. & Papadimitrakopoulos, F. Carbon nanotubes for electronic and electrochemical detection of biomolecules. Advanced materials 19, 3214–3228 (2007).

  22. 22.

    Song, Y. et al. Label-free colorimetric detection of single nucleotide polymorphism by using single-walled carbon nanotube intrinsic peroxidase-like activity. Chemistry–A European Journal 16, 3617–3621 (2010).

  23. 23.

    He, B., Morrow, T. J. & Keating, C. D. Nanowire sensors for multiplexed detection of biomolecules. Current opinion in chemical biology 12, 522–528 (2008).

  24. 24.

    Bakalova, R., Zhelev, Z., Ohba, H. & Baba, Y. Quantum dot-based western blot technology for ultrasensitive detection of tracer proteins. Journal of the American Chemical Society 127, 9328–9329 (2005).

  25. 25.

    An, F. F. & Zhang, X. H. Strategies for preparing albumin-based nanoparticles for multifunctional bioimaging and drug delivery. Theranostics 7, 3667–3689 (2017).

  26. 26.

    Xie, L. et al. Bovine serum albumin nanoparticles modified with multilayers and aptamers for pH-responsive and targeted anti-cancer drug delivery. Journal of Materials Chemistry 22, 6053–6060 (2012).

  27. 27.

    Zhang, L. L. et al. Highly sensitive detection of bovine serum albumin based on the aggregation of triangular silver nanoplates. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 154, 98–102 (2016).

  28. 28.

    Li, Z. et al. Label-free detection of bovine serum albumin based on an in-fiber Mach-Zehnder interferometric biosensor. Optics express 25, 17105–17113 (2017).

  29. 29.

    Yang, J. J. et al. Memristive switching mechanism for metal/oxide/metal nanodevices. Nature nanotechnology 3, 429–433 (2008).

  30. 30.

    Zhuge, F., Fu, B. & Cao., H. Advances in Resistive Switching Memories Based on Graphene Oxide. In New Progress on Graphene Research. IntechOpen (2013)

  31. 31.

    Nadzirah, S., Hashim, U. & Rusop, M. Development of DNA biosensor based on TiO2 nanoparticles. AIP Conference Proceedings 1963, 020062 (2018).

  32. 32.

    Chen, P. & Taniguchi, A. Detection of DNA damage response caused by different forms of titanium dioxide nanoparticles using sensor cells. Journal of Biosensors & Bioelectronics 3 (2012).

  33. 33.

    Li, Y., Liu, X., Yuan, H. & Xiao, D. Glucose biosensor based on the room-temperature phosphorescence of TiO2/SiO2 nanocomposite. Biosensors and bioelectronics 24, 3706–3710 (2009).

  34. 34.

    Mun, K. S., Alvarez, S. D., Choi, W. Y. & Sailor, M. J. A stable, label-free optical interferometric biosensor based on TiO2 nanotube arrays. Acs Nano 4, 2070–2076 (2010).

  35. 35.

    Wang, Y. & Yongnian, N. New insight into protein–nanomaterial interactions with UV-visible spectroscopy and chemometrics: human serum albumin and silver nanoparticles. Analyst 139, 416–424 (2014).

  36. 36.

    Ge, J. & Chaker, M. Oxygen vacancies control transition of resistive switching mode in single-crystal TiO2 memory device. ACS applied materials & interfaces 9, 16327–16334 (2017).

  37. 37.

    Lee, J. S., Lee, S. & Noh, T. W. Resistive switching phenomena: A review of statistical physics approaches. Applied Physics Reviews 2, 031303 (2015).

  38. 38.

    Mentovich, E. D., Belgorodsky, B., Kalifa, I., Cohen, H. & Richter, S. Large-scale fabrication of 4-nm-channel vertical protein-based ambipolar transistors. Nano letters 9, 1296–1300 (2009).

  39. 39.

    Pompa, P. P. et al. Charge transport in disordered films of non-redox proteins. J. Chem. Phys. 125, 021103 (2006).

  40. 40.

    Li, Y., Long, S., Liu, Q., Lv, H. & Liu, M. Resistive Switching Performance Improvement via Modulating Nanoscale Conductive Filament, Involving the Application of Two - Dimensional Layered Materials. Small 13, 1604306 (2017).

  41. 41.

    Kopac, T. & Bozgeyik, K. Effect of surface area enhancement on the adsorption of bovine serum albumin onto titanium dioxide. Colloids and Surfaces B: Biointerfaces 76, 265–271 (2010).

  42. 42.

    Givens, B. E., Xu, Z., Fiegel, J. & Grassian, V. H. Bovine serum albumin adsorption on SiO2 and TiO2 nanoparticle surfaces at circumneutral and acidic pH: A tale of two nano-bio surface interactions. Journal of colloid and interface science 493, 334–341 (2017).

  43. 43.

    Oliva, F. Y., Avalle, L. B., Cámara, O. R. & De Pauli, C. P. Adsorption of human serum albumin (HSA) onto colloidal TiO2 particles, Part I. Journal of colloid and interface science 261, 299–311 (2003).

  44. 44.

    Haynes, C. A. & Norde, W. Globular proteins at solid/liquid interfaces. Colloids and Surfaces B: Biointerfaces 2, 517–566 (1994).

  45. 45.

    Arai, T. & Norde, W. The behavior of some model proteins at solid-liquid interfaces 1. Adsorption from single protein solutions. Colloids and Surfaces 51, 1–15 (1990).

  46. 46.

    MacDonald, D. E., Markovic, B., Boskey, A. L. & Somasundaran, P. Physico-chemical properties of human plasma fibronectin binding to well characterized titanium dioxide. Colloids and Surfaces B: Biointerfaces 11, 131–139 (1998).

  47. 47.

    Servagent-Noinville, S., Revault, M., Quiquampoix, H. & Baron, M. H. Conformational changes of bovine serum albumin induced by adsorption on different clay surfaces: FTIR analysis. Journal of Colloid and Interface Science 221, 273–283 (2000).

  48. 48.

    Bharti, B., Kumar, S., Lee, H.-N. & Kumar, R. Formation of oxygen vacancies and Ti3+ state in TiO2 thin film and enhanced optical properties by air plasma treatment. Scientific reports 6, 32355 (2016).

  49. 49.

    Sanjines, R. et al. Electronic structure of anatase TiO2 oxide. Journal of Applied Physics 75, 2945–2951 (1994).

  50. 50.

    Jiang, X., Zhang, Y., Jiang, J. & Rong, Y. Yancheng Wang, Yichu Wu, and Chunxu Pan. Characterization of oxygen vacancy associates within hydrogenated TiO2: a positron annihilation study. The Journal of Physical Chemistry C 116, 22619–22624 (2012).

  51. 51.

    Lohn, A. J., Patrick, R. M. & Matthew, J. M. Mechanism of electrical shorting failure mode in resistive switching. Journal of Applied Physics 116, 034506 (2014).

  52. 52.

    Raghavan, N., Pey, K. L., Frey, D. D. & Bosman, M. Stochastic failure model for endurance degradation in vacancy modulated HfO × RRAM using the percolation cell framework. In 2014 IEEE International Reliability Physics Symposium, IEEE 2014.

  53. 53.

    Kim, K. M. et al. A detailed understanding of the electronic bipolar resistance switching behavior in Pt/TiO2/Pt structure. Nanotechnology 22, 254010 (2011).

  54. 54.

    Seong, D. J., Jo, M., Lee, D. & Hwang, H. HPHA Effect on Reversible Resistive Switching of Pt/Nb-Doped SrTiO3 Schottky Junction for Nonvolatile Memory Application. Electrochemical and solid-state letters 10, H168–H170 (2007).

  55. 55.

    Garbin, D. et al. Resistive memory variability: A simplified trap-assisted tunneling model. Solid-State Electronics 115, 126–132 (2016).

  56. 56.

    Banerjee, W. et al. Variability Improvement of TiOx/Al2O3 Bilayer Nonvolatile Resistive Switching Devices by Interfacial Band Engineering with an Ultrathin Al2O3 Dielectric Material. ACS Omega 2, 6888–6895 (2017).

  57. 57.

    Hummers, W. S. Jr. & Offeman, R. E. Preparation of graphitic oxide. Journal of the American chemical society 80, 1339–1339 (1958).

Download references


We would like to acknowledge Indian Institute of Technology, Hyderabad for providing financial support. The author Dwipak Prasad Sahu is thankful to the Department of Science and Technology, India (DST-INSPIRE) for the award of senior research fellowship (SRF).

Author information

S.N.J. and D.P.S. conceived and conducted the experiments. Both have analyzed the results and written the manuscript.

Correspondence to S. Narayana Jammalamadaka.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Sahu, D.P., Jammalamadaka, S.N. Detection of bovine serum albumin using hybrid TiO2 + graphene oxide based Bio – resistive random access memory device. Sci Rep 9, 16141 (2019).

Download citation


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.