**1. Introduction**

Fast and sensitive infrared (IR) detectors operating at room temperature are of tremendous interest for industrial, scientific, and military applications, including in security, environmental monitoring, remote controls, optical communication, thermography, and astronomy, as well as for the latest technologies, such as in self-driving cars and for obstacle avoidance in robots [1–5]. Traditional bolometers consist of an absorber and a sensor. During detection, thermal radiation is absorbed by the absorber, leading to a temperature increase, subsequently resulting in a change in electrical resistance in the sensor, which can be measured using electrical circuits. Then, through electrical signal processing, the temperature of the target object is obtained. Nowadays, the main commercial uncooled thermistor materials are amorphous silicon (a-Si), vanadium oxide (VO2), and germanium–silicon– oxide [6–8]. However, a-Si shows long response times of tens to hundreds of ms [9,10]. The production of VO2 causes grea<sup>t</sup> environmental pollution [11,12]. Furthermore, the commercialized uncooled bolometers require sophisticated designs such as micro-bridges or thermal insulation layers to obtain good thermal insulation [13,14], as well as an extra IR absorption layer [15] to ensure good photon absorbance. Thus, although high performance can be achieved with the above complex designs, more accessible uncooled bolometric materials with self-absorbing, self-thermal-insulating, and self-sensing properties are in grea<sup>t</sup> demand to increase the application of bolometers in real life.

Due to their broadband IR absorption and fast responses of up to picoseconds resulting from the ultrahigh carrier mobility and weak electron–phonon scattering, carbon nanotubes (CNTs) and their composites have attracted wide attention as some of the most promising

**Citation:** Wang, Q.; Wu, Y.; Deng, X.; Xiang, L.; Xu, K.; Li, Y.; Xie, Y. Preparation and Bolometric Responses of MoS2 Nanoflowers and Multi-Walled Carbon Nanotube Composite Network. *Nanomaterials* **2022**, *12*, 495. https://doi.org/ 10.3390/nano12030495

Academic Editor: Antonio Di Bartolomeo

Received: 29 December 2021 Accepted: 26 January 2022 Published: 31 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

candidates for flexible IR detectors [16–20]. However, the TCR (temperature coefficient of resistance) of CNTs is low, which makes simple resistive read-outs difficult. Itkis et al. reported a large bolometric photoresponse of suspended single-walled CNT (SWCNT) films with TCR values of 1% at 330 K and 2.5% at 100 K [21], which were close to those of a VO2 bolometer [22,23]. However, the large-scale production of single-walled carbon nanotubes (SWNTs) of high quality and purity is expensive and challenging, which limits their application [24,25]. The relatively cheaper price of MWCNTs and compromised but still excellent optical, electrical, and mechanical strength makes them a good candidate for bolometer applications. Randomly assembled MWCNT films synthesized by vacuum filtration is some of the most accessible forms of bulk CNT materials suitable for large-scale production and application. Nevertheless, for MWCNT films, the TCR was reported to be only 0.088%/K [26]. The high *k* and low TCR of pure MWCNT films result in weak temperature sensitivities and lead to poor bolometric performance. It would ideal if CNT bolometers could be used with simple resistive read-outs and could be manufactured without the use of high-quality CNTs or delicate microfabrication processes.

To improve the bolometric performance of CNTs, photothermal materials with high light absorption and TCR can be composited with CNTs such as graphene [17,18] and metal oxides (ZnO, VO2) [20,27–29]. Lu et al. achieved novel exciton dissociation of a graphene–MWCNT hybrid film through heterojunctions self-assembled at the graphene– MWCNT interfaces. This method significantly improved the responsivity of the CNTs in the near-infrared region [18]. Nandi et al. used a spray coating method to prepare a suspended bolometer based on an MWCNT coated with vanadium oxide. The suspended bolometer showed a high TCR of ~ −0.41%/K, which was ~4.86 times higher than that of the previously reported suspended MWCNT film [22]. Recently, it was reported that MoS2 with a flower-like or spiral-like shape showed excellent light absorption performance [30–32]. Tahersima et al. reported on the rolling of Van der Waal heterostructures of molybdenum disulfide (MoS2)–graphene (Gr)–hexagonal boron nitride (hBN) into a spiral solar cell, leading to strong light matter interactions and allowing for solar absorption up to 90% [31]. Yang et al. prepared an ultrathin 2D porous film for solar steam generation based on MoS2 nanosheets and an SWCNT film. Even at an ultra-thin thickness of about 20 nm, its absorption rate across the entire solar spectrum range exceeded 82% [30]. Thus, it is advantageous for CNTs to be composited with flower-like or spiral-like MoS2 to improve the bolometric performance.

In this work, MoS2 nanoflowers are composited with a CNT network via a facile self-assembling strategy. The CNTs act as a thermally and electrically conductive network, while the MoS2 nanoflowers not only enhance the broadband absorbance, but also influence the intertube coupling in the CNT network, resulting in an improved TCR value. The thermal and electrical transport properties over the temperature range of 296 K–320 K are investigated. The figures of merit of the free-standing composite network, including the photothermal performance, resistive responsivity [( *R*on − *R*off)/ *R*off], detection sensitivity to a wide spectrum ranging from ultraviolet to near-infrared, and response times are studied and compared with the pure CNT network in detail.

#### **2. Materials and Methods**

#### *2.1. Preparation of the CNT–MoS2 Composite Network*

The CNT network was purchased from XFNANO and was prepared from CNT powder by vacuum filtration. A piece of the CNT network with lateral dimensions of about 1 cm × 1 cm was cleaned with N2 plasma (200 W, 120 s). Sodium molybdate (Na2MoO4) and thiourea (CH4N2S) were dissolved in deionized water with magnetic stirring for 30 min to form precursors with two different suspension concentrations (Table 1). The resulting solutions were denoted solution 1 and solution 2, respectively. Next, the CNT network and the prepared mixture were put into a 100 mL autoclave and reacted at 200 ◦C for 24 h. The samples were then removed and washed with deionized water and dried in an oven at 60 ◦C for 12 h. This hydrothermal process can be used to assemble MoS2 flakes with flower-like or spiral-like nanostructures in the nm–um size range, which can significantly improve the light absorption performance [30,32–34]. Finally, the CNT–MoS2 composite network was annealed in a tube furnace at 900 ◦C under Ar atmosphere for 2 h with a heating rate of 2 ◦C/min. The reaction routes can be expressed as follows [35]:

Na2MoO4 2H2O + CH4N2S+H2O → MoS2 + NH3 + CH3COOH + NaOH

**Table 1.** Different concentrations of precursors used for synthesizing the CNT–MoS2 composite network.


#### *2.2. Structural Characterization Methods*

In order to characterize the micro-structures of this composite network, we used Xray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), Raman spectroscopy, and scanning electron microscopy (SEM). The SEM images were taken using a JSM-7800F TEAM Octane Plus instrument with a voltage of 10 kV. The XRD spectroscopy was carried out by using an Empyrean diffractometer (PANalytical, the Netherlands) with Cu K*α* radiation (λ = 1.54 Å) at a generator voltage of 45 kV and a generator current of 40 mA. The elemental composition and functional group analysis were tested using a Thermo Scientific K-Alpha XPS instrument. The Raman spectra were obtained using a Horiba LabRAM HR Evolution instrument. The UV–Vis–NIR spectrometer was used to characterize the absorbance of samples in the range of 300–2000 nm. The instrument was equipped with an integrating sphere to measure transmittance (*T*) and total reflectance (*R*), and finally to obtain the absorbance values (*A* = 1 − *T* − *R*).

#### *2.3. Characterization of the Thermal Diffusivity and TCR*

The transient electro-thermal (TET) technique was used to characterize the thermal diffusivity (*α*) of the samples. The CNT–MoS2 composite network was cut into long rectangular strips, then suspended between two aluminium electrodes (the size of the measured samples in this work is presented in Table 2). A small amount of silver paste was used to fix the ends of the strip onto the electrodes and to reduce the contact resistance [1]. Before the measurement, the sample stage was installed on a cold head in a closed-cycle cryostat system (Janis, CCS) where the environmental temperature was controlled from 320 K to 296 K. The environment temperature, provided through the temperature of the cold head of the cryogenic system, was used to for the characterization of the electrical and thermal properties. At the same time a vacuum environment was provided, in which the air pressure was maintained below 10−<sup>2</sup> Pa. The electrodes were connected in parallel with a current source (Keithley 6221) and an oscilloscope (Tektronix MDO 3054).

**Table 2.** Details of the samples measured in this study.


During the measurement, a step current was fed to the sample through a current source, causing a small and fast joule heating. Here, a one-dimensional heat transfer model can be assumed reasonably. Within a small temperature range, it can be assumed that the TCR of the sample is constant. Then, the normalized temperature can be obtained from the normalized voltage profile as: *T*\* = *<sup>V</sup>*\*=(*V*sample − *V*0)/(*V*∞ − *V*0), where *V0* and *V*∞ are the voltage of the sample before the joule heating and when it reaches steady state, respectively. Thus, the averaged normalized temperature *T*\*=[*T*(*t*) − *T*0]/[*T*(*<sup>t</sup>*→∞) − *T*0] can be derived as [36,37]:

$$T^\* = \frac{48}{\pi^4} \sum\_{m=-1}^{\infty} \frac{1 - (-1)^m}{m^2} \frac{1 - \exp\left[-m^2 \pi^2 a\_{massur} t / L^2\right]}{m^2} \tag{1}$$

where *m* is the normalized parameter, *αmeasure* is the thermal diffusivity of sample, *t* is time, and *L* is the suspended length of the sample. Based on Equation (1), the *αmeasure*can be obtained using MATLAB and via the least squares fitting of the *V*-*t* data. Different trial values of *α* are used for the fitting. The fitting errors were determined to be ±10% or better in our previous work based on the TET technique [36]. During the measurement, *R* is measured using the current source and the oscilloscope in 2-point configurations, with a small bias current (*I*) applied and voltage (*V*) probed. *R* is then calculated by *R* = *V*/*I*. TCR is then obtained by differentiating the *R*-*T* curve.

#### *2.4. Test of Bolometric Response*

In this process, the composite sample (S2, details shown in Table 2) is suspended between two silicon electrodes using the same method as that described in the TET characterization. Before the photodetection test, the whole sample is installed in a vacuum chamber, whose optical window is made of fused quartz. During the test, the suspended sample is fully covered by the laser spot. The laser power irradiated on the sample is adjusted using the laser output and an optical filter. The laser power is measured using an optical power meter (from Thorlabs company in this study). The power density is calculated by *<sup>P</sup>*/(π<sup>d</sup>2/4), where *P* is the laser power and *d* is the measured laser beam diameter, as illustrated in Figure S1 in the Supporting Information. The resistance response of the sample is collected using a 712 digital multimeter (KEITHLEY DMM7510). Upon laser irradiation, the resistances when the laser is turned on and off are denoted as *R*on and *R*off, respectively.

#### *2.5. Measurement of the Response Time*

To measure the transient resistive responses to the lasers, the 405 nm, 860 nm, 1064 nm, and 1550 nm laser outputs are used as the optical sources. The laser outputs are modulated to a 0.2 Hz square wave using a function generator. By applying a small DC current to the sample, with which no appreciable heating occurs, the two-point voltage profiles under the square-wave laser illumination can be recorded using the oscilloscope. In the comparative experiment, to measure the transient resistive response to the joule heating, a square-wave current of 16 mA in amplitude and 0.2 Hz in frequency is applied to the sample using the current source to check the response and to compare it with the response to the modulated laser. The transient resistive response (*V*-*T* profiles) is measured using the oscilloscope. Then, the normalized voltage can be obtained from *<sup>V</sup>*\*=(*V*sample − *V*0)/(*V*∞ − *V*0), where *V*0 and *V*∞ are the voltages of the sample before the heating or illumination and when it reaches the steady state, respectively. From the *V*\*-*t* curve, the response time is identified when *V*\* is decreased by 0.95.
