Physica B 585 (2020) 412100

Contents lists available at ScienceDirect

Physica B: Physics of Condensed Matter

journal homepage: http://www.elsevier.com/locate/physb

Experimental evidence of impurity-band transport in p-type polycrystalline

germanium film
Zhaoguo Li *, Liping Peng, Jicheng Zhang, Jia Li, Yong Zeng, Daojian Qi, Zhiqiang Zhan,
Minjie Zhou, Weidong Wu
Research Center of Laser Fusion, China Academy of Engineering Physics, Mianyang, 621900, China

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

Keywords: Herein, we report on the Hall transport properties of a p-type polycrystalline germanium film, which is grown by
Hall effect the DC magnetron sputtering method. The average grain size of 85 nm is calculated. The resistivity and Hall
Impurity-band transport coefficient of the germanium film were measured at temperatures between 100 K and 350 K. At room temper­
Polycrystalline germanium films
ature (T ¼ 300 K), the resistivity of 1.06 Ω cm, the hole concentration of 7.6 � 1017 cm 3 and mobility of 7.76
Two-band conduction model
cm2V 1s 1 are obtained. The temperature dependence of Hall coefficient presents a steep maximum at a specific
temperature (T☆ � 215 K). At the same time, the slope of the log resistivity versus 1/T curve is changed. The
unusually results can be well explained by a two-band conduction model, which includes the valence band with a
high mobility, and the acceptor impurity band with a much lower mobility. This work gives an evidence for
impurity-band conduction in the polycrystalline germanium film.

1. Introduction anomaly transport properties in the Ge film can be well described by a

two-band conduction model, which includes the usual valence band
The electrical transport study of semiconductor is basically impor­ with a high mobility and the acceptor impurity band with a very small
tant for device design and application. Recently, the polycrystalline mobility. The temperature dependence of carrier concentration and
germanium (Ge) films have attracted great research interest due to their mobility are also discussed, which uncovered that the grain boundary
novel electrical and optoelectronic performance, which hold promise for scattering dominates the carrier scattering processes in our poly­
advanced complementary metal-oxide semiconductor (CMOS) transis­ crystalline Ge film.
tors, thin film field-effect transistors and sensors applications [1–7].
Since the discovery of Ge transistor in 1948, the electrical properties of 2. Experimental methods
the Ge crystals and quantum wells are widely and deeply investigated
[8–12]. However, the transport properties of the polycrystalline Ge films By using a DC magnetron sputtering method, a Ge film was deposited
are rarely concerned [13–16]. Therefore, it is necessary to study the on a 500-nm-SiO2/Si wafer. The substrate was kept at 400 � C during the
electrical transport properties of polycrystalline Ge films. film deposition processes, then the Ge film was in situ annealed at 600 � C
The Hall effect is a basic physical phenomenon to determine the in ultra-high vacuum for 1 h. More details should be found in our pre­
carrier type and concentration in the materials. More electrical transport vious work [13,14]. Fig. 1(a) shows the x-ray diffraction pattern of the
parameters can be extracted by combining the Hall transport and other Ge film. The three dominant diffraction peaks are corresponding to
electrical experiments. It is meaningful to reveal the carrier transport (111), (220) and (311) planes of the diamond cubic structure of Ge
mechanisms of the materials. In this work, we study the Hall transport crystals, which demonstrate the polycrystalline structures of our Ge film.
properties of a p-type polycrystalline Ge film. In such a film, we have Fig. 1(b) shows a scanning electron microscopy (SEM) image of the Ge
measured its resistivity and Hall coefficient in the temperature range of film. The small grains with different crystal orientations can be seen,
100–350 K. The temperature dependence of Hall coefficient presents a which further demonstrated the polycrystalline structure of our Ge film.
maximum at a specific temperature of 215 K. At the same critical tem­ The lower inset of Fig. 1(b) shows the statistical histogram of the grain
perature, the slope of the log resistivity versus 1/T curve is changed. The sizes. The Gauss function fitting of the grain size distribution data gives

* Corresponding author.
E-mail address: [email protected] (Z. Li).

https://doi.org/10.1016/j.physb.2020.412100
Received 7 January 2020; Received in revised form 17 February 2020; Accepted 18 February 2020
Available online 4 March 2020
0921-4526/© 2020 Elsevier B.V. All rights reserved.
Z. Li et al. Physica B: Physics of Condensed Matter 585 (2020) 412100

an average size of 85 nm for our polycrystalline Ge film. The upper inset temperature decreasing in the semiconductor materials. Interestingly,
of Fig. 1(b) displays the cross-sectional SEM image of the Ge film, which the abnormal maximum value is observed in the RH–1/T curve of the Ge
reveals the columnar growth mode in the Ge film. The film thickness of t film. It sheds light on the particular transport mechanisms in our poly­
¼ 630 nm is also extracted from the cross-sectional scanning. crystalline Ge film.
To measure the Hall effect of the Ge film, the six-terminal Hall-bar Now, we turn to concentrate on the resistivity of the Ge film, which is
devices were prepared. The Hall-bar shape Ge film was first patterned by also shown in Fig. 3(a) as a ρxx–1/T curve. The resistivity is monoto­
a stand ultraviolet lithography and wet etching. Then, the six Ti/Au (10 nously increased while temperature decreasing in the whole tempera­
nm/150 nm) electrodes were deposited on the Hall-bar Ge film by the ture range. However, the slope of the lnρxx–1/T curve is changed at
second lithography and lift-off process. The optical image of the Hall-bar temperatures lower than a critical temperature T☆. Coincidentally, the
Ge device is shown in the inset of Fig. 2(a). The center-to-center length T☆ value is close to T ¼ 215 K where RH arrived its maximum. Addi­
between the two electrodes is L ¼ 75 μm, the width of the Ge film is W ¼ tionally, the lnρxx–1/T curve has a large slope at T > T☆, and in the lower
50 μm. The resistivity and Hall effect of the Ge device were measured in temperature range a smaller slope. We therefore speculate that the
the Quantum Design Physical Property Measurement System with a transport mechanisms in the temperature ranges of T > T☆ and T < T☆
perpendicular magnetic field configuration. are different, as will be discussed below.
Fig. 3(b) displays the temperature dependence of carrier (hole)
3. Results and discussion concentration p and mobility μ. With the temperature decreasing, the
total carrier concentration first decreases at T > T☆, as expected in case
3.1. Anomalous Hall coefficient of conductivity due to holes thermally activated into the valence band,
and then increases at T < T☆. For the carrier mobility, it is monotonously
The Hall resistivity (ρxy) curves of the Ge film in the magnetic field decreased with decreasing temperature.
(B) between 9 T and 9 T are measured at several fixed temperatures. As
can be seen in Fig. 2(a), the linear ρxy-B curves are displayed. We can 3.2. Negative magnetoresistance
identify that the carriers are p-type (hole) in our polycrystalline Ge film,
which is further confirmed by the Seebeck effect measurement. The Hall Fig. 2(b) shows the magnetoresistance (MR) curves of the Ge film at
coefficient (RH) of the Ge film is obtained by extracting the slope of the various temperatures. A notable feature is that the MR decreased with
ρxy-B curves. For example, at T ¼ 300 K, RH ¼ 8.19 cm3/C. The carrier increasing field, that is negative MR. At T ¼ 100 K, a signature of the
concentration is p ¼ 1/(eRH) ¼ 0.76 � 1018 cm 3, where e is the electron positive MR at high fields is observed. Several theoretical models have
charge. The carrier mobility (μ ¼ 7.76 cm2V 1s 1) is also obtained by been suggested to interpret the negative MR in the strong localization
using ρxx ¼ 1/(peμ), where ρxx is the resistivity of the Ge film. The systems. For example, Fukuyama and Yosida (FY) [17] predicted a large
electrical parameters of the Ge film at various temperatures have been negative MR of Anderson localized states in the doped semiconductor
summarized in Table 1. We can see that the hole mobility of our Ge film caused by an exponential dependence of the hopping rate on the highest
is much lower than that of Ge single-crystals with the same carrier occupied energy levels. The corresponding negative MR is given by
concentration. It may be originated from the additional grain boundary MRðBÞ ¼ ½coshðB=BFY Þ� 1 1, where BFY is a characteristic field. One
scattering and impurity scattering in the polycrystalline Ge film. easily finds that MRð∞Þ ¼ 1, i.e., FY’s theory gives a saturated
Fig. 3(a) shows the Hall coefficient RH as a function of 1/T. We can negative MR at high fields. Shklovskii and Spivak [18] also give a
see that RH first exponentially increases with temperature decreasing saturated negative MR at high fields as MR∝e B=BSS (BSS is a character­
(215 K < T < 350 K), while RH arrives a maximum value at T ¼ 215 K, istic field) by considering the interference contribution of different
and then RH decreases when temperature continues decreasing (T < 215 pffiffiffi
hopping trajectories. Median et al. [19] predicts that MR∝ B by
K). RH does not change the sign throughout the whole temperature
considering the quantum interference of forward-scattering paths.
range. Generally, RH should monotonously increase with decreasing
Obviously, to determine the accurate mechanism of the negative MR in
temperature since the carrier concentration is gradually frozen with
our Ge film should deserves further evidences.

Fig. 1. The structural characterizations of the Ge film. (a) The x-ray diffraction pattern of the Ge film. The dominant diffraction peaks of (111), (220) and (311) are
indicated, which belong to the diamond cubic structure of Ge crystals. (b) The SEM image of the Ge film, the scale bar is 500 nm. The upper inset shows the cross-
sectional SEM image of the Ge film, the scale bar is 500 nm. The lower inset presents the histogram of grain size, the solid curve is a fit to Gauss function which gives
an average grain size of 85 nm.

2
Z. Li et al. Physica B: Physics of Condensed Matter 585 (2020) 412100

Fig. 2. The magnetotransport properties of the Ge film. (a) The Hall resistivity ρxy plotted as a function of magnetic field B at various temperatures of the Ge film. The
inset shows the optical image of a typical device, the scale bar is 25 μm. (b) The magnetoresistance curves at various constant temperatures, where MRðBÞ ¼
ρxx ðBÞ ρxx ðB¼0Þ
ρxx ðB¼0Þ � 100%.

concentrations. At an appropriate doping concentration, the wave

Table 1 function overlap of the adjacent impurities is occurred, which forms an
The basic electrical parameters of the Ge film. ρ: resistivity, RH: Hall coefficient,
impurity band. Due to the finite overlap of the localized states, the
p: carrier (hole) concentration, μ: carrier mobility.
carrier mobility in the impurity band is expected to be smaller compared
T (K) Carrier ρ (Ω cm) RH p (1018 μ (cm2V 1s 1) to that in the valence/conduction band. This explains the small mobility
type (cm3C 1
) cm 3)
which was observed in our polycrystalline Ge film. According to the
300 p 1.056 8.19 0.76 7.76 two-band conduction theory, the total conductivity σ ¼ 1/ρxx and Hall
200 p 5.525 13.08 0.48 2.37
coefficient RH can be expressed as:
100 p 46.198 2.94 2.13 0.06
σ ¼ σval þ σimp (1)

In the below, we try to understand the anomalous Hall transport

Rval σ 2val þ Rimp σ 2imp
properties in the Ge film. RH ¼ �2 (2)
σ val þ σimp

3.3. Two-band conduction model where σ val ¼ epval μval and σimp ¼ epimp μimp are the conductivities from
the valence and acceptor impurity bands respectively, ​ pval and pimp are
The anomalous Hall transport phenomena are widely investigated in the hole concentrations of valence and impurity bands respectively, μval
the single crystalline materials [9,20,21]. The similar properties are also and μimp are the hole mobilities of the valence and impurity bands
observed in the polycrystalline films, such as Ge [15,16], Te [22–24], respectively and μval ≫μimp , Rval ¼ ep1val and Rimp ¼ ep1imp are the Hall co­
CdSe [25], GaSe [26], CuGaSe2 [27,28], GaSbAs [29,30], etc. The
efficients of the valence and impurity bands respectively. In the previous
characteristic features of the above phenomena can be summarized as
work of W. Yeh et al. [16], the similar anomalous Hall transport has
following: with temperature decreasing, the resistivity and Hall coeffi­
been found in an as-deposited Ge film on a Si substrate. They suggested a
cient first increase exponentially; but at a specific temperature, the Hall
dislocation-induced acceptor-band with a 20 meV energy gap to explain
coefficient reaches a maximum value and then decreases sharply, and
the experimental results. In our work, an acceptor impurity energy level
the slope of the lnρxx–1/T curve is changed; at the ultra-low tempera­
with ~100 meV are observed. The possible physical origins of the im­
tures, the Hall coefficient presents a saturation case with a very small
purity band in our polycrystalline Ge film are further discussed.
thermal activation behavior. In our work, we observed the similar
Moreover, the other mechanisms can be excluded in our work. For
phenomenon in the polycrystalline Ge film. Understanding the anoma­
example, the two-layer conduction model can be used to interpret the
lous Hall transport in the polycrystalline Ge film is very important for
similar phenomena [32,33]. Obviously, the two-layer conduction model
their device applications.
is suitable for multi-layer structure systems or inhomogeneous (along
The anomalous variation of the Hall coefficient with temperature [as
the depth direction) films. Therefore, we believe that this model is not
shown in Fig. 3(a)] can be explained by a two-band conduction model,
applicable to our Ge film due to its single-layer and homogeneous fea­
which is first proposed by Hung [31]. Based on the analysis and dis­
tures along the depth direction [Fig. 1(b)]. Another possible mechanism
cussion in the following, we can identify that the one band is the usual
is the thermoelectric effect [34,35]. However, this mechanism can also
valence band with a high mobility, while another band is the acceptor
be excluded since the thermoelectric potential effect has been removed
impurity band with a small mobility in our polycrystalline Ge film. The
in the transport measurement procedures. To obtain the accurate (Hall)
acceptor impurity band is formed by the p-type impurity doping. As a
resistance, we calculated the average value of the measured resistances
result, the impurity-band dominant transport properties are observed in
with applying the positive and negative current directions.
our polycrystalline Ge film. In the semiconductors, the electronic
structure of impurities is generally described as localized energy levels
with respect to the conduction or valence bands at low doping

3
Z. Li et al. Physica B: Physics of Condensed Matter 585 (2020) 412100

Fig. 3. The electrical parameters of the Ge film. (a) The resistivity ρxx (□) and Hall coefficient RH (○) plotted against 1/T. The solid curve is the theoretical calculation
result of RH from the two-band conduction model. The inset is the schematic diagram of the band structure for our polycrystalline Ge film (CB: conduction band; VB:
valence band; IB: impurity band; ε1: the energy gap between the valence and impurity bands; ε3: the width of the impurity band). (b) The temperature dependence of
the carrier (hole) concentration p and the inverse of mobility μ 1. In the upper panel, the solid curve is the best fitting of Eq. (8); the dashed curves show the
contributions from the lattice (μL 1 ), ionized impurity (μI 1 ) and grain-boundary (μGB1 ) scatterings, respectively.

3.4. Explanation of experiments Second, we will concentrate on the temperature dependence of Hall
coefficient, as shown in Fig. 3(a). We have known that μval ≫μimp . It
In the following, we will explain the observations of electrical means that σ val Rval ≫σ imp Rimp is valid in the whole temperature range. As
transport in our polycrystalline Ge film by adopting the two-band con­ discussed above, in the temperature range of T > T☆, σ val ≫σimp . Then,
duction model. � �
First, let us consider the temperature dependence of resistivity, as we can obtain that RH � Rval ¼ ðpval eÞ 1 ∝exp kεB1T from Eq. (2).
shown in Fig. 3(a). We have observed that the slopes of the lnρxx–1/T Therefore, we observed that RH exponentially increases with tempera­
curve in the temperature before and after T☆ are different. It is revealed ture decreasing. At low temperatures (T < T☆), the conductivity is
that the transport mechanisms are different in the ranges of T > T☆ and dominated by the impurity band, that is σ val ≪σ imp . From Eq. (2), we
T < T☆. At high temperature regime (T > T☆), the valence band trans­ R σ 2val σ val
� �
2ε3 ε1
port dominates the conductivity, that is, σval ≫σimp . According to Eq. (1), have RH � valσ 2 ∝σ 2 ∝exp k B T . We therefore observed that RH
imp imp
we have σ � σ val . As we know, the temperature dependence of con­ exponentially decreases with temperature decreasing since 2ε3 ε1 < 0.
ductivity from band carriers in semiconductor can be described by the Moreover, since j2ε3 ε1 j < ε1 , the absolute slope of the RH–1/T curve
� �
thermal activation model, that is, σval ∝exp in T < T☆ is smaller than that in T > T☆. Obviously, RH arrived its
kB T ; where kB is the
ε1

maximum when σval ¼ σimp � σ0 to be Rmax H ¼ 14 ðRval þ Rimp Þ ¼

Boltzmann constant and ε1 is the thermal activation energy gap [see the 1
ðμval þ μimp Þ � 14Rval , where we have used μval ≫μimp . The mobility of
inset of Fig. 3(a)]. Then, we can extract the band energy gap as ε1 ¼ 4σ 0

120 meV by fitting the lnσ–1/T data at high temperatures. Obviously, the valence band carriers can be estimated as μval � 6.3 cm2V 1s 1 at
the value of ε1 is much smaller than the energy gap of single-crystalline T☆.
Ge (εg ~660 meV). We therefore identified that the holes in the valence Finally, the temperature dependence of carrier concentration can be
well explained by the two-band model [Fig. 3(b)]. The total carrier
band are thermally activated from the acceptor impurity band. At low
concentration p can be written as p ¼ pval þ pimp . As we know, the hole
temperatures (T < T☆), σ � σ imp if we assume σ val ≪σ imp . Here, the free
concentration in the valence and acceptor impurity bands exhibit
holes are nearly frozen. And the conductivity is dominated by the hole
different temperature-dependent behaviors: The hole concentration
hopping between the nearest-neighbor localized states. The nearest-
(pimp ) in the acceptor impurity band decreases with temperature
neighbor hopping (NNH) conductivity is also characterized by an
� � increasing, whereas the hole concentration (pval ) from the valence band
Arrhenius-like equation, i.e. σimp ∝exp kB T , where the activation
ε3
increases with temperature increasing. Consequently, the dominance of
valence band conduction at high temperatures (T > T☆) leads to a
energy ε3 is equal to the width of the impurity-band [36] [see the inset of
decrease of hole concentration from 350 K to 215 K. With further
Fig. 3(a)]. Through experimental fitting, ε3 ¼ 30 meV is obtained. Till
decreasing temperature (T < T☆), the contribution from acceptor im­
now, we can conclude that the usual valence band conduction dominates
purity band carriers becomes more important. Then the total hole con­
the high temperature electrical transport (T > T☆), and the significant
centration shows an increasing trend with temperature decreasing.
contribution from the acceptor impurity band conduction is present in
low temperatures (T < T☆).

4
Z. Li et al. Physica B: Physics of Condensed Matter 585 (2020) 412100

3.5. Analysis model below]. Similar to the fitting results of RH ðTÞ, the good fitting is achieved
in the temperature range of T≲300 ​ K. The calculated results confirmed
According to the two-band conduction model, we have identified the the discussions in Sec. 3.4.
impurity-band transport in the polycrystalline Ge film. Now, we can Now, we concentrate on the fitting parameters (Table 2) of our
obtain the basic semiconductor electronic parameters by fitting the polycrystalline Ge film. Clearly, the value of ε1 extracted from RH ðTÞ is
RH(T) curve. In this section, we will discuss this issue. comparable with that from ρxx(T) fitting. The noteworthy value of ε1 e
In a semiconductor which contains both valence and acceptor im­ 100 meV is obtained both from the fittings of ρxx(T) and RH ðTÞ. It
purity bands, the concentration pval of free holes can be described by demonstrated the existence of the acceptor impurity band gap. More­
[37]. over, the large concentration of NA is also shown in Table 2. The large p-
type impurity concentrations in our Ge film may be originated from the
NA
pval þ ND ¼ � � (3) impurity doping of Mo, Fe, Al, etc. However, further evidence is required
1 þ gNV pVval exp kεB1T to determine the exact doping elements.

where NA is the concentrations of the acceptor impurities; ND is the 3.6. Carrier scattering mechanism
concentration of the compensating donors; gV is the degeneracy factor
(gV ¼ 4 for the valence band of Ge, where the heavy and light hole bands In this section, we will discuss the carrier scattering mechanisms by
3=2 considering the scaling law of μ(T). In the polycrystalline Ge film, the
are both considered); N V ¼ 2ð2π m*h kB T=h2 Þ is the density of states in
carrier scattering mechanisms could include many factors, such as voids,
the valence band; m*h is the density-of-states hole mass, for Ge, m*h ¼ defects, impurities and grain boundaries, etc. By using Matthiessen’s
3=2 3=2 2=3
ðmhh þ mlh Þ ¼ 0:36m0 (where the effective hole masses of heavy approximation, the total mobility can be calculated as[38,39] μ 1 ¼
hole mhh ¼ 0:35m0 and light hole mlh ¼ 0:043m0 are used, m0 is the free μL 1 þ μI 1 þ μGB1 , where μL , μI and μGB are the lattice, ionized impurity
electron mass) [37]; h is the Planck constant. and grain boundary scattering-dependent mobilities respectively. For
By defining x ¼ pimp =pval and b ¼ μval =μimp , Eqs. (1) and (2) can be p-type Ge, μL ∝T 2:33 which has considered the acoustical and optical
rewritten as modes of lattice scattering[40]. For ionized impurity scattering, μI ∝T 3=2
� x� [37]. For grain-boundary scattering effect, the mobility can be described
σ ¼ σval 1 þ (4) � �
eφ
b by μGB ∝T 1=2 exp kB T , if we assume a uniform potential barrier

x þ b2 height φ for scattering at the grain boundaries [41]. Therefore, the

RH ¼ Rval (5)
ðx þ bÞ2 complete theoretical description of μ(T) for our Ge film is
� �
Generally, the temperature dependence of b is considerably weaker eφ
μ 1 ðTÞ ¼ ​ AT 2:33 þ BT 3=2 þ CT 1=2 exp (8)
than that of x. We therefore roughly consider b to be constant. kB T
If we preset the values of ND , NA , ε1 and b, then we can extract the
value of pval from Eq. (3). Furtherly, pimp can be calculated by following where A, B and C are the fitting parameters. By applying Eq. (8) to the
equation. experimental data, we obtain a perfect fitting result with the parameters
of A ¼ 3:7 � 10 8 K 2:33 , B ¼ 0:01 K3=2 , C ¼ 4:0 � 10 4 K 1=2 , and φ ¼
NA 0:073V. The fitting results are shown in the upper panel of Fig. 3(b),
pimp ¼ NA � � (6)
1þ gV pval
exp kεB1T where the mobility components of μL , μI and μGB are also shown. Ac­
cording to the fitting results, we can conclude the following features. In
NV

Now, RH can be calculated by Eq. (5). Repeating the above proced­ the temperature range of 100–350 K for our polycrystalline Ge film, (i)
ures by changing the values of ND , NA , ε1 and b, until the calculated RH the ionized impurity scattering effect can be ignored; (ii) the dominant
agrees to the measured RH. The best fitting result is shown in Fig. 3(a) as scattering mechanism is the grain-boundary scattering with an average
a solid curve. The fitting parameters are summarized in Table 2. We can potential barrier height of eφ ¼ 73 meV. This value is comparable to the
see that the calculated RH agrees with the experimental data at the previous studies [38,39]; and (iii) the lattice scattering provides a weak
regime of T≲300 ​ K. At the high temperatures, the calculated results correction to μ(T) at high temperatures, and this weak correction cannot
evidently deviate from the measured values. Its reason may be ascribed be ignored.
to that the temperature variation of b is not considered in the fitting
process. 4. Conclusion
According to the above fitting results, the temperature dependence
of pval and pimp can be further obtained. The results are shown in Fig. 4 In conclusion, the resistivity and Hall coefficient are measured in the
(a). We can see that pval increases with temperature increasing, and that polycrystalline Ge film which is grown by a DC magnetron sputtering
pimp reacts against that. method. The average grain size of 85 nm is extracted. The typical re­
Combining Eqs. (4) and (5), we can obtain sistivity, hole concentration and mobility at room temperature are 1.06
Ω cm, 7.6 � 1017 cm 3 and 7.76 cm2V 1s 1 respectively. There exists a
μval ¼ μ
bðx þ bÞ
(7) critical temperature (denoted as T☆), above which the resistivity and
x þ b2 Hall coefficient increase exponentially with temperature decreasing. At
Taking account of μimp ¼ μval =b, μval ðTÞ and μimp ðTÞ are extracted. temperatures lower than T☆, the Hall coefficient decreases exponen­
Furtherly, σ val ¼ epval μval and σ imp ¼ epimp μimp are calculated, as shown in tially with temperature decreasing, and the resistivity remains
Fig. 4(b) [Here, the value of μ is calculated from Eq. (8) as discussed increasing with temperature decreasing. However, the slope of log re­
sistivity versus 1/T is smaller than that in the high temperature regime.
Therefore, a maximum of Hall coefficient presents at T☆. For our poly­
Table 2 crystalline Ge film, T☆ � 215 K. These anomalous Hall transport prop­
The fitting parameters of the Ge film. erties can be well explained by a two-band conduction model: the one
NA (cm 3
) ND (cm 3
) ND/NA ε1 (meV) b band is the usual valence band with a high carrier mobility, another
18 14 5
band is the acceptor impurity band with a very low mobility. The
6.9 � 10 5.0 � 10 7.2 � 10 100 58
mobility ratio of valence and impurity bands is b ¼ μval =μimp ¼ 58. The

5
Z. Li et al. Physica B: Physics of Condensed Matter 585 (2020) 412100

Fig. 4. Analysis of the transport properties of the Ge film. (a) The calculated hole concentration pval and pimp as a function of temperature for our polycrystalline Ge
film. (b) The conductivity (σ ¼ 1/ρxx) as a function of temperature. The solid curve is the theoretical calculation result (σ ¼ σ val þ σ imp). The dashed curves present the
contributions from the valence (σval) and impurity (σimp) bands, respectively.

temperature dependence of carrier concentration is also well understood [6] W. Takeuchi, N. Taoka, M. Kurosawa, M. Sakashita, O. Nakatsuka, S. Zaima, Appl.
Phys. Lett. 107 (2015), 022103.
by the two-band theory. By analyzing the temperature dependence of
[7] A. Hara, Y. Nishimura, H. Ohsawa, Jpn. J. Appl. Phys. 56 (2017), 03BB01.
hole mobility, the grain-boundary scattering as the predominant [8] I. Estermann, A. Foner, Phys. Rev. 79 (1950) 365.
mechanism is revealed. [9] C.S. Hung, J.R. Gliessman, Phys. Rev. 79 (1950) 726.
The extracted energy distance between the valence and impurity [10] H. Roth, W.D. Straub, W. Bernard, J.J.E. Mulhern, Phys. Rev. Lett. 11 (1963) 328.
[11] O.A. Mironov, N. d’Ambrumenil, A. Dobbie, D.R. Leadley, A.V. Suslov, E. Green,
bands is in the order of 100 meV for our polycrystalline Ge film. How­ Phys. Rev. Lett. 116 (2016) 176802.
ever, more evidence is required to determine the exact doping elements. [12] N.W. Hendrickx, D.P. Franke, A. Sammak, M. Kouwenhoven, D. Sabbagh, L. Yeoh,
In a word, our work gives an experimental evidence of impurity-band R. Li, M.L.V. Tagliaferri, M. Virgilio, G. Capellini, G. Scappucci, M. Veldhorst, Nat.
Commun. 9 (2018) 2835.
transport in the conductivity of the polycrystalline Ge film, which may [13] Z. Li, L. Peng, J. Zhang, J. Li, Y. Zeng, Y. Luo, Z. Zhan, L. Meng, M. Zhou, W. Wu,
be useful for future Ge-based device applications. Semicond. Sci. Technol. 32 (2017), 035010.
[14] Z. Li, L. Peng, J. Zhang, J. Li, Y. Zeng, Z. Zhan, W. Wu, Philos. Mag. 98 (2018)
1525.
Credit author statement [15] A.P. Klimenko, L.A. Matveeva, Y.A. Tkhorik, Czech. J. Phys. B 24 (1974) 1139.
[16] W. Yeh, A. Matsumoto, K. Sugihara, H. Hayase, ECS J. Solid State Sci. Technol. 3
Z. L. conceived the work and wrote the paper. L. P. and J. Z. grown (2014) Q195.
[17] H. Fukuyama, K. Yosida, J. Phys. Soc. Jpn. 46 (1979) 102.
the Ge films. J. L. and D. Q. performed the XRD measurements. Y. Z. and [18] M. Pollak, B. Shklovskii, Hopping Transport in Solids, North Holland, Amsterdam,
M. Z. performed the SEM measurements. Z. L. and Z. Z. performed the 1991.
microfabrication experiments. Z. L. performed the transport measure­ [19] E. Medina, M. Kardar, Y. Shipir, X.R. Wang, Phys. Rev. Lett. 64 (1990) 1816.
[20] E. Arushanov, C. Kloc, E. Bucher, Phys. Rev. B 50 (1994) 2653.
ments and analyzed the data, W. W. assisted with interpretation of the
[21] P. Pampili, D.V. Dinh, V.Z. Zubialevich, P.J. Parbrook, J. Phys. D Appl. Phys. 51
data. All authors commented on the manuscript. (2018), 06LT01.
[22] M.J. Capers, M. White, Thin Solid Films 15 (1973) 5.
Declaration of competing interest [23] A. Goswami, S.M. Ojha, Thin Solid Films 16 (1973) 187.
[24] A.M. Phahle, Thin Solid Films 41 (1977) 235.
[25] T.K. Gupta, J. Doh, Sol. Energy Mater. Sol. Cell. 27 (1992) 327.
The authors declare no competing financial interest. [26] T. Colakoglu, M. Parlak, Thin Solid Films 492 (2005) 52.
[27] S. Schuler, S. Siebentritt, S. Nishiwaki, N. Rega, J. Beckmann, S. Brehme, M.C. Lux-
Steiner, Phys. Rev. B 69 (2004), 045210.
Acknowledgments [28] E. Arushanov, S. Siebentritt, T. Schedel-Niedrig, M.C. Lux-Steiner, J. Phys.
Condens. Matter 17 (2005) 2699.
This work is financially supported by the National Natural Science [29] Y. Kajikawa, K. Okamura, Phys. Status Solidi (c) 9 (2012) 274.
[30] Y. Kajikawa, K. Okamura, T. Okuzako, Y. Matsui, Thin Solid Films 545 (2013) 161.
Foundation of China (Grant No. 11604310) and the Key Laboratory of [31] C.S. Hung, Phys. Rev. 79 (1950) 727.
Ultra-Precision Machining Technology Foundation of CAEP (Grant No. [32] C. Mavroidis, J.J. Harris, R.B. Jackman, I. Harrison, B.J. Ansell, Z. Bougrioua,
ZD18001). I. Moerman, J. Appl. Phys. 91 (2002) 9835.
[33] C. Mavroidis, J.J. Harris, M.J. Kappers, C.J. Humphreys, Z. Bougrioua, J. Appl.
Phys. 93 (2003) 9095.
References [34] D.I. Jones, W.E. Spear, P.G.L. Comber, J. Non-Cryst. Solids 20 (1976) 259.
[35] D.I. Jones, W.E. Spear, P.G.L. Comber, S. Li, R. Martins, Philos. Mag. B 39 (1979)
[1] Y.-H. Kuo, Y.K. Lee, Y. Ge, S. Ren, J.E. Roth, T.I. Kamins, D.A.B. Miller, J.S. Harris, 147.
Nature 437 (2005) 1334. [36] N.A. Poklonski, S.A. Vyrko, O.N. Poklonskaya, A.G. Zabrodskii, J. Appl. Phys. 110
[2] V.F. Mitin, P.C. McDonald, F. Pavese, N.S. Boltovets, V.V. Kholevchuk, I.Y. Nemish, (2011) 123702.
V.V. Basanets, V.K. Dugaev, P.V. Sorokin, R.V. Konakova, E.F. Venger, E.V. Mitin, [37] J.S. Blakemore, Semiconductor Statistics, Pergamon Press, Oxford, 1962.
Cryogenics 47 (2007) 474. [38] A. Barna, P.B. Barna, Z. Bodo, J.F. Pocza, I. Pozsgai, G. Radnoczi, Thin Solid Films
[3] L. Tang, S.E. Kocabas, S. Latif, A.K. Olyay, D.-S. Ly-Gagnon, K.C. Saraswat, D.A. 23 (1974) 49.
B. Miller, Nat. Photon. 2 (2008) 226. [39] V. Dutta, P. Nath, K.L. Chopra, Phys. Status Solidi (a) 48 (1978) 257.
[4] Z.C. Holman, C.-Y. Liu, U.R. Kortshagen, Nano Lett. 10 (2010) 2661. [40] F.J. Morin, J.P. Maita, Phys. Rev. 94 (1954) 1525.
[5] J. Chen, H.-G. Piao, Z. Luo, X. Zhang, Appl. Phys. Lett. 106 (2015) 173503. [41] R.L. Petritz, Phys. Rev. 104 (1956) 1508.