Broadband photodetection using one-step CVD-fabricated MoS2/MoO2 microflower/microfiber heterostructures

Molybdenum disulfide (MoS2) has been combined so far with other photodetecting semiconductors as an enhancing agent owing to its optical and electronic properties. Existing approaches demonstrated MoS2-incorporated photodetector devices using complex and costly fabrication processes. Here, we report on simplified one-step on the chemical vapor deposition (CVD) based synthesis of a unique microfiber/microflower MoS2-based heterostructure formed by capturing MoO2 intermediate material during the CVD process. This particular morphology engenders a material chemical and electronic interplay exalting the heterostructure absorption up to ~ 98% over a large spectral range between 200 and 1500 nm. An arsenal of characterization methods were used to elucidate the properties of these novel heterostructures including Raman spectroscopy, X-ray diffraction, X-ray photoelectron spectrometry, high-resolution transmission and scanning electron microscopies, and Kelvin probe force microscopy. Our findings revealed that the MoS2 and the MoO2 crystallize in the hexagonal and monoclinic lattices, respectively. The integration of the MoS2/MoO2 heterostructures into functional photodetectors revealed a strong photoresponse under both standard sun illumination AM1.5G and blue light excitation at 450 nm. Responsivity and detectivity values as high as 0.75 mA W−1 and 1.45 × 107 Jones, respectively, were obtained with the lowest light intensity of 20 mW cm−2 at only 1 V bias. These results demonstrate the high performances achieved by the unique MoS2/MoO2 heterostructure for broadband light harvesting and pave the way for their adoption in photodetection applications.

Undeniably, molybdenum disulfide (MoS 2 ) has proven to be an excellent material for opto-electronic applications among the family of transition metal dichalcogenides (TMDs) two-dimensional (2D) materials 1 . Its outstanding electrical and optical properties [2][3][4][5][6] have made MoS 2 one of the most promising candidates for visible light-driven photodetectors 7 . Nonetheless, carriers' recombinations in pristine MoS 2 have limited its photodetecion efficiency, despite its good light absorption performance 8,9 . This has driven a general trend during the last decade combining MoS 2 with other semiconductors with already known and proven photodetecting capabilities. Efforts done in this direction mainly exploited the high light absorption properties of MoS 2 to improve the photodetecting performances of the other semiconductor by favoring carriers separation through the creation of a built-in electric field [6][7][8][9] . As example, ZnS/MoS 2 heterostructures have been shown to exhibit a large photodetection capability, where ZnS has served as local electric field generator, achieving an increased optoelectronic performances 10 . The the ultra-violet (UV) detection property of ZnS, combined with the strong visible light absorption of MoS 2 , have led to the fabrication of photodetectors spanning the whole spectrum from UV to the near infrared range. Another strategy consists of using plasmonic metal to create Schottky contact with MoS 2 for hot electrons injection leading to enhanced optical absorption and good photodetection properties [11][12][13] . Nonetheless, these approaches adopted so far have merely used MoS 2 as a supporting agent to catalyze light absorption of other semiconducting photodetector materials. Despite their advantages, these methods still suffer from several challenging limitations to their efficient integration in large scale broadband photodetector applications. Most prominently, fabrication processes of the used semiconductors, such as ZnO or GaN, are highly demanding in terms of resources, time, and cost. Additionally, the resulting functional devices remain restricted to very small dimensions and their efficient photodetection yield is bound to the application of high voltages (i.e. ~ 20 V) 14 . Therefore, innovative solutions to develop large-scale MoS 2 -based structures with high photodetection performances using affordable and efficient fabrication approaches are required to propel the integration of broadband MoS 2 -based photodetectors 15,16 . For this, exploiting MoS 2 -derived semiconductors (such as MoO 2 and MoO 3 ), naturally occurring during fabrication processes, constitutes a novel viable route to develop new heterostructures with enhance photodetecting performances. This is motivated by recently reported results showing a great potential of MoO 2 as a suitable candidate for optoelectronic applications based heterostructure 17,18 . Recent findings strongly suggest a significant impact of the MoO 2 /MoS 2 and MoO 3 /MoS 2 morphology on the increase of optical absorption properties leading to amplified photodetection performances, as found in vertically aligned MoS 2 needle-like structures 19,20 . Hence, controlling the morphology of MoS 2 is a key for improving photodetection performances 21,22 . Chemical vapor deposition (CVD) has proven to be one of the most promising fabrication techniques to achieve a controlled growth of MoS 2 23 , owing to its ease of implementation, low-cost and scalability [24][25][26][27][28] . Nevertheless, this requires the control of multiple parameters including pressure, temperature, heating rate, carrier gas flow rate, substrate, precursors positions, and reaction dwell time. All these processing parameters affect the morphology, crystallinity, and thickness of the MoS 2 heterostructures, which have been documented elsewhere for the synthesis of high-quality, large-surface-area single and few-layers MoS 2 [29][30][31] . Interestingly, MoO 2 is intermediate material created during the MoS 2 CVD process 32 . Therefore, "capturing" the formation of these materials along with CVD MoS 2 would enable an innovative approach to create seamless MoS 2 /MoO 2 heterostructures in a one single step process.
In this study, we present pioneering results on a single step large-scale controlled CVD growth of a unique MoS 2 /MoO 2 heterostructure showing giant light absorption (i.e. ~ 98%) over the full spectral range from 200 to 1500 nm. The as-grown heterostructures present a special MoO 2 microflowers/MoS 2 microfibers morphology, which was found to exhibit a huge specific area along with a strong broadband light absorption spanning from UV to near IR. An arsenal of characterization methods were used to investigate the optical, structural, crystallographic, chemical, electric, and photoelectric properties of the novel heterostructures. We have further integrated the fabricated heterostructure films into a photodetector test configuration to demonstrate their high potential for broadband applications. Applied voltages, one order of magnitude lower that those used for MoS 2 / GaN structures, enabled a comparable photodetectivity for the unique heterostructures morphology.

Experimental section
MoS 2 /MoO 2 synthesis. The CVD growth conditions were optimized to control the morphology and ratio of the MoS 2 /MoO 2 in the heterostructure grown on silicon substrates. First, the intrinsic Si substrate was successively cleaned with acetone and ethanol, rinsed with deionized water, and the dried with a nitrogen jet before introducing it into the furnace. Molybdenum trioxide (MoO 3 , 99.99%) and sulfur (S, 99.5%) powders were used as reactant and precursor materials, respectively. The cleaned Si substrate was immersed into a mixture solution consisting of 50 mg of S, 50 mg of MoO 3 , and ethanol. The mixture was kept in an ultrasonicator for 10 min with the Si substrate immersed in it. Then, the substrate was removed from solution and few droplets were added onto its surface before introduction into the horizontal quartz tube of the CVD reactor. A ceramic boat with 200 mg of sulfur was placed upstream in the low-temperature zone of the furnace, 27.5 cm from the flow inlet. Another 2 cm-thick boat was placed face-down downstream in the center of the furnace to exploit its thickness in our growth process. An excess of 20 mg MoO 3 was added on the top of the boat at the hot zone (50 cm from the flow inlet). The Si substrate (1 cm 2 ) was placed on the top of this boat 1 cm from the MoO 3 powder, as shown in Fig. 1a. Ultra-high purity Ar gas was flown in the furnace at the rate of 70 sccm during the whole growth process. The center of the furnace was heated from room temperature to 850 °C with a rate of 20 °C/min to achieve a nonhomogeneous temperature profile inside the quartz tube with an incomplete transformation from MoO 3 powder to MoS 2 film (Fig. 1b). The furnace was kept at the growth temperature for 30 min. All syntheses were done at atmospheric pressure. Finally, the furnace was allowed to cool down naturally to room temperature with 70 sccm Ar flow. The optical image of the sample surface is given in Fig. 1c showing the homogeneous deposition.
Characterization. An Olympus BX51M optical microscope was used to observe the morphology of the MoS 2 /MoO 2 using bright field mode. Scanning electron microscopy (SEM) (Quanta 200 FEG, ThermofisherScientific) was used to observe the microstructure and examine the nucleation mechanism of our samples growth. The energy-dispersive X-ray spectroscopy (EDS) mappings were carried out in a Scios 2 dual beam system (ThermofisherScientific) equipped with an EDS system (Oxford Instruments). Electron beam parameters of 10 kV energy and 1.6 nA beam current were implemented during investigations. Transmission electron microscopy (TEM) analyses were performed using a Titan G2 (ThermofisherScientific) operating at 300 kV. The tool has Cs corrected beam optics to reduce spherical aberration for ultra-high resolution imaging. A micro-Raman spectrometer (Renishaw) at an excitation wavelength of 532 nm and X-ray diffraction (XRD) using a D8 Discover diffractometry (Bruker) (K αCu = 1.54 Å) were employed to study the vibrational modes and the crystalline quality of the MoS 2 /MoO 2 heterostructures, respectively. X-ray photoelectron spectroscopy (XPS) analyses were carried out using a ThermofisherScientific K-alpha spectrometer and a PHI VersaProbe III scanning XPS microprobe to investigate the surface composition of our samples. The optical reflectance was measured using To perform photo-electric measurements on our samples, we first exfoliate the MoS 2 /MoO 2 nano-fibers and then transfer them onto a substrate with a circular interdigitated electrodes pattern 33 with 10 µm spacing. The effective detection area of our sample is equal to 5 × 10 -3 cm 2 .Atomic force microscopy (AFM) measurements were carried out in ambient conditions using a Dimension Icon system (Bruker, Santa Barbara, USA) in the peak force Kelvin probe force microscopy (PF-KPFM) method. Conductive Platinum coated silicon AFM probes (Spark150, NuNano, Bristol, UK), with a spring constant k = 20 ± 0.2 N/m, were used. The PF KPFM measurement enables the simultaneous characterization of the sample surface topography and surface potential variations.

Results and discussion
The CVD grown MoS 2 /MoO 2 heterostructure was first examined in SEM. The samples display a unique morphology consisting of microflowers attached to microfibers , as shown in Fig. 2a. A close examination of these structures (see sections below) indicate that the microflowers mainly consist of MoO 2 with an average diameter size of ~ 50 μm, and the microfibers correspond to MoS 2 with a length reaching up to few hundreds of micrometers. Intermixing phases were also observed, especially at the boundaries, as it will be identified later by EDS, HRTEM and KPFM. The inset of Fig. 2b provides a closer look at the morphology of the fabricated MoS 2 /MoO 2 heterostructures, showing MoO 2 microflowers seemingly at the sites of MoS 2 microfibers nucleation. This is well illustrated in the EDX elemental maps in Fig. 3. Individual elemental maps for Si, O, Mo, and S are separately shown, whereas their combined mapping was overlaid (color contrast) on the SEM image (bottom-left). The corresponding contrast provides clear indications that the microflowers are rich in oxygen whereas microfibers have a rather high sulfur content, which is signature of MoS 2 compound.
To examine the mechanism of MoS 2 /MoO 2 heterostructure nucleation and growth, the CVD reaction was interrupted at different times while maintaining the same growth conditions. This allows monitoring the CVD reaction kenitics and analyzing the evolving compound miscrostructure. Figure 4. depicts the MoS 2 /MoO 2 heterostructure after 5 min, 20 min and 30 min reaction time.
One can notice in Fig. 4a that there is first a nucleation sites, two in this micrograph, then random fibers are generated. 20 min later (Fig. 4b) several thicker fibers impinge from the nucleation sites and begin to make the connections between the nucleation sites. At last stage of the CVD reaction 30 min (Fig. 4c), the Si substrate is fully covered by very thick and long fibers as also shown in Fig. 3.    Optical properties of the MoS 2 /MoO 2 heterostructure samples were investigated by measuring their reflectance in the wavelength range of 200-1500 nm at room temperature. Figure 8a shows that the samples exhibit a very low reflectance over the entire investigated spectral range, increasing from 1% at 200 nm to only ~ 5.5% at 1500 nm. This very low reflectance is could be attributed to the change of MoS 2 morphology in the presence of MoO 2 , as previously reported 38,39 . A weak reduction is observed on the reflectance spectrum around 430 nm attributed to the electron transitions, which occur in the optical band gap. The obtained overall reflectance values were used to evaluate the optical band gap using the Kubelka-Munk function (F(R)), plotted using light absorbance in Fig. 8b. The plot of (F(R)hν) 2 versus photon energy (hν) and an approximation for direct band gaps of MoS 2 /MoO 2 allowed extracting two potential values of direct bandgaps, namely E g = 2.8 eV and E g = 1.8 eV, as shown in Fig. 8c.
The band gap E g = 2.8 eV could be ascribed to MoO 2 18,40 , whereas E g = 1.8 eV is consistent with the values quoted for direct bandgap of mono to few-layers of MoS 2 41,42 . The presence of two band gap energies would enhance the probability for electrons to jump to both MoO 2 and MoS 2 conduction bands once excited by an external light source. This might have likely led to the higher broadband light absorption recorded during the optical absorption measurements. Taking advantage of the presence of band gap energies, Z-scheme electron www.nature.com/scientificreports/ excitation may be also activated. The consequence of such optical properties is discussed further in the photodetection measurement section.
KPFM was used to characterize the surface potential the MoS 2 /MoO 2 heterostructure. Figure 9 shows topography maps acquired at different scan sizes. The flower-like structure previously observed in SEM is identified on the large scan are (80 µm × 80 µm) in Fig. 9a. Central volcano-like structures, distributed over the surface, from which elongated petal-like structures emanate in all directions forming microflower-like islands can be clearly observed.
The large topography map reveals domain boundaries between islands, which points towards the growth dynamics dictated by surface energies and thermodynamic conditions. Fade dashed lines delimiting one island boundaries (lower right part of the image in Fig. 9a) were overlaid on the topography map to guide the eye. Interestingly, the magnified topography map (scan area: 30 µm × 30 µm) in Fig. 9b reveals Fig. 9c. To investigate the local physical-chemically dependent characteristics of the coexisting structures, surface potential measurements were performed on the same spot in Fig. 9c. The surface potential (or contact potential difference, V cpd ) stems from the differences in work functions between the AFM probe and the sample. It is an extreme surface-dependent property, highly sensitive to minute variations in the surface chemistry, electronic and crystallographic properties. As described in the experimental section. The AFM electrical measurements, shown here, were conducted in the PF-KPFM mode enabling simultaneous characterization of the surface adhesion to the AFM probe. Figure 10a shows the surface potential variations on the same spot as in Fig. 9c. The observed contrast indicates the coexistence of three different materials with different surface potential values.   Fig. 10a). The surface potential measurements provide a nanoscale electrical property signature, confirming the observation made above regarding the co-presence of three regions forming the heterostructure.
To investigate the photoelectric properties of our MoS 2 /MoO 2 heterostructures, the samples were integrated into photoactive devices, and exposed to solar excitation using a solar simulator equivalent to one sun (i.e. 100 mW.cm -2 with AM 1.5G filter). Figure 11a shows J-V curves recorded under both dark (J dark ) and solar excitation (J light ). It is clearly seen that the J light under sun irradiation is higher than its dark counterpart, confirming the photoactivity of our MoS 2 /MoO 2 material. Figure 11b depicts the transient photocurrent density (ΔJ = J light − J dark ) response obtained at 1 V bias with successive ON/OFF cycles. A maximum photocurrent density is obtained after 20 s of sun exposure with a stable photocurrent density of 22 µA cm −2 indicating the high stable photoactivity of the heterostructure. On the J-V curve, a symmetry in the ON/OFF cycles with respect to the origin is observed, suggesting the presence of an ohmic contact between the heterostructure and the gold electrodes, in accordance with previously reported works. This reveals a good electrons' injection at the MoS 2 /Au interface compatible with the presence of an ohmic contact 12,13 , which can be partially explained by the good chemical affinity between the gold and sulfur atoms leading to very weak injection barrier. Nonetheless, J-V curves exhibit a linear variation for low voltages followed by a curvature for higher voltages. Assuming the contacts are indeed ohmic, such curvature is probably due to space-charge-limited current effects 43 . In addition, Fig. 11b shows that the photocurrent undergoes a slow increase under standard sun illumination, and conversely, it shows a slow decay under darkness condition. The observed slow kinetic behavior is an indicator of carrier trapping within the heterostructure under illumination followed by a thermal detrapping of carriers manifested by the slow decay 43 .
According to the absorbance spectrum obtained (see Fig. 8a,b) the MoS 2 /MoO 2 heterostructure exhibits high absorbance capacity in the blue region. To elucidate this optical behavior, we carried photoresponse investigations of our MoS 2 /MoO 2 -based device under 450 nm laser excitation. Figure 12a shows the typical J-V curves recorded in dark and under illumination at variable laser power densities. Our results show that the photogenerated current density under blue light excitation is continuously increasing with increasing power density in the full range between −1 and 1 V. At a power density of 125 mW cm −2 , the photogenerated current density is five times higher than the one recorded in dark. Hence, the high optical absorption of our MoS 2 /MoO 2 heterostructure is translated to higher photoconversion capacity. The stability and performance of our heterostructure-based photodetector were then further investigated. Figure 12b illustrates the transient photocurrent response steps during light ON/OFF cycles at increasing light power densities φ 0 . Results clearly show that the MoS 2 /MoO 2 heterostructure is stable over time and its photogenerated current is continuously increasing with increasing light power density and/or applied bias. To examine the behavior of the photocurrent change induced by the variation of the incident light intensity, the φ 0 dependence of the photocurrent is depicted in Fig. 12c. Two main behaviors can be identified as follows: (1) For φ 0 < 80 mW cm −2 the photocurrent density (ΔJ) follows the classical power law ΔJ = αΦ 0 n , where α is a wavelength-dependent constant and n ≤ 1 is a constant.
(2) For φ 0 ≥ 80 mW cm −2 there is a clear change in the slope of the photocurrent a steep increase of the photocurrent with increasing φ 0 . The obtained photocurrent was subsequently used to determine the responsivity and detectivity of our MoS 2 / MoO 2 -based photodetector according to the following equations: www.nature.com/scientificreports/ where I p is the generated photocurrent (i.e. difference between the current under illumination and dark current), A is the effective irradiated area, q is the electron charge and I Dark is the dark current.
The obtained responsivity and detectivity as function of the light power densities φ 0 are depicted in Fig. 12d. Our MoS 2 /MoO 2 -based photodetector exhibits maximum values for both responsivity and detectivity at 0.75 mA W −1 and 1.45 × 10 7 Jones, respectively, achieved at very low blue light intensity excitation of 20 mW cm −2 . It is woth noting that high optoelectronic performances based on MoS 2 were reported 44 , however, the fabrication routes cited in this study are tedious, consisting of several fabrication steps. Additionally their photodection responses were achieved on very small active area (~ 10 -7 cm 2 ) and at high applied bias (~ 20 V) compared to our measurements. Generally, our findings concur with the recently reported data as summarized in Table 1.
In this table, we have summarized MoS 2 /MoO 2 based photodetector performances while providing precision on the fabrication technique used, the excitation energy and the bias voltage. One can notice that our results is comparable to reported data, which indicates the ability of the one-step CVD process to compete with other fabrication techniques.
Nonetheless, Fig. 12d shows that as the incident φ 0 increases, both responsivity and detectivity are decreasing until reaching their minimum values at φ 0 = 80 mW cm −2 . Surprisingly, this decrease was followed by a slight augmentation in both the responsivity and detectivity values for higher incident light intensities. This could  contributes to the photocurrent generated when the power density of the light is high enough to pass through the oxide layer. Thus, up to 80 mW cm −2 , we observe a power law (e.g. Fig. 12c), beyond which we observe the contribution of the MoO 2 /MoS 2 heterojunction. Furthermore, The radiation-responsivity and detectivity of the MoS 2 /MoO 2 heterostructure was studied using various excitation wavelengths at 40 mW.cm -2 light power density and 1 V bias as shown in Fig. 13.
The heterostructure photodetection performance is validate for the entire visible spectral range. Nonetheless, it is higher for shorter wavelengths, which is in good agreement with the aforementioned optical measurements. Indeed, the heterostructure is six times more responsive at 450 nm than at 630 nm wavelength. Finally, it is worth noting that the photoresponse performances are measured on the entire sample (microflowers and microfibers). We believe that the photoresponse could be improved if it is recorded using the sole microfibers. Hence different techniques could be put in place to isolate one or few microfibers using mechanical and/or chemical exfoliation or nanofabrication using FIB-SEM to allow developing these microfibers onto photoconductive device.

Conclusion
In summary, complex MoS 2 /MoO 2 heterostructures were successfully synthesized using a facile one-step CVD process. The synthesized heterostructures consist mainly of MoO 2 microflowers of few 10 s microns in diameter, from which MoS 2 microfibers up to 100 s microns-long emanate in all directions. X-ray and electron diffraction techniques have revealed that the crystalline nature of MoS 2 and MoO 2 in the respective hexagonal 2H-MoS 2 and monoclinic m-MoO 2 structures. These MoS 2 /MoO 2 heterostructures were found to exhibit high broadband optical absorption over the entire 200-1500 nm spectral range. This high optical performance is ascribed to the presence of two bandgap energy values measured at 1.8 and 2.8 eV, consistent with those quoted for MoS 2 and MoO 2 , respectively. The strong optical absorption was exploited by integrating the heterostructure samples into functional photodetectors, and interestingly found to exhibit high photoresponsive over the 450-630 nm range. The highest responsivity and detectivity values of 0.75 mA W −2 and 1.45 × 10 7 Jones, respectively, were obtained under the blue light excitation at very low light illumination of 20 mW cm −2 . These results highlight the potential of these CVD-grown MoS 2 /MoO 2 heterostructures for strong broadband light harvesting and photodetection applications.