Low-Power Graphene/ZnO Schottky UV Photodiodes with Enhanced Lateral Schottky Barrier Homogeneity

Abstract

The low-power, high-performance graphene/ZnO Schottky photodiodes were demonstrated through the direct sputter-growth of ZnO onto the thermally-cleaned graphene/SiO₂/Si substrate at room temperature. Prior to the growth of ZnO, a thermal treatment of the graphene surface was performed at 280 °C for 10 min in a vacuum to desorb chemical residues that may serve as trap sites at the interface between graphene and ZnO. The device clearly showed a rectifying behavior with the Schottky barrier of ≈0.61 eV and an ideality factor of 1.16. Under UV illumination, the device exhibited the excellent photoresponse characteristics in both forward and reverse bias regions. When illuminating UV light with the optical power density of 0.62 mW/cm², the device revealed a high on/off current ratio of >10³ even at a low bias voltage of 0.1 V. For the transient characteristics upon switching of UV light pulses, the device represented a fast and stable photoresponse (i.e., rise time: 0.16 s, decay time: 0.19 s). From the temperature-dependent current–voltage characteristics, such an outstanding photoresponse characteristic was found to arise from the enhanced Schottky barrier homogeneity via the thermal treatment of the graphene surface. The results suggest that the ZnO/graphene Schottky diode holds promise for the application in high-performance low-power UV photodetectors.

1. Introduction

Graphene-based hybrid and heterostructures with other inorganic and/or organic materials render more fascinating functionalities compared to conventional electronic and optoelectronic devices [1,2,3]. Among various graphene-based hetero-architectures, the graphene/inorganic semiconductor contacts have attracted much attention because of their ample potential for high-performance Schottky photodiodes (PDs) [4]. For example, the enhanced photoresponse characteristics with high sensitivity and fast response were demonstrated on various graphene-based Schottky PDs that were composed of typical inorganic semiconductor materials (e.g., Si [5,6], Ge [7], GaAs [8,9], CdSe [10], ZnO [11,12,13,14,15,16,17,18,19,20,21], etc.). Among them, ZnO is one of the most prospective materials for high-performance ultraviolet (UV) PDs because of its wide band gap and excitonic properties [22,23]. Due to the c-axis preference of wurtzite ZnO, moreover, ZnO thin films [18,19] or ZnO nanorods [20] can be easily grown on the defective graphene sheet that had been grown using chemical vapor deposition (CVD). Namely, the presence of C–O nucleation sites on CVD graphene allows us to fabricate a simple device scheme of the graphene/ZnO Schottky PD through the direct growth of ZnO on graphene. Furthermore, since the work function of CVD graphene could be controlled by thermal [24,25] and chemical [26] treatments, one can easily manipulate the Schottky barrier height at the graphene/ZnO interface. Owing to such advantages, very recently, high-performance graphene/ZnO Schottky UV PDs have been realized through various techniques; for example, direct growth of ZnO onto graphene by using CVD [16], chemical bath deposition [11], radio frequency (r.f.) magnetron sputtering [18,19], hydrothermal methods [20], and dispersion of ZnO nanorods onto graphene [12,13,14]. When using graphene as an active layer of the electronic and the optoelectronic devices, there are critical issues from chemical adsorbates and residues that might remain on the graphene surface during the graphene transfer step [24]. For instance, during the growth of ZnO on graphene, the chemical adsorbates and residues will degrade the lateral homogeneity of the Schottky barrier because they act as unnecessary dopants and/or contaminants at the graphene/ZnO interface. In Schottky PDs, the inhomogeneous barrier underneath the photon collection area may degrade the ideality factor (i.e., transport characteristics) [4], and will eventually restrict the photoresponse characteristics of the Schottky PDs. To realize the high-performance graphene/ZnO Schottky UV PDs, therefore, enhancing the lateral Schottky barrier homogeneity is essential. In other words, the chemical adsorbates and residues should be effectively removed prior to the growth of ZnO onto the graphene surface.

In light of all the above, we have investigated the enhancement of the lateral Schottky barrier homogeneity in the graphene/ZnO Schottky PDs. The devices were fabricated by direct sputtering of ZnO onto thermally-cleaned single-layer graphene (SLG). Namely, to improve the ZnO/SLG interface properties, in situ thermal cleaning of CVD SLG was performed just prior to the sputtering of ZnO in a single chamber. This simple method uses neither chemical nor physical treatments that may provide additional adsorption of gas molecules during subsequent handling of the sample in air ambience. Despite such advantages, according to our best survey, no previous works have reported on the in situ thermal treatment for fabricating a high-performance SLG/ZnO Schottky PD. Through the temperature-dependent electrical characterization, we analyzed the Schottky barrier homogeneity of the fabricated ZnO/SLG Schottky PDs. In addition, we thoroughly assessed the photoresponse characteristics of the devices due to varying UV powers.

2. Experimental Section

2.1. Preparation of SLG/SiO₂/Si Substrate

Figure 1 schematically illustrates the device fabrication procedures for the SLG/ZnO Schottky PD. First, the SLG sheet was grown on Cu foil using CVD and transferred onto the SiO₂/Si substrate by using a poly(methyl methacrylate) (PMMA) transfer method (Figure 1a). Next, the sample was mounted in the sputtering chamber and was thermally cleaned at 280 °C for 10 min in a high vacuum (≈10⁻⁶ Torr) to eliminate chemical residues and/or molecular oxygens that might be adsorbed onto the SLG surface during the transfer process (Figure 1b).

2.2. Fabrication of ZnO/SLG Schottky PDs

The ZnO/SLG Schottky contacts were formed via the direct growth of the 200-nm-thick ZnO layer onto the SLG/SiO₂/Si substrate through r.f. magnetron sputtering (Figure 1c). The sputtering process was performed at room temperature in Ar plasma ambient under the following conditions: Ar gas flow rate = 30 sccm, working pressure = 30 mTorr, and r.f. power = 150 W. After the growth of the ZnO layer onto SLG, the active area (w ≈ 10 μm, l ≈ 30 μm) was defined by conventional photolithography techniques. Finally, the Al (t_Al ≈ 100 nm) and Ti/Al (t_Ti ≈ 10, t_Al ≈ 100 nm) ohmic electrodes for ZnO and SLG layer were formed using electron-beam evaporation and standard lift-off processes (Figure 1d). We here note that two different types of ZnO/SLG Schottky PDs were prepared by using as-transferred SLG and thermally-cleaned SLG so as to examine the effects of thermal cleaning on the device characteristics. For convenience, we refer the former and the later as a ZnO/AT-SLG Schottky PD and a ZnO/TC-SLG Schottky PD, respectively.

2.3. Measurements of Material and Device Characteristics

The Raman scattering characteristics of the as-transferred and thermally-cleaned SLG samples were measured by using a Renishaw Micro Raman spectrometer (Renishaw, Wotton-under-Edge, UK) under green laser excitation (λ = 514 nm). The topographic cross-section and the crystal structure of the ZnO layer were monitored through scanning electron microscopy (SEM) using an FE SEM XL-30 system (Phillips, Eindhoven, The Netherlands) and X-ray diffractometry (XRD) using a Bede D3 system (Bede Scientific Instruments Ltd., Durham, UK), respectively. The temperature-dependent electrical characteristics of the ZnO/SLG Schottky PDs were assessed at 300–400 K by using a Keysight B1500A semiconductor device parameter analyzer (Keysight Technologies, Santa Rosa, CA, USA). The photoresponse characteristics of the PDs were examined under light illumination using a 365-nm UV light emitting diode, wherein the UV power density (P_UV) was varied from 0–0.77 mW/cm² during photoresponse measurements.

3. Results and Discussion

Figure 2a schematically illustrates the final form of the fabricated ZnO/SLG Schottky PD, and Figure 2b shows the Raman spectra of the SLG sheets used for the device fabrication. Both the as-transferred and the thermally-cleaned SLG sheets revealed two predominant Raman features of G and 2D bands from high-quality graphene. For as-transferred SLG, the G and 2D peaks appeared at ≈1588 and 2684 cm⁻¹, respectively, and the observed positions were blue-shifted from pristine graphene (i.e., G_pristine ≈ 1580–1585 cm⁻¹, 2D_pristine ≈ 2635–2645 cm⁻¹) [25]. This implies that our as-transferred SLG was doped by acceptor impurities from oxygen molecules [27] and/or chemical residues [28,29] adsorbed during the graphene transfer step. After thermal cleaning at 280 °C, both G and 2D peaks were shifted by 9 and 13 cm⁻¹ toward their pristine graphene positions, respectively. Such a red-shift of both the G and 2D peaks depicts the decrease in charge trapping effects because the unintentional acceptors were effectively removed via thermal annealing at 280 °C [24,25]. Furthermore, no appearance of the defect-mediated D band at ≈1350 cm⁻¹ indicates that vacuum-annealing at 280 °C caused no damage to the sp² carbon bonds in SLG.

Onto the surface of high quality SLG, we deposited a 200-nm-thick ZnO layer at room temperature by using an r.f. magnetron sputtering technique. As shown in Figure 2c, the cross-sectional SEM image shows that the ZnO layer was effectively grown on SLG with a well-merged c-axis preferential columnar structure. The ZnO layers grown on the as-transferred and the thermally-cleaned SLG sheets exhibited a typical XRD pattern with the (000l) lattice phase (Figure 2d), which is indicative of the wurtzite structure of typical ZnO. In such a ZnO/SLG structure, the Schottky barrier (ϕ_B = Φ_SLG − χ_ZnO) would be formed at the interface between SLG and ZnO (Figure 2e) because the work function of CVD SLG (Φ_SLG > ≈4.5 + α eV) [30] is greater than electron affinity of ZnO (χ_ZnO ≈ 4.1 eV) [31]. Here, the magnitude of α depends on the difference between the Dirac point and the Fermi level in CVD SLG, and it could be caused by the p-type doping effect from oxygen molecules and/or chemical residues. Thanks to the formation of the Schottky barrier, the SLG/ZnO Schottky PDs show a good rectifying behavior (Figure 2f).

To characterize the Schottky barrier homogeneity of the ZnO/AT-SLG and ZnO/TC-SLG Schottky PDs, as a primary task, we measured the current–voltage (I–V) characteristics for both samples at temperatures ranging from 300 to 400 K in a dark chamber. For both samples, the diode current increased with increasing temperature (Figure 3a,b). Particularly, the increase in the reverse saturation current became significant because of thermally-activated carrier conduction (TACC) at elevated temperatures (see also the inset of Figure 3a). According to the thermionic emission theory [32], the I–V relationship of the Schottky diode at V > 3 kT/q can be expressed as:

$$J=J_0\exp \left(q\mathrm{V}/\eta k\mathrm{T}\right)$$

(1)

$$J_0={\mathrm{AA}}^{*}{\mathrm{T}}^2\exp \left(-q{\varphi }_B/k\mathrm{T}\right),$$

(2)

where J₀ is the reverse saturation current, q is the electron charge, η is the ideality factor, k is the Boltzmann constant, T is the absolute temperature, A is the contact area, and A* is the Richardson constant. Based upon Equations (1) and (2), we represented the Richardson plots (i.e., ln(J₀/T²) vs. 1000/T) to determine the effective value of A* for our devices (Figure 4). From linear best fitting, the values of A* were calculated to be 0.495 and 0.628 A·cm⁻²K⁻² for the ZnO/AT-SLG and the ZnO/TC-SLG Schottky PDs, respectively. The obtained A* values are much smaller than the theoretical value of the ZnO Schottky diode (≈32 A·cm⁻²K⁻²) because of the Schottky barrier inhomogeneity [33,34,35]. In addition, the ultrathin insulating layer at the Schottky interface could be a possible scenario that may degrade the A* value because the chemical adsorbates and oxygen molecules on SLG could locally form inadvertent insulating potential barriers [36,37,38]. Although both samples showed a smaller value of A* than the theoretical calculation, the effective value of A* was higher for the ZnO/TC-SLG Schottky PD (i.e., A* = 0.628 A·cm⁻²K⁻²) than the ZnO/AT-SLG Schottky PD (i.e., A* = 0.495 A·cm⁻²K⁻²). From this result, one can conjecture that the Schottky barrier homogeneity was enhanced in the ZnO/TC-SLG Schottky PD because the thermally-cleaned SLG exhibited a red-shift of G and 2D peaks, which is attributed to the effective elimination of oxygen molecules and/or chemical residues via thermal annealing at 280 °C (see also Figure 2b).

To examine further insight into the Schottky barrier homogeneity, we investigated the temperature dependencies of ϕ_B and η for both samples. At 300 K, the ZnO/AT-SLG Schottky PD revealed a ϕ_B of 0.58 eV and η of 1.91 (Figure 5a). As we explained earlier, the Schottky barrier with ϕ_B = 0.58 eV was effectively formed due to the difference between Φ_SLG and χ_ZnO. The high magnitude of η (= 1.91) is thought to be responsible for multiple recombination channels [39], which might arise from chemical residues residing at the ZnO/SLG interface. As the temperature increased, ϕ_B slightly increased while η suddenly decreased. These behaviors (i.e., ϕ_B ↗ and η ↓ with T ↗) are commonly observed in typical metal/semiconductor Schottky diodes, and are known to be attributed to the lateral Schottky barrier inhomogeneity [40,41]. We therefore expect our ZnO/AT-SLG Schottky PD to have inhomogeneous Schottky barriers with a significant ϕ_B fluctuation along the surface direction, particularly at impurity and/or defect sites attributing to chemical residues (see also Figure 6a). In such a device, the effective values of ϕ_B and η would increase and decrease with increasing temperature, respectively, because TACC through higher Schottky barriers became serious at elevated temperatures (see also Figure 6b,c).

For the ZnO/TC-SLG Schottky PD, ϕ_B and η were determined to be 0.61 eV and 1.16, respectively, at 300 K. In Schottky diodes, the low η manifested a clean interface between metal and semiconductor for the device, resulting in a dominance of thermionic emission rather than recombination [42]. One possible reason could be an effective removal of chemical residues from the SLG surface through the thermal cleaning process. Furthermore, the magnitudes of both ϕ_B and η were almost independent of temperature (Figure 5b), which is totally different from the thermodynamic behavior of the ZnO/AT-SLG Schottky PD. In a homogeneous Schottky barrier system, due to the weak fluctuation of ϕ_B along the surface direction, thermionic emission dominates its carrier transport whereas TACC becomes insignificant at elevated temperatures (see also Figure 6d,f). As a consequence, an enhanced Schottky barrier homogeneity will result in the almost invariance of both ϕ_B and η upon varying the environmental temperature. Therefore, we expected that the Schottky barrier homogeneity was enhanced in our ZnO/TC-SLG Schottky PD via thermal cleaning (i.e., formation of uniform ϕ_B along the surface direction by the effective removal of the chemical residues from the SLG surface).

In Schottky PDs, the Schottky barrier homogeneity is one of the most crucial factors that strongly affects the photoresponse characteristics. To confirm the effect of the lateral homogeneity of the Schottky barrier, we assessed and compared the photoresponse properties of the ZnO/AT-SLG and ZnO/TC-SLG Schottky PDs under UV light illumination with a P_UV of 0–0.77 mW/cm². For the ZnO/AT-SLG Schottky PD, as shown in Figure 7a, the current exponentially increased with increasing P_UV even in both the positive and the negative voltage regions. Figure 7b,c shows the P_UV dependence of the steady state photocurrent (i.e., I_Ph = I_Light − I_Dark) at various bias voltages and its corresponding I_Ph/I_Dark ratio, respectively. The device exhibited a large difference between I_Light and I_Dark, resulting in a high I_Ph/I_Dark ratio. Since the current level of I_Dark was sufficiently low at V = 0.1 V, the high I_Ph/I_Dark ratio of ≈1100 was achievable at 0.1 V.

When illuminating the UV light onto the ZnO/TC-SLG Schottky PD, a similar behavior to the above was observed (Figure 7d). As can be seen from Figure 7e, however, the device exhibited a slightly lower current level of I_Dark (= 0.12 nA) at 0.1 V, compared to the ZnO/AT-SLG Schottky PD (I_Dark = 0.35 nA). We ascribe this feature to the lower contribution of carrier recombination at the small bias voltage in our ZnO/TC-SLG Schottky PD because the device revealed a quite low magnitude of η (= 1.16). Accordingly, the I_Ph/I_Dark ratio was increased by a factor of ≈4 (i.e., I_Ph/I_Dark ≈ 4200) at 0.1 V (Figure 7f), compared to the ZnO/AT-SLG Schottky PD. The low operating voltage was advantageous in both demonstrating a high on/off ratio and reducing the power consumption [43]; hence, we believe the ZnO/SLG Schottky structure is preferable for the application of low power UV photodetectors.

Figure 8 shows the transient waveforms of I_Ph for the ZnO/AT-SLG and ZnO/TC-SLG Schottky PDs under UV light illumination (P_UV = 0.77 mW/cm²) at an optical switching frequency of 0.5 Hz. Both the ZnO/AT-SLG and the ZnO/TC-SLG Schottky PDs exhibited a distinct, stable, and repeatable switching characteristic of I_Ph upon turning on and off the UV signal. However, the on-state photocurrent was unstable in the ZnO/AT-SLG Schottky PD (Figure 8a), whereas the ZnO/TC-SLG Schottky PD revealed a stable photoresponse of I_Ph at the UV on state (Figure 8b). The rising time (τ_r) and the decay time (τ_d) of the ZnO/TC-SLG Schottky PD were τ_r = 0.16 s and τ_d = 0.19 s, respectively; and those of the ZnO/AT-SLG Schottky PD were τ_r = 0.41 s and τ_d = 0.31 s, respectively. When considering the on- and off-state resistance (R_PD = 0.2–500 MΩ) and the electrostatic capacitance (C_PD ≈ 320 pF) of our devices, the time constant (i.e., τ_RC = R_PDC_PD) was determined to be 64 μs–160 ms. Therefore, the observed photoresponse time was close to the maximum τ_RC and is comparable to other graphene/ZnO Schottky PDs [11,12,13,14,15,16,44] (see also

Table 1. Comparison of the photoresponse characteristics for various types of graphene/ZnO-based Schottky PDs.

Materials and Structures		λUV (nm)	VB (V)	τr (s)	τd (s)	G	R	IPh/IDark	Ref.
SLG/ZnO TF	AT	365	0.1	0.41	0.31	≈347	≈101	1100	This Work
	TC		0.1	0.16	0.19	≈381	≈111	4200
MLG/ZnO NRs		365	−5	0.7 × 10−3	3.6 × 10−3	385	113	<100	[12]
SLG/AlZnO NRs		380	SP	37 × 10−6	330 × 10−6	-	0.039		[11]
SLG/ZnO NRs		325	−1	0.3	0.5	-	21.26	≈30	[13]
SLG/ZnO NRs		325	−2	0.52	3.4		1.92	4	[14]
rGO/ZnO NRs		N/A	1	2.42	0.2	≈72	≈2.0	>>103	[15]
rGO/ZnO NRs		365	SP	0.1	0.2			5	[16]
GQDs/ZnO NRs		340	−1						[44]

Note: TF, thin film; NRs, nanorods; SP, self-powered; GQDs, graphene quantum dots; λ_UV, wavelength of UV light; V_B, bias voltage.

). For further improvement of the photoresponse time, a study on defect natures in ZnO can be the next step, for example, engineering of point defects and their lifetimes.

Here, one should focus on the fact that the photoresponse time of the ZnO/TC-SLG Schottky PD was outstandingly shorter than those of the ZnO/AT-SLG Schottky PD. We attribute such a discrepancy to the difference in Schottky barrier homogeneity between two devices. In the case of the inhomogeneous Schottky-barrier system (e.g., ZnO/AT-SLG Schottky PD), as illustrated in the right-hand-side inset of Figure 8a, the photocarriers would irregularly jump over the Schottky barrier due to the ϕ_B fluctuation along the surface direction. In addition, some of photocarriers might suffer from charge-trapping at the ZnO/SLG interface due to the presence of residual chemical constituents and/or oxygen molecules. These cause the unstable on-state photocurrent and retard the photoresponse time. Contrariwise, such behaviors could be effectively diminished in the homogeneous Schottky barrier system (e.g., ZnO/TC-SLG Schottky PD) because the Schottky barrier undulation was suppressed via elimination of trap sites through thermal cleaning of the SLG surface (see the right-hand-side inset of Figure 8b). Owing to the enhanced Schottky barrier homogeneity, more stable and prompter photoresponse characteristics could be achievable from the ZnO/TC-SLG Schottky PD.

Finally, we discuss the responsivity (R) and the gain (G) of the prepared ZnO/SLG Schottky PDs. In PDs, R is defined by the ratio of the photocurrent to the optical power of incident light (P_opt), and G is given by the ratio of the photogenerated carriers (N_e) to the incident photons (N_Ph). These can be described using following equations [45]:

$$R=\frac{{\mathrm{I}}_{\mathrm{Ph}}}{{\mathrm{P}}_{\mathrm{opt}}}=\frac{{\mathrm{I}}_{\mathrm{Light}}-{\mathrm{I}}_{\mathrm{Dark}}}{{\mathrm{P}}_{\mathrm{opt}}},$$

(3)

$$G=\frac{{\mathrm{N}}_{\mathrm{e}}}{{\mathrm{N}}_{\mathrm{Ph}}}=\frac{{\mathrm{I}}_{\mathrm{Ph}}}{{\mathrm{P}}_{\mathrm{opt}}}\times \frac{h\upsilon }{q},$$

(4)

where hν is the photon energy of the incident light. Using Equations (3) and (4), we calculated R and G for both the ZnO/AT-SLG and the ZnO/TC-SLG Schottky PDs. We here note that the average I_Ph value during a single pulse duration was used for calculations of R and G. In the case of the ZnO/AT-SLG Schottky PD, R and G were calculated to be ≈101 A/W and ≈347, respectively, and these are comparable to those of the state-of-the-art graphene/ZnO nanorod Schottky PDs [12,15] (see also Table 1). Compared to the ZnO/AT-SLG Schottky PD, both R and G were increased by ≈10% for the ZnO/TC-SLG Schottky PD (i.e., R ≈ 111 A/W and G ≈ 381). We attribute the increased R and G values to the enhanced Schottky barrier homogeneity in the ZnO/TC-SLG Schottky PD. In photocarrier conduction, G can also be defined as the ratio of the photocarrier lifetime (τ_PC) in the photocarrier generation region to the carrier transit time (τ_TR) in the photocarrier collection region [3]:

$$G\approx \frac{{\mathsf{\tau }}_{\mathrm{PC}}}{{\mathsf{\tau }}_{\mathrm{TR}}}.$$

(5)

From this relation, we can expect τ_PC at the SLG/ZnO interface to be increased in the ZnO/TC-SLG Schottky PD, when assuming that τ_TR is identical in SLG. This specifies that the photocarrier diffusion length (L_PC) is increased in the ZnO/TC-SLG Schottky PD because of the following relationship [32]:

$$L_{\mathrm{PC}}=\sqrt{D_{\mathrm{PC}}·{\mathsf{\tau }}_{\mathrm{PC}}}$$

(6)

where D_PC is the diffusion coefficient of the photocarrier. In other words, L_PC is increased due to the decreased interface trap density resulting from the effective removal of the chemical residues and/or oxygen molecules at the SLG/ZnO interface via thermal cleaning of the SLG surface prior to the deposition of ZnO onto SLG.

4. Summary and Conclusions

We fabricated high-performance, low-power ZnO/SLG Schottky UV PDs by sputtering ZnO onto thermally-cleaned SLG sheets. The device clearly showed a rectifying behavior and exhibited an excellent photoresponse under UV illumination. When the UV light (P_UV = 0.62 mW/cm²) was irradiated onto the device, the I_Ph/I_Dark ratio was recorded to be more than 4 × 10³. For the transient characteristics, the device represented a fast and stable photoresponse (i.e., τ_r = 0.16 s and τ_d = 0.19 s) upon switching of the UV light pulse. From analyses of the temperature-dependent I–V characteristics, we found such an outstanding photoresponse performance to arise from the enhanced Schottky barrier homogeneity due to the reduced interfacial trap density (i.e., effective removal of chemical residues and/or oxygen molecules at the ZnO/SLG interface via thermal cleaning).