Applications to Chemical Gas Sensing
by Asha Rani
M.S in Electronics, May 2003, Banasthali
Vidyapith, India
Mason University, USA
of The George Washington University
in partial fulfillment of the requirements
for the degree of Doctor of Engineering
August 31, 2019
Dissertation directed by
ii
The School of Engineering and Applied Science of The George Washington University
certifies that Asha Rani has passed the Final Examination for the degree of Doctor of
Philosophy as of April 25, 2019 This is the final and approved form of the dissertation.
Synthesis and Characterization of 2H-MoTe2 for Electrical and Gas Sensing
Properties with Applications to Chemical Gas Sensing
Asha Rani
Mona E. Zaghloul, Professor of Engineering and Applied Science, Dissertation
Director
Can Korman, Professor of Engineering and Applied Science, Committee Member
Shahrokh Ahmadi, Assistant Research Professor for Engineering, Committee
Member
Albert V. Davydov, Project Leader, National Institute of Standards and
Technology, Committee Member
iii
All rights reserved
To my parents, who supported and guided me when needed.
v
Acknowledgement
Over the past years I have received support and encouragement from great number
of individuals. I would like to express my sincere gratitude to all of them. First of all, I am
extremely grateful to my academic advisor, Professor Mona Zaghloul, for her patient
guidance, enthusiastic encouragement and useful critiques of this research work. This work
was possible only because of the unconditional support provided by her with an amicable
and positive disposition and I consider it as a great opportunity to do my doctoral program
under her guidance and to learn from her research expertise.
I would like to express my very great appreciation to Dr. Albert Davydov for his
valuable and constructive suggestions during the planning and development of this research
work. His willingness to give his time so generously has been very much appreciated. Long
hours of brainstorming discussion on results, helped me learn the details of this dissertation.
Without his guidance and persistent help this work would not have been possible. I want
to thank him for his excellent cooperation and for all opportunities I was given to conduct
my research at National Institute of Science and Technology (NIST) to further my research.
Similar, profound gratitude goes to Dr. Sergiy Krylyuk, who spent countless hours
proofreading, guiding , insightful comments and listening to my talks about experiments
and results. His guidance helped me in all the time of research and planning of experiments.
I would like to thank Dr. Abhishek Motayed, who helped me through instructions
and invaluable advice to understand the sensors and its working techniques. His knowledge
and experience helped me plan my research work and guided me with positive energy to
move forward.
vi
I would like to thank my friends in ECE department at GWU and NIST for all the
great times we have shared. I am particularly thankful to Kyle DiCamillo, Dr. Ting Xie,
Dr. Guannan Liu, Dr. Baomei Wen, Dr. Ratan Debnath, Irina Kalish and Dr. Kalisadhan
Mukherjee in many technical discussions and issues. My special thanks to N5 Sensors Inc.
for an internship opportunity, which helped in my gas sensor project. Special thanks to
Audie Castello and Dr. Aveek Gangopadhyay for their support and help in understanding
the fabrication related issues and guidance in technical issues during my project.
I am very thankful to all committee members, each of whom has provided patient
advice and guidance throughout the research process. Thank you all for your unwavering
support.
Finally, I am grateful to my family for their endless support and letting me pursue
my career goals. I am blessed to meet all the friends and people who I have worked in this
wonderful journey.
Synthesis and Characterization of 2H-MoTe2 for Electrical and Gas Sensing
Properties with Applications to Chemical Gas Sensing
Molybdenum ditelluride (MoTe2) is a member of two-dimensional (2D) transition
metal dichalcogenides (TMDCs) family, which have two different stable structures: semi-
metallic (2H) and metallic (1T’) phase. Theoretically, 2H-MoTe2 has direct bandgap [1]
(~1.10 eV) for single and indirect bandgap [2, 3] of (~1 eV) for bulk layer respectively.
Single crystals of MoTe2 have been grown by chemical vapor transport (CVT) method [4]
using tellurium tetrachloride (TeCl4) and iodine (I2) as transport agent (TA). For simplicity,
MoTe2 crystals grown by TeCl4 and I2 TA are referred to as MoTe2-Cl and MoTe2-I2 in
this work respectively.
The CVT grown MoTe2 crystals were characterized by optical microscopy, Raman
spectroscopy, Scanning Electron Microscopy (SEM), Atomic Force Microscopy (AFM),
X-Ray Diffraction (XRD) and Secondary Ion Mass Spectroscopy (SIMS). All electronic
measurements of MoTe2 FETs were performed in ambient environment using probe station
and custom-built gas sensing setup.
Until now, a detail study of channel thickness in range of ~5nm – 70nm for MoTe2
devices is not found in literatures [5]. Thickness of channel plays an important role in
electrical properties of semiconducting materials. In this thesis, electronic properties of
field effect transistors (FETs) fabricated from exfoliated MoTe2 single crystals are
investigated as a function of channel thickness. The conductivity type in FETs gradually
changes from n-type for thick MoTe2 layers (>50nm) to ambipolar behavior for
intermediate MoTe2 thickness (between 50 - 12 nm) to p-type for thin layer (<12 nm) [6].
viii
Change in polarity of FET was not observed in MoTe2-I2 devices which showed only p-
type conducting behavior >5nm channel thickness. Change in polarity of MoTe2-Cl devices
as function of channel thickness was also verified by ammonia (NH3) sensing.
This thesis presents Gas Sensing by MoTe2 flakes. We studied the effect of
ultraviolet (UV) light on MoTe2 channel for gas sensing applications. An anomalous
behavior of current collapse is observed in all devices in presence of UV light. This study
explains the probable cause of current collapse in MoTe2 devices in presence of UV light.
We further investigated the sensing properties of fabricated device by exposing oxidizing
(NO2) and reducing (NH3) gases on MoTe2 surface. The study is conducted in compressed
air and N2 as carrier gas. The behavior of target gas (NO2 and NH3) sensing in presence
and absence of UV illumination is also studied to improve the sensing response.
ix
Chapter 2 …………………………………………………………………………….…....5
2.2 MoTe2 crystals and flake characterization …………………………………….…...8
2.2.1 X-Ray Diffraction (XRD) ..........................................................................…...8
2.2.3 Raman Spectroscopy ………………………………………………………...11
2.2.5 Transmission Electron Microscopy (TEM) and Scanning Electron
Microscopy (SEM) …………………………………………………………..14
2.3.1 Alignment marks on substrate ………………………………………...….….17
2.3.2 Exfoliation method of MoTe2 into flakes ……………………………….......20
2.3.3 Device Fabrication after exfoliation ……………………………………..….27
2.3.4 Annealing of samples ………………………………………………………..29
2.3.4 Conclusion ………………………………………………………………......30
Chapter 3……………………………………………………………………………...….32
3.1 Introduction ………………………………………………………………………………..32
x
3.3 Channel Thickness effect of MoTe2 FET prepared from TeCl4 and I2 transport
agent by CVT method………………………..…………………….…….……….47
3.4 Effect of contact metal on MoTe2 surface ……………………………….………..53
3.5 Conclusion ……………………………………………………………….………..58
Chapter 4 ………………………………………………………………………………..60
4.1 introduction ……………………………………………………………………….60
4.3 Results and Discussion ………………………………………………....…………66
4.3 Conclusion ………………………………………………………………….…......82
Chapter 5 ………………………………………………………………………………...83
5.1 Introduction ……………………………………………………………………….83
5.3 Gas Sensing Mechanism…….. …………………………………………………...87
5.4 Experiment …………………………………………………………………...91
5.5 Results and Discussion …………………………………………………………...99
5.5.1 NO2 Sensing behavior with N2 carrier gas ……………………….…….…...100
5.5.2 NH3 Sensing behavior with N2 carrier gas ……..…….………….………….106
5.5.3 NH3 sensing behavior with air as carrier gas ……….………….….…..……110
5.5.4 Sensing behavior under UV illumination ………………………………......112
5.5.5 Channel thickness sensing effect by MoTe2 flakes…………….…………...113
5.6 Conclusion …….……………………………………………………....…………114
Chapter 6…….………………………………………………………………………….116
6.1 Conclusion ………………………………………………………………….116
1.1 Schematic representation of the periodic table with highlighted transition metal
(blue) and chalcogen (yellow) elements that form layered MX2 materials……...…….2
2.1 Pictorial representation of crystal growth inside a horizontal chamber showing hot
and cold zone area. …………………………………………………………………....6
2.2 Crystals of MoTe2 inside ampoule after taking out from the chamber. ………..….……6
2.3 Schematic showing two phases (2H and 1T’) of MoTe2 ……………………………..7
2.4 (a) Left: The hexagonal surface of α-MoTe2, the top Te layer, and the middle Mo
layer are shown schematically (a = 3.52 Å); right: the Mo atoms are trigonal
prismatic coordinated by Te atoms. (b) Left: Distorted octahedral Te coordination
in β-MoTe2; right: schematic drawing showing of β-MoTe2; the surface Te layer
and the subsurface Mo layer are shown. The lattice parameters are a = 56.33 Å and
b = 3.47 Å M1 = 4.01 Å and m2 = 2.32 Å and t1 = 3.11 Å and t2 = Å are Mo-Mo
and Te-Te distances projected into the x-y plane ………………………………..........7
2.5 XRD patterns of polycrystalline MoTe2 powders synthesized at 1000 0C. 2H (red
curve) and 1T’ (blue curve) crystals phases were obtained by slow cooling or
quenching, respectively............…………………………………………..……………9
2.6 XRD patterns of 2H-MoTe2. Inlet shows the MoTe2 crystals. …………………………...10
2.7 XPS spectra of (a) Mo 3d and (b) Te 3d binding energies of MoTe2 (XPS : courtesy Berc
Kalanyan (NIST) ………………………………………………………………….......11
2.8 Raman spectra of 2H- MoTe2 – Cl flake as function of thickness which is in
agreement of few and bulk layer MoTe2 …………………………………….………12
2.9 Result of MoTe2 flake thickness via AFM …………………………………….…….14
2.10 High magnification HAADF-STEM image of 2H-MoTe2 single crystal, together
with the corresponding diffraction pattern in [001] zone axis and the overlapped
structural model. The red spheres represent the Te atoms while the green spheres
the Mo atoms. (Courtesy Alina Bruma, NIST.)…………………………….….…....16
2.11 SEM image of flake with Ti/Au channel …………………………………….….....16
2.12 Fabrication process steps for alignment marks on Si/SiO2 substrate ……………....18
2.13 Mask of patterned samples for exfoliation…………………………………………..19
xii
2.14 Schematic illustration of Au- exfoliation method………………………………..21
2.15 Schematic of process steps of plasma cleaned substrate exfoliation method……23
2.16 (a): Process flow of sample cleaning. (b): Plasma cleaner at NIST. (c): Process
flow of mechanical exfoliation using blue dicing tape………………………...…24
2.17 Steps showing completion of exfoliation method………………………………..25
2.18 Optical images of flake with varying thickness after exfoliation on pre-
patterned Si/SiO2 substrate………………………………………….…………....25
2.19 AFM images showing difference in Au-assisted and Plasma enhanced
exfoliation method. ………………………………………………………….…...26
2.20 Schematic of fabricated MoTe2 for back-gated FET measurement……...………28
2.21(a) Mask design (2.5 mm X 2.5 mm) of device size used for testing………………..28
2.21(b) SEM image of fabricated device. Inlet shows 70nm channel device with flake
used for measurements…………………………………………………………...29
2.22 Optical image of fabricated MoTe2 channel device for back-gate FET
measurement. (a) 50 nm MoTe2 channel, (b) 8nm MoTe2 channel…………...…29
3.1 Schematic diagram of device connection for back-gate MoTe2 FET…………....35
3.2 Current-Voltage (I-V) showing non-linear behavior of 4.3 nm channel MoTe2
device…………………………………………………………………………….35
3.3 The transfer characteristics showing the typical behavior of a (a) p-type, (b)
ambipolar, and (c) n-type MoTe2 FET used in this work………………………..36
3.4 The output characteristics of a typical (a) p-type, (b) ambipolar, and (c) n-type
MoTe2 FET used in this work……………………………………………………37
3.5 Current-Voltage (I-V) showing improvement in current value after vacuum
annealing at 350 C for 5 minutes. Thickness of MoTe2 flake is 4.3 nm .............38
3. 6 Transfer characteristic (Ids-Vgs) at Vds = 2V of p-type MoTe2 FET (7nm
Channel thickness) showing improved on/off ratio of MoTe2 FET after
vacuum annealing at 350 C…………………………………………………..….39
3.7 Transfer characteristic (Ids-Vgs) at Vds = 2V of ambipolar type MoTe2 FET
(19.2nm) channel thickness) showing improved on/off ratio of MoTe2 FET
after vacuum annealing at 350 o C for 5 minutes…………………………………..39
xiii
3.8 Transfer characteristic (Ids-Vgs) at Vds = 2V of p-type MoTe2 FET (70.8 nm
Channel thickness) showing improved on/off ratio of MoTe2 FET after vacuum
annealing at 350 C for 5 minutes…………………...………………...................40
3.9 Raman spectra of MoTe2 flakes of different thickness exfoliated from bulk
MoTe2 Single crystals grown by CVT with TeCl4 transport agent. Inset shows
plan-view optical image of FET device with a 50 nm thick channel. ……….…..41
3.10 Field effect mobility of electrons and holes vs channel thickness for MoTe2
FETs. Encircled pairs of data points corresponds to the ambipolar devices
that exhibit ambipolar conductivity....…………………………...………….…...42
3.11(a) Transfer characteristic of thin channel MoTe2 showing p-type conduction
Behavior …………………………………………………………………….…...43
behavior…………..…………….………………………………………………..43
3.11(c) Transfer characteristic of thick channel MoTe2 showing n-type conducting
behavior ……………………………………………….………………………………..43
3.12 Secondary ion mass spectroscopy (SIMS) measurement showing the presence
of Cl in MoTe2 flakes……………………………………………………...……..44
3.13 Field effect mobility of electrons and holes vs channel thickness for MoTe2
FETs. Encircled pairs of data points corresponds to the ambipolar devices
that exhibit Ambipolar conductivity…….………….………………....................46
3.14 Raman spectra of MoTe2 flakes grown using TeCl4 and I2 transport agents…….48
3.15 Transfer characteristics at Vds = 2 V of (a) TeCl4-grown and (b) I2-grown
MoTe2 FETs with different channel thicknesses as indicated. Drain current
was normalized by the channel length to account for devices with 4µm and
5µm gaps between source and drain contacts……………………………………49
3.16 SIMS measurement of (a) TeCl4 and (b) I2 assisted growth……………………..50
3.17 Field-effect mobility of electron and holes vs channel thickness for (a) MoTe2-
Cl and (b) MoTe2-I2 FETs. Encircled pairs of data points correspond to the
devices that exhibit ambipolar conductivity ……………………………….........51
3.18 Response of 500 ppm NH3 in air of 68 nm (top x-axis) and 5.6 nm (bottom
x-axis) thick MoTe2-Cl devices confirming the n- and p-type conductivity
receptively. Note much faster recovery time for the thin flake...………..………52
xiv
3.19 Transfer curve showing polarity change as a function of MoTe2 channel
thickness. As the thickness of channel increases, polarity of back-gate FET
changes from p-, ambipolar to n-type conduction behavior. Figure (a) shows
Ti/Au, (b) shows Cr/Au and (c) shows Pd/Au …….………………………....….55
3.20 A qualitative schematic of a semiconductor with surface states that pin the
Fermi level in the case of (a) no doping, (b) p-type doping, (c) n-type doping.....58
4.1 (a) Raman spectra, (b) AFM image and (c) height profile of flake used in this
work…….…………………………………………………………………...…...61
4.2 (a) Cryogenic Probe Station and (b) Agilent B1500A semiconductor parameter
analyzer…………………………………………………………………………..62
4.3 (a) An optical image of packaged MoTe2 FET on ceramic chip carrier. Ruler in
cm shown for scale (b) An optical image of a wire bonded MoTe2 FET
showing the gold wires connecting the die bond pads to the chip holder bond
pads…....................................................................................................................63
4.4 Custom built UV sensing setup used for measurement. (a) top-view and (b)
side view…………………………………………………………………………63
4.5 Process step to deposit Al2O3 for passivation layer……………………………...64
4.6 Three – dimensional schematic view of the backgate FET based on MoTe2……66
4.7 Comparison of (a) Current-voltage(I-V) and (b) transient curve of ambipolar
MoTe2 flake before and after UV illumination. (c) Transient curve of MoTe2
with ambipolar conducting behavior showing sharp increase and gradual
decrease in the current value in ambient environment …………………………..68
4.8 (a) Output and (b) Transfer characteristic of 28.3 nm MoTe2 channel showing
p- type conducting behavior……………………………………………………...69
4.9 IV characteristic of 28nm thick MoTe2 channel measured in air, N2 and
N2+UV…………………………………………………………………………...70
4.10 UV response of 5.6 nm channel MoTe2 device (a) Current decreases at UV
Illumination and saturates. (b) UV increases the current once device is
Saturated…………………………………………………………………………71
4.11 Transient response of MoTe2 channel in dark, ambient and presence of UV……72
4.12 Transient response of MoTe2 channel at 5V bias voltage as function of intensity
(71 mW/cm2, 142 mW/cm2 and 213 mW/cm2). (a) Response versus time
showing 213 mW/cm2 intensity stabilizes faster after UV illumination
xv
compared to lower intensity. (b) Current versus time curve showing increase
in dark current as functionof intensity……………………...……….....................74
4.13 (a) Transient response of MoTe2 channel at 71mW/cm2 intensity as function of
Bias voltage. (b) Transient current curve for 28 nm channel thickness. ……………..….75
4.14 Transient curve showing UV response when sample is reset for (a) 24 hr, 12 hr
and 2 hr at 71mW/cm2 intensity and (b) 12 hr and 4hr at 213 mW/cm2 at
5V respectively…………………………………………………………………...75
4.15 (a) Transient curve showing UV response when sample is reset for 12 hr, 4 hr
and (b)Transient response curve showing UV response……………….….……..76
4.16 Hysteresis of devices measured for various channel thickness……………….....78
4.17 (a) Response vs time curve for un-passivated device showing UV
illumination effect. ……………………………………………………………...79
4.18 (a) Response (%) vs time curve of Al2O3 passivated device (b) Transfer curve
of passivated device showing n-type behavior………………………...…………80
4.19 Schematic diagram of the carrier transfer in the MoTe2 adsorbed by moisture
molecules in dark and UV light………………………………………………….81
5.1 Schematic diagram of sensing behavior…………………………………………89
5.2 Schematic of the charge transfer processes for MoTe2 in the presence of NO2
(a) and NH3 (b) molecules…………………………………..…………………...90
5.3 Energy band diagram of (a) NO2 sensing and (b) NH3 sensing by p-type MoTe2
surface…………………………………………………. ………………………..91
5.4 (a) Packaged device for gas sensing experiment. (b) An optical image of a wire
bonded MoTe2 FET showing the gold wires connecting the die bond pads to
the chip holder bond pads…………………………..………………...…….........92
5.5 Side picture of the gas sensing chamber showing the breadboard holding
the sensing device with the stainless-steel chamber fixed over the top. The UV
lamp head used to illuminate samples is also pictured…………………….……94
5.6 Top view picture of the stainless-steel sensing chamber with an MoTe2 device
packaged on a ceramic chip holder inside………………………...……………..95
5.7 Schematic of gas sensing measurement setup……………………………...……95
5.8 A sensing cycle marking the response and recovery time parameters as
xvi
described in this section……………………………………………………….....98
5.9 Current-Voltage curve of 28 nm device in air, N2 and N2+UV environment…………..100
5.10 Gas sensing mechanism of NO2 on p-type MoTe2 surface. ……………...…….101
5.11 The NO2 molecules are acceptors of electrons………………………………………....101
5.12 Transfer curve of 21.9 nm MoTe2 device …………………………..………….102
5.13 Response (%) of NO2 by 21.9 nm thick MoTe2 device in N2 as carrier gas…...103
5.14 Transfer curve of 21.9 nm MoTe2 device………………………………………104
5.15 Response (%) of NO2 by 28.3 nm thick MoTe2 device in N2 as carrier gas…....104
5.16 Transfer curve of 10 nm MoTe2 device……………..………………………….105
5.17 Response (%) of NO2 by 10 nm thick MoTe2 device in N2 as carrier gas…..….105
5.18 Response (%) of NH3 by 21.9 nm thick MoTe2 device in N2 as carrier gas…....107
5.19 Response (%) of NH3 by 28.3 nm thick MoTe2 device in N2 as carrier gas…....108
5.20 Response (%) of NH3 by 10 nm thick MoTe2 device in N2 as carrier gas….…..109
5.21 Response (%) of NH3 by 21.9 nm thick MoTe2 device in N2 as carrier gas……110
5.22 Comparison of response (%) for 21.9 nm MoTe2 device for 300 ppm
Concentration of NH3 sensing in air and N2 environment……………………...111
5.23 Response (%) of 5 and 10ppm concentration NO2 sensing in (a) N2 and (b)
N2+UV illumination by 21.9nm MoTe2 flake……………………………….....111
5.24 Effect of MoTe2 channel thickness on NO2 sensing…………...……………….114
5.25 NO2 sensing in different environment. Sensitivity value is obtained along the
line in the graph……………………………………………………...………....115
5.26 NH3 sensing in different environment. Sensitivity value is obtained along
the line in the graph…………………………………………………………......117
xvii
List of Tables
Table 2.1 Chemical Vapor Transport growth of MoTe2-Cl and MoTe2-I2….………….…6
Table 3.1 Summary of polarity change in MoTe2-Cl and MoTe2-I2 with varying
Channel thickness ………….………………………………………………....49
Table 4.1 Summary of UV illumination testing of MoTe2 flake in different condition....72
Table 5.1 Summary of concentration values used for NO2 sensing at CH1 and CH2 in
LabView Program ……………………………………………………………96
Table 5.2 Summary of concentration values used for NH3 sensing at CH1 and CH2 in
LabView Program …………………………………………………………...96
Table 5.3 Summary of 21.9 nm response to NO2 in N2 as carrier gas ………………....103
Table 5.4 Summary of 28.3 nm response to NO2 in N2 as carrier gas ………………....105
Table 5.5 Summary of 10 nm thick MoTe2 device in N2 as carrier gas. ……………....106
Table 5.6 Summary of NH3 by 21.9 nm thick MoTe2 device in N2 as carrier gas ….....107
Table 5.7 Summary of NH3 by 28.3 nm thick MoTe2 device in N2 as carrier gas…..…108
Table 5.8 Summary of NH3 by 10 nm thick MoTe2 device in N2 as carrier gas……….109
Table 5.9 Summary of NH3 by 21.9 nm thick MoTe2 device in N2 as carrier gas......…110
Table 5.10: Summary of NO2 sensing in different environment………………….…....115
Table 5.11 Summary of NH3 sensing in different environment......................................117
1
1.1 Background
With the exfoliation of graphene, there is world-wide research interest in the atomically
thin layered materials over the last decade. The rapid pace of progress in graphene and
methodology developed to obtain ultrathin layers has led to the exploration of two-
dimensional (2D) materials. Particularly, single layers of transition metal dichalcogenides
(TMDs) similar to graphite have received significant attention because some of these
materials are semiconductor with sizable band gap. The absence of surface dangling bonds,
excellent gate electrostatics, dielectric-mediated mobility and reduced short channel effects
are all desirable attributes of TMDs for transistors applications. The absence of band gap
for graphene limited its usage in logic circuits. Due to absence of band gap, it is difficult
to control the on/off current in graphene based transistors. Though, graphene has ultrahigh
mobility, [7]. Graphene while being fundamentally and technologically interesting for a
variety of applications, is chemically inert and can only be made active by functionalization
with desired molecules, which in turn results in the loss of some of its exotic properties [8].
In contrast, single layered 2D TMDs exhibits versatile chemistry. It offers opportunities
for fundamental and technological research in a variety of fields including catalysis, energy
storage, sensing and electronic devices such as field-effect transistors (FETs) and logic
circuits [9, 10]. Advances and limitations in graphene research attracted researchers
towards layered transition metal-dichalcogenide (TMDs). The layers in these materials are
governed by weak van der Waals, (vdW) forces which can be thinned to few or monolayer
by appropriate methods [11]. TMDs are in the form of MX2, where M is transition metal
2
and X is S, Se, or Te (Fig. 1) [12]. Many of these compounds have a band-gap enabling
strong ON/OFF current modulation. These materials have a typical structure with
sandwiched layers of the form X-M-X separated by vdW type bonding between the
adjacent layers. Each layer are covalently bounded while layer stacks are coupled by van
der Waals forces (vdW) [11, 13]. These weak vdW forces enable exfoliation of TMDs in
layers from bulk crystals.
Figure 1.1: Schematic representation of the periodic table with highlighted transition metal
(blue) and chalcogen (yellow) elements that form layered MX2 materials [12].
As a member of TMD family, Molybdenum Ditelluride (MoTe2) has a favorable band gap,
particularly for tunnel field-effect transistors (TFETs) [14, 15, 16]. Theoretical band gaps
for bulk and single layer MoTe2 are ~1 eV (indirect) and 1.10 eV (direct), respectively [1-
3]. Due to single and few layers MoTe2 holds promise for use in easily controllable
ambipolar FETs and extends the operating range of TMD optoelectronic devices from the
visible to the near-infrared range. Particularly, the band gap, which is close to that of Si
(~1.1 eV), the strong absorption throughout solar spectrum and the strong spin-orbit
3
coupling suggest that MoTe2 is highly attractive materials for use in electronic,
photovoltaic, and spintronic and valley-optoelectronic devices [17]. Recent
photoluminescence experiments show that the exciton binding energy is around 0.6 eV,
placing the direct bandgap for mono-layer (ML) MoTe2 at 1.8 eV [1]. A high on/off current
modulation of up to 106 with low subthreshold swing (SS) of ~ 140 mV dec-1 has been
reported [16]. Recent work has demonstrated realization of integrated logic circuits and
amplifiers based on layered MoTe2 transistors [15]. Photoresponsivitiy have been observed
in MoTe2 transistors [15]. Out of 14 TMDs Compounds [18], MoTe2, HfSe2 and ZrSe2 are
more promising 2D semiconductors compared to MoS2 due to their possible high carrier
mobility and finite band gap. The phonon limited mobility can be above 2500 cm2V-1s-1 at
room temperature.
1.2 Research Dissertation and Outline
In this thesis work, electrical and gas sensing properties of MoTe2 is studied.
Although many work has been done in the material characterization, field-effect transistors
(FETs) behavior and photosensitivity of photodetectors of MoTe2, very few study related
to the electrical properties and gas sensing behavior of varying channel thickness of MoTe2
devices can be found. In the current literature of MoTe2 FETs, there is no mention of
change in polarity from p- to ambipolar to n-type behavior with increasing thickness. This
thesis explains the polarity change of MoTe2 back-gate devices along the results are shown
with experimental measurements [6]. In addition, gas sensing properties of MoTe2 are
studied. MoTe2 bulk crystals were prepared by chemical vapor transport (CVT) method
using two different transport agents: TeCl4 (tellurium tetra chloride) and I2 (Iodine) referred
to a MoTe2-Cl and MoTe2-I2 in this thesis respectively. In chapter-2, material
4
characterization of MoTe2 is examined. Growth of MoTe2 crystals using CVT is described
followed by characterization using X-ray diffraction (XRD), Raman Spectroscopy and
Atomic force microscopy (AFM). Exfoliation method using plasma assisted is explained
followed by device fabrication process for electronic measurements. Chapter-3 describes
about the electrical characteristics of the fabricated devices. I-V characteristic curves, and
transfer input/output curves are used to explain the FET conducting behavior of devices
with varying channel thickness. Control of polarity as a function of channel thickness is
studied. Also, the effect of different contact metal is described to understand the
conductivity of MoTe2-Cl and MoTe2-I2 FETs. In chapter-4, effect of UV illumination on
MoTe2 surface is explained in N2 and in air environment using transient curves. Chapter-
5, describes about the gas sensing measurement technique of MoTe2 devices. Response to
gases based on FET polarity is shown in the transient curves obtained from the sensing
measurements. Improvement of sensing behavior is observed in presence of UV light. The
thesis is concluded with chapter-6, as conclusion and future work of the studied results in
this thesis.
Chapter 2: MoTe2 Crystal Material Preparations and Characterizations
This chapter covers the growth of the MoTe2 crystals for two difference phases [19]
semiconducting (2H) and semi-metallic (1T’) forms, material characterization using X-ray
diffraction, XPS, Raman Spectroscopy, Atomic Force Microscopy , Transmission Electron
Microscopy and Scanning Electron Microscopy, exfoliation method of MoTe2 flakes and
fabrication of devices for electronic measurement.
2.1 Growth and structure of MoTe2 crystals
MoTe2 single crystals were grown by CVT method using TeCl4 (MoTe2-Cl) and I2 (MoTe2-
I2) as transport agents. First, MoTe2 powders were synthesized by heating a stoichiometric
mixture of high purity molybdenum (Mo) and tellurium (Te) powders in sealed evacuated
quartz ampoules. The ampoules were slowly heated to 1000 C, kept at that temperature
for 72 hours and then either slowly cooled at a rate of 20 C/h to room temperature (RT)
or quenched in ice-water. To obtain MoTe2 single crystals, approximately, 2 gms of
polycrystalline MoTe2 powder and a small amount of TeCl4 (6mg/cm3) or I2 (4 mg/cm3)
were sealed in evacuated 17 cm long quartz ampoules. The ampoules were placed in a
furnace containing a temperature gradient so that poly-MoTe2 charge was kept at 1000 C
and temperature at the opposite end of the ampoule was about 800 C (MoTe2-Cl) and 940
C (MoTe2-I2) respectively. The 2H (semiconducting) or 1T’ (semi-metallic) phase
crystals are obtained after 7 days of growth by slow cooling (20 C/h) or ice-water
quenching of ampoules respectively. Cooling rate defines a crystal structure of MoTe2
single crystals, regardless of initial crystal phase of the poly-MoTe2 powders. Slow and
6
fast cooling rate yields 2H and 1T’ phases. Figure 2.3 shows the lattice structure of 2H and
1T’. Table-1 summarizes the CVT method growth conditions of 2H phase MoTe2-Cl and
MoTe2-I2 crystals. Figure 2.1 shows the schematic of chamber for making MoTe2 crystals.
Figure 2.1 shows the crystals growing inside the ampoule and fig. 2.2 shows the
representative picture of CVT grown MoTe2 platelets. The 2H structure comprises of two
triangles in the same orientation whereas, in 1T’, the two triangles are in the opposite
orientation around the central Mo atoms as shown in fig. 2.3.
Table 2.1: Chemical Vapor Transport growth of MoTe2-Cl and MoTe2-I2
MoTe2 Crystal
MoTe2-I2 I2 (5mg/cm3) 940 C 179
Figure 2.1: Pictorial representation of crystal growth inside a horizontal chamber showing hot and
cold zone area.
Figure 2.2: Crystals of MoTe2 inside ampoule after taking out from the chamber.
7
Figure 2.3: Schematic showing two phases (2H and 1T’) of MoTe2.
Figure 2.4 (a) Left: The hexagonal surface of α-MoTe2, the top Te layer, and the middle Mo layer
are shown schematically (a = 3.52 Å); right: the Mo atoms are trigonal prismatic coordinated by
Te atoms. (b) Left: Distorted octahedral Te coordination in β-MoTe2; right: schematic drawing of
β-MoTe2; the surface Te layer and the subsurface Mo layer are shown. The lattice parameters are
a = 56.33 Å and b = 3.47 Å. M1 = 4.01 Å and m2 = 2.32 Å and t1 = 3.11 Å and t2 = Å are
Mo-Mo and Te-Te distances projected into the x,y plane[20].
(a)
(b)
8
Figure 2.4 (a) shows the α-MoTe2 structure in which one sandwich consists of three two-
dimensionally, hexagonal packed layers: a top and a bottom Te layer and a central Mo
layer. They all have the same in-plane lattice constants a = 3.52 Å. The Mo atoms in the
center are trigonal prismatic coordinated with Te atoms. In 1T’ phase, the layered structure
is conserved but the coordination of the Mo atoms in now a distorted octahedron of Te
atoms, where the Mo atoms are slightly shifted from the octahedron’s center. The Mo rows
within the subsurface are shifted towards each other, forming slightly buckled paired
zigzag chains along the b axis leading to a buckled Te surface layer, because the Te atoms
above the paired chains are shifted upward out of the hexagonal plane. This clustering of
the Mo atoms is responsible for the metallic behavior of β-MoTe2.
2.2 MoTe2 crystals and flake Characterization
This section explains about the characterization tools which were used for study of
structure and morphology of MoTe2 crystals and flakes. The 2H-phase MoTe2 crystals were
in-house grown at NIST (courtesy Sergiy Krylyuk, MML, NIST) in a custom-built
laboratory setup. Material characterization of the crystals are important to determine the
quality and phase (2H or 1T’) of flakes.
2.2.1 X-Ray Diffraction (XRD)
For the powder XRD study, MoTe2 crystals were finely grounded using an agate
mortar. The θ - 2θ XRD patterns were derived from Norelco Philips Diffractometer with
the Bragg-Brentano geometry. Lattice parameters were refined using the MDI-JADE 6.5
software package [3]. Figure 2.5 shows the XRD patterns acquired for raw MoTe2 powders.
9
The 1T’ phase was identified for quenched powders whereas slow cooling yielded mostly
the 2H phase crystals. The platelets were milled in order to reliably distinguish the two
planes. For example, a characteristic feature of the 1T’ phase is a large number of low –
intensity peaks around 35 C, which are not inherent to the 2H phase.
Figure 2.5: XRD patterns of polycrystalline MoTe2 powders synthesized at 1000 0C. 2H (red curve)
and 1T’ (blue curve) crystal phases were obtained by slow cooling or quenching, respectively.
Figure 2.6: XRD pattern of 2H-MoTe2. Inlet shows the MoTe2 crystals.
10
2.2.2 X-ray photoelectron spectroscopy (XPS):
The surface chemistry of the exfoliated MoTe2 crystals is investigated with XPS as
shown in fig.2.7. Sharp peaks for Mo-Te binding energy is present at 228.2 (Mo 3d5/2),
231.4 (Mo 2d3/2), 572.9 (Te 3d5/2), and 583.3 eV (Te 3d3/2) [34, 35]. The Te/Mo atomic
ratios obtained from the XPS analysis of the Te 3d5/2 and Mo 3D core level regions vary
within the range of 2.1-2.4, thus suggesting either a tellurium – rich (e.g, Te interstitials or
intevalates) or a molybdenum-deficient (e.g., Mo vacancies or substitutions) chemical
environment [5].
Figure 2.7: XPS spectra of (a) Mo 3d and (b) Te 3d binding energies of MoTe2. (XPS: courtesy
Berc Kalanyan (NIST)
2.2.3 Raman Spectroscopy
The crystal structure of MoTe2 is characterized by the strong ion-covalent bonds within the
planes of hexagonally arranged M (transition metal from group IV-VII, Mo, W Nb and Re)
and X (chalcogen elements, S, Se and Te) atoms and by weak out-of-plane vdW
interactions between the planes [21]. Raman scattering is among the most conventional and
11
fundamental techniques for studying TMDCs [22-23]. It can determine the structure and
layer number in an easy and nondestructive manner. The Raman scattering spectra of few-
layer TMDs strongly depend on the flake thickness. Raman spectroscopy provides the
information on the electron-phonon interactions, in which the resonance between
excitation or scattered photons and electronic excitation is exploited.
For this study, Raman spectra were acquired from the 2H-MoTe2 flakes exfoliated
onto SiO2/Si substrate. Optical spectroscopic measurements were performed using Horiba
Scientific tool at NIST, in the backscattering geometry using 532 nm laser and 600 grating
parameters at room temperature. There are three major peaks observed in that range, which
are due to first-order transitions at the Γ point of the Brillouin zone: the out-of-plane mode
A1g at ~150 cm-1 [24], the in-plane mode E1 2g at ~234 cm-1 and the bulk inactive mode B1
2g
at ~290 cm-1. The B1 2g vibrational mode is optically inactive in bulk MoTe2. The energy of
this mode systematically increases with decreasing layer’s thickness as shown in fig.2.8
for MoTe2-Cl flakes. In that mode, both Te atoms in each layer move at particular time in
the same direction while the central Mo atom moves in the opposite direction. The increase
in the number of layers results in increase of possible vibrational modes. From previous
literatures, it is observed that B1 2g is absent in bulk and single layer flakes. It is observed
to have maximum intensity for the bilayer and decreases with increasing thickness. Similar
to the other dichalcogenides, the E1 2g is dominant mode and downshifts with decreasing
thickness [25]. The relative intensities of E1 2g and B2g peaks can be used to determine the
thickness.
12
Figure 2.8: Raman spectra of 2H- MoTe2 – Cl flake as function of thickness which is in
agreement of few and bulk layer MoTe2.
The Raman spectra for the 2H flakes in fig. 2.8 exhibits an out-of-plane A1g mode
around 170 cm-1, an in-plane E1 2g mode around 235 cm-1 and an out-of-plane B1
2g mode at
289 cm-1. All lines are in excellent agreement with Raman studies reported in early
literatures [26, 27].
2.2.4. Atomic Force Microscopy(AFM)
AFM plays a major role in exploring the basic principles underlying the
functionality of 2D materials. The transitional contact and tapping modes provide stable
control of ultralow imaging force, thus preserving both the AFM tips and fragile samples.
In basic form, AFM involves scanning a sharp probe across a sample surface while
capturing information about the surface properties. In dynamic modes, such as tapping,
non-contact and peak-force tapping, the top oscillates and the feedback is given by the
13
amplitude, frequency, or maximum force at the contact point. The most common mode
used to image and measure the thickness of layered material is tapping mode AFM. The
exfoliated MoTe2 layers were transferred onto SiO2 substrates. The thickness of the flakes
were determine using AFM. Channel thickness of all devices after fabrication were
measured using AFM. Figure 2.9 shows and example of AFM image and height profile of
of the MoTe2 flake exfoliated on SiO2 substrate.
Figure 2.9: Result of MoTe2 flake thickness via AFM.
2.2.5. Transmission Electron Microscopy (TEM) and Scanning Electron microscopy
(SEM)
Nanoparticle size characterization forms an important step in nanotechnology research and
development (R & D) and quality check (QC). Within the development toolkit, electron
microscopy is one of the most powerful methods for determining these critical performance
defining attributes. Microscopy also plays and important role in validating the reliability
of other routine particle sizing techniques, such as laser diffraction or dynamic light
scattering. Numerous microscopy techniques are commercially available, however
Transmission Electron Microscopy (TEM) and Scanning Electron Microscopy (SEM) are
14
arguably the most popular for nanoparticle analysis. Electron microscopy works by
bombarding a sample with a stream of electrons and monitoring either the resulting TEM
or SEM effects. These electrons are detected and converted into magnified images of
particles in the sample dispersion. Image analysis software used this information to
generate particle size data for individual particles, number based size distributions for the
entire dispersion and various shape and morphological parameters. The primary difference
in data output between the two techniques is the way in which the nanoparticle images are
resolved. SEM produces accurate 3D images of particles in the dispersion while TEM
produces 2D images that require further interpretation. However, although the images
rendered are two dimensional, TEM systems are capable of delivering much greater
resolution. By monitoring electrons as they transmit from the sample, TEM also derives
internal composition details, such as a particle’s crystalline and lattice structure. Figure
2.10 shows high magnification HAADF-STEM image of 2H-MoTe2 single crystal,
together with the corresponding diffraction pattern in [001] zone axis and the overlapped
structural model. The red spheres represent the Te atoms while the green spheres the Mo
atoms. SEM also provides this information, but is well suited to looking at sample’s surface
characteristics. SEM is an electron probe, scans over a specific area of the sample and the
size of the area determines the magnification. The main difference in data output between
the two techniques is the way in which the nanoparticle images are resolved. Figure 2.11
shows an SEM image of MoTe2 flake with contact pads.
15
Figure 2.10: High magnification HAADF-STEM image of 2H-MoTe2 single crystal,
together with the corresponding diffraction pattern in [001] zone axis and the overlapped
structural model. The red spheres represent the Te atoms while the green spheres the Mo
atoms. (Courtesy Alina Bruma, NIST.)
Figure 2.11: SEM image of flake with Ti/Au channel
16
2.3. Device Preparation for Electronic Measurements:
Before exfoliation of flakes, the SiO2 substrate was prepared with alignment marks.
These alignment marks helps to detect target flakes to fabricate MoTe2 FETs. Following
section describe the fabrication process of making metal alignment marks on the substrate.
2.3.1 Alignment marks on substrate
Pictorial representation of fabrication process of alignment marks on 6” SiO2 wafer is
shown in figure. 2.12
The fabrication process of the alignment marks with photolithography step is
detailed below:
Step 1(fig.2.12a): The 6 inch SiO2 (285nm) substrate is spin coated with LOR 3A and SPR
220.3 to support the lift-off process for the pattern marks. LOR 3A was spin coated at 4500
rpm for 50 seconds followed by a soft bake at 115 C for 2 minutes. The SPR 220.3
photoresist was then spin coated at 4500 rpm and soft baked at 115 C for 2 minutes.
Step 2(fig. 2.12b): The SiO2 spin coated substrate was exposed to UV illumination through
patterned mask in SUSSMA6 tool. The exposed sample was post baked after UV
illumination at 115 C for 90 seconds.
Step 3(fig. 2.12c): Finally, the exposed SiO2 substrate was developed in CD-26A developer
for 25 seconds followed by a rinse in DI water. This step open the window for the pattern
marks to deposit metals.
Step 4(fig. 2.12d): Ti (20nm) and Au (300 nm) was deposited on the substrate using
electron beam evaporator (Denton Infinity 22 E-beam evaporator) tool.
17
Step 5(fig. 2.12e): Following the metal deposition, the substrate was immersed in 1165
remover at 80 C to allow for the metal lift-off.
Figure 2.12: Fabrication process steps for alignment marks on Si/SiO2 substrate
After metal deposit for pattern marks on SiO2 the wafer was diced in 10mm X 10
mm dies to make FET. Before dicing the wafer was spin coated with SPR 220.3 to protect
18
the surface quality of the wafer during the dicing procedure. After the completion of the
dicing process, the samples are cleaned in boiling 1165, followed by acetone, IPA and DI
water to completely remove the photoresist. The mask for 6” wafer alignment mark is
shown in fig. 2.13.
2.3.2 Exfoliation method of MoTe2 into flakes
6 inch wafer mask
19
First graphene monolayer was exfoliated in 2004 [24, 25]. Since then other 2D
materials (MoS2, SnS2, h-BN etc.) has gained attentions due to their distinctive properties
such as electrical, mechanical, optical and sensing properties. Mechanical exfoliation of
2D materials from bulk crystals to monolayers or few-layers flakes has played a significant
role in the development of 2D materials to explore its properties and applications. More
than 10 years after the first isolation of graphene, the highest quality samples are still
produced by original mechanical exfoliation method with relatively low yield and small
size of the resulting single- or few-layer flakes. Though efforts have been made to improve
the exfoliation yield, but the thinnest resulting flakes typically have a thickness of tens of
nanometers. Besides the quality of the layered bulk crystal, the competition of vdW forces
between the sheets in a layered crystal and between outermost sheet and a substrate is
critical for exfoliation of thin flakes with high yield and large area. Exfoliation method
presented in the work has been proved successful on graphene. In this work, preliminary
results shows that modified exfoliation method also works for MoTe2 materials with
variation in temperature and time.
For comparison of different exfoliation methods, we investigated the Au-exfoliation [26]
and plasma cleaning surface method [27] also. The process flow is shown in fig.2.14. This
method involves thermal tape evaporation and cleaning of 2D material surface using O2
plasma and wet etch (KI/I2).
20
In plasma enhanced mechanical exfoliation method, prior to exfoliation, the adsorbates are
effectively removed from the patterned substrate by oxygen plasma cleaning and an
additional heat treatment maximizes the uniform contact area at the interface between the
source crystal and the substrate. After careful observation of AFM images of Au-assisted
and plasma cleaning sample exfoliation method, the difference is seen between the two
methods. The Au-assisted method leaves ample amount of residue on flakes which
provides unpredictable results for MoTe2 channel applications. Hence to obtain pristine
and clean MoTe2 flakes, plasma enhanced method was used for this work. Exfoliation
method presented in this study has been proved successful on graphene. Schematic
representation of exfoliation method is shown in figure 2.15.
Prior to exfoliation, SiO2/Si substrate with pattern marks are diced to 10mm X
10mm and were ultrasonically cleaned in acetone, 2-propanol and deionized (DI) water
and subjected to O2 plasma cleaning to remove ambient adsorbates from its surface. Once
the substrate is plasma cleaned, it is ready for mechanical exfoliation. Details description
of process steps and equipment used for exfoliation in this work is shown in figure 2.16.
21
The process flow of the substrate cleaning prior to mechanical exfoliation is shown in fig
2.16a. After sonification cleaning of substrate with alignment marks, it is plasma cleaned
for 5 minutes in O2 environment shown in fig.2.16b. Figure 2.16c shows the process flow
of mechanical exfoliation. The MoTe2 crystals were mechanically exfoliated using blue
dicing tape. Regular scotch tape method left ample amount of residues on the substrate.
Following the plasma cleaning steps, the MoTe2 loaded tape is brought in contact with the
substrate, as shown in fig. 2.17a. The substrate with attached tape is heated for 2 minutes
at ~100 C in air on a conventional laboratory hot plate (fig. 2.17b). This removes the gas
molecules between MoTe2 and substrate via, the edge driven by an increase in pressure at
the interface. Once the substrate come at room temperature, the blue tape is slowly peeled
off (fig. 2.17c) leaving only the substrate with flakes (fig. 2.18d). This completes the
exfoliation method. The completion of exfoliation steps is shown in fig.2.17. Figure 2.18
shows the optical contrast of MoTe2 after exfoliation. The exfoliated bulk regions may be
attributed to cleaving at potential defect sites in the source crystals.
22
Figure 2.15: Schematic of process steps of plasma cleaned substrate exfoliation method
23
Step 1: Sonification of substrate in Acetone, IPA and DIW
Figure 2.16 (a): Process flow of sample cleaning.
Step 2: Plasma cleaning of samples: After sonification of samples, they were plasma
cleaned for 10 minutes in Ar/O2 environment.
Figure 2.16(b): Plasma cleaner at NIST
Step 3: Exfoliation of MoTe2 flakes
Figure 2.16 (c): Process flow of mechanical exfoliation using blue dicing tape.
(a) (b) (c)
Acetone 2-propanol Deionized
Figure 2.17: Steps showing completion of exfoliation method.
Figure 2.18: Optical images of flake with varying thickness after exfoliation on pre-patterned
Si/SiO2 substrate
25
Once exfoliation is complete, quality of flake is determined by Raman
Spectroscopy and AFM as explained in section 2.2 of this chapter. Optical images of flakes
after exfoliation gives the idea of flake thickness by contrast light. Thickness of flakes were
confirmed by AFM.
For this work, we explored different exfoliation methods and found that plasma
cleaned substrate method with blue dicing tape resulted in clean and pristine flakes
compared to other methods. Images of flakes after exfoliation using two methods (Au –
assisted and plasma clean) is shown in fig.2.19. As seen from AFM images, there are lot
of dust particles on top of flakes in Au-assisted method compared to plasma clean method.
Figure 2.19: AFM images showing difference in Au-assisted and Plasma enhanced exfoliation
method.
26
After careful observation, in this work, plasma clean substrate method is used for
mechanical exfoliation of flakes to obtain pristine and clean flakes for device fabrication.
2.3.3. Device Fabrication after exfoliation
For device fabrication of 2D materials, earlier literatures has mentioned e-beam
lithography method. In this study, we used simple photolithography method, which is also
utilized for mass production in semiconductor device fabrication. After plasma assisted
mechanical exfoliation, source-drain pattern was created using photolithography steps.
Same fabrication process steps (alignment marks process steps) is used to make source-
drain pattern for measurement. Figure 2.20 shows the schematic representation of
fabricated device. Ti /Au contact metal is used for contact pads. The first layer in contact
with the semiconductor material as channel, is called the contact layer. The contact metal
(Ti) should form low resistance contact with MoTe2 channel. The top layer (Au) prevents
the contacts from oxidation and corrosion. Ti /Au metal layer has been successfully used
in fabrication of MoTe2 based FETs. Process steps for source-drain fabrication on
exfoliated MoTe2 flake substrate is shown in fig. 2.12. The mask design used for sourc-
drain pattern is shown in fig. 2.21a. Fabricated MoTe2 devices are 2.5 mm X 2.5 mm in
dimension. Figure 2.21b shows the SEM image of MoTe2 samples after fabrication. Silver
paste is used for back-gate metal contact. Figure shows the device with two contacts
(source-drain) for electrical measurement. Figure 2.22 shows the optical image of 50nm
(2.22a) and 8nm (2.22b) samples.
27
Figure 2.20: Schematic of fabricated MoTe2 for back-gated FET measurement.
Figure 2.21(a): Mask design (2.5 mm X 2.5 mm) of device size used for testing.
28
Figure 2.21(b): SEM image of fabricated device. Inlet shows 70nm channel device with flake
used for measurements.
Figure 2.22: Optical image of fabricated MoTe2 channel device for back-gate FET measurement.
(a) 50 nm MoTe2 channel, (b) 8nm MoTe2 channel.
2.3.4. Annealing of samples
Annealing is an important step to improve the contact area of metal-pads with
channel. According to past work [28]. A vacuum annealing is employed after deposit of
metal contact, to reduce the contact resistance between the metal and MoTe2 channel. The
29
annealing step provides necessary reactions between metal and channel to form a low
resistive contact. The contacts in previous study is annealed at 300 C for 3 h in forming
gas [29] and 350 C for 5h [16]. It has been noted in early literatures that, annealing the
MoTe2 flakes at 400 C and above, results in dramatic change in the surface morphology.
At high temperatures (above 400 C), the Te loss is accelerated making it impossible to
maintain the 2H structure. The Te loss can occur deeply down to a few micrometers (µm)
at high temperature, probably facilitated by the voids formed inside. The Te deficiency at
higher temperatures leads to the formation of either Mo6Te6 NWs or substantial pits/holes
which makes it difficult to maintain the hexagonal (2H) lattice structure for MoTe2.
After careful literature study [29, 30] and annealing in different environment
(forming gas, Ar, O2, Ar/O2 and vacuum) at various temperature, best results were obtained
at 350 C in vacuum. In this study, the samples after metal deposit were annealed at 350
C in vacuum for 5 minutes. After annealing, we observed that the sample performance
improved for electrical measurement. The phase and polarity of MoTe2 channel remained
unchanged after annealing process.
on MoTe2 devices.
2.4 Conclusion :
We were able to fabricate MoTe2 based FET by plasma assisted mechanical
exfoliation method. MoTe2 crystals were grown by two different transport agents i.e. TeCl4
and I2. The crystals and exfoliated flakes were characterized by AFM, TEM and Raman
spectroscopy to verify the quality and 2H phase of the materials. The crystals were in-built
30
and after careful materials analysis, the quality and 2H-phase matches with the previous
literature confirming the good crystalline nature of the samples. An improvement in the
Ti/MoTe2 junction is observed when samples were thermally annealed in vacuum.
31
3.1 Introduction
To understand the advantages of MoTe2 channel as semiconductors in FETs, the
important electrical characteristics of the devices must be covered. The n+ doped Si
substrate is used as the global back gate with the 285 nm SiO2 layer serving as the gate
dielectric. Electrical measurements are necessary to demonstrate the channel conduction
type and a good ohmic contact between metal and MoTe2. In this chapter, we have focused
on electrical measurements and change in polarity as a function of channel thickness [31].
We note that, few work is done in the area of channel thickness variation effect on polarity
change of MoTe2 devices. 2H-MoTe2 is intrinsically p-doped but can also exhibit
ambipolar behavior [32]. In early literature, conductivity change from p- to n-type in
MoTe2 devices was observed after annealing in forming gas (10% H2 in N2) for 2 hours
[33]. Recently, it was demonstrated that MoTe2 FET polarity can be tuned by using dual
top gates geometry or by selecting metals with appropriate work functions for drain and
source contacts [34-36].
In this section, we demonstrate the control over n-, ambipolar and p-type
conductivity in MoTe2 back-gated FETs by reducing the channel thickness from thick
(above ≈ 60 nm and 15 nm) to thin (below ≈ 10nm), respectively. MoTe2 crystals were
prepared by chemical vapor transport (CVT) method using TeCl4. All devices were
thermally annealed in vacuum at 350 C for 5 minutes and no change in polarity was
observed in any of the fabricated devices as mentioned in early report [33]. From our
measurements, we observed that MoTe2-Cl FETs showed p-, ambipolar and n-type polarity
32
with increasing flake thickness. No additional steps for extrinsic doping is applied during
the material and device fabrication process. The NH3 gas sensing results of MoTe2-Cl
FETs, further confirms the change in polarity as a function of flake thickness.
To further understand the polarity change behaviour of MoTe2 channel, devices
prepared from iodine based transport agent crystals, were also tested. For simplicity we
will refer MoTe2 FETs fabricated by TeCl4 and I2 transport agent as MoTe2-Cl and MoTe2-
I2, respectively. Contrarily, MoTe2-I2 FETs showed ambipolar and p-type polarity with
increasing flake thickness. This variation in electronic properties can be due to variation in
stoichiometry of the MoTe2 crystals prepared from TeCl4 and I2 transport agent (TA).We
also demonstrated that by changing the contact metal electrode (Cr/Au and Pd/Au), showed
similar conducting behaviour when compared to MoTe2-Cl and MoTe2-I2 devices with
Ti/Au as contact metal. The thickness modulated transport properties of MoTe2 FETs can
open many possibilities for digital and analog circuits by providing guidance for fabricating
p-, ambipolar and n-type devices, merely by tuning the channel thickness.
3.2 Control of Polarity in Multilayer MoTe2 Field-Effect Transistors by Channel
Thickness
Few-layered MoTe2 flakes were exfoliated from semiconducting 2H-MoTe2 bulk crystals
and then transferred onto a 285 nm SiO2/N+ doped Si substrate. Metal contacts were
defined by photolithography followed by subsequent deposition of 30/300nm Ti/Au
through E-beam evaporation. Detail description of device fabrication steps is explained in
chapter-2. After fabrication, the output (Ids as function of applied Vds) characteristics and
transfer (Ids as function of applied Vgs for a fixed Vds) characteristics of the devices were
33
measured using Lakeshore 4-point probe station. The source/drain contacts were made by
needle point probes, with one connected to ground and the other to a source measuring unit
(SMU) that provided source/drain voltage. The copper base of the probe station was
connected to a separate SMU to provide a back-gate bias voltage to the silicon backing on
the chips. A small scratch was made to the backside of the chip to remove the native oxide
and a small amount of the silver conductive paint was applied to allow easier conduction
of the electric field from the back-gate electrode. The purpose of measuring the output and
transfer characteristic of MoTe2 devices is to determine their polarity and gain an estimate
of the mobility of each device. In the output and transfer characteristic of each device, an
increasing current with increasing negative gate-bias voltage was classified as “p-type”,
(hole dominating the conduction) whereas devices in which current increased with
increasing positive gate bias voltage is classified as “n-type” (electron dominating the
conduction). Devices with conduction showing increase in current in both negative and
positive gate voltage (Vbg), were classified as “ambipolar” (both charge carriers types are
capable of contributing to the conduction). Fig. 3.1 shows the schematic image of
fabricated device for back-gate FET measurement setup. Current –voltage curve (fig. 3.2)
confirmed the continuity of the channel and also represented that if metal/semiconductor contact is
ohmic or non-linear (Schottky). In our case, all the devices showed non-linear I-V curve, which is
attributed to presence of Schottky barrier between metal/semiconductor junctions due to differences
in work-function of Ti (4.33eV) and electron affinity of MoTe2 (4.3 ±0.1 eV) [35,37] . Figure 3.3
shows the typical transfer curve from our devices for each category (n- , p- and ambipolar type) of
device. The MoTe2 FETs were measured over source/drain biases ranging from -2V to +2V and
from back-gate voltage from -60V to +60V. Any higher back-gate voltages resulted in device
burnout. This does leave the question if the devices were intrinsically p or n-type or ambipolar. We
34
are unable to ascertain this for every measured devices, but classified them as p- or n- or ambipolar
type in the operating range of voltages for gas sensing applications, there is no functional difference
between the response of an intrinsic p, n- or ambipolar devices. Figure 3.4 shows the typical output
characteristics for devices in each category.
Figure 3.1. Schematic diagram of device connection for back-gate MoTe2 FET.
Figure 3.2 Current-Voltage (I-V) showing non-linear behavior of 4.3 nm channel MoTe2
device.
35
Figure 3.3 The transfer characteristics showing the typical behavior of a (a) p-type, (b)
ambipolar, and (c) n-type MoTe2 FET used in this work.
36
Figure 3.4 The output characteristics of a typical (a) p-type, (b) ambipolar, and (c) n-type
MoTe2 FET used in this work.
37
The devices were measured before and after vacuum anneal at 350 C for 5 minutes to
confirm the electrical conductivity of the MoTe2 flake. In vacuum annealing, the devices
were vacuum sealed in an ampoule and inserted inside a furnace in which the temperature
was raised to 350 C with ramp speed of 20 C per 10 minutes [33]. Thermal annealing is
often used to improve the performance of FET devices by removing adsorbates from the
semiconducting material and the metal contacts through the application of heat in an inert
gas vacuum environment. During this process, the contact between the metal electrodes
and the semiconducting material is improved. The heat relaxed the metal and makes good
adhesion contact between the metal and semiconductor, lowering the contact resistance
[38, 39]. The current – voltage (I-V) curve of 4.3 nm MoTe2 channel is shown in fig. 3.5.
Figure 3.5. Current-Voltage (I-V) showing improvement in current value after vacuum annealing
at 350 C for 5 minutes. Thickness of MoTe2 flake is 4.3nm.
38
Figure. 3.6 : Transfer characteristic (Ids-Vgs) at Vds = 2V of p-type MoTe2 FET (7nm
channel thickness) showing improved on/off ratio of MoTe2 FET after vacuum annealing
at 350 C for 5 minutes.
Figure. 3.7: Transfer characteristic (Ids-Vgs) at Vds = 2V of ambipolar type MoTe2 FET
(19.2 nm) channel thickness) showing improved on/off ratio of MoTe2 FET after vacuum
annealing at 350 C for 5 minutes.
39
Figure. 3.8 : Transfer characteristic (Ids-Vgs) at Vds = 2V of p-type MoTe2 FET (70.8 nm
channel thickness) showing improved on/off ratio of MoTe2 FET after vacuum annealing
at 350 C for 5 minutes.
Raman spectroscopy was used for material characterization after fabrication
process to test the 2H phase of MoTe2 channel. The spectra for the 2H-MoTe2 (fig. 3.4)
exhibit characteristic A1g at 170 cm-1, E1 2g at 235 cm-1 and B1
2g at ≈ 288 cm-1 modes (the
latter is active in thin layers only). Comparison of Raman spectra for varying thickness in
fig. 3.4 is in good agreement with Raman studies reported in the literature [40], which
confirms the 2H-phase and good crystalline quality of MoTe2 layers used in FETs. The
inset in fig. 3.4 shows plane-view of a typical FET device.
40
Figure 3.9: Raman spectra of MoTe2 flakes of different thickness exfoliated from bulk MoTe2
single crystals grown by CVT with TeCl4 transport agent. Inset shows plan-view optical image of
FET device with a 50 nm thick channel.
Determination of polarity based on the effect of channel thickness and the method of
growth, that is, MoTe2-Cl and MoTe2-I2 FETs is made by measuring the output and transfer
characteristics using a variety of different flake thickness. The results from output and
transfer curve shows that by adjusting the thickness of the MoTe2 channel, it is possible to
let holes or electrons dominate the channel conduction, resulting in p- or n-type FET
behavior. The output curves at varying Vbg are shown in fig. 3.10 (a) for thin, 5.6 nm and
fig. 3.10 (b) for thick, 78 nm, channels. Clear p-type , i.e., decrease in Ids with increasing
Vbg, and n-type, increase in Ids with decreasing Vbg, transport behavior is observed for thin
vs. thick MoTe2 FETs respectively. Note that although the output characteristics appear to
be nearly linear, a small Schottky barrier at the metal/semiconductor junction is expected
41
due to difference between work function of Ti (4.33 eV) and electron affinity of MoTe2
(4.3 ± 0.1eV) [35,37].
We further analyzed the transfer behavior of the FETs by Ids-Vbg curves at Vds =
2V, which showed n-, ambipolar and p-type behavior for thick, medium and thin channel
respectively. The gate leakage current Igs in all the devices was negligible in pA range. All
Ids valued in the transfer curves are normalized by the channel width (W). The maximum
Ids ON/OFF ratio was about 1 × 103, which decreased with increasing channel thickness.
Multiple devices were fabricated to reproduce the transfer characteristic as a function of
MoTe2 channel thickness. For channel thicker than ≈ 65 nm FETs showed consistent
unipolar n-type behavior. For a medium channel thickness, from ≈ 60 nm to 15 nm, FET
behavior was ambipolar, while FETs with very thin channel, from ≈ 10 nm down to 5 nm,
were all p-type. Figure 3.11 (a-c) shows the transfer curves of p-type, ambipolar and n-type
conducting behavior of thin, medium and thick channel MoTe2 FETs. Effect of channel
thickness on various transport properties in FETs was previously described in black
phosphorous, WSe2, MoS2 and MoTe2 FETs [41-45]. Polarity variation as a function of
MoTe2 flake thickness was not studied in early literatures for wide range of channel
thickness (4nm – 70nm).
Figure 3. 10: Output characteristics of MoTe2 FETs under different back-gate voltages, Vbg. FETs
with channel thickness of (a) 5.6nm and (b) 78nm show p- and n-type conductivity, respectively.
42
Figure 3.11 (a) Transfer characteristic of thin channel MoTe2 showing p-type conduction behavior.
Figure 3.11 (b): Transfer characteristic of medium channel MoTe2 showing ambipolar behavior.
Figure 3.11 (c): Transfer characteristic of thick channel MoTe2 showing n-type conducting
behavior.
43
In reference to n-type behavior, the n-doping in TMDCs has been observed when
using TeCl4 as the transport agent in the CVT growth [42] and during post-growth chloride
molecular doping of TMDC compounds [43]. In the present work, since secondary ion
mass spectroscopy (SIMS) measurement (fig. 3.12) detected a presence of Cl impurity in
the CVT grown MoTe2 crystals, we assume that Cl doping is a dominant factor for n-type
behavior in our thick channel devices. The switching of conductivity to p-type in thinner
channel FETs suggests that the effectiveness of Cl doping diminishes upon MoTe2 layer
thinning. This can be associated with increasing role of surface defects and adsorbates in
ultra-thin layers, a phenomenon which is exemplified in [44] for atomically thin MoS2
FETs. In addition to possible detrimental effect of reduced MoTe2 channel thickness on the
n-type doping efficiency, we also speculate that the n- to p- polarity switching in thinner
layers may be caused by the modulation of Schottky barrier height and corresponding band
alignment and band-bending at the metal/MoTe2 interface [45,46,47].
Figure 3.12: secondary ion mass spectroscopy (SIMS) measurement showing the presence
of Cl in MoTe2 flakes.
44
The charge carrier mobility plays a vital role in semiconductor science and
technology, because the efficiency of semiconductor devices generally improves as charge
mobility increases. The importance of charge mobility, µ, in FET, stems from the fact that
the higher the mobility, the greater the source-drain current (Ids), realized in a FET within
a certain span on the back-gate voltage (Vbg), at a given source-drain voltage (Vds) [38, 39].
For μ to be truly meaningful parameter, the linearity of FET transfer characteristics
dependencies in the linear and saturation regimes, is very important. The equations for Ids
(Vbg, Vds) dependences in a FET (also called Schokley equations) used for mobility
extraction are derived within the gradual channel approximation [48, 49] based on the
specific assumption that:
a. The transverse gate electric field is much greater than the longitudinal source-drain
electric field.
b. The mobility is carrier density independent.
The assumption let a FET model with linear transfer characteristics in which the linear
regime mobility can be expressed as shown in equation (1).
=
) (1)
Where Cox is a silicon oxide capacitance per unit area, L is the channel length, and W is
the channel width. The Cox value is 1.151 x 10-8 F/cm2 for 300 nm SiO2 thickness.
In this study, we investigated the thickness-dependent field-effect carrier mobility
(µFE) of the fabricated devices, which is extracted from the transfer characteristics shown
in fig.3.11. Figure 3.13 shows the overall trend of mobility increasing with increasing
channel thickness. The influence of thickness in MoTe2 devices can be related to Coulomb
45
scattering and quantum confinement [50, 51]. The scattering of the carriers is weakened by
the Coulomb interaction resulting in higher mobility in thicker vs thinner channel as it was
demonstrated for both MoTe2 [41] and MoS2 [52] FETs.
Figure 3.13: Field effect mobility of electrons and holes vs channel thickness for MoTe2 FETs.
Encircled pairs of data points corresponds to the ambipolar devices that exhibit ambipolar
conductivity.
In this study, we have demonstrated a simple, yet effective way to control the
conductivity type in MoTe2 FETs by tuning the channel thickness. FETs were fabricated
from the CVT-grown and mechanically exfoliated MoTe2 single crystals. The transport
properties were examined for varying channel thickness using the output and transfer
characteristics. FETs showed a change in polarity from n-type through ambipolar to p-type
by reducing the channel thickness from 78 nm to 5 nm. The n-type conductivity in thick
MoTe2 layers is attributed to chlorine doping from TeCl4 transport agent used in the CVT
growth. The switch of polarity by thinning the FET channel may be associated with
increasing role of states in ultra-thin layers, which can influence charge carrier
46
concentration and dynamics in the channel by modulating Schottky barrier height and
associated band alignment and band-bending at the metal/semiconductor interface.
3.3.Channel thickness effect of MoTe2 FETs prepared from TeCl4 and I2 transport
agents by CVT method
To further understand the mechanism behind the change in polarity as function of
channel thickness in MoTe2 FETs, we studied the electrical properties of flakes prepared
from two different transport agents. For this study, MoTe2 crystals were prepared by CVT
method using TeCl4 and I2 as transport agent. For simplicity, MoTe2 FETs fabricated by
TeCl4 and I2 transport agents are referred to as MoTe2-Cl and MoTe2-I2, respectively. Here
we report the transition in polarity from p- to ambipolar to n-type as the channel thickness
is varied from thin to thick flakes with no additional doping. From our repeated
measurements of thin, medium and thick channel devices, we observed that MoTe2-Cl
devices showed p- , ambipolar and n-type transfer behavior respectively. Contrarily,
MoTe2-I2 devices showed ambipolar curve for thin (≤ 5nm) and p-type for thick channel
FETs. The NH3 gas sensing results of MoTe2-Cl devices further confirms the change in
polarity from p- to n-type as channel thickness increases. This property of MoTe2 FETs
can create many possibilities in the field of digital and analog circuits.
The lattice vibrational modes of exfoliated 2H-MoTe2 flakes were identified using
Raman Spectroscopy with a laser source of wavelength 532 nm. Comparison of Raman
spectra of MoTe2-Cl and MoTe2-I2 shown in fig.3.10, which is in excellent agreement with
Raman studies reported in early literatures [32,35,36], which confirms the 2H-