Journal of Solid State Chemistry 290 (2020) 121576

Contents lists available at ScienceDirect

Journal of Solid State Chemistry

journal homepage: www.elsevier.com/locate/jssc

Tauc-plot scale and extrapolation effect on bandgap estimation from

UV–vis–NIR data – A case study of β-Ga2O3
P.R. Jubu a, b, F.K. Yam a, *, V.M. Igba b, K.P. Beh a
a
School of Physics, Universiti Sains Malaysia (USM), 11800, Penang, Malaysia
b
Department of Physics, University of Agriculture Makurdi (UAM), P. M. B. 2373, Makurdi, Benue State, Nigeria

A R T I C L E I N F O A B S T R A C T

Keywords: The effect of the Tauc plot scale and bandgap extrapolation technique on estimated optical bandgap values is
Gallium oxide often not strongly emphasized. Herein, we demonstrate that the scale of the conventional Tauc plot and the
Bandgap extrapolation technique poses significant influence on the determined value of bandgap. The nanomaterial used
Tauc plot
for this study was prepared by the CVD method using different H2 reducing gas flow rates. XRD measurements
Extrapolation
UV–vis–NIR
confirmed polycrystalline β-Ga2O3 with high crystalline quality. FE-SEM analysis revealed nanohump-like
morphology. UV–vis–NIR measurements generally showed increasing bandgap with rising H2 supply into the
horizontal reactor. Specifically, it was observed that the proper extrapolation technique for bandgap estimation is
more appropriate as compared to the values obtained from the traditional direct extrapolation method. It is
established that direct extrapolation technique has the potential to either under - or over-estimate bandgap values,
whereas the proper extrapolation method is promising for more reasonable values of optical energy gap.

1. Introduction nanomaterial can be deduced from different measurement techniques

including ultraviolet–visible near-infrared spectroscopy (UV–vis–NIR)
Research into semiconductor nanomaterials is rapidly growing due to [7], photoluminescence spectroscopy (PLS) [8], reflectance spectro-
their unique electrical, optical, photonic, chemical, structural, thermal scopic ellipsometry [9], valence electron energy-loss spectroscopy
and mechanical properties [1]. Bandgap of a semiconductor nano- (VEELS) using monochromated electrons [10] and photoacoustic spec-
material is a key optical property in the optoelectronic, electronic and troscopy (PAS) technique [11], respectively. Among these techniques,
solar industries hence, its accurate determination is of high importance the UV–vis–NIR and PLS are the widely deployed techniques for optical
[2]. The term “band gap” refers to the energy difference between the top bandgap measurement, among which the former is the simplest, cheap-
of the valence band to the bottom of the conduction band [3,4]. The est, commonest and most preferred method for bandgap determination of
energy gap is a pivotal characteristic of the electronic structure of ma- films [4]. The optical gap energy is determined according to Tauc method
terials and ultimately decides about their potential application in devices by simply fitting a straight line (Tauc line) to the linear (or steepest)
[2,4,5] because of its strong influence on materials characteristics [6]. region of the optical spectrum to intersect the photon energy (hv) axis [5,
The present paper attempts to summarize the various factors that influ- 7]. The UV–vis–NIR optical characterization technique in combination
ence the bandgap of a semiconductor nanomaterial, and possibly make a with the Tauc method is the most widely applied procedure for bandgap
little addition to existing established factors. determination [5,12].
Roughly, the optical properties of a material such as transmission, Aside other physical and chemical properties of nanomaterials, the
absorption, emission, reflection, refraction, diffraction, and scattering change in the bandgap of a semiconductor film from its bulk value
effects are determined by its interaction with electromagnetic (EM) ra- happens due to various effects such as doping, parameters of the syn-
diation. For semiconductors, these optical properties (most of which are thesis technology [5,13,14], strain due to mismatch with the lattice of the
in the ultraviolet, or visible range) are related to the characteristics of the substrate [14,15], crystallite size and film thickness [1,16], annea-
electronic bands of the semiconductors or ultimately to their atomic ling/deposition temperature [17,18], microstructure and composition
structure and chemical bonding [3]. The value of the energy gap of a [19–21]. However, it is very difficult to isolate the contributions of

* Corresponding author.
E-mail address: [email protected] (F.K. Yam).

https://doi.org/10.1016/j.jssc.2020.121576
Received 9 May 2020; Received in revised form 1 July 2020; Accepted 8 July 2020
Available online 21 July 2020
0022-4596/© 2020 Elsevier Inc. All rights reserved.
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

takes the behavior (Eq. (2)), which is expected for transitions with
! ! !
negligible changes in the electron wave-vector k ð k f  k i Þ – along
with aspects regarding atomic bonding and selection rules [12,22].


αdir E < Eg ¼ 0 (2)

where αdir is direct optical absorption coefficient and Eg is energy gap of

the nanomaterial.
The Tauc equation for direct bandgap thus, can be expressed as in Eq.
3(a, or b)


n
αdir hv  Eg ¼ A hv  Eg (3a)

Or

1=2
αhv ¼ A hv  Eg (3b)

where hv is photon energy E, n ¼ 1/2 and A is a proportionality constant.

According to Tauc et al. [12,23], the direct bandgap value Eg can be
extrapolated by fitting a straight line to the linear segment to intersect the
E or hv-axis from the plot of α2 versus hv (Eq. (3a)), or the plot of ðαhvÞ2
against hv (see Eq. (3b)). The optical curve exhibit linear and nonlinear
portions which are the characteristics of the allowed transition. The
linear portion of the curve which is mostly in the high energy side of the
Tauc plot characterizes the fundamental absorption of the materials,
whereas the nonlinear portion correspond to a residual absorption
involving defect states [24]. While the Tauc method relies on applying a
linear fit to the main absorption edge at the bandgap, influence from
lower energy defect absorption states are often ignored by not including
the baseline segment to the Tauc line fit thus, affecting the appropri-
ateness of the estimated bandgap value [25].
It could be inferred that Eq. (3b) is useful when absorbance is
Fig. 1. (Fitted) XRD patterns of β-Ga2O3 NSs prepared at different H2 flow rates. measured from UV–vis–NIR spectrometer. The equation changes when
reflectance is measured in the place of absorption. In this case, the
reflectance is converted into absorption coefficient by using the Kubelka-
Munk (K-M) or remission function F(R∞), according to Eqs. (4) and (5),
Table 1 which gives the final relation Eq. (6) [5]:
Comparison between some observed and standard structural parameters for the
prepared β-Ga2O3 films. Rsample
R∞ ¼ (4)
H2 Observed Standard FWHM Observed Standard D Rstandard
flow 2θ ( ) 2θ ( ) (rad) d (Å) d (Å) (nm)
rate 10 3
k ð1  R∞ Þ2
(sccm) FðR∞ Þ ¼ ¼ (5)
s 2R∞
50 33.4810 5.1201 2.674 28.29
100 33.4158 33.459 2.9524 2.679 2.676 49.03
1=2
150 33.4107 5.3020 2.680 27.30 FðR∞ Þ ¼ hv ¼ A hv  Eg (6)
200 33.3803 4.9871 2.682 29.02
where R∞ is the reflectance of the sample with ‘infinite thickness’, since
the contribution of the sample holder is negligible, k and s are the ab-
different possibilities causing changes in the bandgap. sorption and scattering K-M coefficients, respectively. In practice,
The bandgap of a semiconductor nanomaterial can either be direct, thicknesses >2 mm are enough to inhibit contributions from the sample
indirect, direct forbidden, or indirect forbidden allowed transition holder [3]. The diffuse reflectance of the standard material (Rstandard) is
depending on the material type [3,5]. The universal UV–vis–NIR tech- commonly that of polytetrafluoroethylene (PTFE) or BaSO4 pellets. The
nique for optical bandgap measurement allows for estimation in both plot of ½FðR∞ Þhv2 versus hv then leads to an extrapolation that estimates
direct and indirect energy gap materials. The optical absorption of the value of the bandgap.
semiconductor nanomaterials is derived from models in which the elec- This method of direct bandgap estimation is quite simple and
trons receives a quantum mechanics approach and the photons are convenient despite the inadequacies of the semi-classical model such as:
described by classical EM waves. The optical absorption coefficient α, at (i) the consideration of the zero Coulomb attraction which at lower
the photon energy E (¼ ħω) of an electron excited from the valence (VB) temperatures enhances αdir ðEÞ; (ii) the unexpected origination of
to the conduction (CB) band, is given by the transition rate (Eq. (1)) [3]. extrinsic absorption (due to defect states and impurities) which may
affect αdir ; and (iii) the validation of the parabolic-like
2π 2 !
WVBCB ¼ ½M gðEÞ (1)
ħ αdir ðE  Eg Þ ¼ AðE  Eg Þ1=2 shape only at k  0 [22]. For indirect (or
non-vertical) optical transitions involving a photon and (at least) one
where M and g(E) stand for the (coupling) transition matrix element and phonon in order to comply with the conservation of momentum, the
the (joint electron-hole) density of states. transition rates WVB→CB taking place in indirect energy gap semi-
In the case of direct (or vertical) allowed transitions, the energy- conductors are smaller than those typically verified in the direct
dependent absorption coefficient α(E) of a semiconductor nanomaterial bandgap. The corresponding α(E) can be written as in Eq. (7) [26].

2
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

Fig. 2. FE-SEM micrographs of the produced β-Ga2O3 films at different H2 flow rates (sccm): (a) 50; (b) 100; (c) 150; (d) 200.

2
αhv ¼ A hv  Eg (8b)

where ħ℧ denote the energy of a phonon being emitted (þħ℧) or

absorbed (  ħ℧) and (n ¼ 2).
We can infer from the preceding paragraphs that the bandgap of a
semiconductor can be influenced by several factors. Nevertheless, there
are two more factors affecting the bandgap value, which we would
attempt to explore in the present report. These external or post-synthesis
or data-processing factors are rarely emphasized by researchers, yet they
are worth taking into high consideration. They are (i) the scale of the
Tauc plot, and (ii) the method of extrapolation of the energy gap value to
the x-axis. Apparently, these two additional factors influencing the
bandgap value are not related to the nanomaterial’s synthesis technique,
instrument measurement technique or other physical and chemical
properties of the nanomaterial and have rarely been discussed in detail in
published articles.
While carrying out a similar study recently, Escobedo-Morales et al.
[5] elaborated on the effect of extrapolation style on energy gap value by
admonishing that, “the estimation of bandgap by intersecting the
straight-line fit of the Tauc segment with the hv axis is not appropriate for
samples containing more than one prominent optical absorbing center/or
phase because systematic underestimation of the bandgap value may
Fig. 3. UV–vis–NIR optical reflectance spectra of the β-Ga2O3 NSs prepared at occur”. He therefore affirmed that, “the automated method will account
different H2 flow rates. for such cases by introducing a baseline function such that a vertical line

(or ordinate) is then dropped to intersect the hv axis”. Their emphasis in
αind E < Eg ¼ 0 (7) the report was for the spectra with more than one long linear absorbing
edge/or center. However, the authors did not mention the effect of scale
where αind is indirect optical absorption coefficient. of the Tauc plot and did not recommend the use of the baseline segment
The Tauc equation for indirect bandgap can therefore be expressed as for other samples which have a single long linear absorption edge. Also,
Eq. (8a), or b) Dolgonos et al. [27] observed that the traditional method of directly


n extrapolating the linear segment of the Tauc plot for direct bandgap
αind E  Eg ¼ A E  ħ℧  Eg (8a) semiconductor nanomaterials is inappropriate for use on degenerate
Or simply semiconductors, where the occupation of conduction band energy states
cannot be ignored. He strongly emphasized that the application of the

3
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

Fig. 4. Tauc plot of the produced β-Ga2O3 NSs drawn by using different scale size, (a) 2.0–7.0; (b) 2.0–6.0; (c) 3.4–5.4; (d) 4.0–5.4. The dotted lines indicate the
traditional DE technique.

Fig. 5. Tauc plot of the produced β-Ga2O3 NSs drawn by using different scale size, (a) 2.0–7.0; (b) 2.0–6.0; (c) 3.4–5.4; (d) 4.0–5.4. The dotted ordinates indicate the
proposed PE technique.

4
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

Table 2
Bandgap values for DE and PE from different scales of the Tauc plot.
(a) Tauc plot scale: 2.0–7.0 (b) Tauc plot scale: 2.0–6.0

H2 flow rate Eg; DE (eV) Eg; PE H2 flow rate Eg; DE (eV) Eg; PE
(sccm) (eV) (sccm) (eV)

50 4.86 4.75 50 4.88 4.75

100 4.90 4.71 100 4.94 4.71
150 4.96 4.75 150 4.96 4.78
200 5.17 4.97 200 5.19 4.97

(c) Tauc plot scale: 3.4–5.4 (d) Tauc plot scale: 4.0–5.4
H2 flow rate Eg; DE Eg; PE H2 flow rate Eg; DE Eg; PE
(sccm) (eV) (eV) (sccm) (eV) (eV)

50 4.86 4.74 50 4.85 4.73

100 4.90 4.69 100 4.89 4.68
150 4.95 4.75 150 4.95 4.77
200 5.20 4.96 200 5.16 4.95

Notes: Eg; DE : directly extrapolated bandgap, Eg; PE : properly extrapolated

bandgap.

Table 3
Average bandgap values for DE and PE from Tauc plot.
(a) Bandgap by DE
Fig. 7. Schematic diagram showing different bandgap extrapolation techniques.
H2 flow rate (sccm) Average Eg (eV)

50 4.86
100 4.91
spectra of Sn-doped In2O3 (ITO). Dolgonos and co-workers did not
150 4.96
200 5.18
emphasize on the combination of the extrapolated linear fit to the optical
curve with a baseline segment for onward tracing to the x-axis by using
(b) Bandgap by PE an ordinate (or a vertical line).
H2 flow rate (sccm) Average Eg (eV)
In this research paper, we have applied the most preferred Tauc
method for bandgap determination called ‘direct extrapolation’ (DE), in
50 4.74
which the bandgap is evaluated simply by fitting a straight line to the
100 4.70
150 4.76 linear portion of the absorption spectrum [23]. We then compare those
200 4.96 results with the ones obtained from the present proposed method ‘proper
extrapolation’ (PE), which is done by simply drawing a vertical line
perpendicular to the point of intersection of the Tauc linear fit and the
Tauc method (or linear fit) to crystalline materials is highly rooted in baseline [5]. It was observed that DE overestimated the bandgap value,
misconception– because it is often applied for all semiconductor mate- whereas PE gave a more reasonable value. Optical spectra with one or
rials including those which do not have vertical allowed transition. more prominent optical absorbing centers are considered and suggested
Dolgonos and co-researchers thereby proposed a new method for that, PE should be applied for all kinds of allowed transition in order to
extracting the direct optical bandgap from absorption spectra of degen- obtain the most reasonable value of optical bandgap. Meanwhile, the
erately doped bulk semiconductors and applied to pseudo-absorption scale of the Tauc plot was found to equally affect energy gap values
determined from both DE and PE technique. The films used for this study
are those of gallium oxide (Ga2O3) nanostructures (NSs) deposited by the
reducing-ambient chemical vapor deposition technique (RA-CVD).

2. Materials and methods

Ga2O3 NSs were synthesized on Si substrates by the RA-CVD tech-

nique. 58 mg Ga2O3 powder was used as the precursor. N2 gas flow rate of
600 sccm was flowed as the carrier gas, while 50 sccm H2 gas was sup-
plied as the reducing gas to facilitate decomposition of Ga2O3 powder.
The Si substrate was cleaned according to the Radio Corporation of
America (RCA) procedure and placed vertically in a ceramic boat located
at 22 cm away from the source boat which was positioned at the center of
the horizontal furnace. The furnace was flushed with flowing N2 gas for
10 min to remove residual gases in the horizontal reactor before initi-
ating the synthesis. Temperature of the furnace was ramped from room
temperature (RT) to 1000  C and sustained for dwell time of 30 min for
film deposition, after which the furnace was shut down and allowed to
cool down gradually to RT. The experiment was repeated for H2 flow
rates of 100, 150 and 200 sccm to grow three other samples. A similar
experimental procedure is also available in our previous report [21].
The grown films were characterized using high-resolution X-ray
Fig. 6. Dependence of bandgap on H2 gas flow rate; plotted from the average diffraction (HR-XRD) PAnalytical X’Pert Pro, field-emission scanning
bandgap values in Table 3(a, b).

5
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

electron microscope (FE-SEM) FEI Nova NanoSEM 450 and UV–vis–NIR 3.2. Surface morphology
spectrometer Agilent-Cary 5000 for structural, morphological and optical
properties, respectively. The FE-SEM micrographs of the prepared films are shown in Fig. 2.
The NSs exhibit nanohump-like morphology. Fig. 2(a) consists of nano-
3. Results and discussion chunks at surface probably due to the lower gas flow rate. The dense
nature of the films is an indication of the robustness of the growth
3.1. Crystal structure method.

Fig. 1 shows XRD patterns of the Ga2O3 NSs synthesized under 3.2.1. Growth mechanism of the β-Ga2O3 NSs
different H2 flow rates. All indexed crystallographic planes could be Ga2O3 powder melts congruently at 1800  C in oxidizing atmosphere
traced to the β-Ga2O3 phase (JCPDS file no. 041–1103), except for the (*) [35]. However, it decomposes readily into volatile lower oxides at 1200

diffraction peaks which are proposed to belong to the Si substrate, ac- C when heated under O2-deficient environment [36]. Meanwhile, in H2
cording to JCPDS file no. 01-075-0589. Similar peaks have also been reducing atmosphere without flowing O2, Ga2O3 powder vaporizes
traced to Si substrate by other literature reports [28,29]. The intensity of readily at relatively lower temperatures (850  150  C) [18]. However,
the XRD peaks was generally observed to fluctuate with rising H2 supply the decomposition can be greatly suppressed when heated in H2 ambient
into the reactor. However, the intensity of the (111), (310) and (511) with flowing O2.
diffraction lines were seen to increase steadily with H2 flow rate, sug- The proposed growth mechanism for the reducing-ambient CVD-
gesting improvement in crystalline quality. The increasing peak intensity gown Ga2O3 NSs follow the vapor-solid (VS) process in which films are
could be ascribed to the additional Ga2O3 material deposited on the deposited directly on an unseeded substrate. The H2 reducing gas reacted
substrate due to the increase in evaporation of the precursor with rising with Ga2O3 powder at 1000  C to release Ga2O and H2O vapor (Eq. (11))
reducing H2 ambient in the reaction chamber. Meanwhile, the appear- [18,37–40]. The produced volatile reactants were transferred to the
ance of many diffraction peaks suggests the polycrystalline nature of the lower temperature region of the horizontal reactor by the N2 (600 sccm)
material due to the randomly orientated nanocrystals. carrier gas and temperature gradient along the horizontal growth line.
It could be seen that the preferred growth orientation of the NSs is This volatile species (Ga2O) reacted with H2 in the gas phase to generate
changing probably due to increase in the impact probability [21]. This is Ga atoms (Eq. (12)) [37]. The newly arrived Ga vapor was deposited on
because the speed of the vaporized species to the substrate is increasing the surface of the substrate and then backreact with H2O vapor and re-
with rising H2 flow rate thereby, raising the impact probability to the sidual O2 in the reactor to form small solid Ga2O3 seeds (Eq. (13)). The
surface. The impact probability causes the surface energy to increase and proposed reactions involved for the growth of the β-Ga2O3 NSs is based
the migration of adatoms is favored to preferred crystallographic planes on supposition. Further research may be carried out to provide the
which have higher energy. According to Lu et al. [30], the preferred peak experimental evidence.
orientation is due to the higher number of dangling bonds in this crystal
1000  C
plane which has greater attractive force to reactive atoms and so, reactive Ga2 O3 ðsÞ þ 2H2 ðgÞ ! Ga2 OðgÞ þ 2H2 OðgÞ (11)
atoms of weaker energy migrate to this site.
XRD peak profile analysis for all samples was performed on the (111) < 1000  C
Ga2 OðgÞ þ H2 ðgÞ ! 2GaðgÞ þ H2 OðgÞ (12)
diffraction plane, which appears to be the predominant peak on the
highest crystalline sample (Fig. 1(d). This peak has been selected for the < 1000  C
XRD profile analysis because it is the only consistent preferred peak 2GaðgÞ þ O2 þ H2 OðgÞ ! Ga2 O3 ðsÞ þ H2 ðgÞ (13)
orientation that feature on two samples (Fig. 1(a, d). The interplanar Meanwhile, the decomposition of the source material and release of
spacing d was calculated using Eq. (9) [31]. the volatile reactants to the substrate (in the downstream) was in prog-
ress. Therefore, continuous adsorption of Ga atoms at the nucleation
nλ ¼ 2dsin θ (9)
sites, followed by oxidation kept the small nuclei growing until super-
saturated nanostructures were form. It should be noted that, the rate of
where λ is the X-ray wavelength, n ¼ 1 for the first diffraction order and θ
evaporation of the source material increased with rising H2 supply into
is the peak position.
the reactor therefore, the growth rate became faster. Not all the precursor
The estimated d for the film prepared by flowing 50 sccm H2 gas was
evaporated at the lower H2 flow rate (50 sccm), but total decomposition
found to be smaller than the standard value (2.676 Å), indicating the
of the source was observed at higher flow rates (100–200 sccm). The
shrinkage of the lattice constant because the peak position (2θ) shifted to
weak XRD peak intensity (Fig. 1(a) observed for the film prepared at 50
the right (see Table 1). On the other hand, the value of d for the samples
sccm could be due to the lower amount of Ga2O3 on the substrate because
produced at higher flow rates (100, 150 and 200 sccm) were found to be
not all the powder evaporated for growth in a weaker reducing ambient.
bigger than the standard value, which suggests the expansion of the
lattice parameter because the (111) diffraction line shifted to the left
(smaller angle) [14,32,33]. The shifting of 2θ to the right and left was due 3.3. Optical reflectance measurements
to non-uniform and uniform strain, respectively [34].
The crystallite size D was calculated from the (111) plane using Eq. Fig. 3 illustrates the optical reflectance spectral of the β-Ga2O3 NSs
(10) [31]. measured in the wavelength 200–800 nm. The optical spectra exhibit low
reflectance around the solar-blind deep UV (DUV) region of the EM
0:9λ
D¼ (10) spectrum, which is typical for β-Ga2O3. Figs. 4 and 5 are the Tauc plots
ðπβ=180Þcos θ indicating the direct and proper extrapolation techniques applied to es-
timate the optical bandgap of the nanomaterial. The Tauc plots are
where β is the full width at half maxima (FWHM) of the (111) diffraction
derived from data in Fig. 3 by using Eq. (6).
line.
Table 2 presents the various bandgap values evaluated from Figs. 4
The value of D was found to first increase significantly at 50 and 100
and 5. The small error in the bandgap values due to scale of the Tauc plots
sccm H2 flow rates, followed by a remarkable decrease at 150 sccm H2
can be clearly observed for both direct and PE. This suggests that error
supply and finally, increased slightly at 200 sccm flow rate (see Table 1).
due to scale size cannot be avoided for both methods. Nevertheless, the
The irregular trend in the values of D with increasing H2 flow rate could
error observed from DE is slightly bigger as compared to PE. The esti-
be traced to the fluctuation in the values of FWHM.
mated error for DE is about 0.03  0.01, which is slightly bigger as

6
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

compared to the error from PE 0.02  0.01. This is one of the pointers 4. Conclusion
that PE is the preferred technique; although the error due to scale of the
Tauc plot for both methods is negligible. In summary, we have briefly identified the various measurement
By considering the change in the bandgap values due to extrapolation techniques for bandgap determination. The factors influencing the en-
technique, it could be observed that DE (Fig. 4) significantly over- ergy gap of semiconductors such as synthesis method, physical and
estimated the bandgap (see Table 2). This is because the bandgap values chemical properties were also mentioned. In addition to the already re-
recorded from DE are exaggerated (4.86–5.20 eV) when compared to the ported factors affecting the bandgap value of a semiconductor nano-
well-established fundamental bandgap (Eg, F) of β-Ga2O3 (4.90 eV) [14, material, it was pointed out that the scale of the Tauc plot and the
41–44] and the bandgaps obtained by PE. The optical energy gap values extrapolation technique also have significant effect on determined opti-
from proper extrapolation (Fig. 5) are more reasonable (4.69–4.96 eV), cal bandgap value. The effect of the scale of the Tauc plot on bandgap
see Table 2. The values fit well in the widely reported bandgap range for value was found to be negligible. Proper extrapolation was observed to
β-Ga2O3 at room temperature (4.70–4.90 eV) [41,45–51]. give a more reasonable bandgap value, whereas direct extrapolation was
In order to discuss the influence of H2 flow rate on the optical found to lead to either under- or over-estimation, depending on the na-
bandgap of the produced films, we averaged the energy gap values to ture of the optical spectra. The semiconductor films used for this study
correct the effect of scale of the Tauc plot and extrapolation technique on consists of β-Ga2O3 films prepared by the reducing-ambient CVD tech-
the determined values. The mean values are expected to be more nique. XRD analysis revealed the polycrystalline nature of the films.
reasonable as compared to the set of bandgap values obtained from the UV–vis–NIR measurements showed increasing optical bandgap with
individual samples produced at different gas flow rates. The average rising H2 flow rate. FE-SEM analysis revealed dense nanohump-like
bandgap values for both DE and PE are presented in Table 3, and the plot surface morphology. The produced β-Ga2O3 material is promising for
of these values against H2 flow rate is illustrate in Fig. 6. It is obvious that UV detection device, owing to the wide optical bandgap.
the bandgap increased with rising H2 supply into the reactor. The DE
energy gap values are bigger as compared to PE values. It can also be Declaration of competing interest
observed that more than half of the DE values are well-above the Eg, F (see
Fig. 6). The authors declare that they have no known competing financial
Often times, the bandgap of nanomaterials is either overestimated interests or personal relationships that could have appeared to influence
(for samples with the optical absorbing center facing towards the right the work reported in this paper.
(see Fig. 4)) or underestimated (for samples with the optical absorbing
center facing towards the left (see Fig. 7(b and c)) by simply intersecting CRediT authorship contribution statement
the straight-line fit of the Tauc segment with the hv axis. Be it for those
samples containing one or more prominent optical absorbing centers, DE P.R. Jubu: Conceptualization, Methodology, Investigation, Data
is not appropriate because of the proposed systematic under- or over- curation, Software, Writing - original draft. F.K. Yam: Conceptualization,
estimation of the energy gap value. The most reasonable method ac- Methodology, Visualization, Writing - review & editing, Validation, Su-
counting for almost all optical absorption spectra is the PE technique, pervision. V.M. Igba: Software, Writing - review & editing. K.P. Beh:
which is executed by introducing a baseline segment and then taking a Software.
vertical line perpendicularly to the hv axis (see Fig. 5).
In order to gain more insight into the concept of extrapolation, we Acknowledgement
would extend the discussion by providing some sketches with reference
to real absorption spectra reported for different materials in the litera- The authors express their appreciation to Universiti Sains Malaysia
ture. For an optical spectrum such as Fig. 7(a), where an angle of 90 is (USM) for sponsoring this work under the Research University (RU) grant
formed between the fitted Tauc line and the baseline segment, it would scheme (Account No.: 1001/PFIZIK/8011005).
be appropriate to read the energy gap value by DE method. This is
because the fitted Tauc line is parallel to the y-axis and perpendicular to References
the x-axis so that both DE and PE would have the same value. However,
in a situation where the angle between the Tauc line and baseline is less [1] M. Singh, M. Goyal, K. Devlal, Size and shape effects on the band gap of
semiconductor compound nanomaterials, J. Taibah Univ. Sci. 12 (4) (2018)
or greater than 90 (Fig. 7(b), the PE technique would be required to get a 470–475, https://doi.org/10.1080/16583655.2018.1473946.
more reasonable value of energy gap. [2] N. Sangiorgi, L. Aversa, R. Tatti, R. Verucchi, A. Sanson, Spectrophotometric
Fig. 7(c) is the sketch from the real Tauc plot of TiO2 [52]. The au- method for optical band gap and electronic transitions determination of
semiconductor materials, Opt. Mater. 64 (2017) 18–25, https://doi.org/10.1016/
thors estimated the bandgap by taking PE as depicted. This is not sur- j.optmat.2016.11.014 0925–3467.
prising because it is feasible for any researcher to assume the PE due to [3] A.R. Zanatta, Revisiting the optical bandgap of semiconductors and the proposal of
the nature of the optical spectra. However, for a spectrum such as a unified methodology to its determination, Sci. Rep. 9 (2019), https://doi.org/
10.1038/s41598-019-47670-y, 11225–12.
Fig. 7(d) drawn from the actual Tauc plot of ZnO–Al2O3 [5] and ZnO:Er [4] J. Dharma, A. Pisal, Simple Method of Measuring the Band Gap Energy Value of
[53], it is likely that most researchers will use DE instead of PE. The TiO2 in the Powder Form Using a UV/Vis/NIR Spectrometer, PerkinElmer, Inc. USA,
appropriate method to use for such a case is PE as applied by 2019, p. 3. www.perkinelmer.com.
[5] A. Escobedo-Morales, I.I. Ruiz-Lopez, M. deL. Ruiz-Peralta, L. Tepech-Carrillo,
Escobedo-Morales et al. [5] for the ZnO–Al2O3 sample (see Ref. for the
M. Sanchez-Cantú, Automated method for the determination of the band gap energy
real Tauc plot). As it could be seen in Fig. 7(d), there is a big difference of pure and mixed powder samples using diffuse reflectance spectroscopy, Heliyon
between the two expected bandgap values, as DE strongly underestimates 5 (4) (2019), https://doi.org/10.1016/j.heliyon.2019.e01505, 1505–1519.
the bandgap value. Meanwhile, Fig. 7(b) illustrates yet another kind of [6] B.K. Ridley, Large-bandgap semiconductors, Tr. J. Phys. 23 (1999) 577–582.
[7] P.R. Jubu, F.K. Yam, Influence of growth duration and nitrogen-ambient on the
absorption spectrum, which is a sketch to mimic the real Tauc plot of morphological and structural properties of beta-gallium oxide micro- and
indium-tin-oxide (ITO) [54], MoO3 [55], and In2O3 [56], respectively nanostructures, Mater. Chem. Phys. 239 (2020), https://doi.org/10.1016/
(please, see Refs.). Here also, it can be tempting to assume DE but as it j.matchemphys.2019.122043, 122043–4.
[8] G.M. Kimball, A.M. Müller, N.S. Lewis, H.A. Atwater, Photoluminescence-based
may be seen, the expected value is slightly underestimated by DE thereby measurements of the energy gap and diffusion length of Zn3P2, Appl. Phys. Lett. 95
showing a clear difference between the values from the two methods. It is (2009) 112103–112106, https://doi.org/10.1063/1.3225151.
therefore recommended that proper extrapolation should be used at all [9] M.B. Azam, A. Javed, Study of Optical Properties of Isotropic Materials by
Reflection Spectroscopic Ellipsometry, ResearchGate, 2019, pp. 1–50, https://
time to ensure reasonable values of bandgap irrespective of whether the doi.org/10.13140/RG.2.2.25742.00321.
straight edge of the optical spectrum is short or long.

7
P.R. Jubu et al. Journal of Solid State Chemistry 290 (2020) 121576

[10] L. Gu, V. Srot, W. Sigle, C. Koch, P.A. van Aken, VEELS band gap measurements [35] S.J. Pearton, J. Yang, P.H. Cary, F. Ren, J. Kim, M.J. Tadjer, M.A. Mastro, A review
using monochromated electrons, J. Phys.: Conf. Ser. 126 (2008), https://doi.org/ of Ga2O3 materials, processing, and devices, Appl. Phys. Rev. 5 (2018), https://
10.1088/1742-6596/126/1/012005, 012005–4. doi.org/10.1063/1.5006941, 011301–57.
[11] P.P. Gonzalez-Borrero, F. Sato, A.N. Medina, M.L. Baesso, A.C. Bento, Optical band- [36] Z. Galazka, R. Uecker, K. Irmscher, M. Albrecht, D. Klimm, M. Pietsch, M. Brützam,
gap determination of nanostructured WO3 film, Appl. Phys. Lett. 96 (2010), https:// R. Bertram, S. Ganschow, R. Fornari, Czochralski growth and characterization of
doi.org/10.1063/1.3313945, 061909–4. β-Ga2O3 single crystals, Cryst. Res. Technol. 45 (12) (2010) 1229–1236, https://
[12] J. Tauc, R. Grigorovici, A. Vancu, Optical properties and electronic structure of doi.org/10.1002/crat.201000341.
amorphous germanium, Phys. Status Solidi 15 (1966) 627–637, https://doi.org/ [37] L. Fu, Y. Liu, P. Hu, K. Xiao, G. Yu, D. Zhu, Ga2O3 Nanoribbons: synthesis,
10.1002/pssb.19660150224. characterization, and electronic properties, Chem. Mater. 15 (2003) 4287–4291,
[13] R. Korbutowicz, J. Prazmowska, in: Grym Jan (Ed.), “in Wet Thermal Oxidation of https://doi.org/10.1021/cm0343655.
GaAs and GaN”, Semiconductor Technologies, InTech China, 2010, pp. 1–30. [38] G. Wang, J. Park, X. Kong, P.R. Wilson, Z. Chen, J.-h. Ahn, Facile synthesis and
[14] T.S. Ngo, D.D. Le, J.-H. Song, S.-K. Hong, Growth and characterization of gallium characterization of gallium oxide (β-Ga2O3) 1D nanostructures: nanowires,
oxide films grown with nitrogen by plasma-assisted molecular-beam epitaxy, Thin nanoribbons, and nanosheets, Cryst. Growth Des. 8 (6) (2008), https://doi.org/
Solid Films 682 (2019) 93–98, https://doi.org/10.1016/j.tsf.2019.05.029. 10.1021/cg701251j, 1940–1944.
[15] P.S. Vachhani, A.K. Bhatnagar, Oxygen pressure-dependent band gap modification [39] G. Gundiah, A. Govindaraj, C.N.R. Rao, Nanowires, nanobelts and related
in Cu-doped and -undoped ZnO films, Phys. Scripta 87 (2013), https://doi.org/ nanostructures of Ga2O3, Chem. Phys. Lett. 351 (2002) 3–4, https://doi.org/
10.1088/0031-8949/87/04/045702, 045702–5. 10.1016/s0009-2614(01)01372-0, 189–194.
[16] Y. Kokubun, K. Miura, F. Endo, S. Nakagomi, Sol-gel prepared β-Ga2O3 thin films for [40] P.R. Jubu, F.K. Yam, Development and characterization of MSM UV photodetector
ultraviolet photodetectors, Appl. Phys. Lett. 90 (2007), https://doi.org/10.1063/ based on gallium oxide nanostructures, Sensor. Actuator. A 312 (2020)
1.2432946, 031912–4. 112141–112149, https://doi.org/10.1016/j.sna.2020.112141.
[17] F.K. Shan, G.X. Liu, W.J. Lee, G.H. Lee, I.S. Kim, B.C. Shin, Structural, electrical, and [41] K. Akaiwa, S. Fujita, Electrical conductive corundum-structured -Ga2O3 thin films
optical properties of transparent gallium oxide thin films grown by plasma- on sapphire with Tin-doping grown by spray-assisted mist chemical vapor
enhanced atomic layer deposition, J. Appl. Phys. 98 (2005), https://doi.org/ deposition, Jpn. J. Appl. Phys. 51 (2012), https://doi.org/10.1143/
10.1063/1.1980535, 023504–7. JJAP.51.070203, 070203–4.
[18] J.P. Rex, Y.F. Kwong, L.H. San, The influence of deposition temperature on the [42] F. Alema, B. Hertog, O. Ledyaev, D. Volovik, G. Thoma, R. Miller, A. Osinsky,
structural, morphological and optical properties of micro-size structures of beta- P. Mukhopadhyay, S. Bakhshi, H. Ali, W.V. Schoenfeld, Solar blind photodetector
Ga2O3, Results Phys. 14 (2019) 102475–102478, https://doi.org/10.1016/ based on epitaxial zinc doped Ga2O3 thin film, Phys. Status Solidi 5 (2014), https://
j.rinp.2019.102475. doi.org/10.1002/pssa.201600688 (2017) 1600688–7.
[19] K.H. Choi, H.C. Kang, Structural and optical evolution of Ga2O3/glass thin films [43] L. Binet, D. Gourier, Origin of the blue luminescence of β-Ga2O3, J. Phys. Chem.
deposited by radio frequency magnetron sputtering, Mater. Lett. 123 (2014) Solid. 59 (8) (1998) 1241–1249, https://doi.org/10.1016/s0022-3697(98)00047-
160–164, https://doi.org/10.1016/j.matlet.2014.03.038. X.
[20] J. Zhao, X.W. Sun, S.T. Tan, Bandgap-engineered Ga-rich GaZnO thin films for UV [44] M. Orita, H. Ohta, M. Hirano, Deep-ultraviolet transparent conductive β-Ga2O3 thin
transparent electronics, IEEE Trans. Electron. Dev. 56 (12) (2009) 2995–2999. films, Appl. Phys. Lett. 77 (25) (2000) 4166–4168, https://doi.org/10.1063/
[21] P.R. Jubu, F.K. Yam, A.T. Moses, Deposition of gallium oxide nanostructures at low 1.1330559.
substrate temperature by chemical vapor deposition, ECS J. Solid State Sci. Technol. [45] M. Kumar, V. Kumar, R. Singh, Diameter tuning of β-Ga2O3 nanowires using
9 (2020), https://doi.org/10.1149/2162-8777/ab7b47, 035006–7. chemical vapor deposition technique, Nanoscale Res. Lett. 12 (2017), https://
[22] M. Fox, Optical Properties of Solids Chapters 1–3, Oxford Univ. Press, Oxford, 2008, doi.org/10.1186/s11671-017-1915-1, 184 –10.
ISBN 978-0-19-850613-3. [46] T. Zhai, X. Fang, M. Liao, X. Xu, H. Zeng, B. Yoshio, D. Golberg, A Comprehensive
[23] N. Ghobadi, Bandgap determination using absorption spectrum fitting procedure, review of one-dimensional metal-oxide nanostructure photodetectors, Sensors 9
Int. Nano Lett. 3 (2) (2013), https://doi.org/10.1186/2228-5326-3-2, 1–4. (2009) 6504–6529, https://doi.org/10.3390/s90806504.
[24] Brian D. Viezbicke, Shane Patel, Benjamin E. Davis, Dunbar P. Birnie, Evaluation of [47] H.H. Tippins, Optical absorption and photoconductivity in the band edge of
the Tauc method for optical absorption edge determination: ZnO thin films as a β-Ga2O3, Phys. Rev. 140 (1A) (1965), https://doi.org/10.1103/physrev.140.a316.
model system, Phys. Status Solidi B 252 (8) (2015) 1700–1710, https://doi.org/ A316–A319.
10.7282/T3W097T7. Retrieved from. [48] A. Goyal, B.S. Yadav, O.P. Thakur, A.K. Kapoo, Structural and optical
[25] J.B. Coulter, D.P. Birnie, Assessing Tauc plot slope quantification: ZnO thin films as characterization of β-Ga2O3 thin films grown by pulsed laser deposition, Phys.
a model system, Phys. Status Solidi B (2017), https://doi.org/10.1002/ Semicond. Dev. Environ. Sci. Eng. 77 (2014) 77–80, https://doi.org/10.1007/978-
pssb.201700393, 1700393–7. 3-319-03002-918.
[26] P.Y. Yu, M. Cardona, Fundamentals of Semiconductors, Springer, Berlin, 1996, ISBN [49] T. Onuma, S. Fujioka, T. Yamaguchi, M. Higashiwaki, K. Sasaki, T. Masui, T. Honda,
3-540-61461-3 (Chapter 6). Correlation between blue luminescence intensity and resistivity in β-Ga2O3 single
[27] A. Dolgonos, T.O. Mason, K.R. Poeppelmeier, Direct optical band gap measurement crystals, Appl. Phys. Lett. 103 (2013), https://doi.org/10.1063/1.4816759,
in polycrystalline semiconductors: a critical look at the Tauc method, J. Solid State 041910–4.
Chem. 240 (2016) 43–48, https://doi.org/10.1016/j.jssc.2016.05.010. [50] J. Yang, S. Ahn, F. Ren, I.E.E.E. Fellow, S.J. Pearton, I.E.E.E. Fellow, S. Jang,
[28] Q. Zhao, J. Xu, X.Y. Xu, Z. Wang, D.P. Yu, Field emission from AlN nanoneedle A. Kuramata, , High breakdown voltage (-201) β-Ga2O3 Schottky rectifiers, IEEE
arrays, Appl. Phys. Lett. 85 (22) (2004) 5331–5333, https://doi.org/10.1063/ Electron. Device Lett. 38 (2017) 7, https://doi.org/10.1109/LED.2017.2703609,
1.1825620. 906–909.
[29] Y. Su, M. Gao, X. Meng, Y. Chen, Q. Zhou, L. Li, Y. Feng, Synthesis of In-doped [51] M. Higashiwaki, K. Sasaki, H. Murakami, Y. Kumagai, A. Koukitu, K. Akito, Recent
Ga2O3 zigzag-shaped nanowires and optical properties, J. Phys. Chem. Solid. 70 progress in Ga2O3 power devices, Semicond. Sci. Technol. 31 (2016), https://
(2009) 1062–1065, https://doi.org/10.1016/j.jpcs.2009.05.022. doi.org/10.1088/0268-1242/31/3/034001, 034001–11.
[30] Y.M. Lu, X.H. Chen, S. Han, P.J. Cao, F. Jia, Y.X. Zeng, X.K. Liu, W.Y. Xu, W.J. Liu, [52] K.M. Chahrour, F.K. Yam, R. Abdalrheem, High-performance UV photodetector of
D.L. Zhu, Preparation of Ga2O3 thin film solar-blind photodetectors based on mixed- anodic rutile TiO2 nanotube arrays, Mater. Lett. 248 (2019) 161–164, https://
phase structure by pulsed laser deposition, Chin. Phys. B 28 (1) (2019), https:// doi.org/10.1016/j.matlet.2019.04.029.
doi.org/10.1088/1674-1056/28/1/018504, 018504–8. [53] Usha Anitha, Christy Jithin, Characterization Varughese, Thermal effect on optical
[31] P.R. Jubu, F.K. Yam, I. Ahemen, The Effect of hydrogen concentration on chemical band gap energy and Photoluminescence in wurtzite ZnO:Er Nanocrystallites,
vapour deposition synthesis of β-Ga2O3 nanostructures, Solid State Phenom. 301 Mater. Today: SAVE Proc. (2016) 145–151, https://doi.org/10.1016/
(2020), https://doi.org/10.4028/www.scientific.net/SSP.301.27, 27–3. j.matpr.2016.01.046.
[32] S.A.A. Terohid, S. Heidari, A. Jafari, S. Asgary, Effect of growth time on structural, [54] J. Gao, R. Chen, D.H. Li, L. Jiang, J.C. Ye, X.C. Ma, X.D. Chen, Q.H. Xiong, H.D. Sun,
morphological and electrical properties of tungsten oxide nanowire, Appl. Phys. A T. Wu1, UV light emitting transparent conducting tin-doped indium oxide (ITO)
124 (2018) 567–569, https://doi.org/10.1007/s00339-018-1955-0. nanowires, Nanotechnology 22 (2011), https://doi.org/10.1088/0957-4484/22/
[33] X. Zhao, Z. Wu, Y. Zhi, Y. An, W. Cui, L. Li, W. Tan, Improvement for the 19/195706, 195706–10.
performance of solar-blind photodetector based on β-Ga2O3 thin films by doping Zn, [55] A. Kumar, G. Pandey, Synthesis, characterization, effect of temperature on band gap
J. Phys. D Appl. Phys. 50 (2017), https://doi.org/10.1088/1361-6463/aa5758, energy of molybdenum oxide nano rods and their antibacterial activity, Am. J.
085102–6. Appl. Ind. Chem. 3 (3) (2017) 38–42, https://doi.org/10.11648/
[34] S.-L. Ou, D.-S. Wuu, Y.-C. Fu, S.-P. Liu, R.-H. Horng, L. Liu, Z.-C. Feng, Growth and j.ajaic.20170303.12.
etching characteristics of gallium oxide thin films by pulsed laser deposition, Mater. [56] O. Bierwagen, Indium oxide—a transparent, wide-band gap semiconductor for
Chem. Phys. 133 (2012) 700–705, https://doi.org/10.1016/ (opto) electronic applications, Semicond. Sci. Technol. 30 (2015), https://doi.org/
j.matchemphys.2012.01.060. 10.1088/0268-1242/30/2/024001, 024001-16.