Piezoresistive effect in MoO3 nanobelts and its ... · 1 Piezoresistive Effect in MoO3 Nanobelts and its Application for Strain-enhanced Oxygen Sensor Xiaonan Wena†, Weiqing Yangab†,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Nano Res
1
Piezoresistive effect in MoO3 nanobelts and its application
for strain-enhanced oxygen sensor Xiaonan Wen, Weiqing Yang, Yong Ding, Simiao Niu, Zhong Lin Wang ()
Nano Res., Just Accepted Manuscript • DOI: 10.1007/s12274-013-0385-8
http://www.thenanoresearch.com on November 9, 2013
Abstract: MoO3 NBs of different properties are synthesized via PVD method. Characterization measures of XRD, TEM and SEM are adopted to examine their crystallographic structures as well as NB morphologies. Electrical measurement is performed and the profound piezoresistive effect in MoO3 is experimentally studied and verified. Factors that influence the gauge factor, such as NB size, doping concentration and atmosphere composition, is discussed and analyzed. Gas sensing performance is also tested on the device and it is demonstrated that by applying strain to the gas sensor, its sensing performance could be effectively tuned and enhanced. This study provides the first demonstration of significant piezoresistivity in MoO3 NBs and the firstillustration of a generic mechanism how this effect could be coupled with other electronic modulation measures for better device performance and broader material functionality. Key words: molybdenum trioxide; nanobelt; piezoresistive effect; oxygen sensor
2
One dimensional metal oxide nanomaterials have been extensively studied in recent years covering a wide range of materials and their related properties. The dominant oxides studied in the literatures are ZnO and TiO2 nanostructures, but there are only a number of studies aboutmolybdenum trioxide (MoO3), which has many extraordinary properties. First, α-MoO3, the thermodynamically stable form,has a layered structure belonging to theorthorhombiccrystal system (space group Pbnm (62)),[1] whichgives rise to their applications in energy storage devices like lithium batteries[2-4] and supercapacitors.[5-7]MoO3 is also known to be able to reversibly change color in response to applied voltage or electromagnetic radiation, leading to promising applications in smart windows and digital displays.[8-10]In studies of field electron emission, MoO3 hasbeen proven to be wonderful building block materials owning to itslow emission threshold voltage and good thermal stability against high emission current.[11-13]In addition, rich intercalation chemistry of MoO3 makes it a very good candidate for a wide range of chemical sensing applications.[14-16]Novel properties of monolayer, fewlayer MoO3 belts or filmshave also attracted research effort alongside the study of grapheneand graphene-like materials.[17-19]Last but not least,F. Li and Z. Chen recently predictedbandgap modulation via strainin monolayer MoO3by using density functional theory computations,[20] which implies profound piezoresistive effect.[21] If experimentally verified, electrical property modulation by strain can not only be used for electromechanical applications, but can also be potentially coupled into almost any of the above mentioned applications. In this work, MoO3 nanobelts (NBs)are grown by physical vapor deposition (PVD) in a tube furnace under two different atmospheres. Characterization methods of x-ray diffraction (XRD), transmission electron microscopy (TEM) and scanning electron microscopy (SEM) are adopted to examine the crystal structure as well as the morphology of the as-grown NBs under these conditions. Then NBs are distributed on kapton film substrates, after which singleNB devices are made by applying electrodes on both ends of aNB. Owning to the flexibility of the substrates, strain-electrical coupling effects are experimentally studied for the first time on MoO3NBs to confirm the existence of significant piezoresistive effect. The strength of the piezoresistive effect is then tuned by the growth condition as well as the electrical measurement environment and the underlying mechanism is discussed. Lastly, the oxygen sensing performance of the MoO3NBs is studied. It is worth noting that,as restricted by the requirement of sensing temperature, most previous works used hard, inflexible substrates,[14, 15, 22] making it impossible to study effects of external strain on the sensor device. Adoption ofkapton film, which is known to withstand relatively high temperatures, achieved the desired flexibility of the sensor device at 200 oC and as a result, the coupling effects of
3
piezoresistivity and chemical sensing are studiedfor the first time to our best knowledge. It isshown that the piezoresistive effect can be maneuvered to enhance the sensitivity of oxygen sensors based on piezoresistive materials like MoO3NBs. Firstly, ACS grade MoO3 powder (purchased from Sigma Aldrich) is loaded onto an alumina boat. Then the boat is inserted into a quartz tube, before theyare sent to the center of the furnace. In order to produce NBs of different properties for a more comprehensive study, two different gas environments are used in our experiment. One is to connect the synthesis system to external atmosphere, which will be referred to as ‘condition 1’ and the other is to maintain 20 sccm argon flow to the system while pumping it down to 10 Torr, which will be referred to as ‘condition 2’.The furnace is then heated up to 600 oC and maintained for 4 hours before it is naturally cooled down. XRD is firstly used to characterize the as-grown products. Pristine MoO3 powder wasscanned as a reference, shown in Figure 1(a).Then products from both growth conditions are collected from the inner-wall of the quartz tube where the MoO3 vapor nucleated and are transferredonto glass slides for XRD characterization. Spectra from condition 1 and condition 2 products are each shown in Figures 1(b) and 1(c).Both results have all their major peaks correspond to planes normal to the b axisofα-MoO3(a = 3.962 Å; b = 13.858Å; c = 3.697Å), indicating the as-grown MoO3 is also of α-phase. Moreover, both samples have very strongly preferred orientation that is normally seen in crystalline thin film. This implies that products can possibly be plates, flakes, belts etc.[2, 23, 24]that spontaneously align their b-planes in parallel tothe substrate. TEM characterization is subsequently performed on the smaples grown under condition 1 and condition 2. TEM, HRTEM images and diffraction patterns of condition 1 and condition 2 products are each shown in Figures 1(d)-(f) and Figures 1(g)-(i), from which it can be concluded that both products are single crystalline NBs in the form of α-phase MoO3, belonging to the primitive orthorhombic lattice system. Two in-plane axes are (100) and (001), namely the a-axis and c-axis and the longest b-axis, (010), is normal to the plane. Additionally, shadow image based on the diffraction pattern in Figure 1(i)further indicate that the NB grows along its shortest axis, the c-axis. A schematic demonstrating the detailed crystal structure is shown in Figure 1(k). The structure analysis above, however, doesn’t reveal the differences between the two groups of samples. Thus SEM is used to examine the morphology of as-grown NBs.Images of condition 1samples are shown in Figure 2(a)-(b) and images of condition 2 samples are shown in Figure 2(d)-(e). Although they share the same crystallographic structure, the morphologies are rather different.Samples grown under condition 1 tend to have smaller width and thickness compared to samples grown under condition 2. By measuring and statistically analyzing all NBs shown in Figure 2(a) and Figure 2(d), we obtain
4
that the averageNB width is 3.091 μm with their standard deviation at 0.844 μm for condition 1; the average NB width is 14.524μm with their standard deviation at 9.661μm for condition 2. To further reveal their differences via electrical characterization, devices are fabricated viapositioning as-grown NBs onto Kapton film substrates (thickness 125 μm) and covering the two ends of the NB with silver paste as electrodes,after which the as-fabricated devices are baked at 100 oC to remove the solvents from silver paste.Examples of devices made from the two types of NBs are shown by the optical images in Figure 2(c) and Figure 2(f). Due to the low conductivity of MoO3 under room temperature, electrical measurement is conducted at 200 oC. Furthermore, it is also necessary to conduct the measurement under vacuum considering chemically active gas species like O2 and H2O[25, 26] in the air can strongly influence the property of MoO3 at this temperature.Considering the above, a special vacuum chamber is designed asschematically shown in Figure 2(h), which is capable of maintaining vacuum, adjusting environmental gas composition, heating up to specific temperatures andexerting external force through a mechanical feedthrough. One end of the device is fixed to a bracket while the other end is fixed to a movable push rod. Lead wires are connected through feedthroughs to a function generator and current preamplifier. Under vacuum and200 oC, electrical measurements are firstly conducted on condition 1 device. In its strain-free state, the I-V curve is obtained and shown as the black curve in Figure 3(a). By gradually adjusting the position of the push rod, a series of different tensile and compressive strains are loaded onto the Kapton film substrate as well as the NB lying on its surface.The position changes of the rod are converted into a set of strains applied onto the NB[27] along itsc-axis,namely its growth direction. (Tensile strain is defined as positive and compressive strain is defined as negative; expansion or compression along c-axis concurs with compression or expansion along a-axis.)Corresponding I-V characteristics are shown by the colored curves in Figure 3(a). Then current values at -5 V, -3 V, -1 V, 1 V, 3 V and 5 V for each curve are collected and the average value of resistance as well as the standard deviation is calculated for each strain state and plotted as Figure 3(b).It can thus be seen that strain applied on the NB has an obvious effect on the overall resistance of the device with tensile strain decreasing its resistance and compressive strain increasing its resistance. Still under vacuum and 200 oC, condition 2 device is subsequently tested in the same way. ItsI-V characteristics under different strainsand the strain effect on its resistance are shown in Figure 3(c) and Figure 3(d) respectively. The same trend and effect is observed in this device but it differs from the previous one in that the strain freeconductivity is much largerwhile the tuning effect of resistance by strain ismuch smaller.
5
After observing the above phenomenon, an obvious question is whattheunderlying mechanism is.Piezotronic effect,[28-30] geometry effect and piezoresistive effect (sometimes geometry effect is considered as one source of piezoresistive effect) are three major mechanisms thatcan help explain the resistance dependence on strain. Piezotronic effect, however, could firstly be ruled out since α-MoO3 certainly has the center of symmetry in its crystal structure and is not a piezoelectric material. Secondly, geometry effect arises from Ohm’s law‘R = ρl/S.When the NB is subject to tensile strain, increased length l and decreased cross-sectional area S leads to increased resistance; when the NB is subject to compressive strain, resistance decreases for the same reason. Although this effect exists for all materials, it only negatively contributes to the strain modulation effect in our experiment and can also be ruled out. Therefore, it is reliable to conclude that the third mechanism, the piezoresistive effect, which mainly arises from strain induced bandgap modulation, isresponsible for the phenomenon observed in our device. After confirming piezoresistive effect for devices under vacuum, it would be interesting to see if changing the atmosphere will have anyfurther impact. Thus with the samecondition 2 device mounted to the measurement system, ultrahigh purity oxygen is injected into the chamber and raises the pressure to 70 Torr. A period of waiting time is allowed for the temperature to stabilize at 200 oC and for the oxygen to fully interact with the surface of MoO3NB.Then the same measurement method is used to acquire the I-V characteristics under the same set of strain levels, as shown in Figure 3(e).The resistance is then statistically analyzed and plotted in Figure 3(f). Remarkably, the strain modulating effect gets obviously stronger with the introduction of oxygen into the pre-vacuumed system. In order to have a direct,quantitative comparison of the piezoresistive effect of the three cases studied above, the ratio of resistance change to applied strain, namely the gauge factor, is calculated. It should be noted that strain and resistance does not have a linear relationship for devices tested here.Hence the ratio of resistance change to strain for all seven strain levels of the three cases are considered and the averaged gauge factors and corresponding standard deviations are listed in Table 1. It can therefore be concluded that when tested under vacuum,the condition 1device presents a much more profound piezoresistive effect, with its gauge factor averaged at 302, peaked at 407 compared to thecondition 2device, with its gauge factor averaged at 57.8, peaked at 78.7. However, when the chamber pressure is raised to 70 Torr by pure oxygen, the average gauge factor ofcondition 2 device is drastically increased to an average value of 299 and the largest gauge factor recorded in our study is as high as 441.
6
Understanding the phenomenon discovered in our experiment for MoO3NBs could be assisted by previous findings on other piezoresistive materials.Gauge factor of piezoresistive effect could be influenced by many factors such as the type of materials, level of crystallization, orientation of the applied strain, temperature and so on. One factor that is closely related to our work is the doping concentration. For single crystalline silicon, the conclusion is very mature that its gauge factor decreases with increasing doping concentration over a wide range.[31, 32] Several works on polysilicon also reported similar results but anomaly was found for lower doping concentration.[33, 34] With the advent of nano research, a new influential factor was revealed. Studies of silicon nanowires,[35, 36] silicon ultra-thin films[37] and graphene[38] indicate that the piezoresistive effect, largely irrelevant of size in microscale, starts to become stronger when one or more dimensions of the material reduces to sub-micro or nanometer range. Our experimental results could be explained accordingly by referring to the above studies. Firstly, during the growth of NBs in condition 2, the system is constantly purged with pure argonand pumped in the meantime, meaning the oxygen partial pressure will be minimal. Thus vaporized MoO3ispartiallyreduced, leading to significant oxygen vacancies and molybdenum interstitials contributing to a relatively higher level of n-type doping, as is also suggested by several other studies.[11, 39, 40]On the contrary, the growth systemfor condition 1 is connected to the outer atmosphere, meaning plenty of oxygen is in existence during the growth, suppressing the oxygen deficiency in as-grown NBs, thus leading to a much lower doping concentration. Secondly, as indicated by the morphology study in our work, condition 1NBs are on average much smaller in size than condition 2NBs. This is also true for the specifically selected NBs that were used to fabricate the devices.The above differences are schematically reflected in Figure 4(a)-(d) and by referring to the trendings from existing studies, we conclude that the higher gauge factor in condition 1 device is a result of lower doping concentration and smaller size. However, for a more comprehensive conclusion on the gauge factor dependence of size and doping concentrationforthe material of MoO3, more detailed studiesareneeded. Oxygen not only interferes with MoO3 during the growth process, it also actively interacts with MoO3NBs. There are mainly two mechanisms that may contribute to the interaction. One is the chemisorption of O2
- species onto theNBsurface[14, 41] and the other is the incorporation of oxygen atom into its surface crystal lattice.[26, 42] Both of these processes create depletion layers and narrow the conduction channel width, as is schematically shown in Figure 4(c). By the reasoning that smaller size (related to narrower channel width) and lower carrier concentration gives rise to stronger piezoresistive effect, the
7
much large gauge factor we observed for condition 2 device under 70 Torr oxygen is very much expected. Additionally, to interpret this phenomenon in a more sophisticated way, the two-terminal MoO3 NB device,when used as an oxygen sensor, can be considered equivalent to a three-terminal NB based depletion mode field effect transistor (FET). Adsorbing or incorporating oxygen species is equivalent to applying a negative bias to the gate electrode. Channel width is decreasedby the equivalent negative gate voltage, which consequently results in an increased gauge factor, as illustrated by Figure 4(d). Singh et al. studied similar effects on silicon nanowire based three-terminal transistors in 2012, whose results are in full consistence withthe phenomenon as well as our explanation in this work.[43, 44] At this point we have successfully demonstrated a three way coupling effect among electronic, mechanicaland chemical properties. As one can always expect, any coupling of three or more properties will greatly expand the variety of applications. The above discussion is mainly made from the perspective of a strain sensor. From now on, we will discuss about howto apply our findings for the benefit of a gas sensor. Firstly, original oxygen sensing performance of condition 2 device under strain free state is characterized. By gradually injecting oxygen into the pre-vacuumed system,I-V characteristics areconsecutively obtained for oxygen pressures of 10, 30, 50, 70, 90, 110 and 130 Torr, as shown in Figure 5(a).The dynamic response of the device is measured by periodically injecting oxygen into the chamber while the pump is kept at a low but constant pumping speed. The result is shown in the inset of Figure 5(a). By comparing the current values at 2 V, dependence of sensing performance on oxygen pressure is plotted in Figure 5(b). At the pressure of 10 Torr, the gas sensor current drops by 22.2% and at the pressure of 130 Torr, the gas sensor current drops by 88.2%, showing decent gas sensing performance at its strain free state. We already know that existence of oxygen promotes piezoresistive effect. This means when tensile strain is applied to the device, its current value under fixed voltage will increase relatively more in oxygen environment compared to vacuum; when compressive strain is applied to the device, its current value under fixed voltage will decrease relatively more in oxygen environment compared to vacuum. By plotting Figure 3(c) and Figure 3(e) together in the same coordinate system, we get Figure 5(c), in which solid lines are from the condition 2device under vacuum and dotted lines are from the condition 2 device under 70 Torr oxygen.It could be seen that the spread of the I-V curves set for the latter is obviously larger than that of the former under the same set of strains. By collecting the current values at 2 V, dependence of sensing performance on applied strain at fixed oxygen pressure is plotted in Figure 5(d). At zero strain, the sensitivity is 74.24%; at 0.24% strain (tensile), the sensitivity drops to 62.8%; at -0.18% strain (compressive), the sensitivity is increased to 82.6%. Over the range of 0.42% strain, the sensitivity of the gas sensor differs by as much as ~20% and we are
8
thus able to conclude that the device resistance can be maneuvered by external strain to either effectively degrade or enhance the sensing performance of a MoO3NB oxygen sensor. This conclusion can be safely extended tomany otherpiezoresistive semiconductor materials and target gases as long as their interaction modulates the conduction channel width. In summary, two kinds of MoO3NBs are synthesized via PVD methodunder two different conditions. Characterization measures of XRD, TEM and SEM are adopted to detailedly examine their crystallographic structuresas well as NB morphologies. Electrical measurement is performed and the profound piezoresistive effect in MoO3 is experimentally studied and verified. Factors that influence the gauge factor, such as NB size, doping concentration and atmosphere composition, is discussed and analyzed. Gas sensing performance is also tested on the device and it is demonstrated that by applying strain to the gas sensor, its sensing performance could be effectively tuned and enhanced. Two major novelties and contributions of this work are: adding the piezoresistive effect to the already long list of special properties of MoO3; exhibiting and explaining the three-way coupling effect amongelectronic, mechanical and chemical properties, promising more functional and sophisticated applications. The future perspective of MoO3 research is very bright and it would be exciting to see more work done on its electrical (e.g. field emission), optical (e.g. photochromism/electrochromism), mechanical (e.g. piezoresistance) and chemical (e.g. gas sensors/energy storage) properties and also the coupling of the above properties, a largely unexplored territory that offers a lot of research and application opportunities.
Acknowledgements:
References:
[1] King, R. B., Encyclopedia of inorganic chemistry. Wiley: Chichester ; New York, 1994.
[2] Mai, L. Q.; Hu, B.; Chen, W.; Qi, Y. Y.; Lao, C. S.; Yang, R. S.; Dai, Y.; Wang, Z. L. Lithiated MoO3
nanobelts with greatly improved performance for lithium batteries. Adv Mater 2007, 19, 3712‐+.
[3] Lee, S. H.; Kim, Y. H.; Deshpande, R.; Parilla, P. A.; Whitney, E.; Gillaspie, D. T.; Jones, K. M.; Mahan,
A. H.; Zhang, S. B.; Dillon, A. C. Reversible Lithium‐Ion Insertion in Molybdenum Oxide Nanoparticles.
Adv Mater 2008, 20, 3627‐+.
[4] Julien, C. M. Lithium intercalated compounds ‐ Charge transfer and related properties. Mat Sci
Eng R 2003, 40, 47‐102.
[5] Murugan, A. V.; Viswanath, A. K.; Gopinath, C. S.; Vijayamohanan, K. Highly efficient